{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 500,
  "global_step": 32653,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1.780480121956888,
      "learning_rate": 2.0408163265306127e-08,
      "loss": 0.3878,
      "step": 1
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8596089734163604,
      "learning_rate": 4.0816326530612253e-08,
      "loss": 0.2716,
      "step": 2
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3428179382529704,
      "learning_rate": 6.122448979591837e-08,
      "loss": 0.4325,
      "step": 3
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1236103886921316,
      "learning_rate": 8.163265306122451e-08,
      "loss": 0.4865,
      "step": 4
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1099347674267723,
      "learning_rate": 1.0204081632653061e-07,
      "loss": 0.7276,
      "step": 5
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9887845543675161,
      "learning_rate": 1.2244897959183673e-07,
      "loss": 0.2388,
      "step": 6
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7872455250044791,
      "learning_rate": 1.4285714285714287e-07,
      "loss": 0.1738,
      "step": 7
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0811710524318654,
      "learning_rate": 1.6326530612244901e-07,
      "loss": 0.287,
      "step": 8
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.180808447937352,
      "learning_rate": 1.8367346938775513e-07,
      "loss": 0.6155,
      "step": 9
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.315398982518123,
      "learning_rate": 2.0408163265306121e-07,
      "loss": 0.1911,
      "step": 10
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.170097160218178,
      "learning_rate": 2.2448979591836735e-07,
      "loss": 0.3578,
      "step": 11
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6592318434008781,
      "learning_rate": 2.4489795918367347e-07,
      "loss": 0.4703,
      "step": 12
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.944868774574638,
      "learning_rate": 2.653061224489796e-07,
      "loss": 0.6012,
      "step": 13
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0621634321711595,
      "learning_rate": 2.8571428571428575e-07,
      "loss": 0.333,
      "step": 14
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9525397379921536,
      "learning_rate": 3.0612244897959183e-07,
      "loss": 0.2316,
      "step": 15
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.307535748742227,
      "learning_rate": 3.2653061224489803e-07,
      "loss": 0.6861,
      "step": 16
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7257345733932313,
      "learning_rate": 3.469387755102041e-07,
      "loss": 0.1454,
      "step": 17
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.502992416402804,
      "learning_rate": 3.6734693877551025e-07,
      "loss": 0.572,
      "step": 18
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0824542024475314,
      "learning_rate": 3.8775510204081634e-07,
      "loss": 0.2601,
      "step": 19
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6604050999619862,
      "learning_rate": 4.0816326530612243e-07,
      "loss": 0.3532,
      "step": 20
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0366153695848612,
      "learning_rate": 4.285714285714286e-07,
      "loss": 0.3603,
      "step": 21
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.36548089336536,
      "learning_rate": 4.489795918367347e-07,
      "loss": 0.6585,
      "step": 22
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7627096929231691,
      "learning_rate": 4.6938775510204085e-07,
      "loss": 0.312,
      "step": 23
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1933260606713953,
      "learning_rate": 4.897959183673469e-07,
      "loss": 0.4082,
      "step": 24
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8565251483116453,
      "learning_rate": 5.102040816326531e-07,
      "loss": 0.2091,
      "step": 25
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8134823003916904,
      "learning_rate": 5.306122448979592e-07,
      "loss": 0.2835,
      "step": 26
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2177507605558398,
      "learning_rate": 5.510204081632654e-07,
      "loss": 0.4802,
      "step": 27
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.298537169204794,
      "learning_rate": 5.714285714285715e-07,
      "loss": 0.298,
      "step": 28
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2387269668697407,
      "learning_rate": 5.918367346938776e-07,
      "loss": 0.3141,
      "step": 29
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5354282144307474,
      "learning_rate": 6.122448979591837e-07,
      "loss": 0.4614,
      "step": 30
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0607623647661122,
      "learning_rate": 6.326530612244898e-07,
      "loss": 0.3414,
      "step": 31
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.787299448334333,
      "learning_rate": 6.530612244897961e-07,
      "loss": 0.542,
      "step": 32
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.037829539682342,
      "learning_rate": 6.734693877551021e-07,
      "loss": 0.3007,
      "step": 33
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.007299734019227,
      "learning_rate": 6.938775510204082e-07,
      "loss": 0.2167,
      "step": 34
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2621661844536913,
      "learning_rate": 7.142857142857143e-07,
      "loss": 0.232,
      "step": 35
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.195410294490337,
      "learning_rate": 7.346938775510205e-07,
      "loss": 0.4475,
      "step": 36
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1609084764223312,
      "learning_rate": 7.551020408163266e-07,
      "loss": 0.3403,
      "step": 37
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9540152075849876,
      "learning_rate": 7.755102040816327e-07,
      "loss": 0.2608,
      "step": 38
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.5374859257697926,
      "learning_rate": 7.959183673469388e-07,
      "loss": 0.5479,
      "step": 39
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.941044288522346,
      "learning_rate": 8.163265306122449e-07,
      "loss": 0.4084,
      "step": 40
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7142246569308917,
      "learning_rate": 8.367346938775512e-07,
      "loss": 0.6073,
      "step": 41
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0071081969611768,
      "learning_rate": 8.571428571428572e-07,
      "loss": 0.2776,
      "step": 42
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.882454639032991,
      "learning_rate": 8.775510204081633e-07,
      "loss": 0.2627,
      "step": 43
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7950647900814132,
      "learning_rate": 8.979591836734694e-07,
      "loss": 0.2724,
      "step": 44
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5291344086357352,
      "learning_rate": 9.183673469387756e-07,
      "loss": 0.3252,
      "step": 45
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8385505682392278,
      "learning_rate": 9.387755102040817e-07,
      "loss": 0.4175,
      "step": 46
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1393476767466593,
      "learning_rate": 9.591836734693878e-07,
      "loss": 0.2231,
      "step": 47
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6086906627786306,
      "learning_rate": 9.795918367346939e-07,
      "loss": 0.4602,
      "step": 48
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0789409928420188,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.3356,
      "step": 49
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.840653562563349,
      "learning_rate": 1.0204081632653063e-06,
      "loss": 0.7744,
      "step": 50
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9379449736128572,
      "learning_rate": 1.0408163265306123e-06,
      "loss": 0.0833,
      "step": 51
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2101495928700217,
      "learning_rate": 1.0612244897959184e-06,
      "loss": 0.3604,
      "step": 52
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.9810466616142736,
      "learning_rate": 1.0816326530612247e-06,
      "loss": 0.1741,
      "step": 53
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6438890623933389,
      "learning_rate": 1.1020408163265308e-06,
      "loss": 0.399,
      "step": 54
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6226069100607723,
      "learning_rate": 1.122448979591837e-06,
      "loss": 0.428,
      "step": 55
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0752441682688718,
      "learning_rate": 1.142857142857143e-06,
      "loss": 0.2267,
      "step": 56
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.2982027212727165,
      "learning_rate": 1.163265306122449e-06,
      "loss": 0.6274,
      "step": 57
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.679151021294251,
      "learning_rate": 1.1836734693877552e-06,
      "loss": 0.6137,
      "step": 58
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.262814241952776,
      "learning_rate": 1.2040816326530612e-06,
      "loss": 0.7477,
      "step": 59
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0808129049709783,
      "learning_rate": 1.2244897959183673e-06,
      "loss": 0.0848,
      "step": 60
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7754412403466995,
      "learning_rate": 1.2448979591836734e-06,
      "loss": 0.3305,
      "step": 61
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1922272323259306,
      "learning_rate": 1.2653061224489795e-06,
      "loss": 0.3596,
      "step": 62
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1817597028565467,
      "learning_rate": 1.2857142857142856e-06,
      "loss": 0.5622,
      "step": 63
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8720655916093096,
      "learning_rate": 1.3061224489795921e-06,
      "loss": 0.1569,
      "step": 64
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1207452742642028,
      "learning_rate": 1.3265306122448982e-06,
      "loss": 0.3091,
      "step": 65
    },
    {
      "epoch": 0.0,
      "grad_norm": 6.634935520728172,
      "learning_rate": 1.3469387755102043e-06,
      "loss": 0.4139,
      "step": 66
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2091122581514053,
      "learning_rate": 1.3673469387755104e-06,
      "loss": 0.3149,
      "step": 67
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.122370688234995,
      "learning_rate": 1.3877551020408165e-06,
      "loss": 0.8033,
      "step": 68
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1629863123548712,
      "learning_rate": 1.4081632653061225e-06,
      "loss": 0.233,
      "step": 69
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1578879537705133,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.4368,
      "step": 70
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.431290633085253,
      "learning_rate": 1.4489795918367347e-06,
      "loss": 0.4894,
      "step": 71
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.7640003315408648,
      "learning_rate": 1.469387755102041e-06,
      "loss": 0.5662,
      "step": 72
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8762924336139231,
      "learning_rate": 1.489795918367347e-06,
      "loss": 0.2256,
      "step": 73
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3679832827338383,
      "learning_rate": 1.5102040816326532e-06,
      "loss": 0.3294,
      "step": 74
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2303510815260184,
      "learning_rate": 1.5306122448979593e-06,
      "loss": 0.1821,
      "step": 75
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.646638828434809,
      "learning_rate": 1.5510204081632654e-06,
      "loss": 0.7282,
      "step": 76
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.182575689199579,
      "learning_rate": 1.5714285714285714e-06,
      "loss": 0.5401,
      "step": 77
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.015380023270868,
      "learning_rate": 1.5918367346938775e-06,
      "loss": 0.5986,
      "step": 78
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0254251816589568,
      "learning_rate": 1.6122448979591836e-06,
      "loss": 0.2295,
      "step": 79
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5090751923232333,
      "learning_rate": 1.6326530612244897e-06,
      "loss": 0.3735,
      "step": 80
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.022103342892828,
      "learning_rate": 1.6530612244897962e-06,
      "loss": 0.5706,
      "step": 81
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.5554693128781054,
      "learning_rate": 1.6734693877551023e-06,
      "loss": 0.0779,
      "step": 82
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.119436160574601,
      "learning_rate": 1.6938775510204084e-06,
      "loss": 0.3003,
      "step": 83
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.827809560826234,
      "learning_rate": 1.7142857142857145e-06,
      "loss": 0.3777,
      "step": 84
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0355642392509532,
      "learning_rate": 1.7346938775510206e-06,
      "loss": 0.3572,
      "step": 85
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.054709893966784,
      "learning_rate": 1.7551020408163267e-06,
      "loss": 0.5884,
      "step": 86
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8218500898469976,
      "learning_rate": 1.7755102040816327e-06,
      "loss": 0.4665,
      "step": 87
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1840064925470757,
      "learning_rate": 1.7959183673469388e-06,
      "loss": 0.298,
      "step": 88
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.886848884648453,
      "learning_rate": 1.8163265306122451e-06,
      "loss": 0.563,
      "step": 89
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.243255682497521,
      "learning_rate": 1.8367346938775512e-06,
      "loss": 0.0354,
      "step": 90
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0956168563386228,
      "learning_rate": 1.8571428571428573e-06,
      "loss": 0.3305,
      "step": 91
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.7536863120793184,
      "learning_rate": 1.8775510204081634e-06,
      "loss": 0.1446,
      "step": 92
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1077576462183807,
      "learning_rate": 1.8979591836734695e-06,
      "loss": 0.1951,
      "step": 93
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.955107096804397,
      "learning_rate": 1.9183673469387756e-06,
      "loss": 0.7935,
      "step": 94
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.863079398851591,
      "learning_rate": 1.938775510204082e-06,
      "loss": 0.5303,
      "step": 95
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.655120308268914,
      "learning_rate": 1.9591836734693877e-06,
      "loss": 0.3952,
      "step": 96
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.047859048568176,
      "learning_rate": 1.979591836734694e-06,
      "loss": 0.2834,
      "step": 97
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3665086436949123,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.4287,
      "step": 98
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2648797130712586,
      "learning_rate": 2.020408163265306e-06,
      "loss": 0.5736,
      "step": 99
    },
    {
      "epoch": 0.0,
      "grad_norm": 0.8919768561989714,
      "learning_rate": 2.0408163265306125e-06,
      "loss": 0.1669,
      "step": 100
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0370118283980951,
      "learning_rate": 2.0612244897959184e-06,
      "loss": 0.2012,
      "step": 101
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.735468646570244,
      "learning_rate": 2.0816326530612247e-06,
      "loss": 0.3682,
      "step": 102
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2679934157753558,
      "learning_rate": 2.1020408163265306e-06,
      "loss": 0.3266,
      "step": 103
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.341456042731228,
      "learning_rate": 2.122448979591837e-06,
      "loss": 0.6012,
      "step": 104
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.056435102116227,
      "learning_rate": 2.1428571428571427e-06,
      "loss": 0.4977,
      "step": 105
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0443406683545513,
      "learning_rate": 2.1632653061224495e-06,
      "loss": 0.2326,
      "step": 106
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1892774980075713,
      "learning_rate": 2.1836734693877553e-06,
      "loss": 0.5876,
      "step": 107
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0877649502244326,
      "learning_rate": 2.2040816326530616e-06,
      "loss": 0.3168,
      "step": 108
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.3299597230420965,
      "learning_rate": 2.2244897959183675e-06,
      "loss": 0.7532,
      "step": 109
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1404313452803199,
      "learning_rate": 2.244897959183674e-06,
      "loss": 0.0778,
      "step": 110
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.0775967474666386,
      "learning_rate": 2.2653061224489797e-06,
      "loss": 0.2509,
      "step": 111
    },
    {
      "epoch": 0.0,
      "grad_norm": 19.51308692535771,
      "learning_rate": 2.285714285714286e-06,
      "loss": 0.5967,
      "step": 112
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.99100060709999,
      "learning_rate": 2.306122448979592e-06,
      "loss": 0.5602,
      "step": 113
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.6274754638600852,
      "learning_rate": 2.326530612244898e-06,
      "loss": 0.3475,
      "step": 114
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2270027105188364,
      "learning_rate": 2.3469387755102044e-06,
      "loss": 0.3153,
      "step": 115
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.836915214420922,
      "learning_rate": 2.3673469387755103e-06,
      "loss": 0.3595,
      "step": 116
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.692560862218641,
      "learning_rate": 2.3877551020408166e-06,
      "loss": 0.7197,
      "step": 117
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.580140275392278,
      "learning_rate": 2.4081632653061225e-06,
      "loss": 0.5704,
      "step": 118
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7430257235928395,
      "learning_rate": 2.428571428571429e-06,
      "loss": 0.158,
      "step": 119
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.9477610888550567,
      "learning_rate": 2.4489795918367347e-06,
      "loss": 0.405,
      "step": 120
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1776610161582513,
      "learning_rate": 2.469387755102041e-06,
      "loss": 0.3081,
      "step": 121
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.1194087517931415,
      "learning_rate": 2.489795918367347e-06,
      "loss": 0.5826,
      "step": 122
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0274896993502294,
      "learning_rate": 2.5102040816326536e-06,
      "loss": 0.4368,
      "step": 123
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8173499585065778,
      "learning_rate": 2.530612244897959e-06,
      "loss": 0.3115,
      "step": 124
    },
    {
      "epoch": 0.0,
      "grad_norm": 5.057295810289334,
      "learning_rate": 2.5510204081632657e-06,
      "loss": 0.0917,
      "step": 125
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4404640606157797,
      "learning_rate": 2.571428571428571e-06,
      "loss": 0.4056,
      "step": 126
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.000347785848145,
      "learning_rate": 2.591836734693878e-06,
      "loss": 0.344,
      "step": 127
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6492172974725134,
      "learning_rate": 2.6122448979591842e-06,
      "loss": 0.2979,
      "step": 128
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.499527485385493,
      "learning_rate": 2.63265306122449e-06,
      "loss": 0.2253,
      "step": 129
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.369497520350193,
      "learning_rate": 2.6530612244897964e-06,
      "loss": 0.6332,
      "step": 130
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.2533983344673576,
      "learning_rate": 2.6734693877551023e-06,
      "loss": 0.5473,
      "step": 131
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3539209272579367,
      "learning_rate": 2.6938775510204086e-06,
      "loss": 0.3151,
      "step": 132
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0867483037805616,
      "learning_rate": 2.7142857142857144e-06,
      "loss": 0.3108,
      "step": 133
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.702633219739125,
      "learning_rate": 2.7346938775510207e-06,
      "loss": 0.2489,
      "step": 134
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.7243959671074967,
      "learning_rate": 2.7551020408163266e-06,
      "loss": 0.7117,
      "step": 135
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.001570755066192,
      "learning_rate": 2.775510204081633e-06,
      "loss": 0.4598,
      "step": 136
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.1245689675561743,
      "learning_rate": 2.795918367346939e-06,
      "loss": 0.4249,
      "step": 137
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3807048589735762,
      "learning_rate": 2.816326530612245e-06,
      "loss": 0.303,
      "step": 138
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.304276843423313,
      "learning_rate": 2.8367346938775514e-06,
      "loss": 0.4009,
      "step": 139
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.325574736646889,
      "learning_rate": 2.8571428571428573e-06,
      "loss": 0.5476,
      "step": 140
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.9217859560715835,
      "learning_rate": 2.8775510204081636e-06,
      "loss": 0.8032,
      "step": 141
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.4142626859601246,
      "learning_rate": 2.8979591836734694e-06,
      "loss": 0.173,
      "step": 142
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3227142737155693,
      "learning_rate": 2.9183673469387757e-06,
      "loss": 0.2525,
      "step": 143
    },
    {
      "epoch": 0.0,
      "grad_norm": 8.17316725554456,
      "learning_rate": 2.938775510204082e-06,
      "loss": 0.0848,
      "step": 144
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.114704442239561,
      "learning_rate": 2.959183673469388e-06,
      "loss": 0.3355,
      "step": 145
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.7533060614172895,
      "learning_rate": 2.979591836734694e-06,
      "loss": 0.3254,
      "step": 146
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.1642939970907549,
      "learning_rate": 3e-06,
      "loss": 0.3681,
      "step": 147
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.786392276579405,
      "learning_rate": 3.0204081632653064e-06,
      "loss": 0.6742,
      "step": 148
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.750095227014342,
      "learning_rate": 3.0408163265306122e-06,
      "loss": 0.608,
      "step": 149
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.3790061419221935,
      "learning_rate": 3.0612244897959185e-06,
      "loss": 0.3964,
      "step": 150
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.587936159384553,
      "learning_rate": 3.0816326530612244e-06,
      "loss": 0.2461,
      "step": 151
    },
    {
      "epoch": 0.0,
      "grad_norm": 11.063459132712284,
      "learning_rate": 3.1020408163265307e-06,
      "loss": 0.3529,
      "step": 152
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.8333617793607992,
      "learning_rate": 3.1224489795918374e-06,
      "loss": 0.2151,
      "step": 153
    },
    {
      "epoch": 0.0,
      "grad_norm": 3.065750064462318,
      "learning_rate": 3.142857142857143e-06,
      "loss": 0.6608,
      "step": 154
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.206540946144785,
      "learning_rate": 3.1632653061224496e-06,
      "loss": 0.3957,
      "step": 155
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.0742332197051594,
      "learning_rate": 3.183673469387755e-06,
      "loss": 0.3427,
      "step": 156
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.5905926489919318,
      "learning_rate": 3.204081632653062e-06,
      "loss": 0.3405,
      "step": 157
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.6248197788639773,
      "learning_rate": 3.2244897959183672e-06,
      "loss": 0.5978,
      "step": 158
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.560033762793477,
      "learning_rate": 3.244897959183674e-06,
      "loss": 0.7672,
      "step": 159
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.844129014676336,
      "learning_rate": 3.2653061224489794e-06,
      "loss": 0.2283,
      "step": 160
    },
    {
      "epoch": 0.0,
      "grad_norm": 2.776022831240341,
      "learning_rate": 3.285714285714286e-06,
      "loss": 0.2319,
      "step": 161
    },
    {
      "epoch": 0.0,
      "grad_norm": 1.2903407147187096,
      "learning_rate": 3.3061224489795924e-06,
      "loss": 0.3278,
      "step": 162
    },
    {
      "epoch": 0.0,
      "grad_norm": 4.9414259875855,
      "learning_rate": 3.3265306122448983e-06,
      "loss": 0.6505,
      "step": 163
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1443687906129596,
      "learning_rate": 3.3469387755102046e-06,
      "loss": 0.4273,
      "step": 164
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.483185767995319,
      "learning_rate": 3.3673469387755105e-06,
      "loss": 0.3213,
      "step": 165
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6306571364800826,
      "learning_rate": 3.3877551020408168e-06,
      "loss": 0.5633,
      "step": 166
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.01848492989349,
      "learning_rate": 3.4081632653061227e-06,
      "loss": 0.7974,
      "step": 167
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.485531326816614,
      "learning_rate": 3.428571428571429e-06,
      "loss": 0.3267,
      "step": 168
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6108843858826807,
      "learning_rate": 3.4489795918367353e-06,
      "loss": 0.4111,
      "step": 169
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2869073422769843,
      "learning_rate": 3.469387755102041e-06,
      "loss": 0.2553,
      "step": 170
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.3864923250538124,
      "learning_rate": 3.4897959183673474e-06,
      "loss": 0.1288,
      "step": 171
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1525805963665094,
      "learning_rate": 3.5102040816326533e-06,
      "loss": 0.5534,
      "step": 172
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7426498097809973,
      "learning_rate": 3.5306122448979596e-06,
      "loss": 0.3811,
      "step": 173
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0330933753733014,
      "learning_rate": 3.5510204081632655e-06,
      "loss": 0.3345,
      "step": 174
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6517975095388353,
      "learning_rate": 3.5714285714285718e-06,
      "loss": 0.2929,
      "step": 175
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.283568771629904,
      "learning_rate": 3.5918367346938777e-06,
      "loss": 0.839,
      "step": 176
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1754429747781465,
      "learning_rate": 3.612244897959184e-06,
      "loss": 0.6089,
      "step": 177
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7020657973610738,
      "learning_rate": 3.6326530612244903e-06,
      "loss": 0.3091,
      "step": 178
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9508027515352977,
      "learning_rate": 3.653061224489796e-06,
      "loss": 0.2035,
      "step": 179
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1176771082111991,
      "learning_rate": 3.6734693877551024e-06,
      "loss": 0.3047,
      "step": 180
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.325915779391523,
      "learning_rate": 3.6938775510204083e-06,
      "loss": 0.4014,
      "step": 181
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5057896145771836,
      "learning_rate": 3.7142857142857146e-06,
      "loss": 0.4924,
      "step": 182
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.162401232270206,
      "learning_rate": 3.7346938775510205e-06,
      "loss": 0.2262,
      "step": 183
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.522008395882638,
      "learning_rate": 3.7551020408163268e-06,
      "loss": 0.3633,
      "step": 184
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.2621344608361165,
      "learning_rate": 3.7755102040816327e-06,
      "loss": 0.7275,
      "step": 185
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0522703054050482,
      "learning_rate": 3.795918367346939e-06,
      "loss": 0.2982,
      "step": 186
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4302667498712207,
      "learning_rate": 3.816326530612245e-06,
      "loss": 0.4671,
      "step": 187
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.073230615525919,
      "learning_rate": 3.836734693877551e-06,
      "loss": 0.255,
      "step": 188
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2391193188839138,
      "learning_rate": 3.857142857142858e-06,
      "loss": 0.2193,
      "step": 189
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5856600327557113,
      "learning_rate": 3.877551020408164e-06,
      "loss": 0.4993,
      "step": 190
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.291768051348699,
      "learning_rate": 3.89795918367347e-06,
      "loss": 0.4027,
      "step": 191
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0799157101258194,
      "learning_rate": 3.9183673469387755e-06,
      "loss": 0.242,
      "step": 192
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.241645132635435,
      "learning_rate": 3.938775510204082e-06,
      "loss": 0.4303,
      "step": 193
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.746409727553965,
      "learning_rate": 3.959183673469388e-06,
      "loss": 0.5362,
      "step": 194
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.185726689616376,
      "learning_rate": 3.979591836734694e-06,
      "loss": 0.8051,
      "step": 195
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9892515798715403,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0846,
      "step": 196
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.007621121700785,
      "learning_rate": 4.0204081632653065e-06,
      "loss": 0.275,
      "step": 197
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4265515876701185,
      "learning_rate": 4.040816326530612e-06,
      "loss": 0.3604,
      "step": 198
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4072887054654353,
      "learning_rate": 4.061224489795918e-06,
      "loss": 0.4305,
      "step": 199
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.212225841231732,
      "learning_rate": 4.081632653061225e-06,
      "loss": 0.2963,
      "step": 200
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9484591179553828,
      "learning_rate": 4.102040816326531e-06,
      "loss": 0.2375,
      "step": 201
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6156215104709313,
      "learning_rate": 4.122448979591837e-06,
      "loss": 0.654,
      "step": 202
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.6474191555313,
      "learning_rate": 4.1428571428571435e-06,
      "loss": 0.7221,
      "step": 203
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9819805050070208,
      "learning_rate": 4.163265306122449e-06,
      "loss": 0.3544,
      "step": 204
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9479151929788296,
      "learning_rate": 4.183673469387755e-06,
      "loss": 0.2048,
      "step": 205
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5674435329706171,
      "learning_rate": 4.204081632653061e-06,
      "loss": 0.3779,
      "step": 206
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.152444766266446,
      "learning_rate": 4.224489795918368e-06,
      "loss": 0.5075,
      "step": 207
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3551524167007711,
      "learning_rate": 4.244897959183674e-06,
      "loss": 0.4053,
      "step": 208
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8013290880275719,
      "learning_rate": 4.2653061224489804e-06,
      "loss": 0.244,
      "step": 209
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8588115046503881,
      "learning_rate": 4.2857142857142855e-06,
      "loss": 0.3899,
      "step": 210
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1962949233743365,
      "learning_rate": 4.306122448979592e-06,
      "loss": 0.3042,
      "step": 211
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.542145021702998,
      "learning_rate": 4.326530612244899e-06,
      "loss": 0.7942,
      "step": 212
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.856411389262536,
      "learning_rate": 4.346938775510205e-06,
      "loss": 0.5934,
      "step": 213
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0991386720782352,
      "learning_rate": 4.367346938775511e-06,
      "loss": 0.3706,
      "step": 214
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1170030095264376,
      "learning_rate": 4.3877551020408165e-06,
      "loss": 0.3045,
      "step": 215
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1466690981639331,
      "learning_rate": 4.408163265306123e-06,
      "loss": 0.3895,
      "step": 216
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8915687401979833,
      "learning_rate": 4.428571428571429e-06,
      "loss": 0.651,
      "step": 217
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7434988019477953,
      "learning_rate": 4.448979591836735e-06,
      "loss": 0.1219,
      "step": 218
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.582743017069143,
      "learning_rate": 4.469387755102041e-06,
      "loss": 0.2459,
      "step": 219
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.134938891765576,
      "learning_rate": 4.489795918367348e-06,
      "loss": 0.311,
      "step": 220
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.740803894122949,
      "learning_rate": 4.5102040816326535e-06,
      "loss": 0.8005,
      "step": 221
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9161948653638429,
      "learning_rate": 4.530612244897959e-06,
      "loss": 0.3033,
      "step": 222
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5658665453617013,
      "learning_rate": 4.551020408163266e-06,
      "loss": 0.442,
      "step": 223
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9326140198974505,
      "learning_rate": 4.571428571428572e-06,
      "loss": 0.302,
      "step": 224
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4958009125689316,
      "learning_rate": 4.591836734693878e-06,
      "loss": 0.5207,
      "step": 225
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5771066785880485,
      "learning_rate": 4.612244897959184e-06,
      "loss": 0.6692,
      "step": 226
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7504234654891411,
      "learning_rate": 4.63265306122449e-06,
      "loss": 0.2043,
      "step": 227
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1110566312133083,
      "learning_rate": 4.653061224489796e-06,
      "loss": 0.292,
      "step": 228
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.327023179730912,
      "learning_rate": 4.673469387755102e-06,
      "loss": 0.3258,
      "step": 229
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.859482501608939,
      "learning_rate": 4.693877551020409e-06,
      "loss": 0.8724,
      "step": 230
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0139463633430643,
      "learning_rate": 4.714285714285715e-06,
      "loss": 0.5159,
      "step": 231
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3558364515963839,
      "learning_rate": 4.734693877551021e-06,
      "loss": 0.4798,
      "step": 232
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.157552912542442,
      "learning_rate": 4.7551020408163265e-06,
      "loss": 0.2384,
      "step": 233
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8918909177323217,
      "learning_rate": 4.775510204081633e-06,
      "loss": 0.3537,
      "step": 234
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.691911984165747,
      "learning_rate": 4.795918367346939e-06,
      "loss": 0.4944,
      "step": 235
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0231256564756903,
      "learning_rate": 4.816326530612245e-06,
      "loss": 0.1795,
      "step": 236
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0943349999421723,
      "learning_rate": 4.836734693877552e-06,
      "loss": 0.0863,
      "step": 237
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3192667850528783,
      "learning_rate": 4.857142857142858e-06,
      "loss": 0.3091,
      "step": 238
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2255186881624454,
      "learning_rate": 4.8775510204081635e-06,
      "loss": 0.4484,
      "step": 239
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.070174099880665,
      "learning_rate": 4.897959183673469e-06,
      "loss": 0.3822,
      "step": 240
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.965694992803717,
      "learning_rate": 4.918367346938776e-06,
      "loss": 0.5885,
      "step": 241
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.205467878567083,
      "learning_rate": 4.938775510204082e-06,
      "loss": 0.2399,
      "step": 242
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.705410020802552,
      "learning_rate": 4.959183673469388e-06,
      "loss": 0.6021,
      "step": 243
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.167074048571217,
      "learning_rate": 4.979591836734694e-06,
      "loss": 0.4097,
      "step": 244
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8428928690161919,
      "learning_rate": 5e-06,
      "loss": 0.2468,
      "step": 245
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8632332989779945,
      "learning_rate": 5.020408163265307e-06,
      "loss": 0.1835,
      "step": 246
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1729911811157556,
      "learning_rate": 5.040816326530613e-06,
      "loss": 0.3321,
      "step": 247
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.604173314076025,
      "learning_rate": 5.061224489795918e-06,
      "loss": 0.6545,
      "step": 248
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8743685083167203,
      "learning_rate": 5.081632653061225e-06,
      "loss": 0.5772,
      "step": 249
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0846822270675114,
      "learning_rate": 5.1020408163265315e-06,
      "loss": 0.3988,
      "step": 250
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9042355050749824,
      "learning_rate": 5.122448979591837e-06,
      "loss": 0.2442,
      "step": 251
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.197690013908971,
      "learning_rate": 5.142857142857142e-06,
      "loss": 0.4067,
      "step": 252
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0152399980517,
      "learning_rate": 5.163265306122449e-06,
      "loss": 0.5409,
      "step": 253
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.896430805498563,
      "learning_rate": 5.183673469387756e-06,
      "loss": 0.276,
      "step": 254
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9218299176442906,
      "learning_rate": 5.204081632653062e-06,
      "loss": 0.2201,
      "step": 255
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5763570981816888,
      "learning_rate": 5.2244897959183684e-06,
      "loss": 0.378,
      "step": 256
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.050369577110174,
      "learning_rate": 5.2448979591836735e-06,
      "loss": 0.339,
      "step": 257
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3425900890160039,
      "learning_rate": 5.26530612244898e-06,
      "loss": 0.4489,
      "step": 258
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.46767593005749,
      "learning_rate": 5.285714285714286e-06,
      "loss": 0.2887,
      "step": 259
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7732761664264869,
      "learning_rate": 5.306122448979593e-06,
      "loss": 0.4048,
      "step": 260
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0644367659598406,
      "learning_rate": 5.326530612244898e-06,
      "loss": 0.3114,
      "step": 261
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6629909792244613,
      "learning_rate": 5.3469387755102045e-06,
      "loss": 0.7208,
      "step": 262
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9986082524845771,
      "learning_rate": 5.36734693877551e-06,
      "loss": 0.333,
      "step": 263
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6818142721083178,
      "learning_rate": 5.387755102040817e-06,
      "loss": 0.0812,
      "step": 264
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9156654658726437,
      "learning_rate": 5.408163265306123e-06,
      "loss": 0.3204,
      "step": 265
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9212747040847982,
      "learning_rate": 5.428571428571429e-06,
      "loss": 0.5217,
      "step": 266
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.845876560091224,
      "learning_rate": 5.448979591836735e-06,
      "loss": 0.6111,
      "step": 267
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8061387208200511,
      "learning_rate": 5.4693877551020415e-06,
      "loss": 0.1155,
      "step": 268
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2082679547196598,
      "learning_rate": 5.489795918367347e-06,
      "loss": 0.2974,
      "step": 269
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8933671567938759,
      "learning_rate": 5.510204081632653e-06,
      "loss": 0.3042,
      "step": 270
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.426042621469968,
      "learning_rate": 5.530612244897959e-06,
      "loss": 0.6899,
      "step": 271
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6882158603445183,
      "learning_rate": 5.551020408163266e-06,
      "loss": 0.1222,
      "step": 272
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8754440456231853,
      "learning_rate": 5.571428571428572e-06,
      "loss": 0.218,
      "step": 273
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1484390683297379,
      "learning_rate": 5.591836734693878e-06,
      "loss": 0.2989,
      "step": 274
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0080744648669873,
      "learning_rate": 5.6122448979591834e-06,
      "loss": 0.3745,
      "step": 275
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.043766371500851,
      "learning_rate": 5.63265306122449e-06,
      "loss": 0.5371,
      "step": 276
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8659428330194543,
      "learning_rate": 5.653061224489796e-06,
      "loss": 0.1835,
      "step": 277
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9873912253326571,
      "learning_rate": 5.673469387755103e-06,
      "loss": 0.3065,
      "step": 278
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3278038589029486,
      "learning_rate": 5.6938775510204095e-06,
      "loss": 0.3626,
      "step": 279
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.121814614849602,
      "learning_rate": 5.7142857142857145e-06,
      "loss": 0.697,
      "step": 280
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9010069541340373,
      "learning_rate": 5.73469387755102e-06,
      "loss": 0.3442,
      "step": 281
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.638941354971352,
      "learning_rate": 5.755102040816327e-06,
      "loss": 0.4571,
      "step": 282
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0004029396597296,
      "learning_rate": 5.775510204081634e-06,
      "loss": 0.2292,
      "step": 283
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4557765199577897,
      "learning_rate": 5.795918367346939e-06,
      "loss": 0.4728,
      "step": 284
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4137984068512988,
      "learning_rate": 5.816326530612246e-06,
      "loss": 0.6163,
      "step": 285
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.854142891000127,
      "learning_rate": 5.8367346938775515e-06,
      "loss": 0.2909,
      "step": 286
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1993432378844826,
      "learning_rate": 5.857142857142858e-06,
      "loss": 0.0899,
      "step": 287
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1243978417022815,
      "learning_rate": 5.877551020408164e-06,
      "loss": 0.3552,
      "step": 288
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2316991941784299,
      "learning_rate": 5.89795918367347e-06,
      "loss": 0.1791,
      "step": 289
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6303872567543398,
      "learning_rate": 5.918367346938776e-06,
      "loss": 0.4771,
      "step": 290
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7161764985808474,
      "learning_rate": 5.9387755102040825e-06,
      "loss": 0.5743,
      "step": 291
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8926001426839676,
      "learning_rate": 5.959183673469388e-06,
      "loss": 0.2449,
      "step": 292
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.840375872521032,
      "learning_rate": 5.979591836734694e-06,
      "loss": 0.369,
      "step": 293
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7988494535844217,
      "learning_rate": 6e-06,
      "loss": 0.5762,
      "step": 294
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3247543889620306,
      "learning_rate": 6.020408163265307e-06,
      "loss": 0.7422,
      "step": 295
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.093059385628771,
      "learning_rate": 6.040816326530613e-06,
      "loss": 0.0877,
      "step": 296
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.0044475860394866,
      "learning_rate": 6.0612244897959195e-06,
      "loss": 0.3277,
      "step": 297
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2980005899972635,
      "learning_rate": 6.0816326530612245e-06,
      "loss": 0.3588,
      "step": 298
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2105271556901898,
      "learning_rate": 6.102040816326531e-06,
      "loss": 0.3458,
      "step": 299
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.453368600011727,
      "learning_rate": 6.122448979591837e-06,
      "loss": 0.4418,
      "step": 300
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0473783288848246,
      "learning_rate": 6.142857142857144e-06,
      "loss": 0.3235,
      "step": 301
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4261696047208892,
      "learning_rate": 6.163265306122449e-06,
      "loss": 0.5228,
      "step": 302
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.4167794379990846,
      "learning_rate": 6.1836734693877556e-06,
      "loss": 0.5662,
      "step": 303
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.712374930421919,
      "learning_rate": 6.2040816326530614e-06,
      "loss": 0.3972,
      "step": 304
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8019398624302515,
      "learning_rate": 6.224489795918368e-06,
      "loss": 0.2558,
      "step": 305
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.911584076128742,
      "learning_rate": 6.244897959183675e-06,
      "loss": 0.2422,
      "step": 306
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.079644641146538,
      "learning_rate": 6.26530612244898e-06,
      "loss": 0.1538,
      "step": 307
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7728496150202429,
      "learning_rate": 6.285714285714286e-06,
      "loss": 0.6461,
      "step": 308
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.636041620974596,
      "learning_rate": 6.3061224489795925e-06,
      "loss": 0.4644,
      "step": 309
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1974359805462127,
      "learning_rate": 6.326530612244899e-06,
      "loss": 0.3342,
      "step": 310
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9469712335219249,
      "learning_rate": 6.346938775510204e-06,
      "loss": 0.3181,
      "step": 311
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1626373785262802,
      "learning_rate": 6.36734693877551e-06,
      "loss": 0.4364,
      "step": 312
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.9151757563690432,
      "learning_rate": 6.387755102040817e-06,
      "loss": 0.6016,
      "step": 313
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4517337859100172,
      "learning_rate": 6.408163265306124e-06,
      "loss": 0.3144,
      "step": 314
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7840401641836314,
      "learning_rate": 6.4285714285714295e-06,
      "loss": 0.2189,
      "step": 315
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7703151744075702,
      "learning_rate": 6.4489795918367345e-06,
      "loss": 0.2168,
      "step": 316
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1102752026442635,
      "learning_rate": 6.469387755102041e-06,
      "loss": 0.4501,
      "step": 317
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6830202598798778,
      "learning_rate": 6.489795918367348e-06,
      "loss": 0.5134,
      "step": 318
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9783326206369412,
      "learning_rate": 6.510204081632654e-06,
      "loss": 0.2929,
      "step": 319
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3646047895716307,
      "learning_rate": 6.530612244897959e-06,
      "loss": 0.3103,
      "step": 320
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5905815851167238,
      "learning_rate": 6.5510204081632656e-06,
      "loss": 0.8237,
      "step": 321
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.145022450931165,
      "learning_rate": 6.571428571428572e-06,
      "loss": 0.3001,
      "step": 322
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.110118905017628,
      "learning_rate": 6.591836734693878e-06,
      "loss": 0.3686,
      "step": 323
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9581493816013513,
      "learning_rate": 6.612244897959185e-06,
      "loss": 0.2111,
      "step": 324
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0166428874195688,
      "learning_rate": 6.63265306122449e-06,
      "loss": 0.3426,
      "step": 325
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7448804649218044,
      "learning_rate": 6.653061224489797e-06,
      "loss": 0.4768,
      "step": 326
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8555958292654826,
      "learning_rate": 6.6734693877551025e-06,
      "loss": 0.6089,
      "step": 327
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1966338975264694,
      "learning_rate": 6.693877551020409e-06,
      "loss": 0.2453,
      "step": 328
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8559762129263041,
      "learning_rate": 6.714285714285714e-06,
      "loss": 0.306,
      "step": 329
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.139430821726984,
      "learning_rate": 6.734693877551021e-06,
      "loss": 0.8511,
      "step": 330
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.4718801366947534,
      "learning_rate": 6.755102040816327e-06,
      "loss": 0.6201,
      "step": 331
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5475201080854903,
      "learning_rate": 6.7755102040816336e-06,
      "loss": 0.3682,
      "step": 332
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.926527411018724,
      "learning_rate": 6.7959183673469394e-06,
      "loss": 0.2238,
      "step": 333
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9292740189606314,
      "learning_rate": 6.816326530612245e-06,
      "loss": 0.3302,
      "step": 334
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8785540060849909,
      "learning_rate": 6.836734693877551e-06,
      "loss": 0.3502,
      "step": 335
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7558179794598143,
      "learning_rate": 6.857142857142858e-06,
      "loss": 0.557,
      "step": 336
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9721334421405073,
      "learning_rate": 6.877551020408164e-06,
      "loss": 0.207,
      "step": 337
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.327347425860215,
      "learning_rate": 6.8979591836734705e-06,
      "loss": 0.3982,
      "step": 338
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.1500586095043284,
      "learning_rate": 6.9183673469387755e-06,
      "loss": 0.7342,
      "step": 339
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8616177835209609,
      "learning_rate": 6.938775510204082e-06,
      "loss": 0.3521,
      "step": 340
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6540676805718268,
      "learning_rate": 6.959183673469388e-06,
      "loss": 0.2675,
      "step": 341
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9787147784468877,
      "learning_rate": 6.979591836734695e-06,
      "loss": 0.2522,
      "step": 342
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.914284262975302,
      "learning_rate": 7e-06,
      "loss": 0.3663,
      "step": 343
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9302441443063613,
      "learning_rate": 7.020408163265307e-06,
      "loss": 0.3548,
      "step": 344
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.5620036471623,
      "learning_rate": 7.0408163265306125e-06,
      "loss": 0.6857,
      "step": 345
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8026152292370166,
      "learning_rate": 7.061224489795919e-06,
      "loss": 0.2431,
      "step": 346
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8541684819857843,
      "learning_rate": 7.081632653061226e-06,
      "loss": 0.3455,
      "step": 347
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2984751376228005,
      "learning_rate": 7.102040816326531e-06,
      "loss": 0.7778,
      "step": 348
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.0839401664529964,
      "learning_rate": 7.122448979591837e-06,
      "loss": 0.7137,
      "step": 349
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5448918602207042,
      "learning_rate": 7.1428571428571436e-06,
      "loss": 0.3474,
      "step": 350
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9258615230700239,
      "learning_rate": 7.16326530612245e-06,
      "loss": 0.3278,
      "step": 351
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9898229948056005,
      "learning_rate": 7.183673469387755e-06,
      "loss": 0.4257,
      "step": 352
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1459091102899905,
      "learning_rate": 7.204081632653061e-06,
      "loss": 0.3707,
      "step": 353
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7798935703060488,
      "learning_rate": 7.224489795918368e-06,
      "loss": 0.1297,
      "step": 354
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.795966379404617,
      "learning_rate": 7.244897959183675e-06,
      "loss": 0.2091,
      "step": 355
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2525550320778687,
      "learning_rate": 7.2653061224489805e-06,
      "loss": 0.423,
      "step": 356
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.785383094912085,
      "learning_rate": 7.285714285714286e-06,
      "loss": 0.6352,
      "step": 357
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8567761276411174,
      "learning_rate": 7.306122448979592e-06,
      "loss": 0.3631,
      "step": 358
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7043497720304703,
      "learning_rate": 7.326530612244899e-06,
      "loss": 0.4441,
      "step": 359
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9485829161154159,
      "learning_rate": 7.346938775510205e-06,
      "loss": 0.3164,
      "step": 360
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8650560265118774,
      "learning_rate": 7.367346938775511e-06,
      "loss": 0.5163,
      "step": 361
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6375158910524807,
      "learning_rate": 7.387755102040817e-06,
      "loss": 0.5686,
      "step": 362
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.68501608635458,
      "learning_rate": 7.408163265306123e-06,
      "loss": 0.1868,
      "step": 363
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9769504359176036,
      "learning_rate": 7.428571428571429e-06,
      "loss": 0.2956,
      "step": 364
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0582791259129578,
      "learning_rate": 7.448979591836736e-06,
      "loss": 0.287,
      "step": 365
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9544417199878605,
      "learning_rate": 7.469387755102041e-06,
      "loss": 0.81,
      "step": 366
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.449715972913052,
      "learning_rate": 7.489795918367348e-06,
      "loss": 0.5841,
      "step": 367
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8414082054097125,
      "learning_rate": 7.5102040816326536e-06,
      "loss": 0.5127,
      "step": 368
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9848358666435301,
      "learning_rate": 7.53061224489796e-06,
      "loss": 0.2879,
      "step": 369
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.946927324126308,
      "learning_rate": 7.551020408163265e-06,
      "loss": 0.3421,
      "step": 370
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.053006667644627,
      "learning_rate": 7.571428571428572e-06,
      "loss": 0.3484,
      "step": 371
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6641321280346989,
      "learning_rate": 7.591836734693878e-06,
      "loss": 0.0812,
      "step": 372
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6624548353756894,
      "learning_rate": 7.612244897959185e-06,
      "loss": 0.3709,
      "step": 373
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0532881496502668,
      "learning_rate": 7.63265306122449e-06,
      "loss": 0.2677,
      "step": 374
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.8400598869827127,
      "learning_rate": 7.653061224489796e-06,
      "loss": 0.7644,
      "step": 375
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9558010818327125,
      "learning_rate": 7.673469387755102e-06,
      "loss": 0.3249,
      "step": 376
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.771086324389035,
      "learning_rate": 7.693877551020409e-06,
      "loss": 0.6117,
      "step": 377
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7960090071541908,
      "learning_rate": 7.714285714285716e-06,
      "loss": 0.2408,
      "step": 378
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5486832536690764,
      "learning_rate": 7.73469387755102e-06,
      "loss": 0.638,
      "step": 379
    },
    {
      "epoch": 0.01,
      "grad_norm": 7.193006517321874,
      "learning_rate": 7.755102040816327e-06,
      "loss": 0.5569,
      "step": 380
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.339605053672799,
      "learning_rate": 7.775510204081632e-06,
      "loss": 0.1934,
      "step": 381
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.471553925355099,
      "learning_rate": 7.79591836734694e-06,
      "loss": 0.2991,
      "step": 382
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4973886897466264,
      "learning_rate": 7.816326530612246e-06,
      "loss": 0.3057,
      "step": 383
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.326411910901716,
      "learning_rate": 7.836734693877551e-06,
      "loss": 0.7042,
      "step": 384
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.3827902453260204,
      "learning_rate": 7.857142857142858e-06,
      "loss": 0.5786,
      "step": 385
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7185694994860188,
      "learning_rate": 7.877551020408164e-06,
      "loss": 0.5989,
      "step": 386
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.359681023057485,
      "learning_rate": 7.897959183673471e-06,
      "loss": 0.2557,
      "step": 387
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.318548158944632,
      "learning_rate": 7.918367346938776e-06,
      "loss": 0.6283,
      "step": 388
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9520737136866437,
      "learning_rate": 7.938775510204081e-06,
      "loss": 0.2957,
      "step": 389
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2125476120437042,
      "learning_rate": 7.959183673469388e-06,
      "loss": 0.1704,
      "step": 390
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7974204717077,
      "learning_rate": 7.979591836734695e-06,
      "loss": 0.1909,
      "step": 391
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.410727026566612,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.4181,
      "step": 392
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.417799214685939,
      "learning_rate": 8.020408163265306e-06,
      "loss": 0.4058,
      "step": 393
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0266303963672991,
      "learning_rate": 8.040816326530613e-06,
      "loss": 0.379,
      "step": 394
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.7124833480645405,
      "learning_rate": 8.06122448979592e-06,
      "loss": 0.5695,
      "step": 395
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1103107971790354,
      "learning_rate": 8.081632653061225e-06,
      "loss": 0.207,
      "step": 396
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.335531250661806,
      "learning_rate": 8.102040816326532e-06,
      "loss": 0.4057,
      "step": 397
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.6696803524401296,
      "learning_rate": 8.122448979591837e-06,
      "loss": 0.5475,
      "step": 398
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0401530049576941,
      "learning_rate": 8.142857142857143e-06,
      "loss": 0.2731,
      "step": 399
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8924394167297713,
      "learning_rate": 8.16326530612245e-06,
      "loss": 0.266,
      "step": 400
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0775221886675748,
      "learning_rate": 8.183673469387757e-06,
      "loss": 0.3275,
      "step": 401
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2985216930782817,
      "learning_rate": 8.204081632653062e-06,
      "loss": 0.5328,
      "step": 402
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.463966422410408,
      "learning_rate": 8.224489795918369e-06,
      "loss": 0.6018,
      "step": 403
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8904623151275716,
      "learning_rate": 8.244897959183674e-06,
      "loss": 0.1295,
      "step": 404
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.436695791736262,
      "learning_rate": 8.26530612244898e-06,
      "loss": 0.313,
      "step": 405
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1405631269256995,
      "learning_rate": 8.285714285714287e-06,
      "loss": 0.3066,
      "step": 406
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.445485072708934,
      "learning_rate": 8.306122448979592e-06,
      "loss": 0.7119,
      "step": 407
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.469258466876562,
      "learning_rate": 8.326530612244899e-06,
      "loss": 0.6055,
      "step": 408
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0905524724979083,
      "learning_rate": 8.346938775510205e-06,
      "loss": 0.1904,
      "step": 409
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0467754286979944,
      "learning_rate": 8.36734693877551e-06,
      "loss": 0.363,
      "step": 410
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.7423127774910927,
      "learning_rate": 8.387755102040817e-06,
      "loss": 0.536,
      "step": 411
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9986711216702835,
      "learning_rate": 8.408163265306122e-06,
      "loss": 0.411,
      "step": 412
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9299139634911142,
      "learning_rate": 8.428571428571429e-06,
      "loss": 0.1866,
      "step": 413
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9832423734287986,
      "learning_rate": 8.448979591836736e-06,
      "loss": 0.3022,
      "step": 414
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.2589095331031617,
      "learning_rate": 8.469387755102042e-06,
      "loss": 0.3644,
      "step": 415
    },
    {
      "epoch": 0.01,
      "grad_norm": 29.653607335411433,
      "learning_rate": 8.489795918367347e-06,
      "loss": 0.8725,
      "step": 416
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7979560867831784,
      "learning_rate": 8.510204081632654e-06,
      "loss": 0.31,
      "step": 417
    },
    {
      "epoch": 0.01,
      "grad_norm": 4.713856019176225,
      "learning_rate": 8.530612244897961e-06,
      "loss": 0.6077,
      "step": 418
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.072554584288203,
      "learning_rate": 8.551020408163266e-06,
      "loss": 0.2415,
      "step": 419
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5886736093535614,
      "learning_rate": 8.571428571428571e-06,
      "loss": 0.4891,
      "step": 420
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.053432396517986,
      "learning_rate": 8.591836734693878e-06,
      "loss": 0.5827,
      "step": 421
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5620032794556709,
      "learning_rate": 8.612244897959184e-06,
      "loss": 0.0966,
      "step": 422
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3376759607465594,
      "learning_rate": 8.632653061224491e-06,
      "loss": 0.3547,
      "step": 423
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0501690052260575,
      "learning_rate": 8.653061224489798e-06,
      "loss": 0.29,
      "step": 424
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.586704964284831,
      "learning_rate": 8.673469387755103e-06,
      "loss": 0.6864,
      "step": 425
    },
    {
      "epoch": 0.01,
      "grad_norm": 5.357650722594848,
      "learning_rate": 8.69387755102041e-06,
      "loss": 0.7085,
      "step": 426
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.467938500489913,
      "learning_rate": 8.714285714285715e-06,
      "loss": 0.5944,
      "step": 427
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.303217724337933,
      "learning_rate": 8.734693877551021e-06,
      "loss": 0.2758,
      "step": 428
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.5191049236978227,
      "learning_rate": 8.755102040816326e-06,
      "loss": 0.6381,
      "step": 429
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.203619238242545,
      "learning_rate": 8.775510204081633e-06,
      "loss": 0.3261,
      "step": 430
    },
    {
      "epoch": 0.01,
      "grad_norm": 6.013266088562742,
      "learning_rate": 8.79591836734694e-06,
      "loss": 0.8133,
      "step": 431
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6902571172145242,
      "learning_rate": 8.816326530612247e-06,
      "loss": 0.0999,
      "step": 432
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.203311960250954,
      "learning_rate": 8.836734693877552e-06,
      "loss": 0.4342,
      "step": 433
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.8337737909549232,
      "learning_rate": 8.857142857142858e-06,
      "loss": 0.0438,
      "step": 434
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1557750211806916,
      "learning_rate": 8.877551020408163e-06,
      "loss": 0.3424,
      "step": 435
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.285330722925525,
      "learning_rate": 8.89795918367347e-06,
      "loss": 0.4579,
      "step": 436
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.246686459910577,
      "learning_rate": 8.918367346938777e-06,
      "loss": 0.2629,
      "step": 437
    },
    {
      "epoch": 0.01,
      "grad_norm": 12.07319077679582,
      "learning_rate": 8.938775510204082e-06,
      "loss": 0.5909,
      "step": 438
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4501153244024376,
      "learning_rate": 8.959183673469388e-06,
      "loss": 0.5597,
      "step": 439
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3425802483105473,
      "learning_rate": 8.979591836734695e-06,
      "loss": 0.239,
      "step": 440
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9701766825203533,
      "learning_rate": 9e-06,
      "loss": 0.2594,
      "step": 441
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.6776795172346208,
      "learning_rate": 9.020408163265307e-06,
      "loss": 0.4694,
      "step": 442
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1465127541181317,
      "learning_rate": 9.040816326530612e-06,
      "loss": 0.4183,
      "step": 443
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.2032574249081502,
      "learning_rate": 9.061224489795919e-06,
      "loss": 0.5845,
      "step": 444
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.122131326516552,
      "learning_rate": 9.081632653061225e-06,
      "loss": 0.4377,
      "step": 445
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.25109698042287,
      "learning_rate": 9.102040816326532e-06,
      "loss": 0.379,
      "step": 446
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.355565589565279,
      "learning_rate": 9.122448979591837e-06,
      "loss": 0.3345,
      "step": 447
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9822270771460858,
      "learning_rate": 9.142857142857144e-06,
      "loss": 0.3517,
      "step": 448
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1865146633848498,
      "learning_rate": 9.163265306122449e-06,
      "loss": 0.2351,
      "step": 449
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.397457358277249,
      "learning_rate": 9.183673469387756e-06,
      "loss": 0.0999,
      "step": 450
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9501830445915749,
      "learning_rate": 9.204081632653062e-06,
      "loss": 0.3297,
      "step": 451
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1882126617297986,
      "learning_rate": 9.224489795918367e-06,
      "loss": 0.5561,
      "step": 452
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8613813853222919,
      "learning_rate": 9.244897959183674e-06,
      "loss": 0.3865,
      "step": 453
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4659546102345893,
      "learning_rate": 9.26530612244898e-06,
      "loss": 0.5271,
      "step": 454
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.95747342129825,
      "learning_rate": 9.285714285714288e-06,
      "loss": 0.3005,
      "step": 455
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0244093833563812,
      "learning_rate": 9.306122448979593e-06,
      "loss": 0.333,
      "step": 456
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.7577217789077593,
      "learning_rate": 9.326530612244898e-06,
      "loss": 0.7961,
      "step": 457
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.341648047148871,
      "learning_rate": 9.346938775510204e-06,
      "loss": 0.4844,
      "step": 458
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8638383506179236,
      "learning_rate": 9.367346938775511e-06,
      "loss": 0.3339,
      "step": 459
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7724755614350028,
      "learning_rate": 9.387755102040818e-06,
      "loss": 0.2248,
      "step": 460
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.2848690259872393,
      "learning_rate": 9.408163265306123e-06,
      "loss": 0.1773,
      "step": 461
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4658964656572397,
      "learning_rate": 9.42857142857143e-06,
      "loss": 0.6152,
      "step": 462
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.4774844098937383,
      "learning_rate": 9.448979591836736e-06,
      "loss": 0.5786,
      "step": 463
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.169272995675832,
      "learning_rate": 9.469387755102041e-06,
      "loss": 0.2889,
      "step": 464
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.3124899787142414,
      "learning_rate": 9.489795918367348e-06,
      "loss": 0.3182,
      "step": 465
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9318385708694921,
      "learning_rate": 9.510204081632653e-06,
      "loss": 0.3547,
      "step": 466
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.1602823370082325,
      "learning_rate": 9.53061224489796e-06,
      "loss": 0.5687,
      "step": 467
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1578220582580245,
      "learning_rate": 9.551020408163266e-06,
      "loss": 0.1708,
      "step": 468
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0304128706434008,
      "learning_rate": 9.571428571428573e-06,
      "loss": 0.2427,
      "step": 469
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0437703186697416,
      "learning_rate": 9.591836734693878e-06,
      "loss": 0.3941,
      "step": 470
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8111175024233086,
      "learning_rate": 9.612244897959185e-06,
      "loss": 0.3783,
      "step": 471
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.522437224920067,
      "learning_rate": 9.63265306122449e-06,
      "loss": 0.6646,
      "step": 472
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1154346661195935,
      "learning_rate": 9.653061224489797e-06,
      "loss": 0.0884,
      "step": 473
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8428593599544706,
      "learning_rate": 9.673469387755103e-06,
      "loss": 0.3066,
      "step": 474
    },
    {
      "epoch": 0.01,
      "grad_norm": 3.5045150769644904,
      "learning_rate": 9.693877551020408e-06,
      "loss": 0.7865,
      "step": 475
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.0233216355470593,
      "learning_rate": 9.714285714285715e-06,
      "loss": 0.3464,
      "step": 476
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.004880892417148,
      "learning_rate": 9.734693877551022e-06,
      "loss": 0.3989,
      "step": 477
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6693782814020133,
      "learning_rate": 9.755102040816327e-06,
      "loss": 0.2329,
      "step": 478
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8341334078168989,
      "learning_rate": 9.775510204081634e-06,
      "loss": 0.3553,
      "step": 479
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7627561817968833,
      "learning_rate": 9.795918367346939e-06,
      "loss": 0.31,
      "step": 480
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.9889502046681873,
      "learning_rate": 9.816326530612245e-06,
      "loss": 0.7644,
      "step": 481
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7397559424067022,
      "learning_rate": 9.836734693877552e-06,
      "loss": 0.1843,
      "step": 482
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.8098924411282891,
      "learning_rate": 9.857142857142859e-06,
      "loss": 0.3275,
      "step": 483
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.9902833305135174,
      "learning_rate": 9.877551020408164e-06,
      "loss": 0.3509,
      "step": 484
    },
    {
      "epoch": 0.01,
      "grad_norm": 2.54574116841437,
      "learning_rate": 9.89795918367347e-06,
      "loss": 0.8097,
      "step": 485
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.5225103657508605,
      "learning_rate": 9.918367346938776e-06,
      "loss": 0.1372,
      "step": 486
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.6976266925325001,
      "learning_rate": 9.938775510204082e-06,
      "loss": 0.3011,
      "step": 487
    },
    {
      "epoch": 0.01,
      "grad_norm": 1.1720046672331839,
      "learning_rate": 9.959183673469387e-06,
      "loss": 0.597,
      "step": 488
    },
    {
      "epoch": 0.01,
      "grad_norm": 0.7791319598517659,
      "learning_rate": 9.979591836734694e-06,
      "loss": 0.3455,
      "step": 489
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9937189349661096,
      "learning_rate": 1e-05,
      "loss": 0.3395,
      "step": 490
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6045541373165747,
      "learning_rate": 1.0020408163265308e-05,
      "loss": 0.2095,
      "step": 491
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.009100689569268,
      "learning_rate": 1.0040816326530614e-05,
      "loss": 0.3963,
      "step": 492
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7644143954573062,
      "learning_rate": 1.006122448979592e-05,
      "loss": 0.529,
      "step": 493
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.560794911426429,
      "learning_rate": 1.0081632653061226e-05,
      "loss": 0.7086,
      "step": 494
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7853754149266833,
      "learning_rate": 1.0102040816326531e-05,
      "loss": 0.3119,
      "step": 495
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7407466576489805,
      "learning_rate": 1.0122448979591836e-05,
      "loss": 0.402,
      "step": 496
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.83553777540171,
      "learning_rate": 1.0142857142857143e-05,
      "loss": 0.3525,
      "step": 497
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7533877447896665,
      "learning_rate": 1.016326530612245e-05,
      "loss": 0.2914,
      "step": 498
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7737621386159419,
      "learning_rate": 1.0183673469387756e-05,
      "loss": 0.2116,
      "step": 499
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7604578577125376,
      "learning_rate": 1.0204081632653063e-05,
      "loss": 0.2538,
      "step": 500
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9062879906214945,
      "learning_rate": 1.0224489795918368e-05,
      "loss": 0.3095,
      "step": 501
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9426639304963061,
      "learning_rate": 1.0244897959183675e-05,
      "loss": 0.3164,
      "step": 502
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6706324986586005,
      "learning_rate": 1.0265306122448981e-05,
      "loss": 0.5856,
      "step": 503
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0071504504438047,
      "learning_rate": 1.0285714285714285e-05,
      "loss": 0.4507,
      "step": 504
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8096213797333377,
      "learning_rate": 1.0306122448979591e-05,
      "loss": 0.2933,
      "step": 505
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7558430897080136,
      "learning_rate": 1.0326530612244898e-05,
      "loss": 0.3581,
      "step": 506
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7696572469918393,
      "learning_rate": 1.0346938775510205e-05,
      "loss": 0.3546,
      "step": 507
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8058023551613069,
      "learning_rate": 1.0367346938775512e-05,
      "loss": 0.0881,
      "step": 508
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5635359806041575,
      "learning_rate": 1.0387755102040817e-05,
      "loss": 0.4639,
      "step": 509
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8363381161274385,
      "learning_rate": 1.0408163265306123e-05,
      "loss": 0.2667,
      "step": 510
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7582908516719031,
      "learning_rate": 1.042857142857143e-05,
      "loss": 0.6928,
      "step": 511
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.381947357443399,
      "learning_rate": 1.0448979591836737e-05,
      "loss": 0.4904,
      "step": 512
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6901114203690888,
      "learning_rate": 1.046938775510204e-05,
      "loss": 0.3619,
      "step": 513
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9385042218290874,
      "learning_rate": 1.0489795918367347e-05,
      "loss": 0.2997,
      "step": 514
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9657804974343484,
      "learning_rate": 1.0510204081632654e-05,
      "loss": 0.3621,
      "step": 515
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.690874555328071,
      "learning_rate": 1.053061224489796e-05,
      "loss": 0.7449,
      "step": 516
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.837269341999424,
      "learning_rate": 1.0551020408163265e-05,
      "loss": 0.1112,
      "step": 517
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.221796002402705,
      "learning_rate": 1.0571428571428572e-05,
      "loss": 0.335,
      "step": 518
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7413347265015331,
      "learning_rate": 1.0591836734693879e-05,
      "loss": 0.2739,
      "step": 519
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9207910023488637,
      "learning_rate": 1.0612244897959186e-05,
      "loss": 0.3749,
      "step": 520
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3878992304267919,
      "learning_rate": 1.0632653061224492e-05,
      "loss": 0.5046,
      "step": 521
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3509375004358515,
      "learning_rate": 1.0653061224489796e-05,
      "loss": 0.565,
      "step": 522
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4177886104570563,
      "learning_rate": 1.0673469387755102e-05,
      "loss": 0.0947,
      "step": 523
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9456717794850137,
      "learning_rate": 1.0693877551020409e-05,
      "loss": 0.3626,
      "step": 524
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7431683574525995,
      "learning_rate": 1.0714285714285714e-05,
      "loss": 0.2913,
      "step": 525
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.47957989210453744,
      "learning_rate": 1.073469387755102e-05,
      "loss": 0.0975,
      "step": 526
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.6618933469285837,
      "learning_rate": 1.0755102040816328e-05,
      "loss": 0.6741,
      "step": 527
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6940892961399557,
      "learning_rate": 1.0775510204081634e-05,
      "loss": 0.2295,
      "step": 528
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7650444959432459,
      "learning_rate": 1.0795918367346941e-05,
      "loss": 0.6473,
      "step": 529
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7126080044458347,
      "learning_rate": 1.0816326530612246e-05,
      "loss": 0.3717,
      "step": 530
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5612438285201367,
      "learning_rate": 1.0836734693877551e-05,
      "loss": 0.6499,
      "step": 531
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6022831904170782,
      "learning_rate": 1.0857142857142858e-05,
      "loss": 0.1938,
      "step": 532
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8449326863262158,
      "learning_rate": 1.0877551020408163e-05,
      "loss": 0.3758,
      "step": 533
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7999100720488339,
      "learning_rate": 1.089795918367347e-05,
      "loss": 0.0853,
      "step": 534
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8598781383784274,
      "learning_rate": 1.0918367346938776e-05,
      "loss": 0.2537,
      "step": 535
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.76164651845109,
      "learning_rate": 1.0938775510204083e-05,
      "loss": 0.2975,
      "step": 536
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7728178025933216,
      "learning_rate": 1.095918367346939e-05,
      "loss": 0.3041,
      "step": 537
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1013087777682644,
      "learning_rate": 1.0979591836734695e-05,
      "loss": 0.3731,
      "step": 538
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.817097510712764,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 0.5222,
      "step": 539
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3616762601063905,
      "learning_rate": 1.1020408163265306e-05,
      "loss": 0.6297,
      "step": 540
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0082364740097924,
      "learning_rate": 1.1040816326530611e-05,
      "loss": 0.2008,
      "step": 541
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7646677037424645,
      "learning_rate": 1.1061224489795918e-05,
      "loss": 0.3271,
      "step": 542
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9222702747862551,
      "learning_rate": 1.1081632653061225e-05,
      "loss": 0.3272,
      "step": 543
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0768555032198512,
      "learning_rate": 1.1102040816326532e-05,
      "loss": 0.3455,
      "step": 544
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1684125465265602,
      "learning_rate": 1.1122448979591838e-05,
      "loss": 0.322,
      "step": 545
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0403702733988704,
      "learning_rate": 1.1142857142857143e-05,
      "loss": 0.3452,
      "step": 546
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2250660924804364,
      "learning_rate": 1.116326530612245e-05,
      "loss": 0.4024,
      "step": 547
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.206887508640137,
      "learning_rate": 1.1183673469387757e-05,
      "loss": 0.4559,
      "step": 548
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8928256211760339,
      "learning_rate": 1.1204081632653062e-05,
      "loss": 0.2426,
      "step": 549
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8663722310678121,
      "learning_rate": 1.1224489795918367e-05,
      "loss": 0.3491,
      "step": 550
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7784096894647445,
      "learning_rate": 1.1244897959183674e-05,
      "loss": 0.2761,
      "step": 551
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.434624203764921,
      "learning_rate": 1.126530612244898e-05,
      "loss": 0.0675,
      "step": 552
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8619429210831706,
      "learning_rate": 1.1285714285714287e-05,
      "loss": 0.2906,
      "step": 553
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6956201395061768,
      "learning_rate": 1.1306122448979592e-05,
      "loss": 0.3026,
      "step": 554
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8369523372522975,
      "learning_rate": 1.1326530612244899e-05,
      "loss": 0.2775,
      "step": 555
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9032435034771735,
      "learning_rate": 1.1346938775510206e-05,
      "loss": 0.4159,
      "step": 556
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4913491262138383,
      "learning_rate": 1.1367346938775512e-05,
      "loss": 0.6384,
      "step": 557
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6226424000114711,
      "learning_rate": 1.1387755102040819e-05,
      "loss": 0.1436,
      "step": 558
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4700602921486106,
      "learning_rate": 1.1408163265306122e-05,
      "loss": 0.4445,
      "step": 559
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7161959834831972,
      "learning_rate": 1.1428571428571429e-05,
      "loss": 0.2731,
      "step": 560
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.733105970307493,
      "learning_rate": 1.1448979591836736e-05,
      "loss": 0.3583,
      "step": 561
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6071003139866922,
      "learning_rate": 1.146938775510204e-05,
      "loss": 0.0527,
      "step": 562
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6423343496971055,
      "learning_rate": 1.1489795918367347e-05,
      "loss": 0.642,
      "step": 563
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7654202632340549,
      "learning_rate": 1.1510204081632654e-05,
      "loss": 0.2852,
      "step": 564
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.856586308529009,
      "learning_rate": 1.1530612244897961e-05,
      "loss": 0.4491,
      "step": 565
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.842608721854325,
      "learning_rate": 1.1551020408163268e-05,
      "loss": 0.2951,
      "step": 566
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5980088067577748,
      "learning_rate": 1.1571428571428573e-05,
      "loss": 0.1708,
      "step": 567
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1972022645655782,
      "learning_rate": 1.1591836734693878e-05,
      "loss": 0.3383,
      "step": 568
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8030633746505018,
      "learning_rate": 1.1612244897959184e-05,
      "loss": 0.2825,
      "step": 569
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3316535183052802,
      "learning_rate": 1.1632653061224491e-05,
      "loss": 0.7835,
      "step": 570
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4668653366817679,
      "learning_rate": 1.1653061224489796e-05,
      "loss": 0.5529,
      "step": 571
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.715250680896333,
      "learning_rate": 1.1673469387755103e-05,
      "loss": 0.3823,
      "step": 572
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7467920079247107,
      "learning_rate": 1.169387755102041e-05,
      "loss": 0.199,
      "step": 573
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0096759494774032,
      "learning_rate": 1.1714285714285716e-05,
      "loss": 0.4363,
      "step": 574
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1783176678899872,
      "learning_rate": 1.1734693877551021e-05,
      "loss": 0.5482,
      "step": 575
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.879355314214938,
      "learning_rate": 1.1755102040816328e-05,
      "loss": 0.167,
      "step": 576
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8994797624564155,
      "learning_rate": 1.1775510204081633e-05,
      "loss": 0.0887,
      "step": 577
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.111743010468325,
      "learning_rate": 1.179591836734694e-05,
      "loss": 0.3493,
      "step": 578
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.692865046290311,
      "learning_rate": 1.1816326530612245e-05,
      "loss": 0.293,
      "step": 579
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3938108318953575,
      "learning_rate": 1.1836734693877552e-05,
      "loss": 0.5194,
      "step": 580
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2181849923441486,
      "learning_rate": 1.1857142857142858e-05,
      "loss": 0.5382,
      "step": 581
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9543724269318221,
      "learning_rate": 1.1877551020408165e-05,
      "loss": 0.2003,
      "step": 582
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0719134406193573,
      "learning_rate": 1.189795918367347e-05,
      "loss": 0.4089,
      "step": 583
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7270536957588328,
      "learning_rate": 1.1918367346938777e-05,
      "loss": 0.3218,
      "step": 584
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.845958302803179,
      "learning_rate": 1.1938775510204084e-05,
      "loss": 0.8293,
      "step": 585
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.888972899958462,
      "learning_rate": 1.1959183673469389e-05,
      "loss": 0.1256,
      "step": 586
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8303253922170286,
      "learning_rate": 1.1979591836734694e-05,
      "loss": 0.3186,
      "step": 587
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.309814338363089,
      "learning_rate": 1.2e-05,
      "loss": 0.527,
      "step": 588
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.69753072942486,
      "learning_rate": 1.2020408163265307e-05,
      "loss": 0.6028,
      "step": 589
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7050378650966671,
      "learning_rate": 1.2040816326530614e-05,
      "loss": 0.3598,
      "step": 590
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9223172209156314,
      "learning_rate": 1.206122448979592e-05,
      "loss": 0.3803,
      "step": 591
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9577932668169392,
      "learning_rate": 1.2081632653061225e-05,
      "loss": 0.3119,
      "step": 592
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.580865689458526,
      "learning_rate": 1.2102040816326532e-05,
      "loss": 0.5991,
      "step": 593
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0436123048958816,
      "learning_rate": 1.2122448979591839e-05,
      "loss": 0.3147,
      "step": 594
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.41633276964914795,
      "learning_rate": 1.2142857142857142e-05,
      "loss": 0.08,
      "step": 595
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7352840848191786,
      "learning_rate": 1.2163265306122449e-05,
      "loss": 0.3245,
      "step": 596
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8567484145041376,
      "learning_rate": 1.2183673469387756e-05,
      "loss": 0.2987,
      "step": 597
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.596813506743633,
      "learning_rate": 1.2204081632653062e-05,
      "loss": 0.6104,
      "step": 598
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.576457355378549,
      "learning_rate": 1.222448979591837e-05,
      "loss": 0.5721,
      "step": 599
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7869128928714801,
      "learning_rate": 1.2244897959183674e-05,
      "loss": 0.3022,
      "step": 600
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8987861325347973,
      "learning_rate": 1.2265306122448981e-05,
      "loss": 0.3045,
      "step": 601
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9082381269622332,
      "learning_rate": 1.2285714285714288e-05,
      "loss": 0.4422,
      "step": 602
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.59767699578339,
      "learning_rate": 1.2306122448979594e-05,
      "loss": 0.2476,
      "step": 603
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7034945170066397,
      "learning_rate": 1.2326530612244898e-05,
      "loss": 0.2638,
      "step": 604
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8424512797672575,
      "learning_rate": 1.2346938775510204e-05,
      "loss": 0.2449,
      "step": 605
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1713053322481788,
      "learning_rate": 1.2367346938775511e-05,
      "loss": 0.4232,
      "step": 606
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9828242022163833,
      "learning_rate": 1.2387755102040818e-05,
      "loss": 0.4555,
      "step": 607
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8884156256637892,
      "learning_rate": 1.2408163265306123e-05,
      "loss": 0.3301,
      "step": 608
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6075487419152823,
      "learning_rate": 1.242857142857143e-05,
      "loss": 0.4472,
      "step": 609
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.631929093518685,
      "learning_rate": 1.2448979591836736e-05,
      "loss": 0.2609,
      "step": 610
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0455608950316746,
      "learning_rate": 1.2469387755102043e-05,
      "loss": 0.7362,
      "step": 611
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4324895010113472,
      "learning_rate": 1.248979591836735e-05,
      "loss": 0.4838,
      "step": 612
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5500853361460135,
      "learning_rate": 1.2510204081632653e-05,
      "loss": 0.2356,
      "step": 613
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6989940147185083,
      "learning_rate": 1.253061224489796e-05,
      "loss": 0.2401,
      "step": 614
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.818399802416431,
      "learning_rate": 1.2551020408163267e-05,
      "loss": 0.4451,
      "step": 615
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0831957606624758,
      "learning_rate": 1.2571428571428572e-05,
      "loss": 0.5172,
      "step": 616
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5333095905638339,
      "learning_rate": 1.2591836734693878e-05,
      "loss": 0.6124,
      "step": 617
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6778478940444286,
      "learning_rate": 1.2612244897959185e-05,
      "loss": 0.193,
      "step": 618
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0925629060400408,
      "learning_rate": 1.2632653061224492e-05,
      "loss": 0.3086,
      "step": 619
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9850691765720584,
      "learning_rate": 1.2653061224489798e-05,
      "loss": 0.434,
      "step": 620
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9031676939584062,
      "learning_rate": 1.2673469387755104e-05,
      "loss": 0.3703,
      "step": 621
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8266968205339826,
      "learning_rate": 1.2693877551020409e-05,
      "loss": 0.3109,
      "step": 622
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7854942430289334,
      "learning_rate": 1.2714285714285715e-05,
      "loss": 0.2335,
      "step": 623
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4820976601274658,
      "learning_rate": 1.273469387755102e-05,
      "loss": 0.5875,
      "step": 624
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6031845953648253,
      "learning_rate": 1.2755102040816327e-05,
      "loss": 0.2911,
      "step": 625
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8344951291417452,
      "learning_rate": 1.2775510204081634e-05,
      "loss": 0.4204,
      "step": 626
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.729387171527938,
      "learning_rate": 1.279591836734694e-05,
      "loss": 0.2145,
      "step": 627
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1323766742985937,
      "learning_rate": 1.2816326530612247e-05,
      "loss": 0.3927,
      "step": 628
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1024657454685545,
      "learning_rate": 1.2836734693877552e-05,
      "loss": 0.6437,
      "step": 629
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.975507501185196,
      "learning_rate": 1.2857142857142859e-05,
      "loss": 0.7572,
      "step": 630
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.726425931867654,
      "learning_rate": 1.2877551020408164e-05,
      "loss": 0.3091,
      "step": 631
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9291410214516044,
      "learning_rate": 1.2897959183673469e-05,
      "loss": 0.2278,
      "step": 632
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9264858626129938,
      "learning_rate": 1.2918367346938776e-05,
      "loss": 0.4725,
      "step": 633
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6374197678150193,
      "learning_rate": 1.2938775510204082e-05,
      "loss": 0.1787,
      "step": 634
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6558408006021987,
      "learning_rate": 1.2959183673469389e-05,
      "loss": 0.7794,
      "step": 635
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9402174823446985,
      "learning_rate": 1.2979591836734696e-05,
      "loss": 0.2634,
      "step": 636
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9265800140716998,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 0.3049,
      "step": 637
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7502440013305205,
      "learning_rate": 1.3020408163265308e-05,
      "loss": 0.3318,
      "step": 638
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4847867376441406,
      "learning_rate": 1.3040816326530614e-05,
      "loss": 0.7814,
      "step": 639
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3733064557938348,
      "learning_rate": 1.3061224489795918e-05,
      "loss": 0.5272,
      "step": 640
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9031524391585722,
      "learning_rate": 1.3081632653061224e-05,
      "loss": 0.4135,
      "step": 641
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7327520302776948,
      "learning_rate": 1.3102040816326531e-05,
      "loss": 0.3367,
      "step": 642
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6980511649913749,
      "learning_rate": 1.3122448979591838e-05,
      "loss": 0.1893,
      "step": 643
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7245389043648406,
      "learning_rate": 1.3142857142857145e-05,
      "loss": 0.3072,
      "step": 644
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6204351238113122,
      "learning_rate": 1.316326530612245e-05,
      "loss": 0.1841,
      "step": 645
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8420780239691985,
      "learning_rate": 1.3183673469387756e-05,
      "loss": 0.3576,
      "step": 646
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7175767306993932,
      "learning_rate": 1.3204081632653063e-05,
      "loss": 0.4014,
      "step": 647
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7125622886682474,
      "learning_rate": 1.322448979591837e-05,
      "loss": 0.6399,
      "step": 648
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6064339957862633,
      "learning_rate": 1.3244897959183673e-05,
      "loss": 0.357,
      "step": 649
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7748752094300059,
      "learning_rate": 1.326530612244898e-05,
      "loss": 0.3265,
      "step": 650
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9625351522807984,
      "learning_rate": 1.3285714285714287e-05,
      "loss": 0.3481,
      "step": 651
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.0467899928082174,
      "learning_rate": 1.3306122448979593e-05,
      "loss": 0.243,
      "step": 652
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.657005549990083,
      "learning_rate": 1.3326530612244898e-05,
      "loss": 0.207,
      "step": 653
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4443296298659127,
      "learning_rate": 1.3346938775510205e-05,
      "loss": 0.4299,
      "step": 654
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7203525519997779,
      "learning_rate": 1.3367346938775512e-05,
      "loss": 0.2262,
      "step": 655
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5863897705357113,
      "learning_rate": 1.3387755102040818e-05,
      "loss": 0.3114,
      "step": 656
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2137119706446255,
      "learning_rate": 1.3408163265306125e-05,
      "loss": 0.4905,
      "step": 657
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8826813783499218,
      "learning_rate": 1.3428571428571429e-05,
      "loss": 0.3802,
      "step": 658
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5894523434909502,
      "learning_rate": 1.3448979591836735e-05,
      "loss": 0.4004,
      "step": 659
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7176531643513427,
      "learning_rate": 1.3469387755102042e-05,
      "loss": 0.2948,
      "step": 660
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6370431234878117,
      "learning_rate": 1.3489795918367347e-05,
      "loss": 0.2855,
      "step": 661
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5832392036436793,
      "learning_rate": 1.3510204081632654e-05,
      "loss": 0.2348,
      "step": 662
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.678625064832223,
      "learning_rate": 1.353061224489796e-05,
      "loss": 0.8945,
      "step": 663
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6396504072186845,
      "learning_rate": 1.3551020408163267e-05,
      "loss": 0.2362,
      "step": 664
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.151468644412274,
      "learning_rate": 1.3571428571428574e-05,
      "loss": 0.7715,
      "step": 665
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.776945744774685,
      "learning_rate": 1.3591836734693879e-05,
      "loss": 0.5475,
      "step": 666
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8116536971386127,
      "learning_rate": 1.3612244897959184e-05,
      "loss": 0.3331,
      "step": 667
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7643897829821308,
      "learning_rate": 1.363265306122449e-05,
      "loss": 0.2707,
      "step": 668
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8267328422980034,
      "learning_rate": 1.3653061224489796e-05,
      "loss": 0.3496,
      "step": 669
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7893438561235766,
      "learning_rate": 1.3673469387755102e-05,
      "loss": 0.4401,
      "step": 670
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5117925619875184,
      "learning_rate": 1.3693877551020409e-05,
      "loss": 0.1492,
      "step": 671
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8465228794021626,
      "learning_rate": 1.3714285714285716e-05,
      "loss": 0.3852,
      "step": 672
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6330812566935348,
      "learning_rate": 1.3734693877551023e-05,
      "loss": 0.2269,
      "step": 673
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8965110226861079,
      "learning_rate": 1.3755102040816328e-05,
      "loss": 0.3888,
      "step": 674
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0694424731592551,
      "learning_rate": 1.3775510204081634e-05,
      "loss": 0.359,
      "step": 675
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6361634753975307,
      "learning_rate": 1.3795918367346941e-05,
      "loss": 0.6581,
      "step": 676
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7383324905949138,
      "learning_rate": 1.3816326530612244e-05,
      "loss": 0.2463,
      "step": 677
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9877911914782169,
      "learning_rate": 1.3836734693877551e-05,
      "loss": 0.3851,
      "step": 678
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8451017468319793,
      "learning_rate": 1.3857142857142858e-05,
      "loss": 0.3271,
      "step": 679
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5836317253433491,
      "learning_rate": 1.3877551020408165e-05,
      "loss": 0.2639,
      "step": 680
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7352905835419441,
      "learning_rate": 1.3897959183673471e-05,
      "loss": 0.3167,
      "step": 681
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6678481359894013,
      "learning_rate": 1.3918367346938776e-05,
      "loss": 0.4685,
      "step": 682
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.79795158256387,
      "learning_rate": 1.3938775510204083e-05,
      "loss": 0.3624,
      "step": 683
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7750642683219922,
      "learning_rate": 1.395918367346939e-05,
      "loss": 0.3161,
      "step": 684
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6824094842377565,
      "learning_rate": 1.3979591836734696e-05,
      "loss": 0.3598,
      "step": 685
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0774955004330544,
      "learning_rate": 1.4e-05,
      "loss": 0.0944,
      "step": 686
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7144257726463071,
      "learning_rate": 1.4020408163265307e-05,
      "loss": 0.3285,
      "step": 687
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9885393783728982,
      "learning_rate": 1.4040816326530613e-05,
      "loss": 0.586,
      "step": 688
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3082365093998183,
      "learning_rate": 1.406122448979592e-05,
      "loss": 0.8838,
      "step": 689
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8081373138849134,
      "learning_rate": 1.4081632653061225e-05,
      "loss": 0.3102,
      "step": 690
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.847544019023508,
      "learning_rate": 1.4102040816326532e-05,
      "loss": 0.2794,
      "step": 691
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7703591377374095,
      "learning_rate": 1.4122448979591838e-05,
      "loss": 0.3407,
      "step": 692
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6584464406311035,
      "learning_rate": 1.4142857142857145e-05,
      "loss": 0.3698,
      "step": 693
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6108992069440955,
      "learning_rate": 1.4163265306122452e-05,
      "loss": 0.1521,
      "step": 694
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1968929066677747,
      "learning_rate": 1.4183673469387755e-05,
      "loss": 0.5098,
      "step": 695
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8058056837079421,
      "learning_rate": 1.4204081632653062e-05,
      "loss": 0.2749,
      "step": 696
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.198676257549707,
      "learning_rate": 1.4224489795918369e-05,
      "loss": 0.2949,
      "step": 697
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.441803657564317,
      "learning_rate": 1.4244897959183674e-05,
      "loss": 0.7757,
      "step": 698
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9455550026788422,
      "learning_rate": 1.426530612244898e-05,
      "loss": 0.5221,
      "step": 699
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8737843898414316,
      "learning_rate": 1.4285714285714287e-05,
      "loss": 0.3351,
      "step": 700
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8719234296243357,
      "learning_rate": 1.4306122448979594e-05,
      "loss": 0.336,
      "step": 701
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7834254943019043,
      "learning_rate": 1.43265306122449e-05,
      "loss": 0.3241,
      "step": 702
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6403735312479718,
      "learning_rate": 1.4346938775510206e-05,
      "loss": 0.2678,
      "step": 703
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4502511291311904,
      "learning_rate": 1.436734693877551e-05,
      "loss": 0.5229,
      "step": 704
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7151565987693118,
      "learning_rate": 1.4387755102040817e-05,
      "loss": 0.2298,
      "step": 705
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1268833729777474,
      "learning_rate": 1.4408163265306122e-05,
      "loss": 0.4159,
      "step": 706
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2601320788053645,
      "learning_rate": 1.4428571428571429e-05,
      "loss": 0.5345,
      "step": 707
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6359765785720934,
      "learning_rate": 1.4448979591836736e-05,
      "loss": 0.3487,
      "step": 708
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8255849713946286,
      "learning_rate": 1.4469387755102043e-05,
      "loss": 0.0862,
      "step": 709
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6956036228372331,
      "learning_rate": 1.448979591836735e-05,
      "loss": 0.3478,
      "step": 710
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1410699902067343,
      "learning_rate": 1.4510204081632654e-05,
      "loss": 0.7799,
      "step": 711
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9132478094017972,
      "learning_rate": 1.4530612244897961e-05,
      "loss": 0.122,
      "step": 712
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0274144241797285,
      "learning_rate": 1.4551020408163266e-05,
      "loss": 0.7717,
      "step": 713
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6736485704274043,
      "learning_rate": 1.4571428571428573e-05,
      "loss": 0.2286,
      "step": 714
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7131977057591253,
      "learning_rate": 1.4591836734693878e-05,
      "loss": 0.3526,
      "step": 715
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.012783827680085,
      "learning_rate": 1.4612244897959185e-05,
      "loss": 0.5429,
      "step": 716
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.5405033808717155,
      "learning_rate": 1.4632653061224491e-05,
      "loss": 0.6352,
      "step": 717
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8989091162858869,
      "learning_rate": 1.4653061224489798e-05,
      "loss": 0.3277,
      "step": 718
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8260169544722441,
      "learning_rate": 1.4673469387755103e-05,
      "loss": 0.3447,
      "step": 719
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8749162666511164,
      "learning_rate": 1.469387755102041e-05,
      "loss": 0.345,
      "step": 720
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.988090504284863,
      "learning_rate": 1.4714285714285716e-05,
      "loss": 0.4095,
      "step": 721
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7682694467941851,
      "learning_rate": 1.4734693877551021e-05,
      "loss": 0.1222,
      "step": 722
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7001624130681571,
      "learning_rate": 1.4755102040816326e-05,
      "loss": 0.2742,
      "step": 723
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.1251913721555664,
      "learning_rate": 1.4775510204081633e-05,
      "loss": 0.6965,
      "step": 724
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4578562036627014,
      "learning_rate": 1.479591836734694e-05,
      "loss": 0.5767,
      "step": 725
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9043688881950668,
      "learning_rate": 1.4816326530612247e-05,
      "loss": 0.4584,
      "step": 726
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6083041015855989,
      "learning_rate": 1.4836734693877552e-05,
      "loss": 0.2373,
      "step": 727
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9226321238382381,
      "learning_rate": 1.4857142857142858e-05,
      "loss": 0.4022,
      "step": 728
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4686617524375243,
      "learning_rate": 1.4877551020408165e-05,
      "loss": 0.6385,
      "step": 729
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6913828811616523,
      "learning_rate": 1.4897959183673472e-05,
      "loss": 0.1729,
      "step": 730
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9433044306577101,
      "learning_rate": 1.4918367346938775e-05,
      "loss": 0.0998,
      "step": 731
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8872294681627249,
      "learning_rate": 1.4938775510204082e-05,
      "loss": 0.3536,
      "step": 732
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6150343312320375,
      "learning_rate": 1.4959183673469389e-05,
      "loss": 0.2957,
      "step": 733
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4268397428061865,
      "learning_rate": 1.4979591836734695e-05,
      "loss": 0.6032,
      "step": 734
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2423976433706543,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.5718,
      "step": 735
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0277763599182075,
      "learning_rate": 1.5020408163265307e-05,
      "loss": 0.0899,
      "step": 736
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7793252065624419,
      "learning_rate": 1.5040816326530614e-05,
      "loss": 0.3836,
      "step": 737
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8902482105692707,
      "learning_rate": 1.506122448979592e-05,
      "loss": 0.3681,
      "step": 738
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0245785479520604,
      "learning_rate": 1.5081632653061227e-05,
      "loss": 0.3145,
      "step": 739
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5629929067263313,
      "learning_rate": 1.510204081632653e-05,
      "loss": 0.079,
      "step": 740
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7484256807258068,
      "learning_rate": 1.5122448979591837e-05,
      "loss": 0.2777,
      "step": 741
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0212050182359453,
      "learning_rate": 1.5142857142857144e-05,
      "loss": 0.4029,
      "step": 742
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.6930497071321737,
      "learning_rate": 1.516326530612245e-05,
      "loss": 0.5897,
      "step": 743
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8746673944241986,
      "learning_rate": 1.5183673469387756e-05,
      "loss": 0.3256,
      "step": 744
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8213926676990672,
      "learning_rate": 1.5204081632653063e-05,
      "loss": 0.3827,
      "step": 745
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6701806884084368,
      "learning_rate": 1.522448979591837e-05,
      "loss": 0.2571,
      "step": 746
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.2684126733448817,
      "learning_rate": 1.5244897959183676e-05,
      "loss": 0.7432,
      "step": 747
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8512649974488006,
      "learning_rate": 1.526530612244898e-05,
      "loss": 0.0911,
      "step": 748
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6282245981120872,
      "learning_rate": 1.5285714285714286e-05,
      "loss": 0.2601,
      "step": 749
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7416549580316524,
      "learning_rate": 1.530612244897959e-05,
      "loss": 0.2661,
      "step": 750
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6768323627778384,
      "learning_rate": 1.53265306122449e-05,
      "loss": 0.3231,
      "step": 751
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2938447418127466,
      "learning_rate": 1.5346938775510204e-05,
      "loss": 0.5859,
      "step": 752
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1134599413612212,
      "learning_rate": 1.5367346938775513e-05,
      "loss": 0.5285,
      "step": 753
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2093161544737387,
      "learning_rate": 1.5387755102040818e-05,
      "loss": 0.3725,
      "step": 754
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6941830377868214,
      "learning_rate": 1.5408163265306123e-05,
      "loss": 0.3015,
      "step": 755
    },
    {
      "epoch": 0.02,
      "grad_norm": 3.0932579606197517,
      "learning_rate": 1.542857142857143e-05,
      "loss": 0.6245,
      "step": 756
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.834511210412421,
      "learning_rate": 1.5448979591836736e-05,
      "loss": 0.3418,
      "step": 757
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7679095731998528,
      "learning_rate": 1.546938775510204e-05,
      "loss": 0.2275,
      "step": 758
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6816420177386995,
      "learning_rate": 1.5489795918367346e-05,
      "loss": 0.1965,
      "step": 759
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8952275791001809,
      "learning_rate": 1.5510204081632655e-05,
      "loss": 0.4218,
      "step": 760
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2103296666180023,
      "learning_rate": 1.553061224489796e-05,
      "loss": 0.5394,
      "step": 761
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7572005365791613,
      "learning_rate": 1.5551020408163265e-05,
      "loss": 0.3594,
      "step": 762
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.7220364827308026,
      "learning_rate": 1.5571428571428573e-05,
      "loss": 0.5487,
      "step": 763
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6439548664155308,
      "learning_rate": 1.559183673469388e-05,
      "loss": 0.239,
      "step": 764
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.543732893411752,
      "learning_rate": 1.5612244897959187e-05,
      "loss": 0.7657,
      "step": 765
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7283366389719368,
      "learning_rate": 1.5632653061224492e-05,
      "loss": 0.1508,
      "step": 766
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.4911004442679876,
      "learning_rate": 1.5653061224489797e-05,
      "loss": 0.621,
      "step": 767
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.598697182399304,
      "learning_rate": 1.5673469387755102e-05,
      "loss": 0.2828,
      "step": 768
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5765334987449886,
      "learning_rate": 1.569387755102041e-05,
      "loss": 0.3195,
      "step": 769
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.774982962904502,
      "learning_rate": 1.5714285714285715e-05,
      "loss": 0.3027,
      "step": 770
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.3749077988862544,
      "learning_rate": 1.573469387755102e-05,
      "loss": 0.7637,
      "step": 771
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.091903118276975,
      "learning_rate": 1.575510204081633e-05,
      "loss": 0.3414,
      "step": 772
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6943519145557014,
      "learning_rate": 1.5775510204081634e-05,
      "loss": 0.3118,
      "step": 773
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6505188989242852,
      "learning_rate": 1.5795918367346942e-05,
      "loss": 0.3392,
      "step": 774
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.9849115512534612,
      "learning_rate": 1.5816326530612247e-05,
      "loss": 0.7935,
      "step": 775
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.1036072970873652,
      "learning_rate": 1.5836734693877552e-05,
      "loss": 0.4821,
      "step": 776
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5923551500518347,
      "learning_rate": 1.5857142857142857e-05,
      "loss": 0.2658,
      "step": 777
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.554545779308851,
      "learning_rate": 1.5877551020408162e-05,
      "loss": 0.289,
      "step": 778
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5094764328984737,
      "learning_rate": 1.589795918367347e-05,
      "loss": 0.1493,
      "step": 779
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6741023475954389,
      "learning_rate": 1.5918367346938776e-05,
      "loss": 0.3502,
      "step": 780
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8780555616423122,
      "learning_rate": 1.5938775510204084e-05,
      "loss": 0.3397,
      "step": 781
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6826331841515063,
      "learning_rate": 1.595918367346939e-05,
      "loss": 0.2961,
      "step": 782
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.040855967080385,
      "learning_rate": 1.5979591836734694e-05,
      "loss": 0.4382,
      "step": 783
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3965024819046539,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.6629,
      "step": 784
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7976516017074206,
      "learning_rate": 1.6020408163265308e-05,
      "loss": 0.4086,
      "step": 785
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8038520285966642,
      "learning_rate": 1.6040816326530613e-05,
      "loss": 0.303,
      "step": 786
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6915801936791116,
      "learning_rate": 1.6061224489795918e-05,
      "loss": 0.284,
      "step": 787
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2988834472035997,
      "learning_rate": 1.6081632653061226e-05,
      "loss": 0.2226,
      "step": 788
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.536033668265738,
      "learning_rate": 1.610204081632653e-05,
      "loss": 0.1415,
      "step": 789
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.4068953361599585,
      "learning_rate": 1.612244897959184e-05,
      "loss": 0.6108,
      "step": 790
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.772877764715428,
      "learning_rate": 1.6142857142857145e-05,
      "loss": 0.2799,
      "step": 791
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7422176377895493,
      "learning_rate": 1.616326530612245e-05,
      "loss": 0.3045,
      "step": 792
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.2593235956632627,
      "learning_rate": 1.6183673469387758e-05,
      "loss": 0.6425,
      "step": 793
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.195559338084164,
      "learning_rate": 1.6204081632653063e-05,
      "loss": 0.5008,
      "step": 794
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.0993146438956438,
      "learning_rate": 1.6224489795918368e-05,
      "loss": 0.5174,
      "step": 795
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6265316809135901,
      "learning_rate": 1.6244897959183673e-05,
      "loss": 0.2595,
      "step": 796
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8319122569246719,
      "learning_rate": 1.626530612244898e-05,
      "loss": 0.179,
      "step": 797
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7036275322065548,
      "learning_rate": 1.6285714285714287e-05,
      "loss": 0.3238,
      "step": 798
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.6583236317128227,
      "learning_rate": 1.630612244897959e-05,
      "loss": 0.8183,
      "step": 799
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5911093697777315,
      "learning_rate": 1.63265306122449e-05,
      "loss": 0.2323,
      "step": 800
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8307790209909187,
      "learning_rate": 1.6346938775510205e-05,
      "loss": 0.7339,
      "step": 801
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3217508507670699,
      "learning_rate": 1.6367346938775513e-05,
      "loss": 0.5522,
      "step": 802
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.8005980064423034,
      "learning_rate": 1.638775510204082e-05,
      "loss": 0.4176,
      "step": 803
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7730123472774382,
      "learning_rate": 1.6408163265306124e-05,
      "loss": 0.31,
      "step": 804
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7216566561708412,
      "learning_rate": 1.642857142857143e-05,
      "loss": 0.3242,
      "step": 805
    },
    {
      "epoch": 0.02,
      "grad_norm": 2.834045601157214,
      "learning_rate": 1.6448979591836737e-05,
      "loss": 0.7749,
      "step": 806
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6226649709624577,
      "learning_rate": 1.6469387755102042e-05,
      "loss": 0.1262,
      "step": 807
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.8025438769078652,
      "learning_rate": 1.6489795918367347e-05,
      "loss": 0.8089,
      "step": 808
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.5932818027443855,
      "learning_rate": 1.6510204081632655e-05,
      "loss": 0.2403,
      "step": 809
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3292695144818907,
      "learning_rate": 1.653061224489796e-05,
      "loss": 0.5841,
      "step": 810
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7220175573622806,
      "learning_rate": 1.655102040816327e-05,
      "loss": 0.3899,
      "step": 811
    },
    {
      "epoch": 0.02,
      "grad_norm": 1.3407510248632117,
      "learning_rate": 1.6571428571428574e-05,
      "loss": 0.6782,
      "step": 812
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6704335281663879,
      "learning_rate": 1.659183673469388e-05,
      "loss": 0.0756,
      "step": 813
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.9175489686259981,
      "learning_rate": 1.6612244897959184e-05,
      "loss": 0.3695,
      "step": 814
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.7216226015155742,
      "learning_rate": 1.6632653061224492e-05,
      "loss": 0.2248,
      "step": 815
    },
    {
      "epoch": 0.02,
      "grad_norm": 0.6880050731596358,
      "learning_rate": 1.6653061224489797e-05,
      "loss": 0.3231,
      "step": 816
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.808776060870983,
      "learning_rate": 1.6673469387755102e-05,
      "loss": 0.7428,
      "step": 817
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6925293367192004,
      "learning_rate": 1.669387755102041e-05,
      "loss": 0.2022,
      "step": 818
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6651850603356546,
      "learning_rate": 1.6714285714285716e-05,
      "loss": 0.3366,
      "step": 819
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.27822868261775,
      "learning_rate": 1.673469387755102e-05,
      "loss": 0.5303,
      "step": 820
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.960043450808682,
      "learning_rate": 1.675510204081633e-05,
      "loss": 0.3582,
      "step": 821
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9739066156375361,
      "learning_rate": 1.6775510204081634e-05,
      "loss": 0.292,
      "step": 822
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7015656674453727,
      "learning_rate": 1.679591836734694e-05,
      "loss": 0.3128,
      "step": 823
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5316193648219091,
      "learning_rate": 1.6816326530612244e-05,
      "loss": 0.1454,
      "step": 824
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.883131698460159,
      "learning_rate": 1.6836734693877553e-05,
      "loss": 0.7822,
      "step": 825
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4310711565822813,
      "learning_rate": 1.6857142857142858e-05,
      "loss": 0.5716,
      "step": 826
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8653544969992953,
      "learning_rate": 1.6877551020408166e-05,
      "loss": 0.4071,
      "step": 827
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.759507913730268,
      "learning_rate": 1.689795918367347e-05,
      "loss": 0.2908,
      "step": 828
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3551916563235458,
      "learning_rate": 1.6918367346938776e-05,
      "loss": 0.4854,
      "step": 829
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7843280750377268,
      "learning_rate": 1.6938775510204085e-05,
      "loss": 0.3193,
      "step": 830
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7278417924456195,
      "learning_rate": 1.695918367346939e-05,
      "loss": 0.1027,
      "step": 831
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7710982807860293,
      "learning_rate": 1.6979591836734695e-05,
      "loss": 0.3575,
      "step": 832
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9638423124119995,
      "learning_rate": 1.7e-05,
      "loss": 0.4492,
      "step": 833
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6989204098229158,
      "learning_rate": 1.7020408163265308e-05,
      "loss": 0.3501,
      "step": 834
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4744799939070092,
      "learning_rate": 1.7040816326530613e-05,
      "loss": 0.5506,
      "step": 835
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6907830896510864,
      "learning_rate": 1.7061224489795922e-05,
      "loss": 0.2859,
      "step": 836
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0315070808612374,
      "learning_rate": 1.7081632653061227e-05,
      "loss": 0.3515,
      "step": 837
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6815012767127565,
      "learning_rate": 1.7102040816326532e-05,
      "loss": 0.3227,
      "step": 838
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5083962473642297,
      "learning_rate": 1.712244897959184e-05,
      "loss": 0.2313,
      "step": 839
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3502536303217496,
      "learning_rate": 1.7142857142857142e-05,
      "loss": 0.3803,
      "step": 840
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7081358307444915,
      "learning_rate": 1.716326530612245e-05,
      "loss": 0.2321,
      "step": 841
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.181743219455068,
      "learning_rate": 1.7183673469387755e-05,
      "loss": 0.3576,
      "step": 842
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.096706000260005,
      "learning_rate": 1.7204081632653064e-05,
      "loss": 0.8174,
      "step": 843
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3255923897401527,
      "learning_rate": 1.722448979591837e-05,
      "loss": 0.5259,
      "step": 844
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8030962382318613,
      "learning_rate": 1.7244897959183674e-05,
      "loss": 0.3437,
      "step": 845
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6509730544235569,
      "learning_rate": 1.7265306122448982e-05,
      "loss": 0.2822,
      "step": 846
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7598838717826685,
      "learning_rate": 1.7285714285714287e-05,
      "loss": 0.3165,
      "step": 847
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5823625853576421,
      "learning_rate": 1.7306122448979596e-05,
      "loss": 0.0855,
      "step": 848
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4502623667628676,
      "learning_rate": 1.7326530612244897e-05,
      "loss": 0.6815,
      "step": 849
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6632881112560293,
      "learning_rate": 1.7346938775510206e-05,
      "loss": 0.2352,
      "step": 850
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.903934963149226,
      "learning_rate": 1.736734693877551e-05,
      "loss": 0.4119,
      "step": 851
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8311524208902449,
      "learning_rate": 1.738775510204082e-05,
      "loss": 0.3123,
      "step": 852
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4140577704765898,
      "learning_rate": 1.7408163265306124e-05,
      "loss": 0.5923,
      "step": 853
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.908271924164661,
      "learning_rate": 1.742857142857143e-05,
      "loss": 0.3433,
      "step": 854
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7813241542812616,
      "learning_rate": 1.7448979591836738e-05,
      "loss": 0.3065,
      "step": 855
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.7229398379223597,
      "learning_rate": 1.7469387755102043e-05,
      "loss": 0.5795,
      "step": 856
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6202190263652873,
      "learning_rate": 1.748979591836735e-05,
      "loss": 0.2057,
      "step": 857
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8408041170854321,
      "learning_rate": 1.7510204081632653e-05,
      "loss": 0.5997,
      "step": 858
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5751949186729747,
      "learning_rate": 1.753061224489796e-05,
      "loss": 0.2326,
      "step": 859
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4989728034223506,
      "learning_rate": 1.7551020408163266e-05,
      "loss": 0.8725,
      "step": 860
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3770005673756782,
      "learning_rate": 1.757142857142857e-05,
      "loss": 0.5481,
      "step": 861
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2786216787194369,
      "learning_rate": 1.759183673469388e-05,
      "loss": 0.594,
      "step": 862
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5496286864292906,
      "learning_rate": 1.7612244897959185e-05,
      "loss": 0.2515,
      "step": 863
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8922053300627798,
      "learning_rate": 1.7632653061224493e-05,
      "loss": 0.3598,
      "step": 864
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7440051574086852,
      "learning_rate": 1.7653061224489798e-05,
      "loss": 0.3306,
      "step": 865
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.204721314491403,
      "learning_rate": 1.7673469387755103e-05,
      "loss": 0.2045,
      "step": 866
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2133513584330793,
      "learning_rate": 1.7693877551020408e-05,
      "loss": 0.3158,
      "step": 867
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.829966915053445,
      "learning_rate": 1.7714285714285717e-05,
      "loss": 0.2315,
      "step": 868
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.166565119714848,
      "learning_rate": 1.773469387755102e-05,
      "loss": 0.3888,
      "step": 869
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.671046259258887,
      "learning_rate": 1.7755102040816327e-05,
      "loss": 0.3767,
      "step": 870
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1245265300846536,
      "learning_rate": 1.7775510204081635e-05,
      "loss": 0.6509,
      "step": 871
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8484854649894977,
      "learning_rate": 1.779591836734694e-05,
      "loss": 0.177,
      "step": 872
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8177862988419791,
      "learning_rate": 1.781632653061225e-05,
      "loss": 0.3196,
      "step": 873
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6469208998971793,
      "learning_rate": 1.7836734693877553e-05,
      "loss": 0.1701,
      "step": 874
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.69234028891371,
      "learning_rate": 1.785714285714286e-05,
      "loss": 0.3811,
      "step": 875
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6719063695059235,
      "learning_rate": 1.7877551020408164e-05,
      "loss": 0.1939,
      "step": 876
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6937294634027232,
      "learning_rate": 1.789795918367347e-05,
      "loss": 0.3112,
      "step": 877
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.709887303311289,
      "learning_rate": 1.7918367346938777e-05,
      "loss": 0.3568,
      "step": 878
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.180101113567959,
      "learning_rate": 1.7938775510204082e-05,
      "loss": 0.6326,
      "step": 879
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0004140923285,
      "learning_rate": 1.795918367346939e-05,
      "loss": 0.5304,
      "step": 880
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6309213601252178,
      "learning_rate": 1.7979591836734695e-05,
      "loss": 0.2467,
      "step": 881
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7304348269720704,
      "learning_rate": 1.8e-05,
      "loss": 0.3605,
      "step": 882
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.9982908909432027,
      "learning_rate": 1.802040816326531e-05,
      "loss": 0.5025,
      "step": 883
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.752017447998416,
      "learning_rate": 1.8040816326530614e-05,
      "loss": 0.2249,
      "step": 884
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7286945773426504,
      "learning_rate": 1.806122448979592e-05,
      "loss": 0.2759,
      "step": 885
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.655235827601095,
      "learning_rate": 1.8081632653061224e-05,
      "loss": 0.3033,
      "step": 886
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8463634724545689,
      "learning_rate": 1.8102040816326532e-05,
      "loss": 0.3513,
      "step": 887
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6226640455001744,
      "learning_rate": 1.8122448979591837e-05,
      "loss": 0.3626,
      "step": 888
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4773797995246631,
      "learning_rate": 1.8142857142857146e-05,
      "loss": 0.5559,
      "step": 889
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.164783671499591,
      "learning_rate": 1.816326530612245e-05,
      "loss": 0.4915,
      "step": 890
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.638777246868785,
      "learning_rate": 1.8183673469387756e-05,
      "loss": 0.2251,
      "step": 891
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8387061831551179,
      "learning_rate": 1.8204081632653064e-05,
      "loss": 0.3923,
      "step": 892
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.44814233918602764,
      "learning_rate": 1.822448979591837e-05,
      "loss": 0.2543,
      "step": 893
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.39483246953980145,
      "learning_rate": 1.8244897959183674e-05,
      "loss": 0.0789,
      "step": 894
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3048253951280406,
      "learning_rate": 1.826530612244898e-05,
      "loss": 0.5113,
      "step": 895
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7310033416080838,
      "learning_rate": 1.8285714285714288e-05,
      "loss": 0.3642,
      "step": 896
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2672854684036172,
      "learning_rate": 1.8306122448979593e-05,
      "loss": 0.5978,
      "step": 897
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6205723932185567,
      "learning_rate": 1.8326530612244898e-05,
      "loss": 0.332,
      "step": 898
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2484483571751666,
      "learning_rate": 1.8346938775510206e-05,
      "loss": 0.8285,
      "step": 899
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.568592510075185,
      "learning_rate": 1.836734693877551e-05,
      "loss": 0.2319,
      "step": 900
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3703407629936613,
      "learning_rate": 1.838775510204082e-05,
      "loss": 0.8031,
      "step": 901
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.710043474757912,
      "learning_rate": 1.8408163265306125e-05,
      "loss": 0.1917,
      "step": 902
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7079953813881997,
      "learning_rate": 1.842857142857143e-05,
      "loss": 0.3352,
      "step": 903
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8331048294081974,
      "learning_rate": 1.8448979591836735e-05,
      "loss": 0.3661,
      "step": 904
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6343135301108352,
      "learning_rate": 1.8469387755102043e-05,
      "loss": 0.3379,
      "step": 905
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6789846482875188,
      "learning_rate": 1.8489795918367348e-05,
      "loss": 0.3611,
      "step": 906
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.462827887125851,
      "learning_rate": 1.8510204081632653e-05,
      "loss": 0.5402,
      "step": 907
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3645078980350394,
      "learning_rate": 1.853061224489796e-05,
      "loss": 0.4531,
      "step": 908
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6047105424330054,
      "learning_rate": 1.8551020408163267e-05,
      "loss": 0.2523,
      "step": 909
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7696597312098996,
      "learning_rate": 1.8571428571428575e-05,
      "loss": 0.6946,
      "step": 910
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.634351852658061,
      "learning_rate": 1.859183673469388e-05,
      "loss": 0.2866,
      "step": 911
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.271180030714975,
      "learning_rate": 1.8612244897959185e-05,
      "loss": 0.5932,
      "step": 912
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6405816393085645,
      "learning_rate": 1.863265306122449e-05,
      "loss": 0.2682,
      "step": 913
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8862598898945756,
      "learning_rate": 1.8653061224489795e-05,
      "loss": 0.4317,
      "step": 914
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.505899290924887,
      "learning_rate": 1.8673469387755104e-05,
      "loss": 0.1774,
      "step": 915
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7892252750134754,
      "learning_rate": 1.869387755102041e-05,
      "loss": 0.4527,
      "step": 916
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6182889433993982,
      "learning_rate": 1.8714285714285717e-05,
      "loss": 0.2838,
      "step": 917
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2902465929932934,
      "learning_rate": 1.8734693877551022e-05,
      "loss": 0.3994,
      "step": 918
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6325357384453751,
      "learning_rate": 1.8755102040816327e-05,
      "loss": 0.2723,
      "step": 919
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3062932161071525,
      "learning_rate": 1.8775510204081636e-05,
      "loss": 0.523,
      "step": 920
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9652563325825048,
      "learning_rate": 1.879591836734694e-05,
      "loss": 0.5641,
      "step": 921
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7323528366811626,
      "learning_rate": 1.8816326530612246e-05,
      "loss": 0.2325,
      "step": 922
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6921650533343774,
      "learning_rate": 1.883673469387755e-05,
      "loss": 0.3086,
      "step": 923
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4689730347259762,
      "learning_rate": 1.885714285714286e-05,
      "loss": 0.1726,
      "step": 924
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3411315036399847,
      "learning_rate": 1.8877551020408164e-05,
      "loss": 0.8019,
      "step": 925
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7582030791182752,
      "learning_rate": 1.8897959183673473e-05,
      "loss": 0.5092,
      "step": 926
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6647748760719637,
      "learning_rate": 1.8918367346938778e-05,
      "loss": 0.3122,
      "step": 927
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8120821408644816,
      "learning_rate": 1.8938775510204083e-05,
      "loss": 0.3065,
      "step": 928
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6521416313257378,
      "learning_rate": 1.895918367346939e-05,
      "loss": 0.3769,
      "step": 929
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.08818097863161,
      "learning_rate": 1.8979591836734696e-05,
      "loss": 0.4981,
      "step": 930
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9550832605023019,
      "learning_rate": 1.9e-05,
      "loss": 0.4583,
      "step": 931
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.615855122490492,
      "learning_rate": 1.9020408163265306e-05,
      "loss": 0.2495,
      "step": 932
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5892109536326703,
      "learning_rate": 1.9040816326530614e-05,
      "loss": 0.0997,
      "step": 933
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8633101618832848,
      "learning_rate": 1.906122448979592e-05,
      "loss": 0.4088,
      "step": 934
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8127710133459488,
      "learning_rate": 1.9081632653061225e-05,
      "loss": 0.3507,
      "step": 935
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7206769941445378,
      "learning_rate": 1.9102040816326533e-05,
      "loss": 0.2901,
      "step": 936
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9938520647027537,
      "learning_rate": 1.9122448979591838e-05,
      "loss": 0.3556,
      "step": 937
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7991322851347018,
      "learning_rate": 1.9142857142857146e-05,
      "loss": 0.6187,
      "step": 938
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3121531652075071,
      "learning_rate": 1.916326530612245e-05,
      "loss": 0.5062,
      "step": 939
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7346461737216308,
      "learning_rate": 1.9183673469387756e-05,
      "loss": 0.2889,
      "step": 940
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7903221346286118,
      "learning_rate": 1.920408163265306e-05,
      "loss": 0.2767,
      "step": 941
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5725526099713937,
      "learning_rate": 1.922448979591837e-05,
      "loss": 0.2759,
      "step": 942
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6455837774530349,
      "learning_rate": 1.9244897959183675e-05,
      "loss": 0.2028,
      "step": 943
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.45141917295832,
      "learning_rate": 1.926530612244898e-05,
      "loss": 0.854,
      "step": 944
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.42141750855670995,
      "learning_rate": 1.928571428571429e-05,
      "loss": 0.0807,
      "step": 945
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9940066804751351,
      "learning_rate": 1.9306122448979593e-05,
      "loss": 0.3559,
      "step": 946
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3119241547026446,
      "learning_rate": 1.9326530612244902e-05,
      "loss": 0.4146,
      "step": 947
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.17042532183898,
      "learning_rate": 1.9346938775510207e-05,
      "loss": 0.5524,
      "step": 948
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9650363544623228,
      "learning_rate": 1.9367346938775512e-05,
      "loss": 0.6407,
      "step": 949
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5882433118455457,
      "learning_rate": 1.9387755102040817e-05,
      "loss": 0.2398,
      "step": 950
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6676705795217382,
      "learning_rate": 1.9408163265306122e-05,
      "loss": 0.2871,
      "step": 951
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.006704810610391,
      "learning_rate": 1.942857142857143e-05,
      "loss": 0.3402,
      "step": 952
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.2147335683719436,
      "learning_rate": 1.9448979591836735e-05,
      "loss": 0.807,
      "step": 953
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.3914035969272964,
      "learning_rate": 1.9469387755102044e-05,
      "loss": 0.1236,
      "step": 954
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8516240430313606,
      "learning_rate": 1.948979591836735e-05,
      "loss": 0.3783,
      "step": 955
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8174556811344709,
      "learning_rate": 1.9510204081632654e-05,
      "loss": 0.5392,
      "step": 956
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.458921143972807,
      "learning_rate": 1.9530612244897962e-05,
      "loss": 0.8019,
      "step": 957
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8619292116247872,
      "learning_rate": 1.9551020408163267e-05,
      "loss": 0.3039,
      "step": 958
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6202432464247841,
      "learning_rate": 1.9571428571428572e-05,
      "loss": 0.3223,
      "step": 959
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7426277199636495,
      "learning_rate": 1.9591836734693877e-05,
      "loss": 0.5037,
      "step": 960
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7391627215880334,
      "learning_rate": 1.9612244897959186e-05,
      "loss": 0.5717,
      "step": 961
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5635670291403656,
      "learning_rate": 1.963265306122449e-05,
      "loss": 0.2828,
      "step": 962
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6701111033707307,
      "learning_rate": 1.96530612244898e-05,
      "loss": 0.2795,
      "step": 963
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9097993660498097,
      "learning_rate": 1.9673469387755104e-05,
      "loss": 0.4379,
      "step": 964
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7319281301437565,
      "learning_rate": 1.969387755102041e-05,
      "loss": 0.3306,
      "step": 965
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9088508328432178,
      "learning_rate": 1.9714285714285718e-05,
      "loss": 0.3031,
      "step": 966
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6673345311594514,
      "learning_rate": 1.9734693877551023e-05,
      "loss": 0.0842,
      "step": 967
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6886196146602037,
      "learning_rate": 1.9755102040816328e-05,
      "loss": 0.2987,
      "step": 968
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7543583541732772,
      "learning_rate": 1.9775510204081633e-05,
      "loss": 0.2896,
      "step": 969
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8504748084943358,
      "learning_rate": 1.979591836734694e-05,
      "loss": 0.4356,
      "step": 970
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5706062908698174,
      "learning_rate": 1.9816326530612246e-05,
      "loss": 0.2736,
      "step": 971
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8796232530552723,
      "learning_rate": 1.983673469387755e-05,
      "loss": 0.3984,
      "step": 972
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.672471673339248,
      "learning_rate": 1.985714285714286e-05,
      "loss": 0.3226,
      "step": 973
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6076167549298931,
      "learning_rate": 1.9877551020408165e-05,
      "loss": 0.3136,
      "step": 974
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7249424051434074,
      "learning_rate": 1.9897959183673473e-05,
      "loss": 0.4593,
      "step": 975
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7816210227439762,
      "learning_rate": 1.9918367346938775e-05,
      "loss": 0.2993,
      "step": 976
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8879026004982281,
      "learning_rate": 1.9938775510204083e-05,
      "loss": 0.3264,
      "step": 977
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7136992594516219,
      "learning_rate": 1.9959183673469388e-05,
      "loss": 0.294,
      "step": 978
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7490209170815485,
      "learning_rate": 1.9979591836734697e-05,
      "loss": 0.8566,
      "step": 979
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8782760692306917,
      "learning_rate": 2e-05,
      "loss": 0.4852,
      "step": 980
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5547830388496273,
      "learning_rate": 1.9999999950808357e-05,
      "loss": 0.5035,
      "step": 981
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5728194901190884,
      "learning_rate": 1.999999980323342e-05,
      "loss": 0.2681,
      "step": 982
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5973218593332481,
      "learning_rate": 1.99999995572752e-05,
      "loss": 0.2741,
      "step": 983
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6596196972291111,
      "learning_rate": 1.9999999212933687e-05,
      "loss": 0.1327,
      "step": 984
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.0767694024679355,
      "learning_rate": 1.9999998770208895e-05,
      "loss": 0.6398,
      "step": 985
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4986078812167549,
      "learning_rate": 1.9999998229100823e-05,
      "loss": 0.2255,
      "step": 986
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7243788123562949,
      "learning_rate": 1.999999758960948e-05,
      "loss": 0.3415,
      "step": 987
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8372726084551009,
      "learning_rate": 1.9999996851734872e-05,
      "loss": 0.4128,
      "step": 988
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8303898037609888,
      "learning_rate": 1.9999996015477e-05,
      "loss": 0.4171,
      "step": 989
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0715221313277945,
      "learning_rate": 1.999999508083588e-05,
      "loss": 0.3448,
      "step": 990
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7301824036513483,
      "learning_rate": 1.999999404781152e-05,
      "loss": 0.2958,
      "step": 991
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8094890727973251,
      "learning_rate": 1.9999992916403922e-05,
      "loss": 0.2631,
      "step": 992
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.45535916860183284,
      "learning_rate": 1.9999991686613107e-05,
      "loss": 0.081,
      "step": 993
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6892682084741962,
      "learning_rate": 1.9999990358439082e-05,
      "loss": 0.3435,
      "step": 994
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5430199441664911,
      "learning_rate": 1.999998893188186e-05,
      "loss": 0.2012,
      "step": 995
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8747040494814293,
      "learning_rate": 1.9999987406941458e-05,
      "loss": 0.4149,
      "step": 996
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.301868387361056,
      "learning_rate": 1.999998578361789e-05,
      "loss": 0.5953,
      "step": 997
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8877380443060191,
      "learning_rate": 1.999998406191117e-05,
      "loss": 0.6135,
      "step": 998
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7170851614189,
      "learning_rate": 1.999998224182132e-05,
      "loss": 0.278,
      "step": 999
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5419191339735232,
      "learning_rate": 1.999998032334835e-05,
      "loss": 0.2596,
      "step": 1000
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7638306290397909,
      "learning_rate": 1.9999978306492287e-05,
      "loss": 0.4128,
      "step": 1001
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9064811733426041,
      "learning_rate": 1.9999976191253144e-05,
      "loss": 0.0869,
      "step": 1002
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.009162348913056,
      "learning_rate": 1.9999973977630948e-05,
      "loss": 0.7478,
      "step": 1003
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.770268644880684,
      "learning_rate": 1.9999971665625716e-05,
      "loss": 0.253,
      "step": 1004
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8197045611141829,
      "learning_rate": 1.999996925523747e-05,
      "loss": 0.436,
      "step": 1005
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6510165103866182,
      "learning_rate": 1.999996674646624e-05,
      "loss": 0.3793,
      "step": 1006
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2543482561693735,
      "learning_rate": 1.999996413931204e-05,
      "loss": 0.6745,
      "step": 1007
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.846139742781928,
      "learning_rate": 1.9999961433774908e-05,
      "loss": 0.4339,
      "step": 1008
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.62672157666823,
      "learning_rate": 1.9999958629854865e-05,
      "loss": 0.3191,
      "step": 1009
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7975145733029376,
      "learning_rate": 1.999995572755194e-05,
      "loss": 0.185,
      "step": 1010
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7605910418194443,
      "learning_rate": 1.9999952726866155e-05,
      "loss": 0.2497,
      "step": 1011
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5515777370900957,
      "learning_rate": 1.999994962779755e-05,
      "loss": 0.2745,
      "step": 1012
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5297002744029581,
      "learning_rate": 1.9999946430346145e-05,
      "loss": 0.2124,
      "step": 1013
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.822928031909462,
      "learning_rate": 1.999994313451198e-05,
      "loss": 0.4536,
      "step": 1014
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3088855014610214,
      "learning_rate": 1.999993974029508e-05,
      "loss": 0.5504,
      "step": 1015
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1853274462854917,
      "learning_rate": 1.9999936247695486e-05,
      "loss": 0.6986,
      "step": 1016
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7138979269089103,
      "learning_rate": 1.999993265671323e-05,
      "loss": 0.2795,
      "step": 1017
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5977105850992013,
      "learning_rate": 1.9999928967348344e-05,
      "loss": 0.3247,
      "step": 1018
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5289815278617098,
      "learning_rate": 1.9999925179600868e-05,
      "loss": 0.4148,
      "step": 1019
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.095664149257095,
      "learning_rate": 1.9999921293470833e-05,
      "loss": 0.2284,
      "step": 1020
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3202667627204927,
      "learning_rate": 1.9999917308958284e-05,
      "loss": 0.44,
      "step": 1021
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6455063268909108,
      "learning_rate": 1.9999913226063265e-05,
      "loss": 0.31,
      "step": 1022
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7476678203232013,
      "learning_rate": 1.9999909044785803e-05,
      "loss": 0.369,
      "step": 1023
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6711859164206907,
      "learning_rate": 1.9999904765125944e-05,
      "loss": 0.3817,
      "step": 1024
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.265746332987469,
      "learning_rate": 1.9999900387083735e-05,
      "loss": 0.5443,
      "step": 1025
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.6836421704836169,
      "learning_rate": 1.9999895910659214e-05,
      "loss": 0.2611,
      "step": 1026
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8031742320896436,
      "learning_rate": 1.9999891335852426e-05,
      "loss": 0.3188,
      "step": 1027
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8982825507104952,
      "learning_rate": 1.999988666266342e-05,
      "loss": 0.2779,
      "step": 1028
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7918318893737214,
      "learning_rate": 1.9999881891092233e-05,
      "loss": 0.2173,
      "step": 1029
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6842246977429236,
      "learning_rate": 1.9999877021138922e-05,
      "loss": 0.2734,
      "step": 1030
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0043639961643627,
      "learning_rate": 1.9999872052803527e-05,
      "loss": 0.4871,
      "step": 1031
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5401143040517221,
      "learning_rate": 1.9999866986086106e-05,
      "loss": 0.299,
      "step": 1032
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0083960932471092,
      "learning_rate": 1.99998618209867e-05,
      "loss": 0.5963,
      "step": 1033
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8056851280484405,
      "learning_rate": 1.999985655750536e-05,
      "loss": 0.5897,
      "step": 1034
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.870360852235471,
      "learning_rate": 1.9999851195642145e-05,
      "loss": 0.4208,
      "step": 1035
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6775084721122187,
      "learning_rate": 1.9999845735397103e-05,
      "loss": 0.237,
      "step": 1036
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7002838411174412,
      "learning_rate": 1.9999840176770285e-05,
      "loss": 0.3617,
      "step": 1037
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1162944801171883,
      "learning_rate": 1.9999834519761752e-05,
      "loss": 0.095,
      "step": 1038
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6352344754460132,
      "learning_rate": 1.9999828764371555e-05,
      "loss": 0.3055,
      "step": 1039
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8045782101016304,
      "learning_rate": 1.9999822910599753e-05,
      "loss": 0.3867,
      "step": 1040
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6308073922707108,
      "learning_rate": 1.9999816958446404e-05,
      "loss": 0.3248,
      "step": 1041
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8275832349338047,
      "learning_rate": 1.9999810907911565e-05,
      "loss": 0.3835,
      "step": 1042
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.999403155790214,
      "learning_rate": 1.9999804758995296e-05,
      "loss": 0.5996,
      "step": 1043
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.990453011727922,
      "learning_rate": 1.9999798511697654e-05,
      "loss": 0.7974,
      "step": 1044
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8701704433424318,
      "learning_rate": 1.9999792166018708e-05,
      "loss": 0.229,
      "step": 1045
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8352718445980476,
      "learning_rate": 1.9999785721958514e-05,
      "loss": 0.6506,
      "step": 1046
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7456341587744306,
      "learning_rate": 1.9999779179517138e-05,
      "loss": 0.2106,
      "step": 1047
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6737610432593423,
      "learning_rate": 1.9999772538694646e-05,
      "loss": 0.3351,
      "step": 1048
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.664928432799233,
      "learning_rate": 1.99997657994911e-05,
      "loss": 0.2807,
      "step": 1049
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9135949000370222,
      "learning_rate": 1.999975896190657e-05,
      "loss": 0.3909,
      "step": 1050
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2655643752787356,
      "learning_rate": 1.9999752025941116e-05,
      "loss": 0.5107,
      "step": 1051
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9495187189760952,
      "learning_rate": 1.9999744991594815e-05,
      "loss": 0.7653,
      "step": 1052
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7704698857320476,
      "learning_rate": 1.9999737858867728e-05,
      "loss": 0.3417,
      "step": 1053
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6213038829181897,
      "learning_rate": 1.9999730627759935e-05,
      "loss": 0.1995,
      "step": 1054
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.968712859545823,
      "learning_rate": 1.99997232982715e-05,
      "loss": 0.4082,
      "step": 1055
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7903926012830873,
      "learning_rate": 1.9999715870402497e-05,
      "loss": 0.5276,
      "step": 1056
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3243448192039389,
      "learning_rate": 1.9999708344153e-05,
      "loss": 0.5927,
      "step": 1057
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7162822062336244,
      "learning_rate": 1.9999700719523078e-05,
      "loss": 0.2139,
      "step": 1058
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7067521627316493,
      "learning_rate": 1.9999692996512815e-05,
      "loss": 0.319,
      "step": 1059
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5043057622096248,
      "learning_rate": 1.999968517512228e-05,
      "loss": 0.2529,
      "step": 1060
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.7950991824249345,
      "learning_rate": 1.9999677255351548e-05,
      "loss": 0.7696,
      "step": 1061
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4734253696753956,
      "learning_rate": 1.9999669237200704e-05,
      "loss": 0.5595,
      "step": 1062
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8942046863861297,
      "learning_rate": 1.9999661120669826e-05,
      "loss": 0.2968,
      "step": 1063
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6662726762499006,
      "learning_rate": 1.9999652905758988e-05,
      "loss": 0.3206,
      "step": 1064
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7583394074639195,
      "learning_rate": 1.9999644592468275e-05,
      "loss": 0.3747,
      "step": 1065
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7657318318370638,
      "learning_rate": 1.999963618079777e-05,
      "loss": 0.4351,
      "step": 1066
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7976946951932018,
      "learning_rate": 1.9999627670747554e-05,
      "loss": 0.3154,
      "step": 1067
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6968780179158555,
      "learning_rate": 1.999961906231771e-05,
      "loss": 0.2558,
      "step": 1068
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4840313234896065,
      "learning_rate": 1.9999610355508322e-05,
      "loss": 0.1382,
      "step": 1069
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9655155778855933,
      "learning_rate": 1.9999601550319478e-05,
      "loss": 0.971,
      "step": 1070
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5209448075124296,
      "learning_rate": 1.9999592646751264e-05,
      "loss": 0.2989,
      "step": 1071
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6481052589173409,
      "learning_rate": 1.9999583644803768e-05,
      "loss": 0.3294,
      "step": 1072
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6875243489663103,
      "learning_rate": 1.9999574544477074e-05,
      "loss": 0.2991,
      "step": 1073
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1064079114918788,
      "learning_rate": 1.999956534577128e-05,
      "loss": 0.6351,
      "step": 1074
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8559479226846755,
      "learning_rate": 1.9999556048686468e-05,
      "loss": 0.5039,
      "step": 1075
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8466734977541645,
      "learning_rate": 1.9999546653222735e-05,
      "loss": 0.2915,
      "step": 1076
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6802901773729113,
      "learning_rate": 1.999953715938017e-05,
      "loss": 0.2344,
      "step": 1077
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4392290731162533,
      "learning_rate": 1.9999527567158873e-05,
      "loss": 0.2516,
      "step": 1078
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9576248246097359,
      "learning_rate": 1.999951787655893e-05,
      "loss": 0.8641,
      "step": 1079
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.359420212443585,
      "learning_rate": 1.9999508087580438e-05,
      "loss": 0.4679,
      "step": 1080
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9783493480917953,
      "learning_rate": 1.9999498200223503e-05,
      "loss": 0.3366,
      "step": 1081
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5891653933005581,
      "learning_rate": 1.9999488214488205e-05,
      "loss": 0.3265,
      "step": 1082
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6129383211256566,
      "learning_rate": 1.9999478130374658e-05,
      "loss": 0.3807,
      "step": 1083
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.059300388325958,
      "learning_rate": 1.999946794788295e-05,
      "loss": 0.4968,
      "step": 1084
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.3976631059941016,
      "learning_rate": 1.999945766701319e-05,
      "loss": 0.7127,
      "step": 1085
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5080938306088487,
      "learning_rate": 1.9999447287765473e-05,
      "loss": 0.2343,
      "step": 1086
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5912408571935311,
      "learning_rate": 1.9999436810139904e-05,
      "loss": 0.1945,
      "step": 1087
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.1694442406061873,
      "learning_rate": 1.9999426234136584e-05,
      "loss": 0.7045,
      "step": 1088
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6182853834842034,
      "learning_rate": 1.9999415559755623e-05,
      "loss": 0.3425,
      "step": 1089
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6872059535092649,
      "learning_rate": 1.999940478699712e-05,
      "loss": 0.2275,
      "step": 1090
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8091478708561494,
      "learning_rate": 1.999939391586118e-05,
      "loss": 0.3615,
      "step": 1091
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.5689386230437015,
      "learning_rate": 1.999938294634791e-05,
      "loss": 0.7312,
      "step": 1092
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4014645762393512,
      "learning_rate": 1.9999371878457427e-05,
      "loss": 0.4951,
      "step": 1093
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6728527515220405,
      "learning_rate": 1.999936071218983e-05,
      "loss": 0.4032,
      "step": 1094
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4920916810281571,
      "learning_rate": 1.999934944754523e-05,
      "loss": 0.2164,
      "step": 1095
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7438866089247715,
      "learning_rate": 1.9999338084523745e-05,
      "loss": 0.3545,
      "step": 1096
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5557967281209282,
      "learning_rate": 1.9999326623125478e-05,
      "loss": 0.184,
      "step": 1097
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6717444922856545,
      "learning_rate": 1.9999315063350547e-05,
      "loss": 0.2765,
      "step": 1098
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6924528584382185,
      "learning_rate": 1.9999303405199063e-05,
      "loss": 0.294,
      "step": 1099
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.903161562402639,
      "learning_rate": 1.9999291648671142e-05,
      "loss": 0.3869,
      "step": 1100
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7393381027926322,
      "learning_rate": 1.99992797937669e-05,
      "loss": 0.3641,
      "step": 1101
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.4483897114911766,
      "learning_rate": 1.9999267840486453e-05,
      "loss": 0.6252,
      "step": 1102
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.9705716295420145,
      "learning_rate": 1.9999255788829915e-05,
      "loss": 0.348,
      "step": 1103
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.60197655155061,
      "learning_rate": 1.9999243638797416e-05,
      "loss": 0.2266,
      "step": 1104
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5304544177843526,
      "learning_rate": 1.9999231390389064e-05,
      "loss": 0.2894,
      "step": 1105
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.67959337916265,
      "learning_rate": 1.9999219043604984e-05,
      "loss": 0.3432,
      "step": 1106
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5277637970867891,
      "learning_rate": 1.9999206598445294e-05,
      "loss": 0.32,
      "step": 1107
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.919350993561027,
      "learning_rate": 1.9999194054910124e-05,
      "loss": 0.3878,
      "step": 1108
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6042951469098108,
      "learning_rate": 1.999918141299959e-05,
      "loss": 0.3599,
      "step": 1109
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.1379229223651477,
      "learning_rate": 1.9999168672713816e-05,
      "loss": 0.5172,
      "step": 1110
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.8620588487764942,
      "learning_rate": 1.9999155834052937e-05,
      "loss": 0.7588,
      "step": 1111
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6059343515421555,
      "learning_rate": 1.9999142897017067e-05,
      "loss": 0.285,
      "step": 1112
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5728349237219131,
      "learning_rate": 1.9999129861606345e-05,
      "loss": 0.3064,
      "step": 1113
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5699925549217161,
      "learning_rate": 1.999911672782089e-05,
      "loss": 0.3159,
      "step": 1114
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.9638940877580977,
      "learning_rate": 1.9999103495660832e-05,
      "loss": 0.8706,
      "step": 1115
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.3123146230208071,
      "learning_rate": 1.9999090165126307e-05,
      "loss": 0.5083,
      "step": 1116
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8295072399100445,
      "learning_rate": 1.9999076736217443e-05,
      "loss": 0.3593,
      "step": 1117
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49065503079462786,
      "learning_rate": 1.999906320893437e-05,
      "loss": 0.2649,
      "step": 1118
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4734107063473509,
      "learning_rate": 1.9999049583277225e-05,
      "loss": 0.1925,
      "step": 1119
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7126663254985884,
      "learning_rate": 1.9999035859246138e-05,
      "loss": 0.3788,
      "step": 1120
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.232341998026714,
      "learning_rate": 1.9999022036841247e-05,
      "loss": 0.342,
      "step": 1121
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6032656022592455,
      "learning_rate": 1.999900811606269e-05,
      "loss": 0.3135,
      "step": 1122
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7346221856921028,
      "learning_rate": 1.9998994096910594e-05,
      "loss": 0.2967,
      "step": 1123
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.137077630277944,
      "learning_rate": 1.999897997938511e-05,
      "loss": 0.8668,
      "step": 1124
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6560117859498378,
      "learning_rate": 1.9998965763486368e-05,
      "loss": 0.3696,
      "step": 1125
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.0296528399289602,
      "learning_rate": 1.9998951449214513e-05,
      "loss": 0.5158,
      "step": 1126
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6214617301623948,
      "learning_rate": 1.9998937036569682e-05,
      "loss": 0.2235,
      "step": 1127
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5974313463458063,
      "learning_rate": 1.9998922525552017e-05,
      "loss": 0.27,
      "step": 1128
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.4512363670848625,
      "learning_rate": 1.9998907916161663e-05,
      "loss": 0.1787,
      "step": 1129
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.5745980878961452,
      "learning_rate": 1.9998893208398765e-05,
      "loss": 0.3522,
      "step": 1130
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.49200231885839657,
      "learning_rate": 1.9998878402263464e-05,
      "loss": 0.0826,
      "step": 1131
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6899142360945189,
      "learning_rate": 1.9998863497755908e-05,
      "loss": 0.2838,
      "step": 1132
    },
    {
      "epoch": 0.03,
      "grad_norm": 1.2408953025437572,
      "learning_rate": 1.9998848494876242e-05,
      "loss": 0.6784,
      "step": 1133
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.8999475407688425,
      "learning_rate": 1.9998833393624616e-05,
      "loss": 0.514,
      "step": 1134
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.278755282364428,
      "learning_rate": 1.9998818194001175e-05,
      "loss": 0.7825,
      "step": 1135
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6099924501234558,
      "learning_rate": 1.999880289600607e-05,
      "loss": 0.2262,
      "step": 1136
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.501941922060119,
      "learning_rate": 1.9998787499639453e-05,
      "loss": 0.2727,
      "step": 1137
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6280786790524929,
      "learning_rate": 1.9998772004901474e-05,
      "loss": 0.2257,
      "step": 1138
    },
    {
      "epoch": 0.03,
      "grad_norm": 2.4071470556195607,
      "learning_rate": 1.999875641179229e-05,
      "loss": 0.8647,
      "step": 1139
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6676905782142238,
      "learning_rate": 1.9998740720312043e-05,
      "loss": 0.1399,
      "step": 1140
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.6868814043113239,
      "learning_rate": 1.9998724930460903e-05,
      "loss": 0.3647,
      "step": 1141
    },
    {
      "epoch": 0.03,
      "grad_norm": 0.7826674544299556,
      "learning_rate": 1.9998709042239012e-05,
      "loss": 0.3623,
      "step": 1142
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7495052084846726,
      "learning_rate": 1.999869305564653e-05,
      "loss": 0.4296,
      "step": 1143
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7923455083961974,
      "learning_rate": 1.999867697068362e-05,
      "loss": 0.3846,
      "step": 1144
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5146773193170383,
      "learning_rate": 1.9998660787350436e-05,
      "loss": 0.2416,
      "step": 1145
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.761930265097687,
      "learning_rate": 1.9998644505647135e-05,
      "loss": 0.2271,
      "step": 1146
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.54383349349221,
      "learning_rate": 1.999862812557388e-05,
      "loss": 0.1649,
      "step": 1147
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5741856883281703,
      "learning_rate": 1.999861164713083e-05,
      "loss": 0.3626,
      "step": 1148
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5765736129501435,
      "learning_rate": 1.999859507031815e-05,
      "loss": 0.203,
      "step": 1149
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0136622681211997,
      "learning_rate": 1.9998578395136003e-05,
      "loss": 0.4467,
      "step": 1150
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4413425334505015,
      "learning_rate": 1.999856162158455e-05,
      "loss": 0.5233,
      "step": 1151
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0616208707463803,
      "learning_rate": 1.9998544749663957e-05,
      "loss": 0.6148,
      "step": 1152
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7686946860240382,
      "learning_rate": 1.9998527779374394e-05,
      "loss": 0.3202,
      "step": 1153
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7538681006496181,
      "learning_rate": 1.999851071071602e-05,
      "loss": 0.3594,
      "step": 1154
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5273129248844022,
      "learning_rate": 1.999849354368901e-05,
      "loss": 0.2881,
      "step": 1155
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5802658781893455,
      "learning_rate": 1.999847627829353e-05,
      "loss": 0.1798,
      "step": 1156
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.556037648211537,
      "learning_rate": 1.9998458914529754e-05,
      "loss": 0.4195,
      "step": 1157
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1579833231371646,
      "learning_rate": 1.9998441452397848e-05,
      "loss": 0.3793,
      "step": 1158
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5635106344172744,
      "learning_rate": 1.9998423891897983e-05,
      "loss": 0.2955,
      "step": 1159
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6496167964064365,
      "learning_rate": 1.9998406233030336e-05,
      "loss": 0.4194,
      "step": 1160
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6724363700976764,
      "learning_rate": 1.9998388475795075e-05,
      "loss": 0.4252,
      "step": 1161
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5632091381861775,
      "learning_rate": 1.9998370620192384e-05,
      "loss": 0.4005,
      "step": 1162
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.634385838592412,
      "learning_rate": 1.9998352666222428e-05,
      "loss": 0.296,
      "step": 1163
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43017537359321617,
      "learning_rate": 1.9998334613885388e-05,
      "loss": 0.2221,
      "step": 1164
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6032391857473386,
      "learning_rate": 1.9998316463181443e-05,
      "loss": 0.2439,
      "step": 1165
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5383155507837882,
      "learning_rate": 1.9998298214110775e-05,
      "loss": 0.2881,
      "step": 1166
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9594444208557853,
      "learning_rate": 1.9998279866673555e-05,
      "loss": 0.4913,
      "step": 1167
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5590300457360644,
      "learning_rate": 1.999826142086997e-05,
      "loss": 0.3033,
      "step": 1168
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9472870908621503,
      "learning_rate": 1.9998242876700198e-05,
      "loss": 0.6199,
      "step": 1169
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.123477601196464,
      "learning_rate": 1.9998224234164423e-05,
      "loss": 0.5864,
      "step": 1170
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6149702974859186,
      "learning_rate": 1.9998205493262828e-05,
      "loss": 0.3353,
      "step": 1171
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6830453580967766,
      "learning_rate": 1.9998186653995598e-05,
      "loss": 0.2891,
      "step": 1172
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4788296453008641,
      "learning_rate": 1.9998167716362917e-05,
      "loss": 0.2879,
      "step": 1173
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8603470258241641,
      "learning_rate": 1.9998148680364973e-05,
      "loss": 0.2652,
      "step": 1174
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6535491238287979,
      "learning_rate": 1.999812954600195e-05,
      "loss": 0.2264,
      "step": 1175
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8944488082263824,
      "learning_rate": 1.9998110313274043e-05,
      "loss": 0.3965,
      "step": 1176
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6121769724353894,
      "learning_rate": 1.9998090982181435e-05,
      "loss": 0.298,
      "step": 1177
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3631373838148582,
      "learning_rate": 1.9998071552724318e-05,
      "loss": 0.6544,
      "step": 1178
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7028964218485073,
      "learning_rate": 1.9998052024902883e-05,
      "loss": 0.3011,
      "step": 1179
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7787054635567143,
      "learning_rate": 1.9998032398717326e-05,
      "loss": 0.8715,
      "step": 1180
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8725143627339041,
      "learning_rate": 1.999801267416783e-05,
      "loss": 0.0914,
      "step": 1181
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6021261042521378,
      "learning_rate": 1.99979928512546e-05,
      "loss": 0.3381,
      "step": 1182
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5770794323418574,
      "learning_rate": 1.9997972929977823e-05,
      "loss": 0.1146,
      "step": 1183
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5761624360795936,
      "learning_rate": 1.9997952910337702e-05,
      "loss": 0.3635,
      "step": 1184
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.241705342275482,
      "learning_rate": 1.999793279233443e-05,
      "loss": 0.6718,
      "step": 1185
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5218634134241589,
      "learning_rate": 1.9997912575968204e-05,
      "loss": 0.229,
      "step": 1186
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.105485959862523,
      "learning_rate": 1.9997892261239223e-05,
      "loss": 0.5992,
      "step": 1187
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0279749269692078,
      "learning_rate": 1.999787184814769e-05,
      "loss": 0.5863,
      "step": 1188
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.4668809380348056,
      "learning_rate": 1.9997851336693804e-05,
      "loss": 0.6666,
      "step": 1189
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5066081069108681,
      "learning_rate": 1.999783072687777e-05,
      "loss": 0.238,
      "step": 1190
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6290370636593756,
      "learning_rate": 1.9997810018699785e-05,
      "loss": 0.3691,
      "step": 1191
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5535255646166901,
      "learning_rate": 1.9997789212160056e-05,
      "loss": 0.2118,
      "step": 1192
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1298622189220138,
      "learning_rate": 1.9997768307258782e-05,
      "loss": 0.5844,
      "step": 1193
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4144358220437827,
      "learning_rate": 1.9997747303996176e-05,
      "loss": 0.2605,
      "step": 1194
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5660937260207369,
      "learning_rate": 1.999772620237245e-05,
      "loss": 0.3121,
      "step": 1195
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5635054541062666,
      "learning_rate": 1.9997705002387796e-05,
      "loss": 0.3134,
      "step": 1196
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4423823686099118,
      "learning_rate": 1.999768370404243e-05,
      "loss": 0.5528,
      "step": 1197
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.580994608405788,
      "learning_rate": 1.9997662307336567e-05,
      "loss": 0.6934,
      "step": 1198
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4995088252967807,
      "learning_rate": 1.999764081227041e-05,
      "loss": 0.2244,
      "step": 1199
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8535628673706426,
      "learning_rate": 1.9997619218844174e-05,
      "loss": 0.3656,
      "step": 1200
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5524192237494954,
      "learning_rate": 1.9997597527058067e-05,
      "loss": 0.2254,
      "step": 1201
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5562789055773552,
      "learning_rate": 1.999757573691231e-05,
      "loss": 0.3899,
      "step": 1202
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45173046740812367,
      "learning_rate": 1.999755384840711e-05,
      "loss": 0.2681,
      "step": 1203
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5604805706073269,
      "learning_rate": 1.9997531861542687e-05,
      "loss": 0.3035,
      "step": 1204
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.598176538947288,
      "learning_rate": 1.9997509776319256e-05,
      "loss": 0.4086,
      "step": 1205
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.198717184905615,
      "learning_rate": 1.9997487592737035e-05,
      "loss": 0.8345,
      "step": 1206
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5190875548752284,
      "learning_rate": 1.999746531079624e-05,
      "loss": 0.2983,
      "step": 1207
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2522945412333553,
      "learning_rate": 1.999744293049709e-05,
      "loss": 0.2975,
      "step": 1208
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5795556246400131,
      "learning_rate": 1.9997420451839813e-05,
      "loss": 0.2607,
      "step": 1209
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4048967204473182,
      "learning_rate": 1.999739787482462e-05,
      "loss": 0.5188,
      "step": 1210
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9590792217079566,
      "learning_rate": 1.999737519945174e-05,
      "loss": 0.5661,
      "step": 1211
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45261150621794144,
      "learning_rate": 1.9997352425721386e-05,
      "loss": 0.08,
      "step": 1212
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5754192748226463,
      "learning_rate": 1.9997329553633794e-05,
      "loss": 0.2787,
      "step": 1213
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4566391221033436,
      "learning_rate": 1.9997306583189186e-05,
      "loss": 0.2797,
      "step": 1214
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0234503131926664,
      "learning_rate": 1.9997283514387784e-05,
      "loss": 0.8737,
      "step": 1215
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6546469957052874,
      "learning_rate": 1.999726034722982e-05,
      "loss": 0.385,
      "step": 1216
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2717191783539201,
      "learning_rate": 1.9997237081715518e-05,
      "loss": 0.5211,
      "step": 1217
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5462303239064062,
      "learning_rate": 1.9997213717845106e-05,
      "loss": 0.2945,
      "step": 1218
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2169802004245078,
      "learning_rate": 1.9997190255618817e-05,
      "loss": 0.5884,
      "step": 1219
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5406149003277734,
      "learning_rate": 1.999716669503688e-05,
      "loss": 0.2862,
      "step": 1220
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8653823213766045,
      "learning_rate": 1.9997143036099527e-05,
      "loss": 0.6532,
      "step": 1221
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48397546933595226,
      "learning_rate": 1.9997119278806997e-05,
      "loss": 0.2283,
      "step": 1222
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.45237902491493415,
      "learning_rate": 1.9997095423159512e-05,
      "loss": 0.1848,
      "step": 1223
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1963636586051987,
      "learning_rate": 1.9997071469157312e-05,
      "loss": 0.9251,
      "step": 1224
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43989880393666664,
      "learning_rate": 1.9997047416800638e-05,
      "loss": 0.2362,
      "step": 1225
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8003786555942112,
      "learning_rate": 1.999702326608972e-05,
      "loss": 0.3251,
      "step": 1226
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5391514976820437,
      "learning_rate": 1.9996999017024795e-05,
      "loss": 0.3169,
      "step": 1227
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2082093172535795,
      "learning_rate": 1.999697466960611e-05,
      "loss": 0.6547,
      "step": 1228
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8029436199453589,
      "learning_rate": 1.9996950223833896e-05,
      "loss": 0.4633,
      "step": 1229
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6014318774027501,
      "learning_rate": 1.9996925679708395e-05,
      "loss": 0.2474,
      "step": 1230
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47646954461988245,
      "learning_rate": 1.9996901037229847e-05,
      "loss": 0.1917,
      "step": 1231
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5295399234426436,
      "learning_rate": 1.9996876296398505e-05,
      "loss": 0.348,
      "step": 1232
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5865786648959392,
      "learning_rate": 1.99968514572146e-05,
      "loss": 0.3451,
      "step": 1233
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6146363923363461,
      "learning_rate": 1.9996826519678382e-05,
      "loss": 0.2893,
      "step": 1234
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.832808055499121,
      "learning_rate": 1.9996801483790092e-05,
      "loss": 0.419,
      "step": 1235
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5764489641968098,
      "learning_rate": 1.999677634954998e-05,
      "loss": 0.2471,
      "step": 1236
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9021759657809906,
      "learning_rate": 1.9996751116958296e-05,
      "loss": 0.6206,
      "step": 1237
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48557953603691495,
      "learning_rate": 1.9996725786015285e-05,
      "loss": 0.2949,
      "step": 1238
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.749608587091901,
      "learning_rate": 1.9996700356721198e-05,
      "loss": 0.8614,
      "step": 1239
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43053586462562954,
      "learning_rate": 1.999667482907628e-05,
      "loss": 0.1929,
      "step": 1240
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7609120389869627,
      "learning_rate": 1.9996649203080786e-05,
      "loss": 0.4668,
      "step": 1241
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.42914306833855814,
      "learning_rate": 1.999662347873497e-05,
      "loss": 0.1156,
      "step": 1242
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.330669098508489,
      "learning_rate": 1.999659765603908e-05,
      "loss": 0.6563,
      "step": 1243
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4675136417264915,
      "learning_rate": 1.9996571734993373e-05,
      "loss": 0.3018,
      "step": 1244
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5742750948606892,
      "learning_rate": 1.9996545715598107e-05,
      "loss": 0.3165,
      "step": 1245
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8302043754162172,
      "learning_rate": 1.9996519597853533e-05,
      "loss": 0.4948,
      "step": 1246
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.942631475809941,
      "learning_rate": 1.9996493381759913e-05,
      "loss": 0.7085,
      "step": 1247
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1816942641308776,
      "learning_rate": 1.9996467067317497e-05,
      "loss": 0.4527,
      "step": 1248
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44976025667777747,
      "learning_rate": 1.999644065452655e-05,
      "loss": 0.2496,
      "step": 1249
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.558989114422605,
      "learning_rate": 1.9996414143387332e-05,
      "loss": 0.3416,
      "step": 1250
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1418438354170322,
      "learning_rate": 1.99963875339001e-05,
      "loss": 0.5439,
      "step": 1251
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6091193853937803,
      "learning_rate": 1.9996360826065122e-05,
      "loss": 0.2454,
      "step": 1252
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7987470714678535,
      "learning_rate": 1.9996334019882655e-05,
      "loss": 0.3458,
      "step": 1253
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5598831957448555,
      "learning_rate": 1.9996307115352965e-05,
      "loss": 0.3712,
      "step": 1254
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6565380155719573,
      "learning_rate": 1.9996280112476315e-05,
      "loss": 0.3823,
      "step": 1255
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6093426767440445,
      "learning_rate": 1.9996253011252973e-05,
      "loss": 0.3379,
      "step": 1256
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0260277679296945,
      "learning_rate": 1.9996225811683205e-05,
      "loss": 0.4861,
      "step": 1257
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6297829740190505,
      "learning_rate": 1.999619851376728e-05,
      "loss": 0.2709,
      "step": 1258
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6933860977648583,
      "learning_rate": 1.999617111750546e-05,
      "loss": 0.2973,
      "step": 1259
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4924287614213194,
      "learning_rate": 1.9996143622898022e-05,
      "loss": 0.7554,
      "step": 1260
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5277444668101865,
      "learning_rate": 1.9996116029945237e-05,
      "loss": 0.3483,
      "step": 1261
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8907108461679846,
      "learning_rate": 1.9996088338647367e-05,
      "loss": 0.4018,
      "step": 1262
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7762977476975952,
      "learning_rate": 1.9996060549004697e-05,
      "loss": 0.3546,
      "step": 1263
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.38147043368177486,
      "learning_rate": 1.999603266101749e-05,
      "loss": 0.146,
      "step": 1264
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0432317668141775,
      "learning_rate": 1.9996004674686026e-05,
      "loss": 0.6984,
      "step": 1265
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.65176142717371,
      "learning_rate": 1.9995976590010578e-05,
      "loss": 0.341,
      "step": 1266
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.632298047580507,
      "learning_rate": 1.9995948406991424e-05,
      "loss": 0.3578,
      "step": 1267
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5168852505662236,
      "learning_rate": 1.999592012562884e-05,
      "loss": 0.2478,
      "step": 1268
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1956022696967692,
      "learning_rate": 1.9995891745923107e-05,
      "loss": 0.6515,
      "step": 1269
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0144019240960336,
      "learning_rate": 1.99958632678745e-05,
      "loss": 0.4854,
      "step": 1270
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9665514925785662,
      "learning_rate": 1.99958346914833e-05,
      "loss": 0.6219,
      "step": 1271
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.468638098012292,
      "learning_rate": 1.999580601674979e-05,
      "loss": 0.2411,
      "step": 1272
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4367329400958947,
      "learning_rate": 1.999577724367425e-05,
      "loss": 0.2292,
      "step": 1273
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7033832154386801,
      "learning_rate": 1.9995748372256967e-05,
      "loss": 0.2852,
      "step": 1274
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6331266751636648,
      "learning_rate": 1.9995719402498226e-05,
      "loss": 0.4007,
      "step": 1275
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8394922453183385,
      "learning_rate": 1.9995690334398303e-05,
      "loss": 0.3401,
      "step": 1276
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6853011858316851,
      "learning_rate": 1.999566116795749e-05,
      "loss": 0.296,
      "step": 1277
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5724307828160207,
      "learning_rate": 1.9995631903176073e-05,
      "loss": 0.6518,
      "step": 1278
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48540989792142963,
      "learning_rate": 1.9995602540054344e-05,
      "loss": 0.3481,
      "step": 1279
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9188654122641824,
      "learning_rate": 1.9995573078592584e-05,
      "loss": 0.5704,
      "step": 1280
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.44667761630693437,
      "learning_rate": 1.9995543518791088e-05,
      "loss": 0.2411,
      "step": 1281
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7857127728531645,
      "learning_rate": 1.9995513860650148e-05,
      "loss": 0.3201,
      "step": 1282
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47059862904756244,
      "learning_rate": 1.9995484104170055e-05,
      "loss": 0.188,
      "step": 1283
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.352962905579499,
      "learning_rate": 1.9995454249351093e-05,
      "loss": 0.8605,
      "step": 1284
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.46985132362194637,
      "learning_rate": 1.999542429619357e-05,
      "loss": 0.2324,
      "step": 1285
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8345694254632152,
      "learning_rate": 1.999539424469777e-05,
      "loss": 0.3491,
      "step": 1286
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7884250438147068,
      "learning_rate": 1.9995364094863997e-05,
      "loss": 0.4232,
      "step": 1287
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1660875500356505,
      "learning_rate": 1.999533384669254e-05,
      "loss": 0.5457,
      "step": 1288
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2639881953976684,
      "learning_rate": 1.99953035001837e-05,
      "loss": 0.4882,
      "step": 1289
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5572148376764785,
      "learning_rate": 1.9995273055337775e-05,
      "loss": 0.2252,
      "step": 1290
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.48392915534761594,
      "learning_rate": 1.9995242512155063e-05,
      "loss": 0.2745,
      "step": 1291
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4267759043459543,
      "learning_rate": 1.999521187063587e-05,
      "loss": 0.2127,
      "step": 1292
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6679465328043988,
      "learning_rate": 1.9995181130780494e-05,
      "loss": 0.7416,
      "step": 1293
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6432113504386086,
      "learning_rate": 1.9995150292589237e-05,
      "loss": 0.3036,
      "step": 1294
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5369093988084621,
      "learning_rate": 1.9995119356062398e-05,
      "loss": 0.3077,
      "step": 1295
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.869518382020697,
      "learning_rate": 1.9995088321200292e-05,
      "loss": 0.5109,
      "step": 1296
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5728860661537635,
      "learning_rate": 1.999505718800321e-05,
      "loss": 0.3635,
      "step": 1297
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6733429253755863,
      "learning_rate": 1.9995025956471473e-05,
      "loss": 0.5372,
      "step": 1298
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0276051219130926,
      "learning_rate": 1.999499462660538e-05,
      "loss": 0.33,
      "step": 1299
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3357931565189911,
      "learning_rate": 1.9994963198405243e-05,
      "loss": 0.1802,
      "step": 1300
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5106782508117003,
      "learning_rate": 1.999493167187137e-05,
      "loss": 0.4412,
      "step": 1301
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6450497161612232,
      "learning_rate": 1.9994900047004068e-05,
      "loss": 0.7455,
      "step": 1302
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5000492903249766,
      "learning_rate": 1.999486832380365e-05,
      "loss": 0.2369,
      "step": 1303
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5722956394483986,
      "learning_rate": 1.999483650227043e-05,
      "loss": 0.353,
      "step": 1304
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8849664392656205,
      "learning_rate": 1.999480458240472e-05,
      "loss": 0.4731,
      "step": 1305
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3055253726800662,
      "learning_rate": 1.9994772564206834e-05,
      "loss": 0.6599,
      "step": 1306
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8305270409164152,
      "learning_rate": 1.9994740447677085e-05,
      "loss": 0.4164,
      "step": 1307
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6763693467661297,
      "learning_rate": 1.9994708232815793e-05,
      "loss": 0.3379,
      "step": 1308
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5027096762435657,
      "learning_rate": 1.999467591962327e-05,
      "loss": 0.2719,
      "step": 1309
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.448876282303198,
      "learning_rate": 1.999464350809984e-05,
      "loss": 0.2422,
      "step": 1310
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2943030552335877,
      "learning_rate": 1.9994610998245815e-05,
      "loss": 0.5023,
      "step": 1311
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8432564767635976,
      "learning_rate": 1.9994578390061518e-05,
      "loss": 0.4892,
      "step": 1312
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5502838859468773,
      "learning_rate": 1.9994545683547273e-05,
      "loss": 0.2245,
      "step": 1313
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7290765512070475,
      "learning_rate": 1.9994512878703397e-05,
      "loss": 0.4315,
      "step": 1314
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4986796100977506,
      "learning_rate": 1.9994479975530216e-05,
      "loss": 0.2973,
      "step": 1315
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8962854968173631,
      "learning_rate": 1.9994446974028054e-05,
      "loss": 0.2768,
      "step": 1316
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2479962986747895,
      "learning_rate": 1.999441387419723e-05,
      "loss": 0.4658,
      "step": 1317
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5110534096575953,
      "learning_rate": 1.9994380676038078e-05,
      "loss": 0.282,
      "step": 1318
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5046920859638718,
      "learning_rate": 1.9994347379550918e-05,
      "loss": 0.2131,
      "step": 1319
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5728358310257631,
      "learning_rate": 1.999431398473608e-05,
      "loss": 0.3847,
      "step": 1320
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6376197469382938,
      "learning_rate": 1.9994280491593895e-05,
      "loss": 0.4534,
      "step": 1321
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4592432371763356,
      "learning_rate": 1.9994246900124686e-05,
      "loss": 0.2282,
      "step": 1322
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9983098382136614,
      "learning_rate": 1.999421321032879e-05,
      "loss": 0.5966,
      "step": 1323
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.480279287280205,
      "learning_rate": 1.9994179422206535e-05,
      "loss": 0.6718,
      "step": 1324
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.2747158579894964,
      "learning_rate": 1.999414553575825e-05,
      "loss": 0.8253,
      "step": 1325
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6477491995097585,
      "learning_rate": 1.999411155098428e-05,
      "loss": 0.2253,
      "step": 1326
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5024724319283731,
      "learning_rate": 1.999407746788495e-05,
      "loss": 0.3053,
      "step": 1327
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3527662862083162,
      "learning_rate": 1.9994043286460596e-05,
      "loss": 0.1677,
      "step": 1328
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8833822753099975,
      "learning_rate": 1.9994009006711557e-05,
      "loss": 0.4862,
      "step": 1329
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.79708723652566,
      "learning_rate": 1.9993974628638168e-05,
      "loss": 0.4925,
      "step": 1330
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4974309907686763,
      "learning_rate": 1.9993940152240767e-05,
      "loss": 0.2401,
      "step": 1331
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.080826383411322,
      "learning_rate": 1.9993905577519693e-05,
      "loss": 0.8758,
      "step": 1332
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5168651053794442,
      "learning_rate": 1.9993870904475293e-05,
      "loss": 0.3035,
      "step": 1333
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.0849083953469787,
      "learning_rate": 1.9993836133107897e-05,
      "loss": 0.747,
      "step": 1334
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5499670085637272,
      "learning_rate": 1.9993801263417856e-05,
      "loss": 0.1873,
      "step": 1335
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8400183842621682,
      "learning_rate": 1.999376629540551e-05,
      "loss": 0.3532,
      "step": 1336
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5313493944229266,
      "learning_rate": 1.99937312290712e-05,
      "loss": 0.26,
      "step": 1337
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6763415839915471,
      "learning_rate": 1.9993696064415277e-05,
      "loss": 0.4475,
      "step": 1338
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3491805518749588,
      "learning_rate": 1.9993660801438084e-05,
      "loss": 0.2299,
      "step": 1339
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5264636309946262,
      "learning_rate": 1.9993625440139966e-05,
      "loss": 0.4165,
      "step": 1340
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7225968311011395,
      "learning_rate": 1.999358998052127e-05,
      "loss": 0.2833,
      "step": 1341
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3541299333977803,
      "learning_rate": 1.999355442258235e-05,
      "loss": 0.5994,
      "step": 1342
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4784743004453353,
      "learning_rate": 1.9993518766323555e-05,
      "loss": 0.6884,
      "step": 1343
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5617422805941001,
      "learning_rate": 1.999348301174523e-05,
      "loss": 0.2296,
      "step": 1344
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5561790120145921,
      "learning_rate": 1.9993447158847737e-05,
      "loss": 0.3109,
      "step": 1345
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5069237918827939,
      "learning_rate": 1.9993411207631415e-05,
      "loss": 0.3669,
      "step": 1346
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5637155218278,
      "learning_rate": 1.9993375158096632e-05,
      "loss": 0.3924,
      "step": 1347
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4770172164762431,
      "learning_rate": 1.9993339010243736e-05,
      "loss": 0.1336,
      "step": 1348
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5882609910758619,
      "learning_rate": 1.9993302764073076e-05,
      "loss": 0.2972,
      "step": 1349
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5526398003895721,
      "learning_rate": 1.999326641958502e-05,
      "loss": 0.3069,
      "step": 1350
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6983541088388998,
      "learning_rate": 1.999322997677992e-05,
      "loss": 0.4293,
      "step": 1351
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2538335881585465,
      "learning_rate": 1.9993193435658136e-05,
      "loss": 0.3707,
      "step": 1352
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0878358807361594,
      "learning_rate": 1.9993156796220027e-05,
      "loss": 0.4208,
      "step": 1353
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4657145157414994,
      "learning_rate": 1.999312005846595e-05,
      "loss": 0.2465,
      "step": 1354
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3854343524089994,
      "learning_rate": 1.9993083222396273e-05,
      "loss": 0.4846,
      "step": 1355
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7924068242457152,
      "learning_rate": 1.9993046288011354e-05,
      "loss": 0.3954,
      "step": 1356
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.390245703152429,
      "learning_rate": 1.9993009255311554e-05,
      "loss": 0.1669,
      "step": 1357
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5477425936721658,
      "learning_rate": 1.9992972124297247e-05,
      "loss": 0.2954,
      "step": 1358
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6607006333281212,
      "learning_rate": 1.9992934894968787e-05,
      "loss": 0.2986,
      "step": 1359
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.960296233522704,
      "learning_rate": 1.999289756732655e-05,
      "loss": 0.9019,
      "step": 1360
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2988839771755245,
      "learning_rate": 1.9992860141370897e-05,
      "loss": 0.425,
      "step": 1361
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5879168618333596,
      "learning_rate": 1.9992822617102194e-05,
      "loss": 0.2827,
      "step": 1362
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5326595041770746,
      "learning_rate": 1.9992784994520817e-05,
      "loss": 0.2622,
      "step": 1363
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6648665462142312,
      "learning_rate": 1.9992747273627134e-05,
      "loss": 0.4588,
      "step": 1364
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9277622257431031,
      "learning_rate": 1.9992709454421515e-05,
      "loss": 0.4898,
      "step": 1365
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6678087310334865,
      "learning_rate": 1.9992671536904332e-05,
      "loss": 0.2956,
      "step": 1366
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3950691238746098,
      "learning_rate": 1.9992633521075956e-05,
      "loss": 0.0814,
      "step": 1367
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6282032493323312,
      "learning_rate": 1.9992595406936765e-05,
      "loss": 0.3028,
      "step": 1368
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6319309162774499,
      "learning_rate": 1.9992557194487135e-05,
      "loss": 0.3645,
      "step": 1369
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47232542943375677,
      "learning_rate": 1.9992518883727436e-05,
      "loss": 0.2204,
      "step": 1370
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1356794670049817,
      "learning_rate": 1.999248047465805e-05,
      "loss": 0.6695,
      "step": 1371
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4380156958966524,
      "learning_rate": 1.9992441967279355e-05,
      "loss": 0.2247,
      "step": 1372
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9047224043182277,
      "learning_rate": 1.9992403361591725e-05,
      "loss": 0.6097,
      "step": 1373
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4151372988862392,
      "learning_rate": 1.9992364657595543e-05,
      "loss": 0.2886,
      "step": 1374
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6566791944835109,
      "learning_rate": 1.9992325855291187e-05,
      "loss": 0.7863,
      "step": 1375
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4005692994139296,
      "learning_rate": 1.9992286954679048e-05,
      "loss": 0.1084,
      "step": 1376
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5838936981851108,
      "learning_rate": 1.9992247955759496e-05,
      "loss": 0.3429,
      "step": 1377
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.32620554257298007,
      "learning_rate": 1.9992208858532924e-05,
      "loss": 0.1355,
      "step": 1378
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.737849617020263,
      "learning_rate": 1.9992169662999715e-05,
      "loss": 0.6509,
      "step": 1379
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4948337782420355,
      "learning_rate": 1.999213036916025e-05,
      "loss": 0.3015,
      "step": 1380
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5152799042525615,
      "learning_rate": 1.9992090977014917e-05,
      "loss": 0.2906,
      "step": 1381
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.16743160619535,
      "learning_rate": 1.9992051486564108e-05,
      "loss": 0.624,
      "step": 1382
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1694595000853276,
      "learning_rate": 1.9992011897808206e-05,
      "loss": 0.3405,
      "step": 1383
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7051979858432267,
      "learning_rate": 1.9991972210747606e-05,
      "loss": 0.7258,
      "step": 1384
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5088681658614501,
      "learning_rate": 1.9991932425382696e-05,
      "loss": 0.2101,
      "step": 1385
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.745453609318162,
      "learning_rate": 1.9991892541713866e-05,
      "loss": 0.3947,
      "step": 1386
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4888322689394611,
      "learning_rate": 1.9991852559741506e-05,
      "loss": 0.2853,
      "step": 1387
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4907416413267397,
      "learning_rate": 1.999181247946602e-05,
      "loss": 0.2623,
      "step": 1388
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8588233972115484,
      "learning_rate": 1.9991772300887788e-05,
      "loss": 0.4262,
      "step": 1389
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7360090381443061,
      "learning_rate": 1.9991732024007216e-05,
      "loss": 0.3464,
      "step": 1390
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7181604215201794,
      "learning_rate": 1.99916916488247e-05,
      "loss": 0.26,
      "step": 1391
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5099064944930383,
      "learning_rate": 1.9991651175340628e-05,
      "loss": 0.3799,
      "step": 1392
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.4931486524003608,
      "learning_rate": 1.9991610603555404e-05,
      "loss": 0.3414,
      "step": 1393
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7249048994852877,
      "learning_rate": 1.999156993346943e-05,
      "loss": 0.082,
      "step": 1394
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5737999721181831,
      "learning_rate": 1.9991529165083102e-05,
      "loss": 0.3562,
      "step": 1395
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.422708150529981,
      "learning_rate": 1.9991488298396823e-05,
      "loss": 0.2388,
      "step": 1396
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5794624342568249,
      "learning_rate": 1.9991447333410994e-05,
      "loss": 0.4258,
      "step": 1397
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4851722209839137,
      "learning_rate": 1.999140627012602e-05,
      "loss": 0.3243,
      "step": 1398
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5779596168327364,
      "learning_rate": 1.9991365108542302e-05,
      "loss": 0.2594,
      "step": 1399
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.456720729760566,
      "learning_rate": 1.9991323848660246e-05,
      "loss": 0.2139,
      "step": 1400
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.510828885427331,
      "learning_rate": 1.999128249048026e-05,
      "loss": 0.6601,
      "step": 1401
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.5349318463428898,
      "learning_rate": 1.999124103400275e-05,
      "loss": 0.3792,
      "step": 1402
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.092096852211758,
      "learning_rate": 1.9991199479228117e-05,
      "loss": 0.5388,
      "step": 1403
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4469761725948586,
      "learning_rate": 1.9991157826156785e-05,
      "loss": 0.2583,
      "step": 1404
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7069381465289465,
      "learning_rate": 1.999111607478915e-05,
      "loss": 0.4483,
      "step": 1405
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8076819726249251,
      "learning_rate": 1.9991074225125628e-05,
      "loss": 0.4722,
      "step": 1406
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3808045016054495,
      "learning_rate": 1.999103227716663e-05,
      "loss": 0.1988,
      "step": 1407
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5704068434835022,
      "learning_rate": 1.9990990230912567e-05,
      "loss": 0.3191,
      "step": 1408
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6974488975006417,
      "learning_rate": 1.999094808636386e-05,
      "loss": 0.3011,
      "step": 1409
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6562228996811768,
      "learning_rate": 1.999090584352091e-05,
      "loss": 0.4071,
      "step": 1410
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6415372579080091,
      "learning_rate": 1.999086350238415e-05,
      "loss": 0.3487,
      "step": 1411
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8556461034338626,
      "learning_rate": 1.999082106295398e-05,
      "loss": 0.3685,
      "step": 1412
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5740140873591976,
      "learning_rate": 1.999077852523083e-05,
      "loss": 0.2738,
      "step": 1413
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.2500061599752705,
      "learning_rate": 1.9990735889215114e-05,
      "loss": 0.6705,
      "step": 1414
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6206522677536283,
      "learning_rate": 1.9990693154907248e-05,
      "loss": 0.3901,
      "step": 1415
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8387212297025176,
      "learning_rate": 1.9990650322307656e-05,
      "loss": 0.4131,
      "step": 1416
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47733004953699365,
      "learning_rate": 1.9990607391416758e-05,
      "loss": 0.2089,
      "step": 1417
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4770585740072936,
      "learning_rate": 1.9990564362234978e-05,
      "loss": 0.2472,
      "step": 1418
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3676419084619386,
      "learning_rate": 1.999052123476274e-05,
      "loss": 0.1276,
      "step": 1419
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.7920339610339828,
      "learning_rate": 1.9990478009000465e-05,
      "loss": 0.6849,
      "step": 1420
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7514480668127466,
      "learning_rate": 1.9990434684948582e-05,
      "loss": 0.3994,
      "step": 1421
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4902980401353472,
      "learning_rate": 1.999039126260751e-05,
      "loss": 0.2344,
      "step": 1422
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6182073911706732,
      "learning_rate": 1.9990347741977686e-05,
      "loss": 0.4659,
      "step": 1423
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9404651953925338,
      "learning_rate": 1.9990304123059533e-05,
      "loss": 0.5211,
      "step": 1424
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8006646850973358,
      "learning_rate": 1.999026040585348e-05,
      "loss": 0.6672,
      "step": 1425
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3292216857526261,
      "learning_rate": 1.999021659035996e-05,
      "loss": 0.1474,
      "step": 1426
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5285047177361533,
      "learning_rate": 1.99901726765794e-05,
      "loss": 0.2436,
      "step": 1427
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5747279884033586,
      "learning_rate": 1.9990128664512235e-05,
      "loss": 0.2722,
      "step": 1428
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.6464527301704055,
      "learning_rate": 1.9990084554158895e-05,
      "loss": 0.7208,
      "step": 1429
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.9966590428108473,
      "learning_rate": 1.9990040345519816e-05,
      "loss": 0.3411,
      "step": 1430
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5266683458265824,
      "learning_rate": 1.9989996038595435e-05,
      "loss": 0.3137,
      "step": 1431
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.259668404958786,
      "learning_rate": 1.9989951633386185e-05,
      "loss": 0.5224,
      "step": 1432
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6739413212399277,
      "learning_rate": 1.9989907129892505e-05,
      "loss": 0.3193,
      "step": 1433
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6134639657578796,
      "learning_rate": 1.998986252811483e-05,
      "loss": 0.3843,
      "step": 1434
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.47047694793935957,
      "learning_rate": 1.99898178280536e-05,
      "loss": 0.1998,
      "step": 1435
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4012626239291511,
      "learning_rate": 1.998977302970926e-05,
      "loss": 0.2043,
      "step": 1436
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.7107491321511357,
      "learning_rate": 1.9989728133082238e-05,
      "loss": 0.1939,
      "step": 1437
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.8879016702564722,
      "learning_rate": 1.998968313817299e-05,
      "loss": 0.9656,
      "step": 1438
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4521068910844738,
      "learning_rate": 1.998963804498195e-05,
      "loss": 0.3021,
      "step": 1439
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5452619109844327,
      "learning_rate": 1.9989592853509562e-05,
      "loss": 0.3076,
      "step": 1440
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1160803076924084,
      "learning_rate": 1.9989547563756274e-05,
      "loss": 0.4894,
      "step": 1441
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0392863073596246,
      "learning_rate": 1.998950217572253e-05,
      "loss": 0.6466,
      "step": 1442
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5779224227634603,
      "learning_rate": 1.9989456689408778e-05,
      "loss": 0.1268,
      "step": 1443
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6323390268331,
      "learning_rate": 1.998941110481546e-05,
      "loss": 0.3198,
      "step": 1444
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.43794900485947524,
      "learning_rate": 1.9989365421943035e-05,
      "loss": 0.1696,
      "step": 1445
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5075811897411316,
      "learning_rate": 1.998931964079194e-05,
      "loss": 0.3066,
      "step": 1446
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0385450343070224,
      "learning_rate": 1.9989273761362634e-05,
      "loss": 0.6377,
      "step": 1447
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0047209600302134,
      "learning_rate": 1.9989227783655566e-05,
      "loss": 0.4795,
      "step": 1448
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.612294931776603,
      "learning_rate": 1.998918170767119e-05,
      "loss": 0.2825,
      "step": 1449
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5409836260746129,
      "learning_rate": 1.9989135533409953e-05,
      "loss": 0.3703,
      "step": 1450
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6386978309838058,
      "learning_rate": 1.9989089260872315e-05,
      "loss": 0.3962,
      "step": 1451
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6351723112016172,
      "learning_rate": 1.998904289005873e-05,
      "loss": 0.2744,
      "step": 1452
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.3458048871644004,
      "learning_rate": 1.9988996420969656e-05,
      "loss": 0.472,
      "step": 1453
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3869132563009714,
      "learning_rate": 1.9988949853605552e-05,
      "loss": 0.1936,
      "step": 1454
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.7231306387180854,
      "learning_rate": 1.9988903187966866e-05,
      "loss": 0.3158,
      "step": 1455
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.027981347640159,
      "learning_rate": 1.9988856424054066e-05,
      "loss": 0.5094,
      "step": 1456
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.539806079361391,
      "learning_rate": 1.998880956186761e-05,
      "loss": 0.3698,
      "step": 1457
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4959298899142651,
      "learning_rate": 1.9988762601407958e-05,
      "loss": 0.2152,
      "step": 1458
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.6452485261630801,
      "learning_rate": 1.9988715542675573e-05,
      "loss": 0.4954,
      "step": 1459
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1933569384255303,
      "learning_rate": 1.998866838567092e-05,
      "loss": 0.8821,
      "step": 1460
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.1651350084275311,
      "learning_rate": 1.9988621130394455e-05,
      "loss": 0.4096,
      "step": 1461
    },
    {
      "epoch": 0.04,
      "grad_norm": 1.0399146680570537,
      "learning_rate": 1.9988573776846655e-05,
      "loss": 0.3195,
      "step": 1462
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.3919313966134844,
      "learning_rate": 1.9988526325027976e-05,
      "loss": 0.237,
      "step": 1463
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8072461395108451,
      "learning_rate": 1.998847877493889e-05,
      "loss": 0.3975,
      "step": 1464
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.4630135135500908,
      "learning_rate": 1.9988431126579863e-05,
      "loss": 0.264,
      "step": 1465
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.8655802302446115,
      "learning_rate": 1.998838337995136e-05,
      "loss": 0.5267,
      "step": 1466
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.593366038298171,
      "learning_rate": 1.998833553505386e-05,
      "loss": 0.2284,
      "step": 1467
    },
    {
      "epoch": 0.04,
      "grad_norm": 2.1045222590486112,
      "learning_rate": 1.9988287591887827e-05,
      "loss": 0.8352,
      "step": 1468
    },
    {
      "epoch": 0.04,
      "grad_norm": 0.5945837893036405,
      "learning_rate": 1.9988239550453735e-05,
      "loss": 0.303,
      "step": 1469
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.684881371828717,
      "learning_rate": 1.998819141075205e-05,
      "loss": 0.4152,
      "step": 1470
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0481414572120027,
      "learning_rate": 1.998814317278326e-05,
      "loss": 0.0947,
      "step": 1471
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6194733909834154,
      "learning_rate": 1.9988094836547824e-05,
      "loss": 0.2824,
      "step": 1472
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.37927079502765476,
      "learning_rate": 1.998804640204623e-05,
      "loss": 0.2024,
      "step": 1473
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5069073462057333,
      "learning_rate": 1.9987997869278947e-05,
      "loss": 0.3897,
      "step": 1474
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.621456540751678,
      "learning_rate": 1.998794923824645e-05,
      "loss": 0.3619,
      "step": 1475
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6365306940280037,
      "learning_rate": 1.998790050894923e-05,
      "loss": 0.1969,
      "step": 1476
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7768137792404145,
      "learning_rate": 1.9987851681387748e-05,
      "loss": 0.4252,
      "step": 1477
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3038623938779976,
      "learning_rate": 1.99878027555625e-05,
      "loss": 0.5268,
      "step": 1478
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5231912047089917,
      "learning_rate": 1.9987753731473963e-05,
      "loss": 0.6795,
      "step": 1479
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7809770446721771,
      "learning_rate": 1.9987704609122622e-05,
      "loss": 0.2697,
      "step": 1480
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5656075379831901,
      "learning_rate": 1.998765538850895e-05,
      "loss": 0.3233,
      "step": 1481
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5011424552597049,
      "learning_rate": 1.998760606963344e-05,
      "loss": 0.2782,
      "step": 1482
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9755366486610081,
      "learning_rate": 1.998755665249658e-05,
      "loss": 0.589,
      "step": 1483
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5462305232576065,
      "learning_rate": 1.9987507137098846e-05,
      "loss": 0.2273,
      "step": 1484
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5252441637786375,
      "learning_rate": 1.998745752344073e-05,
      "loss": 0.218,
      "step": 1485
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7946785162378907,
      "learning_rate": 1.998740781152272e-05,
      "loss": 0.426,
      "step": 1486
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6224876696752226,
      "learning_rate": 1.9987358001345308e-05,
      "loss": 0.2816,
      "step": 1487
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6789877563560681,
      "learning_rate": 1.998730809290898e-05,
      "loss": 0.3496,
      "step": 1488
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5842365377627423,
      "learning_rate": 1.998725808621423e-05,
      "loss": 0.0772,
      "step": 1489
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6136052593861542,
      "learning_rate": 1.998720798126155e-05,
      "loss": 0.3504,
      "step": 1490
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0263477917144763,
      "learning_rate": 1.998715777805143e-05,
      "loss": 0.458,
      "step": 1491
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5878626154338603,
      "learning_rate": 1.9987107476584368e-05,
      "loss": 0.3815,
      "step": 1492
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3825198943822507,
      "learning_rate": 1.9987057076860853e-05,
      "loss": 0.2296,
      "step": 1493
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5834093701307811,
      "learning_rate": 1.9987006578881388e-05,
      "loss": 0.2782,
      "step": 1494
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6839845552721493,
      "learning_rate": 1.9986955982646463e-05,
      "loss": 0.291,
      "step": 1495
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6284141796042169,
      "learning_rate": 1.998690528815658e-05,
      "loss": 0.7975,
      "step": 1496
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2509906619988524,
      "learning_rate": 1.9986854495412235e-05,
      "loss": 0.4374,
      "step": 1497
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4921841838174593,
      "learning_rate": 1.998680360441393e-05,
      "loss": 0.3064,
      "step": 1498
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.637355315468698,
      "learning_rate": 1.998675261516217e-05,
      "loss": 0.324,
      "step": 1499
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5367375950612352,
      "learning_rate": 1.9986701527657446e-05,
      "loss": 0.3339,
      "step": 1500
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1011920886851114,
      "learning_rate": 1.9986650341900268e-05,
      "loss": 0.5832,
      "step": 1501
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40209519922617376,
      "learning_rate": 1.9986599057891138e-05,
      "loss": 0.1395,
      "step": 1502
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9225006265013247,
      "learning_rate": 1.998654767563056e-05,
      "loss": 0.3514,
      "step": 1503
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5735421601071174,
      "learning_rate": 1.9986496195119042e-05,
      "loss": 0.2823,
      "step": 1504
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4396415988426068,
      "learning_rate": 1.9986444616357086e-05,
      "loss": 0.302,
      "step": 1505
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8314793544761525,
      "learning_rate": 1.99863929393452e-05,
      "loss": 0.439,
      "step": 1506
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3546700320575795,
      "learning_rate": 1.9986341164083896e-05,
      "loss": 0.6709,
      "step": 1507
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41551109973581385,
      "learning_rate": 1.9986289290573682e-05,
      "loss": 0.2233,
      "step": 1508
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9164902487606993,
      "learning_rate": 1.998623731881507e-05,
      "loss": 0.5643,
      "step": 1509
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3209381606604187,
      "learning_rate": 1.9986185248808567e-05,
      "loss": 0.365,
      "step": 1510
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5511719256815497,
      "learning_rate": 1.9986133080554692e-05,
      "loss": 0.3384,
      "step": 1511
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5819786813720418,
      "learning_rate": 1.998608081405395e-05,
      "loss": 0.1806,
      "step": 1512
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5808636513085695,
      "learning_rate": 1.9986028449306856e-05,
      "loss": 0.2832,
      "step": 1513
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6584873994761156,
      "learning_rate": 1.9985975986313932e-05,
      "loss": 0.2679,
      "step": 1514
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6395581950049153,
      "learning_rate": 1.998592342507569e-05,
      "loss": 0.2455,
      "step": 1515
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5277732779829133,
      "learning_rate": 1.998587076559265e-05,
      "loss": 0.4039,
      "step": 1516
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5674202790426692,
      "learning_rate": 1.9985818007865325e-05,
      "loss": 0.2192,
      "step": 1517
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2149747591129203,
      "learning_rate": 1.9985765151894236e-05,
      "loss": 0.6357,
      "step": 1518
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0012566036359238,
      "learning_rate": 1.9985712197679906e-05,
      "loss": 0.4857,
      "step": 1519
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5102873689771588,
      "learning_rate": 1.9985659145222852e-05,
      "loss": 0.7064,
      "step": 1520
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5655153644368442,
      "learning_rate": 1.9985605994523597e-05,
      "loss": 0.2292,
      "step": 1521
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8183887273819932,
      "learning_rate": 1.9985552745582665e-05,
      "loss": 0.415,
      "step": 1522
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3992713048480554,
      "learning_rate": 1.9985499398400582e-05,
      "loss": 0.2267,
      "step": 1523
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6547015999643384,
      "learning_rate": 1.9985445952977866e-05,
      "loss": 0.3055,
      "step": 1524
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1190901167622174,
      "learning_rate": 1.9985392409315053e-05,
      "loss": 0.4406,
      "step": 1525
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5940609750688771,
      "learning_rate": 1.9985338767412657e-05,
      "loss": 0.2304,
      "step": 1526
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8267248300951409,
      "learning_rate": 1.9985285027271217e-05,
      "loss": 0.3753,
      "step": 1527
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5780613975768399,
      "learning_rate": 1.9985231188891255e-05,
      "loss": 0.33,
      "step": 1528
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7099850731547875,
      "learning_rate": 1.9985177252273303e-05,
      "loss": 0.4459,
      "step": 1529
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6449532699025934,
      "learning_rate": 1.9985123217417893e-05,
      "loss": 0.1693,
      "step": 1530
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.52050550727405,
      "learning_rate": 1.9985069084325557e-05,
      "loss": 0.3246,
      "step": 1531
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49991267761328895,
      "learning_rate": 1.9985014852996826e-05,
      "loss": 0.1248,
      "step": 1532
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9506825539328038,
      "learning_rate": 1.998496052343223e-05,
      "loss": 0.5688,
      "step": 1533
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5583341020254391,
      "learning_rate": 1.9984906095632305e-05,
      "loss": 0.3297,
      "step": 1534
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6264401005703292,
      "learning_rate": 1.9984851569597593e-05,
      "loss": 0.276,
      "step": 1535
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.475547095905338,
      "learning_rate": 1.9984796945328625e-05,
      "loss": 0.2848,
      "step": 1536
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.490669745592488,
      "learning_rate": 1.9984742222825935e-05,
      "loss": 0.7078,
      "step": 1537
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.31930927870101,
      "learning_rate": 1.998468740209007e-05,
      "loss": 0.462,
      "step": 1538
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6248756506244505,
      "learning_rate": 1.9984632483121565e-05,
      "loss": 0.1821,
      "step": 1539
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6334634758399651,
      "learning_rate": 1.998457746592096e-05,
      "loss": 0.335,
      "step": 1540
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5365104864345774,
      "learning_rate": 1.9984522350488795e-05,
      "loss": 0.2784,
      "step": 1541
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5928003053473593,
      "learning_rate": 1.9984467136825614e-05,
      "loss": 0.3315,
      "step": 1542
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5968530472781753,
      "learning_rate": 1.998441182493196e-05,
      "loss": 0.2802,
      "step": 1543
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.733892466028286,
      "learning_rate": 1.9984356414808376e-05,
      "loss": 0.3073,
      "step": 1544
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.626699307818908,
      "learning_rate": 1.9984300906455413e-05,
      "loss": 0.2498,
      "step": 1545
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8510655910740117,
      "learning_rate": 1.9984245299873608e-05,
      "loss": 0.7071,
      "step": 1546
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5040176394330041,
      "learning_rate": 1.9984189595063516e-05,
      "loss": 0.3207,
      "step": 1547
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0312506015936036,
      "learning_rate": 1.9984133792025677e-05,
      "loss": 0.5132,
      "step": 1548
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5805832500499607,
      "learning_rate": 1.9984077890760648e-05,
      "loss": 0.2192,
      "step": 1549
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6052039950088977,
      "learning_rate": 1.9984021891268973e-05,
      "loss": 0.4164,
      "step": 1550
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8225163274961733,
      "learning_rate": 1.9983965793551207e-05,
      "loss": 0.4726,
      "step": 1551
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.37316619995900957,
      "learning_rate": 1.9983909597607902e-05,
      "loss": 0.2102,
      "step": 1552
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3409700697457951,
      "learning_rate": 1.9983853303439607e-05,
      "loss": 0.4044,
      "step": 1553
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38597487574248435,
      "learning_rate": 1.998379691104688e-05,
      "loss": 0.2041,
      "step": 1554
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4181155477263008,
      "learning_rate": 1.998374042043027e-05,
      "loss": 0.7707,
      "step": 1555
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3489201694265365,
      "learning_rate": 1.998368383159034e-05,
      "loss": 0.614,
      "step": 1556
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9068232979377222,
      "learning_rate": 1.9983627144527643e-05,
      "loss": 0.7897,
      "step": 1557
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5032481357977806,
      "learning_rate": 1.9983570359242734e-05,
      "loss": 0.2321,
      "step": 1558
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5217874372085312,
      "learning_rate": 1.998351347573618e-05,
      "loss": 0.3759,
      "step": 1559
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7894408253689438,
      "learning_rate": 1.9983456494008533e-05,
      "loss": 0.4763,
      "step": 1560
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7622822096212092,
      "learning_rate": 1.9983399414060353e-05,
      "loss": 0.7287,
      "step": 1561
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.436110628430791,
      "learning_rate": 1.998334223589221e-05,
      "loss": 0.0818,
      "step": 1562
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5605537767318407,
      "learning_rate": 1.9983284959504657e-05,
      "loss": 0.3316,
      "step": 1563
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5657481845347064,
      "learning_rate": 1.9983227584898262e-05,
      "loss": 0.3049,
      "step": 1564
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5213341201026311,
      "learning_rate": 1.9983170112073592e-05,
      "loss": 0.2968,
      "step": 1565
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.287161239448347,
      "learning_rate": 1.9983112541031207e-05,
      "loss": 0.5994,
      "step": 1566
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4288825631871544,
      "learning_rate": 1.9983054871771675e-05,
      "loss": 0.2846,
      "step": 1567
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.026888863925626,
      "learning_rate": 1.998299710429557e-05,
      "loss": 0.5726,
      "step": 1568
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.054404258343606,
      "learning_rate": 1.9982939238603447e-05,
      "loss": 0.4774,
      "step": 1569
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5062613280582487,
      "learning_rate": 1.9982881274695888e-05,
      "loss": 0.2977,
      "step": 1570
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.34328645396159185,
      "learning_rate": 1.9982823212573456e-05,
      "loss": 0.1572,
      "step": 1571
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47305760866476176,
      "learning_rate": 1.9982765052236724e-05,
      "loss": 0.2667,
      "step": 1572
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4605233265590172,
      "learning_rate": 1.9982706793686267e-05,
      "loss": 0.3837,
      "step": 1573
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.013896895022277,
      "learning_rate": 1.9982648436922656e-05,
      "loss": 0.9387,
      "step": 1574
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0650411170376208,
      "learning_rate": 1.9982589981946465e-05,
      "loss": 0.4672,
      "step": 1575
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2319234117209108,
      "learning_rate": 1.9982531428758265e-05,
      "loss": 0.4666,
      "step": 1576
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3871072547680524,
      "learning_rate": 1.998247277735864e-05,
      "loss": 0.2674,
      "step": 1577
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9637424427540703,
      "learning_rate": 1.998241402774816e-05,
      "loss": 0.5199,
      "step": 1578
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5690426395070233,
      "learning_rate": 1.998235517992741e-05,
      "loss": 0.2982,
      "step": 1579
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.25376475079702215,
      "learning_rate": 1.9982296233896963e-05,
      "loss": 0.0744,
      "step": 1580
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5270055686204602,
      "learning_rate": 1.9982237189657403e-05,
      "loss": 0.3336,
      "step": 1581
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42266322751158686,
      "learning_rate": 1.9982178047209306e-05,
      "loss": 0.2774,
      "step": 1582
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3956808401870453,
      "learning_rate": 1.998211880655326e-05,
      "loss": 0.7674,
      "step": 1583
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9033876020796671,
      "learning_rate": 1.9982059467689842e-05,
      "loss": 0.472,
      "step": 1584
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5540904460156968,
      "learning_rate": 1.998200003061964e-05,
      "loss": 0.2657,
      "step": 1585
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5147925315999351,
      "learning_rate": 1.9981940495343236e-05,
      "loss": 0.381,
      "step": 1586
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8238477348717592,
      "learning_rate": 1.9981880861861216e-05,
      "loss": 0.7797,
      "step": 1587
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46834156669898885,
      "learning_rate": 1.9981821130174168e-05,
      "loss": 0.2727,
      "step": 1588
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49639569388713456,
      "learning_rate": 1.9981761300282682e-05,
      "loss": 0.247,
      "step": 1589
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.35347611516369887,
      "learning_rate": 1.9981701372187342e-05,
      "loss": 0.1921,
      "step": 1590
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1188082498188028,
      "learning_rate": 1.9981641345888737e-05,
      "loss": 0.3473,
      "step": 1591
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2693010185095877,
      "learning_rate": 1.9981581221387467e-05,
      "loss": 0.6728,
      "step": 1592
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.570362117843228,
      "learning_rate": 1.998152099868411e-05,
      "loss": 0.3775,
      "step": 1593
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5425369484415422,
      "learning_rate": 1.9981460677779267e-05,
      "loss": 0.2813,
      "step": 1594
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5276954572937389,
      "learning_rate": 1.998140025867353e-05,
      "loss": 0.2822,
      "step": 1595
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.086247178890014,
      "learning_rate": 1.9981339741367493e-05,
      "loss": 0.8802,
      "step": 1596
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1969529607837128,
      "learning_rate": 1.998127912586175e-05,
      "loss": 0.4068,
      "step": 1597
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48356833378664077,
      "learning_rate": 1.9981218412156898e-05,
      "loss": 0.2164,
      "step": 1598
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6040239871163562,
      "learning_rate": 1.9981157600253537e-05,
      "loss": 0.2655,
      "step": 1599
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48094777713258585,
      "learning_rate": 1.9981096690152264e-05,
      "loss": 0.2598,
      "step": 1600
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5844886267774798,
      "learning_rate": 1.998103568185368e-05,
      "loss": 0.374,
      "step": 1601
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9696909117196728,
      "learning_rate": 1.9980974575358378e-05,
      "loss": 0.5761,
      "step": 1602
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4329322639705186,
      "learning_rate": 1.9980913370666963e-05,
      "loss": 0.0822,
      "step": 1603
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5204558252233074,
      "learning_rate": 1.998085206778004e-05,
      "loss": 0.2857,
      "step": 1604
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.367054073968383,
      "learning_rate": 1.9980790666698213e-05,
      "loss": 0.7196,
      "step": 1605
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4872412180797744,
      "learning_rate": 1.998072916742208e-05,
      "loss": 0.2708,
      "step": 1606
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8965230698972193,
      "learning_rate": 1.998066756995225e-05,
      "loss": 0.6697,
      "step": 1607
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.49362880307258333,
      "learning_rate": 1.9980605874289326e-05,
      "loss": 0.2218,
      "step": 1608
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41070736961049875,
      "learning_rate": 1.998054408043392e-05,
      "loss": 0.2085,
      "step": 1609
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.8653601996241491,
      "learning_rate": 1.998048218838664e-05,
      "loss": 0.5284,
      "step": 1610
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.0134606782289777,
      "learning_rate": 1.998042019814809e-05,
      "loss": 0.8191,
      "step": 1611
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41654227467856675,
      "learning_rate": 1.998035810971888e-05,
      "loss": 0.2296,
      "step": 1612
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5066370359818123,
      "learning_rate": 1.9980295923099625e-05,
      "loss": 0.3545,
      "step": 1613
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.249547610021564,
      "learning_rate": 1.9980233638290933e-05,
      "loss": 0.5083,
      "step": 1614
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.9267590521232179,
      "learning_rate": 1.9980171255293423e-05,
      "loss": 0.7351,
      "step": 1615
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.2806259700785918,
      "learning_rate": 1.9980108774107697e-05,
      "loss": 0.078,
      "step": 1616
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.33489913360739965,
      "learning_rate": 1.9980046194734383e-05,
      "loss": 0.1882,
      "step": 1617
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5444779249326102,
      "learning_rate": 1.9979983517174088e-05,
      "loss": 0.3689,
      "step": 1618
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1849193735401322,
      "learning_rate": 1.9979920741427434e-05,
      "loss": 0.4886,
      "step": 1619
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1447108588129047,
      "learning_rate": 1.997985786749503e-05,
      "loss": 0.6728,
      "step": 1620
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48433271672682043,
      "learning_rate": 1.9979794895377507e-05,
      "loss": 0.2138,
      "step": 1621
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7589646667882304,
      "learning_rate": 1.9979731825075474e-05,
      "loss": 0.4271,
      "step": 1622
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3091871367591976,
      "learning_rate": 1.9979668656589556e-05,
      "loss": 0.1246,
      "step": 1623
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5379355577595136,
      "learning_rate": 1.9979605389920377e-05,
      "loss": 0.359,
      "step": 1624
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0660293155049256,
      "learning_rate": 1.9979542025068556e-05,
      "loss": 0.2963,
      "step": 1625
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6406467008056842,
      "learning_rate": 1.9979478562034713e-05,
      "loss": 0.3308,
      "step": 1626
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6003683850188845,
      "learning_rate": 1.997941500081948e-05,
      "loss": 0.3754,
      "step": 1627
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6045857448077248,
      "learning_rate": 1.9979351341423477e-05,
      "loss": 0.3774,
      "step": 1628
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3508176656181679,
      "learning_rate": 1.997928758384733e-05,
      "loss": 0.2329,
      "step": 1629
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48914583847172083,
      "learning_rate": 1.9979223728091672e-05,
      "loss": 0.0805,
      "step": 1630
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5620624747031441,
      "learning_rate": 1.9979159774157124e-05,
      "loss": 0.3509,
      "step": 1631
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4611568250611382,
      "learning_rate": 1.997909572204432e-05,
      "loss": 0.5294,
      "step": 1632
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7909882909608017,
      "learning_rate": 1.9979031571753888e-05,
      "loss": 0.7975,
      "step": 1633
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.999232498067305,
      "learning_rate": 1.997896732328646e-05,
      "loss": 0.4713,
      "step": 1634
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5983166579247996,
      "learning_rate": 1.997890297664267e-05,
      "loss": 0.2836,
      "step": 1635
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45588364214846366,
      "learning_rate": 1.997883853182315e-05,
      "loss": 0.3092,
      "step": 1636
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5283425137148853,
      "learning_rate": 1.997877398882853e-05,
      "loss": 0.3367,
      "step": 1637
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38057281915121394,
      "learning_rate": 1.997870934765945e-05,
      "loss": 0.1146,
      "step": 1638
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1442068767197366,
      "learning_rate": 1.997864460831654e-05,
      "loss": 0.5594,
      "step": 1639
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45184796794555365,
      "learning_rate": 1.9978579770800444e-05,
      "loss": 0.251,
      "step": 1640
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5920980844260962,
      "learning_rate": 1.99785148351118e-05,
      "loss": 0.3849,
      "step": 1641
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5681649042360273,
      "learning_rate": 1.9978449801251236e-05,
      "loss": 0.2717,
      "step": 1642
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7205830766841989,
      "learning_rate": 1.9978384669219405e-05,
      "loss": 0.3667,
      "step": 1643
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5832042501612257,
      "learning_rate": 1.997831943901694e-05,
      "loss": 0.3316,
      "step": 1644
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9043569008702343,
      "learning_rate": 1.9978254110644487e-05,
      "loss": 0.4666,
      "step": 1645
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.7144173485416525,
      "learning_rate": 1.9978188684102685e-05,
      "loss": 0.7227,
      "step": 1646
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.38934922199428645,
      "learning_rate": 1.997812315939218e-05,
      "loss": 0.2274,
      "step": 1647
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8014488932848856,
      "learning_rate": 1.9978057536513617e-05,
      "loss": 0.3499,
      "step": 1648
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5449499392299983,
      "learning_rate": 1.997799181546764e-05,
      "loss": 0.2851,
      "step": 1649
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6492805372533652,
      "learning_rate": 1.9977925996254896e-05,
      "loss": 0.3092,
      "step": 1650
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2288755101336546,
      "learning_rate": 1.9977860078876035e-05,
      "loss": 0.4787,
      "step": 1651
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.921779568800309,
      "learning_rate": 1.9977794063331702e-05,
      "loss": 0.5259,
      "step": 1652
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4645458056004312,
      "learning_rate": 1.9977727949622546e-05,
      "loss": 0.1958,
      "step": 1653
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5073440904486292,
      "learning_rate": 1.997766173774922e-05,
      "loss": 0.3439,
      "step": 1654
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6002023603205643,
      "learning_rate": 1.9977595427712376e-05,
      "loss": 0.2734,
      "step": 1655
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4746279594648723,
      "learning_rate": 1.9977529019512663e-05,
      "loss": 0.1641,
      "step": 1656
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1963137838505264,
      "learning_rate": 1.997746251315074e-05,
      "loss": 0.5343,
      "step": 1657
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.507723211991171,
      "learning_rate": 1.9977395908627258e-05,
      "loss": 0.2405,
      "step": 1658
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.41473513996034544,
      "learning_rate": 1.997732920594287e-05,
      "loss": 0.2573,
      "step": 1659
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7226857066981354,
      "learning_rate": 1.9977262405098236e-05,
      "loss": 0.3731,
      "step": 1660
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2257655979979063,
      "learning_rate": 1.997719550609401e-05,
      "loss": 0.6474,
      "step": 1661
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42277361508043443,
      "learning_rate": 1.997712850893085e-05,
      "loss": 0.1974,
      "step": 1662
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5689713071626074,
      "learning_rate": 1.997706141360942e-05,
      "loss": 0.4181,
      "step": 1663
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9751572941350246,
      "learning_rate": 1.9976994220130374e-05,
      "loss": 0.4251,
      "step": 1664
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6889986444668514,
      "learning_rate": 1.997692692849438e-05,
      "loss": 0.3792,
      "step": 1665
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5395138879711248,
      "learning_rate": 1.997685953870209e-05,
      "loss": 0.2658,
      "step": 1666
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5038867159255613,
      "learning_rate": 1.9976792050754177e-05,
      "loss": 0.2932,
      "step": 1667
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.44128832311643074,
      "learning_rate": 1.99767244646513e-05,
      "loss": 0.0994,
      "step": 1668
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1668941783604931,
      "learning_rate": 1.9976656780394122e-05,
      "loss": 0.5063,
      "step": 1669
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0165405906555296,
      "learning_rate": 1.997658899798332e-05,
      "loss": 0.6118,
      "step": 1670
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3784760409018777,
      "learning_rate": 1.9976521117419544e-05,
      "loss": 0.226,
      "step": 1671
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5181282288961536,
      "learning_rate": 1.9976453138703477e-05,
      "loss": 0.3684,
      "step": 1672
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1589634434925724,
      "learning_rate": 1.9976385061835777e-05,
      "loss": 0.3514,
      "step": 1673
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.374336537790892,
      "learning_rate": 1.997631688681712e-05,
      "loss": 0.6879,
      "step": 1674
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.593848147430315,
      "learning_rate": 1.9976248613648175e-05,
      "loss": 0.1859,
      "step": 1675
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5896472337796588,
      "learning_rate": 1.9976180242329613e-05,
      "loss": 0.3726,
      "step": 1676
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.703918460829394,
      "learning_rate": 1.9976111772862108e-05,
      "loss": 0.2214,
      "step": 1677
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5174678800895354,
      "learning_rate": 1.9976043205246333e-05,
      "loss": 0.376,
      "step": 1678
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5740121800898227,
      "learning_rate": 1.9975974539482964e-05,
      "loss": 0.2726,
      "step": 1679
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5373628935541511,
      "learning_rate": 1.997590577557267e-05,
      "loss": 0.2848,
      "step": 1680
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6547858956960634,
      "learning_rate": 1.9975836913516134e-05,
      "loss": 0.255,
      "step": 1681
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3996822874465955,
      "learning_rate": 1.9975767953314033e-05,
      "loss": 0.4918,
      "step": 1682
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47372024217497527,
      "learning_rate": 1.9975698894967044e-05,
      "loss": 0.3402,
      "step": 1683
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7163129820026878,
      "learning_rate": 1.9975629738475848e-05,
      "loss": 0.2462,
      "step": 1684
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5958232727898585,
      "learning_rate": 1.9975560483841123e-05,
      "loss": 0.3304,
      "step": 1685
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.40104025706757324,
      "learning_rate": 1.9975491131063553e-05,
      "loss": 0.2363,
      "step": 1686
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.758423643483129,
      "learning_rate": 1.9975421680143817e-05,
      "loss": 0.5465,
      "step": 1687
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46929339827470673,
      "learning_rate": 1.99753521310826e-05,
      "loss": 0.1293,
      "step": 1688
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4638358607069192,
      "learning_rate": 1.997528248388059e-05,
      "loss": 0.3184,
      "step": 1689
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43916460031500065,
      "learning_rate": 1.9975212738538466e-05,
      "loss": 0.2388,
      "step": 1690
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3848051474981304,
      "learning_rate": 1.9975142895056916e-05,
      "loss": 0.678,
      "step": 1691
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2279007948153926,
      "learning_rate": 1.997507295343663e-05,
      "loss": 0.552,
      "step": 1692
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4185113645760292,
      "learning_rate": 1.9975002913678292e-05,
      "loss": 0.7594,
      "step": 1693
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45225253850627417,
      "learning_rate": 1.9974932775782594e-05,
      "loss": 0.2246,
      "step": 1694
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5451327004288552,
      "learning_rate": 1.9974862539750224e-05,
      "loss": 0.4269,
      "step": 1695
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5578968404069613,
      "learning_rate": 1.9974792205581878e-05,
      "loss": 0.3308,
      "step": 1696
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46837785364679485,
      "learning_rate": 1.997472177327824e-05,
      "loss": 0.15,
      "step": 1697
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.871192980545675,
      "learning_rate": 1.997465124284001e-05,
      "loss": 0.3435,
      "step": 1698
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5529480414985452,
      "learning_rate": 1.9974580614267875e-05,
      "loss": 0.2525,
      "step": 1699
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.4482650840025277,
      "learning_rate": 1.9974509887562537e-05,
      "loss": 0.7534,
      "step": 1700
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48396556300278976,
      "learning_rate": 1.997443906272469e-05,
      "loss": 0.3004,
      "step": 1701
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0948064897772298,
      "learning_rate": 1.9974368139755028e-05,
      "loss": 0.6495,
      "step": 1702
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4928647808549902,
      "learning_rate": 1.997429711865425e-05,
      "loss": 0.2188,
      "step": 1703
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0987928566199097,
      "learning_rate": 1.9974225999423052e-05,
      "loss": 0.6634,
      "step": 1704
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6798700166129977,
      "learning_rate": 1.997415478206214e-05,
      "loss": 0.4415,
      "step": 1705
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43808215398363715,
      "learning_rate": 1.9974083466572208e-05,
      "loss": 0.2194,
      "step": 1706
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.437620312474001,
      "learning_rate": 1.9974012052953967e-05,
      "loss": 0.2318,
      "step": 1707
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7018590796368072,
      "learning_rate": 1.997394054120811e-05,
      "loss": 0.4148,
      "step": 1708
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1194844799009713,
      "learning_rate": 1.9973868931335346e-05,
      "loss": 0.4581,
      "step": 1709
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.569484980057481,
      "learning_rate": 1.9973797223336374e-05,
      "loss": 0.7139,
      "step": 1710
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8484330422462553,
      "learning_rate": 1.9973725417211906e-05,
      "loss": 0.4957,
      "step": 1711
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5598966826531686,
      "learning_rate": 1.9973653512962644e-05,
      "loss": 0.2205,
      "step": 1712
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4834715259876392,
      "learning_rate": 1.99735815105893e-05,
      "loss": 0.3541,
      "step": 1713
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5374822262178063,
      "learning_rate": 1.9973509410092573e-05,
      "loss": 0.3373,
      "step": 1714
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5911902670602617,
      "learning_rate": 1.9973437211473182e-05,
      "loss": 0.2582,
      "step": 1715
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3499141437861104,
      "learning_rate": 1.9973364914731838e-05,
      "loss": 0.082,
      "step": 1716
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4780785349002867,
      "learning_rate": 1.997329251986924e-05,
      "loss": 0.3044,
      "step": 1717
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.6076643204939705,
      "learning_rate": 1.997322002688612e-05,
      "loss": 0.5901,
      "step": 1718
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5253111068277259,
      "learning_rate": 1.997314743578317e-05,
      "loss": 0.3673,
      "step": 1719
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8048695441058147,
      "learning_rate": 1.9973074746561117e-05,
      "loss": 0.4691,
      "step": 1720
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.765610055237347,
      "learning_rate": 1.997300195922067e-05,
      "loss": 0.3806,
      "step": 1721
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4978203336396717,
      "learning_rate": 1.9972929073762556e-05,
      "loss": 0.3275,
      "step": 1722
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.700083821242781,
      "learning_rate": 1.9972856090187477e-05,
      "loss": 0.4767,
      "step": 1723
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6379352957401104,
      "learning_rate": 1.997278300849616e-05,
      "loss": 0.3952,
      "step": 1724
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.34078001505205346,
      "learning_rate": 1.9972709828689325e-05,
      "loss": 0.1587,
      "step": 1725
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4189473867653872,
      "learning_rate": 1.9972636550767688e-05,
      "loss": 0.2609,
      "step": 1726
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5813642226451838,
      "learning_rate": 1.9972563174731968e-05,
      "loss": 0.3616,
      "step": 1727
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5412248850395818,
      "learning_rate": 1.9972489700582894e-05,
      "loss": 0.6957,
      "step": 1728
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9200951149859589,
      "learning_rate": 1.9972416128321184e-05,
      "loss": 0.4925,
      "step": 1729
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.525068448147769,
      "learning_rate": 1.9972342457947563e-05,
      "loss": 0.3092,
      "step": 1730
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.46551996399511353,
      "learning_rate": 1.9972268689462757e-05,
      "loss": 0.2732,
      "step": 1731
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5692204568640107,
      "learning_rate": 1.997219482286749e-05,
      "loss": 0.3944,
      "step": 1732
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39734215558660674,
      "learning_rate": 1.9972120858162482e-05,
      "loss": 0.1383,
      "step": 1733
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5299671911552777,
      "learning_rate": 1.9972046795348478e-05,
      "loss": 0.2783,
      "step": 1734
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5622312775831285,
      "learning_rate": 1.9971972634426188e-05,
      "loss": 0.2205,
      "step": 1735
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6893847413914425,
      "learning_rate": 1.9971898375396355e-05,
      "loss": 0.3182,
      "step": 1736
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.48900992853105435,
      "learning_rate": 1.9971824018259702e-05,
      "loss": 0.3747,
      "step": 1737
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5484355250371853,
      "learning_rate": 1.9971749563016964e-05,
      "loss": 0.539,
      "step": 1738
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.8402639659331204,
      "learning_rate": 1.9971675009668873e-05,
      "loss": 0.2743,
      "step": 1739
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5110540106649946,
      "learning_rate": 1.997160035821616e-05,
      "loss": 0.2843,
      "step": 1740
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.2143566062433921,
      "learning_rate": 1.997152560865957e-05,
      "loss": 0.6552,
      "step": 1741
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.47028237985456783,
      "learning_rate": 1.997145076099982e-05,
      "loss": 0.2904,
      "step": 1742
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6674003847105897,
      "learning_rate": 1.9971375815237663e-05,
      "loss": 0.2913,
      "step": 1743
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.43652258888384377,
      "learning_rate": 1.9971300771373823e-05,
      "loss": 0.2295,
      "step": 1744
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6242438749592722,
      "learning_rate": 1.997122562940905e-05,
      "loss": 0.2905,
      "step": 1745
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3112807837479952,
      "learning_rate": 1.9971150389344077e-05,
      "loss": 0.5219,
      "step": 1746
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1846795693557104,
      "learning_rate": 1.9971075051179646e-05,
      "loss": 0.6862,
      "step": 1747
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.42485300155899175,
      "learning_rate": 1.9970999614916497e-05,
      "loss": 0.2118,
      "step": 1748
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5072430580499369,
      "learning_rate": 1.9970924080555376e-05,
      "loss": 0.2866,
      "step": 1749
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.292521574588829,
      "learning_rate": 1.9970848448097014e-05,
      "loss": 0.8327,
      "step": 1750
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1536374661683024,
      "learning_rate": 1.9970772717542174e-05,
      "loss": 0.0659,
      "step": 1751
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.782461079058697,
      "learning_rate": 1.9970696888891585e-05,
      "loss": 0.1797,
      "step": 1752
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5512510457187575,
      "learning_rate": 1.9970620962146005e-05,
      "loss": 0.2804,
      "step": 1753
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3028425073789085,
      "learning_rate": 1.997054493730617e-05,
      "loss": 0.6591,
      "step": 1754
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5891816739456746,
      "learning_rate": 1.9970468814372834e-05,
      "loss": 0.3517,
      "step": 1755
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.31172765262477,
      "learning_rate": 1.9970392593346748e-05,
      "loss": 0.7148,
      "step": 1756
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4933784766821851,
      "learning_rate": 1.9970316274228656e-05,
      "loss": 0.2113,
      "step": 1757
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.7243497410602027,
      "learning_rate": 1.9970239857019314e-05,
      "loss": 0.393,
      "step": 1758
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.9539481216013089,
      "learning_rate": 1.997016334171947e-05,
      "loss": 0.4717,
      "step": 1759
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6413195734195973,
      "learning_rate": 1.9970086728329878e-05,
      "loss": 0.3638,
      "step": 1760
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.39499767371724304,
      "learning_rate": 1.9970010016851295e-05,
      "loss": 0.196,
      "step": 1761
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5504598821543647,
      "learning_rate": 1.996993320728447e-05,
      "loss": 0.218,
      "step": 1762
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6137145043472252,
      "learning_rate": 1.9969856299630163e-05,
      "loss": 0.3033,
      "step": 1763
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.5681665607855801,
      "learning_rate": 1.996977929388913e-05,
      "loss": 0.5262,
      "step": 1764
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6613542771819474,
      "learning_rate": 1.9969702190062127e-05,
      "loss": 0.4346,
      "step": 1765
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5076653971355554,
      "learning_rate": 1.9969624988149912e-05,
      "loss": 0.2574,
      "step": 1766
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.45629707859574625,
      "learning_rate": 1.9969547688153247e-05,
      "loss": 0.3042,
      "step": 1767
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.4481211269555483,
      "learning_rate": 1.996947029007289e-05,
      "loss": 0.4935,
      "step": 1768
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.3065026295756057,
      "learning_rate": 1.9969392793909606e-05,
      "loss": 0.6564,
      "step": 1769
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0761331033564878,
      "learning_rate": 1.9969315199664154e-05,
      "loss": 0.4493,
      "step": 1770
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5604601699054695,
      "learning_rate": 1.9969237507337302e-05,
      "loss": 0.2913,
      "step": 1771
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4821230670964276,
      "learning_rate": 1.9969159716929808e-05,
      "loss": 0.3453,
      "step": 1772
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3946686764168399,
      "learning_rate": 1.996908182844244e-05,
      "loss": 0.2749,
      "step": 1773
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.352437366195415,
      "learning_rate": 1.9969003841875966e-05,
      "loss": 0.5562,
      "step": 1774
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.35143993146907093,
      "learning_rate": 1.996892575723115e-05,
      "loss": 0.1814,
      "step": 1775
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4499142944338347,
      "learning_rate": 1.9968847574508766e-05,
      "loss": 0.2568,
      "step": 1776
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.1407313939742139,
      "learning_rate": 1.9968769293709575e-05,
      "loss": 0.3573,
      "step": 1777
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6496411572765395,
      "learning_rate": 1.9968690914834353e-05,
      "loss": 0.4224,
      "step": 1778
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5142026473571419,
      "learning_rate": 1.9968612437883874e-05,
      "loss": 0.3742,
      "step": 1779
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5675537074495811,
      "learning_rate": 1.99685338628589e-05,
      "loss": 0.2993,
      "step": 1780
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5437980158732774,
      "learning_rate": 1.9968455189760216e-05,
      "loss": 0.3311,
      "step": 1781
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5654344820737868,
      "learning_rate": 1.9968376418588583e-05,
      "loss": 0.3666,
      "step": 1782
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.567703564861966,
      "learning_rate": 1.9968297549344785e-05,
      "loss": 0.3292,
      "step": 1783
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5109623440514741,
      "learning_rate": 1.9968218582029598e-05,
      "loss": 0.3146,
      "step": 1784
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.53786838239359,
      "learning_rate": 1.9968139516643792e-05,
      "loss": 0.249,
      "step": 1785
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6295075662577703,
      "learning_rate": 1.996806035318815e-05,
      "loss": 0.3001,
      "step": 1786
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.0335764352121741,
      "learning_rate": 1.9967981091663457e-05,
      "loss": 0.4225,
      "step": 1787
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5955747639578913,
      "learning_rate": 1.996790173207048e-05,
      "loss": 0.4058,
      "step": 1788
    },
    {
      "epoch": 0.05,
      "grad_norm": 1.158035102700958,
      "learning_rate": 1.9967822274410007e-05,
      "loss": 0.4266,
      "step": 1789
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.4261367275268446,
      "learning_rate": 1.996774271868282e-05,
      "loss": 0.2969,
      "step": 1790
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.6592145571007315,
      "learning_rate": 1.99676630648897e-05,
      "loss": 0.4158,
      "step": 1791
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3868852682878928,
      "learning_rate": 1.996758331303143e-05,
      "loss": 0.1762,
      "step": 1792
    },
    {
      "epoch": 0.05,
      "grad_norm": 2.156446188416199,
      "learning_rate": 1.9967503463108796e-05,
      "loss": 0.7933,
      "step": 1793
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.3946052651710403,
      "learning_rate": 1.9967423515122583e-05,
      "loss": 0.2258,
      "step": 1794
    },
    {
      "epoch": 0.05,
      "grad_norm": 0.5500078765645028,
      "learning_rate": 1.9967343469073577e-05,
      "loss": 0.2654,
      "step": 1795
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44587516450320586,
      "learning_rate": 1.9967263324962567e-05,
      "loss": 0.3165,
      "step": 1796
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9644503479031603,
      "learning_rate": 1.9967183082790343e-05,
      "loss": 0.6203,
      "step": 1797
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5129189542856442,
      "learning_rate": 1.996710274255769e-05,
      "loss": 0.315,
      "step": 1798
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47020686154811503,
      "learning_rate": 1.99670223042654e-05,
      "loss": 0.3324,
      "step": 1799
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.39467671819943667,
      "learning_rate": 1.9966941767914265e-05,
      "loss": 0.1212,
      "step": 1800
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3523326457148532,
      "learning_rate": 1.9966861133505078e-05,
      "loss": 0.5412,
      "step": 1801
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43035920424261553,
      "learning_rate": 1.9966780401038634e-05,
      "loss": 0.3144,
      "step": 1802
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.404937911893826,
      "learning_rate": 1.996669957051572e-05,
      "loss": 0.2503,
      "step": 1803
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6548929692099172,
      "learning_rate": 1.9966618641937142e-05,
      "loss": 0.2765,
      "step": 1804
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9857532641330176,
      "learning_rate": 1.9966537615303683e-05,
      "loss": 0.4692,
      "step": 1805
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.801450929988938,
      "learning_rate": 1.9966456490616153e-05,
      "loss": 0.5715,
      "step": 1806
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4230299603537542,
      "learning_rate": 1.9966375267875343e-05,
      "loss": 0.1916,
      "step": 1807
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46080888441974416,
      "learning_rate": 1.9966293947082053e-05,
      "loss": 0.3023,
      "step": 1808
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5856506935547885,
      "learning_rate": 1.9966212528237086e-05,
      "loss": 0.3095,
      "step": 1809
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5065827150090665,
      "learning_rate": 1.996613101134124e-05,
      "loss": 0.7418,
      "step": 1810
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3719115341451015,
      "learning_rate": 1.9966049396395318e-05,
      "loss": 0.1311,
      "step": 1811
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6002229004328484,
      "learning_rate": 1.9965967683400122e-05,
      "loss": 0.3136,
      "step": 1812
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3471096428209259,
      "learning_rate": 1.9965885872356458e-05,
      "loss": 0.1922,
      "step": 1813
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8093784432055173,
      "learning_rate": 1.996580396326513e-05,
      "loss": 0.3442,
      "step": 1814
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0744331471828312,
      "learning_rate": 1.996572195612694e-05,
      "loss": 0.587,
      "step": 1815
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6670345024407223,
      "learning_rate": 1.99656398509427e-05,
      "loss": 0.089,
      "step": 1816
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5442785628400635,
      "learning_rate": 1.9965557647713216e-05,
      "loss": 0.3326,
      "step": 1817
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2383607367638152,
      "learning_rate": 1.9965475346439297e-05,
      "loss": 0.443,
      "step": 1818
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7034476034421672,
      "learning_rate": 1.996539294712175e-05,
      "loss": 0.4529,
      "step": 1819
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5187447181804192,
      "learning_rate": 1.9965310449761394e-05,
      "loss": 0.2641,
      "step": 1820
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6062217427100794,
      "learning_rate": 1.996522785435903e-05,
      "loss": 0.2979,
      "step": 1821
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38311247593127923,
      "learning_rate": 1.9965145160915477e-05,
      "loss": 0.2513,
      "step": 1822
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7394799082886382,
      "learning_rate": 1.996506236943155e-05,
      "loss": 0.5697,
      "step": 1823
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4744369606136454,
      "learning_rate": 1.9964979479908053e-05,
      "loss": 0.0772,
      "step": 1824
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5627045148736144,
      "learning_rate": 1.9964896492345814e-05,
      "loss": 0.3531,
      "step": 1825
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40839462867049703,
      "learning_rate": 1.9964813406745646e-05,
      "loss": 0.2434,
      "step": 1826
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.437575331331514,
      "learning_rate": 1.996473022310836e-05,
      "loss": 0.4448,
      "step": 1827
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.814769922481516,
      "learning_rate": 1.996464694143478e-05,
      "loss": 0.8898,
      "step": 1828
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7622853694547538,
      "learning_rate": 1.9964563561725726e-05,
      "loss": 0.4947,
      "step": 1829
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5795509321971768,
      "learning_rate": 1.996448008398202e-05,
      "loss": 0.2949,
      "step": 1830
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.614776762054701,
      "learning_rate": 1.9964396508204475e-05,
      "loss": 0.3729,
      "step": 1831
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5309030009374321,
      "learning_rate": 1.9964312834393922e-05,
      "loss": 0.3346,
      "step": 1832
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4967568008513103,
      "learning_rate": 1.9964229062551178e-05,
      "loss": 0.1577,
      "step": 1833
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0258167599847319,
      "learning_rate": 1.9964145192677072e-05,
      "loss": 0.4034,
      "step": 1834
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5731909197879361,
      "learning_rate": 1.9964061224772428e-05,
      "loss": 0.2657,
      "step": 1835
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6661121728861366,
      "learning_rate": 1.996397715883807e-05,
      "loss": 0.7277,
      "step": 1836
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5824028537246017,
      "learning_rate": 1.9963892994874822e-05,
      "loss": 0.3059,
      "step": 1837
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6679363822685329,
      "learning_rate": 1.996380873288352e-05,
      "loss": 0.4243,
      "step": 1838
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7605419289223462,
      "learning_rate": 1.996372437286499e-05,
      "loss": 0.0796,
      "step": 1839
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5840856319515085,
      "learning_rate": 1.9963639914820062e-05,
      "loss": 0.3308,
      "step": 1840
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9045107091311789,
      "learning_rate": 1.996355535874956e-05,
      "loss": 0.5536,
      "step": 1841
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.49160207412947887,
      "learning_rate": 1.996347070465433e-05,
      "loss": 0.0771,
      "step": 1842
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6570934150927423,
      "learning_rate": 1.996338595253519e-05,
      "loss": 0.3822,
      "step": 1843
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4119662566381829,
      "learning_rate": 1.9963301102392983e-05,
      "loss": 0.2277,
      "step": 1844
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5973462621471904,
      "learning_rate": 1.996321615422854e-05,
      "loss": 0.4121,
      "step": 1845
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1174512321204013,
      "learning_rate": 1.99631311080427e-05,
      "loss": 0.6879,
      "step": 1846
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.140903821402689,
      "learning_rate": 1.99630459638363e-05,
      "loss": 0.6142,
      "step": 1847
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4814933890761632,
      "learning_rate": 1.9962960721610172e-05,
      "loss": 0.2572,
      "step": 1848
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5698855329756009,
      "learning_rate": 1.996287538136516e-05,
      "loss": 0.4473,
      "step": 1849
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5123990730311891,
      "learning_rate": 1.99627899431021e-05,
      "loss": 0.2694,
      "step": 1850
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6919268697164719,
      "learning_rate": 1.9962704406821836e-05,
      "loss": 0.2444,
      "step": 1851
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45678419339215665,
      "learning_rate": 1.9962618772525205e-05,
      "loss": 0.1207,
      "step": 1852
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47842672189752783,
      "learning_rate": 1.9962533040213056e-05,
      "loss": 0.235,
      "step": 1853
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9117828195175315,
      "learning_rate": 1.9962447209886225e-05,
      "loss": 0.8196,
      "step": 1854
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5811448363113795,
      "learning_rate": 1.9962361281545566e-05,
      "loss": 0.261,
      "step": 1855
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.657507655445899,
      "learning_rate": 1.9962275255191916e-05,
      "loss": 0.4152,
      "step": 1856
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5396611031437624,
      "learning_rate": 1.9962189130826122e-05,
      "loss": 0.2666,
      "step": 1857
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7222697940525817,
      "learning_rate": 1.9962102908449035e-05,
      "loss": 0.4133,
      "step": 1858
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0490408401670306,
      "learning_rate": 1.9962016588061504e-05,
      "loss": 0.3273,
      "step": 1859
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5399829484219572,
      "learning_rate": 1.9961930169664374e-05,
      "loss": 0.207,
      "step": 1860
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40379624649052737,
      "learning_rate": 1.9961843653258494e-05,
      "loss": 0.256,
      "step": 1861
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5031200201029243,
      "learning_rate": 1.996175703884472e-05,
      "loss": 0.3122,
      "step": 1862
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1430586388823063,
      "learning_rate": 1.9961670326423903e-05,
      "loss": 0.3677,
      "step": 1863
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6841502731452473,
      "learning_rate": 1.9961583515996895e-05,
      "loss": 0.8314,
      "step": 1864
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.843865777505154,
      "learning_rate": 1.9961496607564554e-05,
      "loss": 0.4728,
      "step": 1865
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48768008749110275,
      "learning_rate": 1.9961409601127725e-05,
      "loss": 0.08,
      "step": 1866
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5997687691549242,
      "learning_rate": 1.9961322496687278e-05,
      "loss": 0.3383,
      "step": 1867
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48385749094011005,
      "learning_rate": 1.9961235294244057e-05,
      "loss": 0.2779,
      "step": 1868
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5631933578616771,
      "learning_rate": 1.9961147993798925e-05,
      "loss": 0.2917,
      "step": 1869
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.36361981486256834,
      "learning_rate": 1.9961060595352745e-05,
      "loss": 0.1295,
      "step": 1870
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5406810822107273,
      "learning_rate": 1.9960973098906374e-05,
      "loss": 0.326,
      "step": 1871
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6338428607053247,
      "learning_rate": 1.996088550446067e-05,
      "loss": 0.3016,
      "step": 1872
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4824579165374624,
      "learning_rate": 1.9960797812016496e-05,
      "loss": 0.3765,
      "step": 1873
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.213253049347204,
      "learning_rate": 1.9960710021574716e-05,
      "loss": 0.4721,
      "step": 1874
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8913974236024559,
      "learning_rate": 1.9960622133136195e-05,
      "loss": 0.5329,
      "step": 1875
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4628587743833688,
      "learning_rate": 1.9960534146701796e-05,
      "loss": 0.2519,
      "step": 1876
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.391803406530486,
      "learning_rate": 1.9960446062272384e-05,
      "loss": 0.6805,
      "step": 1877
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5918141270036241,
      "learning_rate": 1.9960357879848825e-05,
      "loss": 0.0796,
      "step": 1878
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4270828019245726,
      "learning_rate": 1.9960269599431984e-05,
      "loss": 0.2917,
      "step": 1879
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3913450990885546,
      "learning_rate": 1.996018122102274e-05,
      "loss": 0.2163,
      "step": 1880
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1775636825049893,
      "learning_rate": 1.996009274462195e-05,
      "loss": 0.5166,
      "step": 1881
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2522788325123209,
      "learning_rate": 1.9960004170230498e-05,
      "loss": 0.5418,
      "step": 1882
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7539122203761266,
      "learning_rate": 1.9959915497849245e-05,
      "loss": 0.493,
      "step": 1883
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9064053830790864,
      "learning_rate": 1.995982672747906e-05,
      "loss": 0.3392,
      "step": 1884
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3737733529376645,
      "learning_rate": 1.9959737859120828e-05,
      "loss": 0.2662,
      "step": 1885
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.63914810393596,
      "learning_rate": 1.995964889277542e-05,
      "loss": 0.3656,
      "step": 1886
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1846290134271449,
      "learning_rate": 1.9959559828443704e-05,
      "loss": 0.4468,
      "step": 1887
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5613080468069205,
      "learning_rate": 1.9959470666126563e-05,
      "loss": 0.1643,
      "step": 1888
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4764411733711985,
      "learning_rate": 1.9959381405824875e-05,
      "loss": 0.1986,
      "step": 1889
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7422308250754749,
      "learning_rate": 1.9959292047539513e-05,
      "loss": 0.4279,
      "step": 1890
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4593765443615965,
      "learning_rate": 1.9959202591271358e-05,
      "loss": 0.343,
      "step": 1891
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0258935176429367,
      "learning_rate": 1.995911303702129e-05,
      "loss": 0.5869,
      "step": 1892
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5737193475484144,
      "learning_rate": 1.9959023384790194e-05,
      "loss": 0.2517,
      "step": 1893
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4347712176492951,
      "learning_rate": 1.9958933634578947e-05,
      "loss": 0.2343,
      "step": 1894
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2143353690429004,
      "learning_rate": 1.9958843786388438e-05,
      "loss": 0.6714,
      "step": 1895
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40453426170281725,
      "learning_rate": 1.9958753840219544e-05,
      "loss": 0.2173,
      "step": 1896
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4744377815461775,
      "learning_rate": 1.995866379607315e-05,
      "loss": 0.2393,
      "step": 1897
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42483032352369565,
      "learning_rate": 1.995857365395015e-05,
      "loss": 0.1812,
      "step": 1898
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5937008733444495,
      "learning_rate": 1.995848341385142e-05,
      "loss": 0.4374,
      "step": 1899
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.662249964080098,
      "learning_rate": 1.9958393075777857e-05,
      "loss": 0.5014,
      "step": 1900
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9489891682555234,
      "learning_rate": 1.9958302639730345e-05,
      "loss": 0.866,
      "step": 1901
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5625409090806384,
      "learning_rate": 1.9958212105709777e-05,
      "loss": 0.2454,
      "step": 1902
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5217336890674791,
      "learning_rate": 1.9958121473717042e-05,
      "loss": 0.3216,
      "step": 1903
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4899222686601934,
      "learning_rate": 1.9958030743753027e-05,
      "loss": 0.2623,
      "step": 1904
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5868415272218954,
      "learning_rate": 1.995793991581863e-05,
      "loss": 0.6765,
      "step": 1905
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38821638501011485,
      "learning_rate": 1.995784898991475e-05,
      "loss": 0.2038,
      "step": 1906
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48424098859518394,
      "learning_rate": 1.9957757966042265e-05,
      "loss": 0.2584,
      "step": 1907
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5195881223541609,
      "learning_rate": 1.995766684420209e-05,
      "loss": 0.4312,
      "step": 1908
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4790978451341529,
      "learning_rate": 1.9957575624395106e-05,
      "loss": 0.2827,
      "step": 1909
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6091136024339586,
      "learning_rate": 1.9957484306622216e-05,
      "loss": 0.2571,
      "step": 1910
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.48151761069766913,
      "learning_rate": 1.9957392890884322e-05,
      "loss": 0.1778,
      "step": 1911
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5580361998070088,
      "learning_rate": 1.995730137718232e-05,
      "loss": 0.3701,
      "step": 1912
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6593063908853505,
      "learning_rate": 1.995720976551711e-05,
      "loss": 0.3634,
      "step": 1913
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1869137315554013,
      "learning_rate": 1.9957118055889594e-05,
      "loss": 0.9142,
      "step": 1914
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42543758350060545,
      "learning_rate": 1.9957026248300675e-05,
      "loss": 0.3185,
      "step": 1915
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.552563311620386,
      "learning_rate": 1.9956934342751254e-05,
      "loss": 0.2626,
      "step": 1916
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5536252276919599,
      "learning_rate": 1.9956842339242235e-05,
      "loss": 0.3254,
      "step": 1917
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6928790039935538,
      "learning_rate": 1.995675023777453e-05,
      "loss": 0.362,
      "step": 1918
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.476782746076621,
      "learning_rate": 1.9956658038349034e-05,
      "loss": 0.6979,
      "step": 1919
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38336805816376346,
      "learning_rate": 1.9956565740966667e-05,
      "loss": 0.1752,
      "step": 1920
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.697468381881715,
      "learning_rate": 1.9956473345628325e-05,
      "loss": 0.327,
      "step": 1921
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43626211028952633,
      "learning_rate": 1.9956380852334926e-05,
      "loss": 0.2945,
      "step": 1922
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7573710222399088,
      "learning_rate": 1.995628826108737e-05,
      "loss": 0.7403,
      "step": 1923
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1822603841974768,
      "learning_rate": 1.995619557188658e-05,
      "loss": 0.4713,
      "step": 1924
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.868398820790644,
      "learning_rate": 1.9956102784733456e-05,
      "loss": 0.3532,
      "step": 1925
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.500784607207767,
      "learning_rate": 1.9956009899628924e-05,
      "loss": 0.2997,
      "step": 1926
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38895710034403497,
      "learning_rate": 1.9955916916573885e-05,
      "loss": 0.2763,
      "step": 1927
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.726439493457961,
      "learning_rate": 1.995582383556926e-05,
      "loss": 0.5781,
      "step": 1928
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5634594089174043,
      "learning_rate": 1.9955730656615965e-05,
      "loss": 0.2693,
      "step": 1929
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.47478135085308554,
      "learning_rate": 1.995563737971492e-05,
      "loss": 0.229,
      "step": 1930
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6321436819091222,
      "learning_rate": 1.9955544004867034e-05,
      "loss": 0.6819,
      "step": 1931
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.24000417649613,
      "learning_rate": 1.995545053207323e-05,
      "loss": 0.4748,
      "step": 1932
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4518282220632304,
      "learning_rate": 1.995535696133443e-05,
      "loss": 0.3405,
      "step": 1933
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7233824702601546,
      "learning_rate": 1.995526329265155e-05,
      "loss": 0.3989,
      "step": 1934
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5222596857124335,
      "learning_rate": 1.995516952602552e-05,
      "loss": 0.2923,
      "step": 1935
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.4191710622630915,
      "learning_rate": 1.995507566145725e-05,
      "loss": 0.8168,
      "step": 1936
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8072478938697671,
      "learning_rate": 1.9954981698947674e-05,
      "loss": 0.7419,
      "step": 1937
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.36270207368512714,
      "learning_rate": 1.995488763849771e-05,
      "loss": 0.2726,
      "step": 1938
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43729679684351663,
      "learning_rate": 1.9954793480108288e-05,
      "loss": 0.2294,
      "step": 1939
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7294924050155006,
      "learning_rate": 1.995469922378033e-05,
      "loss": 0.3554,
      "step": 1940
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9419841154993847,
      "learning_rate": 1.9954604869514765e-05,
      "loss": 0.483,
      "step": 1941
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9203194677567909,
      "learning_rate": 1.9954510417312524e-05,
      "loss": 0.6104,
      "step": 1942
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4216355012703606,
      "learning_rate": 1.995441586717453e-05,
      "loss": 0.223,
      "step": 1943
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6745507079444031,
      "learning_rate": 1.9954321219101725e-05,
      "loss": 0.3886,
      "step": 1944
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44942956666735706,
      "learning_rate": 1.9954226473095027e-05,
      "loss": 0.2868,
      "step": 1945
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.555754492037417,
      "learning_rate": 1.9954131629155372e-05,
      "loss": 0.7496,
      "step": 1946
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.43109904375363145,
      "learning_rate": 1.9954036687283696e-05,
      "loss": 0.194,
      "step": 1947
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5260660914226528,
      "learning_rate": 1.9953941647480937e-05,
      "loss": 0.2304,
      "step": 1948
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7476754208411451,
      "learning_rate": 1.9953846509748017e-05,
      "loss": 0.3946,
      "step": 1949
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5480816800912784,
      "learning_rate": 1.9953751274085884e-05,
      "loss": 0.3455,
      "step": 1950
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1052680124392054,
      "learning_rate": 1.9953655940495473e-05,
      "loss": 0.5841,
      "step": 1951
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5216275486614494,
      "learning_rate": 1.9953560508977716e-05,
      "loss": 0.2064,
      "step": 1952
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46727728544505426,
      "learning_rate": 1.9953464979533555e-05,
      "loss": 0.3272,
      "step": 1953
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.867609068053895,
      "learning_rate": 1.9953369352163932e-05,
      "loss": 0.5173,
      "step": 1954
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8937801885337797,
      "learning_rate": 1.9953273626869786e-05,
      "loss": 0.8975,
      "step": 1955
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.44611171729790555,
      "learning_rate": 1.995317780365206e-05,
      "loss": 0.2716,
      "step": 1956
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5478910951371997,
      "learning_rate": 1.9953081882511697e-05,
      "loss": 0.2783,
      "step": 1957
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.33231224853274477,
      "learning_rate": 1.9952985863449634e-05,
      "loss": 0.1859,
      "step": 1958
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9903544241306642,
      "learning_rate": 1.9952889746466826e-05,
      "loss": 0.4762,
      "step": 1959
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0433179852249563,
      "learning_rate": 1.995279353156421e-05,
      "loss": 0.5885,
      "step": 1960
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.38896884685368477,
      "learning_rate": 1.995269721874274e-05,
      "loss": 0.077,
      "step": 1961
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6035021950200821,
      "learning_rate": 1.995260080800336e-05,
      "loss": 0.3512,
      "step": 1962
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5274392350534391,
      "learning_rate": 1.9952504299347014e-05,
      "loss": 0.2587,
      "step": 1963
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6247147085033216,
      "learning_rate": 1.995240769277466e-05,
      "loss": 0.8537,
      "step": 1964
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42741216400254867,
      "learning_rate": 1.995231098828724e-05,
      "loss": 0.1751,
      "step": 1965
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5548185689498005,
      "learning_rate": 1.9952214185885714e-05,
      "loss": 0.2988,
      "step": 1966
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5776270722975998,
      "learning_rate": 1.995211728557103e-05,
      "loss": 0.3225,
      "step": 1967
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8475022948626021,
      "learning_rate": 1.995202028734414e-05,
      "loss": 0.5154,
      "step": 1968
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.42884460543693315,
      "learning_rate": 1.9951923191206003e-05,
      "loss": 0.2302,
      "step": 1969
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9391165752480364,
      "learning_rate": 1.9951825997157566e-05,
      "loss": 0.4686,
      "step": 1970
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5493584185857034,
      "learning_rate": 1.995172870519979e-05,
      "loss": 0.2131,
      "step": 1971
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.65123766665144,
      "learning_rate": 1.9951631315333638e-05,
      "loss": 0.3036,
      "step": 1972
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8530675018969487,
      "learning_rate": 1.995153382756006e-05,
      "loss": 0.8562,
      "step": 1973
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.37264128603982005,
      "learning_rate": 1.9951436241880014e-05,
      "loss": 0.2497,
      "step": 1974
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2907578694623187,
      "learning_rate": 1.995133855829447e-05,
      "loss": 0.4767,
      "step": 1975
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5207156309402101,
      "learning_rate": 1.995124077680438e-05,
      "loss": 0.3341,
      "step": 1976
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.539232323693417,
      "learning_rate": 1.9951142897410707e-05,
      "loss": 0.3415,
      "step": 1977
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.3439127483006497,
      "learning_rate": 1.9951044920114418e-05,
      "loss": 0.1215,
      "step": 1978
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1417542754603898,
      "learning_rate": 1.9950946844916474e-05,
      "loss": 0.728,
      "step": 1979
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.539843411946295,
      "learning_rate": 1.995084867181784e-05,
      "loss": 0.2253,
      "step": 1980
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.46518717682568056,
      "learning_rate": 1.9950750400819482e-05,
      "loss": 0.3087,
      "step": 1981
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.42992004987894,
      "learning_rate": 1.9950652031922372e-05,
      "loss": 0.6259,
      "step": 1982
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.033324108699501,
      "learning_rate": 1.995055356512747e-05,
      "loss": 0.5922,
      "step": 1983
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5964309867205776,
      "learning_rate": 1.9950455000435746e-05,
      "loss": 0.328,
      "step": 1984
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4528421964562943,
      "learning_rate": 1.9950356337848172e-05,
      "loss": 0.3029,
      "step": 1985
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7560514004930133,
      "learning_rate": 1.995025757736572e-05,
      "loss": 0.4175,
      "step": 1986
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.40137307546469464,
      "learning_rate": 1.9950158718989358e-05,
      "loss": 0.1594,
      "step": 1987
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1854635117684085,
      "learning_rate": 1.9950059762720065e-05,
      "loss": 0.4989,
      "step": 1988
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9915113681872759,
      "learning_rate": 1.9949960708558803e-05,
      "loss": 0.2689,
      "step": 1989
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0456141107511563,
      "learning_rate": 1.994986155650656e-05,
      "loss": 0.9107,
      "step": 1990
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4052818530367466,
      "learning_rate": 1.9949762306564303e-05,
      "loss": 0.5368,
      "step": 1991
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3237348806919529,
      "learning_rate": 1.994966295873301e-05,
      "loss": 0.3703,
      "step": 1992
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8217872027412625,
      "learning_rate": 1.9949563513013658e-05,
      "loss": 0.2598,
      "step": 1993
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0992891350868397,
      "learning_rate": 1.9949463969407227e-05,
      "loss": 0.4321,
      "step": 1994
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5774914177905264,
      "learning_rate": 1.9949364327914695e-05,
      "loss": 0.4675,
      "step": 1995
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7696810156502913,
      "learning_rate": 1.9949264588537046e-05,
      "loss": 0.2066,
      "step": 1996
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3385506712557387,
      "learning_rate": 1.9949164751275256e-05,
      "loss": 0.2703,
      "step": 1997
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8189838825743627,
      "learning_rate": 1.994906481613031e-05,
      "loss": 0.2477,
      "step": 1998
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.738218030686773,
      "learning_rate": 1.994896478310319e-05,
      "loss": 0.3754,
      "step": 1999
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0199990631681946,
      "learning_rate": 1.994886465219488e-05,
      "loss": 0.5435,
      "step": 2000
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8680743266589628,
      "learning_rate": 1.994876442340637e-05,
      "loss": 0.5549,
      "step": 2001
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.916988355765509,
      "learning_rate": 1.9948664096738636e-05,
      "loss": 0.2595,
      "step": 2002
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3930337702095263,
      "learning_rate": 1.9948563672192675e-05,
      "loss": 0.3604,
      "step": 2003
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.069160784485521,
      "learning_rate": 1.994846314976947e-05,
      "loss": 0.3275,
      "step": 2004
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1814020243528454,
      "learning_rate": 1.9948362529470012e-05,
      "loss": 0.2644,
      "step": 2005
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.148080918534558,
      "learning_rate": 1.994826181129529e-05,
      "loss": 0.0914,
      "step": 2006
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0711146948514576,
      "learning_rate": 1.9948160995246292e-05,
      "loss": 0.3418,
      "step": 2007
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9413493628529899,
      "learning_rate": 1.9948060081324016e-05,
      "loss": 0.3575,
      "step": 2008
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3950655403096608,
      "learning_rate": 1.9947959069529452e-05,
      "loss": 0.6379,
      "step": 2009
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.111014621746194,
      "learning_rate": 1.994785795986359e-05,
      "loss": 0.3522,
      "step": 2010
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.1662208675196197,
      "learning_rate": 1.994775675232743e-05,
      "loss": 0.3433,
      "step": 2011
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6948298748145851,
      "learning_rate": 1.9947655446921965e-05,
      "loss": 0.3778,
      "step": 2012
    },
    {
      "epoch": 0.06,
      "grad_norm": 41.10755624027442,
      "learning_rate": 1.9947554043648197e-05,
      "loss": 0.6499,
      "step": 2013
    },
    {
      "epoch": 0.06,
      "grad_norm": 43.56586584111511,
      "learning_rate": 1.9947452542507114e-05,
      "loss": 1.7986,
      "step": 2014
    },
    {
      "epoch": 0.06,
      "grad_norm": 6.148074625175885,
      "learning_rate": 1.994735094349972e-05,
      "loss": 0.5274,
      "step": 2015
    },
    {
      "epoch": 0.06,
      "grad_norm": 5.901660565484113,
      "learning_rate": 1.9947249246627017e-05,
      "loss": 0.5515,
      "step": 2016
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.567557220739008,
      "learning_rate": 1.994714745189e-05,
      "loss": 0.4014,
      "step": 2017
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5397815617975337,
      "learning_rate": 1.9947045559289675e-05,
      "loss": 0.7508,
      "step": 2018
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0181502686848132,
      "learning_rate": 1.994694356882704e-05,
      "loss": 0.5375,
      "step": 2019
    },
    {
      "epoch": 0.06,
      "grad_norm": 15.705331072492635,
      "learning_rate": 1.9946841480503108e-05,
      "loss": 0.8862,
      "step": 2020
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.358010811192372,
      "learning_rate": 1.994673929431887e-05,
      "loss": 0.3755,
      "step": 2021
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.011346060140442,
      "learning_rate": 1.994663701027534e-05,
      "loss": 0.6474,
      "step": 2022
    },
    {
      "epoch": 0.06,
      "grad_norm": 8.835594436072988,
      "learning_rate": 1.9946534628373522e-05,
      "loss": 0.6212,
      "step": 2023
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1268186748934004,
      "learning_rate": 1.9946432148614424e-05,
      "loss": 0.4058,
      "step": 2024
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.0205759170093254,
      "learning_rate": 1.9946329570999054e-05,
      "loss": 0.6468,
      "step": 2025
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.337824752801503,
      "learning_rate": 1.9946226895528423e-05,
      "loss": 0.564,
      "step": 2026
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1466427684303735,
      "learning_rate": 1.9946124122203535e-05,
      "loss": 0.594,
      "step": 2027
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.699103466126485,
      "learning_rate": 1.9946021251025406e-05,
      "loss": 0.4788,
      "step": 2028
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.597595868884784,
      "learning_rate": 1.994591828199505e-05,
      "loss": 0.8049,
      "step": 2029
    },
    {
      "epoch": 0.06,
      "grad_norm": 4.4742984906925685,
      "learning_rate": 1.9945815215113472e-05,
      "loss": 0.393,
      "step": 2030
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.9255209041714902,
      "learning_rate": 1.9945712050381697e-05,
      "loss": 0.7357,
      "step": 2031
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.5789220733801383,
      "learning_rate": 1.994560878780073e-05,
      "loss": 0.2633,
      "step": 2032
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.00973164718247,
      "learning_rate": 1.9945505427371593e-05,
      "loss": 0.4555,
      "step": 2033
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4438637450800105,
      "learning_rate": 1.99454019690953e-05,
      "loss": 0.3343,
      "step": 2034
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8035155136103507,
      "learning_rate": 1.994529841297287e-05,
      "loss": 0.4171,
      "step": 2035
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8794523786199613,
      "learning_rate": 1.9945194759005323e-05,
      "loss": 0.479,
      "step": 2036
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4264070889361244,
      "learning_rate": 1.994509100719368e-05,
      "loss": 0.5654,
      "step": 2037
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.7941242441831826,
      "learning_rate": 1.9944987157538954e-05,
      "loss": 0.5347,
      "step": 2038
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7569311364974372,
      "learning_rate": 1.9944883210042175e-05,
      "loss": 0.2988,
      "step": 2039
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0767211066806055,
      "learning_rate": 1.9944779164704366e-05,
      "loss": 0.4248,
      "step": 2040
    },
    {
      "epoch": 0.06,
      "grad_norm": 11.0073304760045,
      "learning_rate": 1.9944675021526543e-05,
      "loss": 0.2436,
      "step": 2041
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.219759724939259,
      "learning_rate": 1.9944570780509734e-05,
      "loss": 0.2986,
      "step": 2042
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6946137894517,
      "learning_rate": 1.9944466441654968e-05,
      "loss": 0.3289,
      "step": 2043
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.6394679376227135,
      "learning_rate": 1.994436200496327e-05,
      "loss": 0.4544,
      "step": 2044
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.791200829115811,
      "learning_rate": 1.9944257470435665e-05,
      "loss": 0.4901,
      "step": 2045
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7519626424774813,
      "learning_rate": 1.9944152838073182e-05,
      "loss": 0.4282,
      "step": 2046
    },
    {
      "epoch": 0.06,
      "grad_norm": 3.1219779651668307,
      "learning_rate": 1.994404810787685e-05,
      "loss": 0.3422,
      "step": 2047
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8716712574212387,
      "learning_rate": 1.9943943279847704e-05,
      "loss": 0.3423,
      "step": 2048
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6961947075879428,
      "learning_rate": 1.9943838353986772e-05,
      "loss": 0.3098,
      "step": 2049
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9089510756087529,
      "learning_rate": 1.994373333029508e-05,
      "loss": 0.2598,
      "step": 2050
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6616959531478191,
      "learning_rate": 1.9943628208773672e-05,
      "loss": 0.4227,
      "step": 2051
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.148727324926537,
      "learning_rate": 1.994352298942358e-05,
      "loss": 0.1476,
      "step": 2052
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7187874418384821,
      "learning_rate": 1.9943417672245834e-05,
      "loss": 0.4137,
      "step": 2053
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9178163466603809,
      "learning_rate": 1.9943312257241476e-05,
      "loss": 0.3328,
      "step": 2054
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.8441616756026922,
      "learning_rate": 1.994320674441154e-05,
      "loss": 0.9694,
      "step": 2055
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.5368305072507513,
      "learning_rate": 1.994310113375706e-05,
      "loss": 0.3828,
      "step": 2056
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6845803396282113,
      "learning_rate": 1.9942995425279083e-05,
      "loss": 0.3245,
      "step": 2057
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5766427926049753,
      "learning_rate": 1.9942889618978645e-05,
      "loss": 0.3082,
      "step": 2058
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0743315104022457,
      "learning_rate": 1.9942783714856788e-05,
      "loss": 0.8622,
      "step": 2059
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8166898849447181,
      "learning_rate": 1.9942677712914552e-05,
      "loss": 0.4773,
      "step": 2060
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8044325865900468,
      "learning_rate": 1.994257161315298e-05,
      "loss": 0.5197,
      "step": 2061
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5770024948689958,
      "learning_rate": 1.994246541557312e-05,
      "loss": 0.2738,
      "step": 2062
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5009233802089982,
      "learning_rate": 1.994235912017601e-05,
      "loss": 0.2778,
      "step": 2063
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0182057598489407,
      "learning_rate": 1.9942252726962702e-05,
      "loss": 1.0328,
      "step": 2064
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0338507176450975,
      "learning_rate": 1.994214623593424e-05,
      "loss": 0.1775,
      "step": 2065
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.539100163069503,
      "learning_rate": 1.994203964709167e-05,
      "loss": 0.3127,
      "step": 2066
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.316652823247976,
      "learning_rate": 1.9941932960436046e-05,
      "loss": 0.3625,
      "step": 2067
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.6323373713159106,
      "learning_rate": 1.9941826175968412e-05,
      "loss": 0.7173,
      "step": 2068
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4930161070335665,
      "learning_rate": 1.994171929368982e-05,
      "loss": 0.3527,
      "step": 2069
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.633552376191326,
      "learning_rate": 1.9941612313601323e-05,
      "loss": 0.417,
      "step": 2070
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5278408600789045,
      "learning_rate": 1.9941505235703973e-05,
      "loss": 0.3126,
      "step": 2071
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0603743616447647,
      "learning_rate": 1.9941398059998823e-05,
      "loss": 0.3129,
      "step": 2072
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2735890831098202,
      "learning_rate": 1.9941290786486928e-05,
      "loss": 0.6958,
      "step": 2073
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4995508858058173,
      "learning_rate": 1.994118341516934e-05,
      "loss": 0.3321,
      "step": 2074
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.658299722731587,
      "learning_rate": 1.9941075946047127e-05,
      "loss": 0.2198,
      "step": 2075
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4700184434677997,
      "learning_rate": 1.994096837912133e-05,
      "loss": 0.3683,
      "step": 2076
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1772454771802667,
      "learning_rate": 1.994086071439302e-05,
      "loss": 0.4568,
      "step": 2077
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9699283262599564,
      "learning_rate": 1.994075295186325e-05,
      "loss": 0.4614,
      "step": 2078
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.0376108738789496,
      "learning_rate": 1.994064509153308e-05,
      "loss": 0.4597,
      "step": 2079
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4610442661901119,
      "learning_rate": 1.9940537133403574e-05,
      "loss": 0.2555,
      "step": 2080
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6963015287857391,
      "learning_rate": 1.994042907747579e-05,
      "loss": 0.423,
      "step": 2081
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6012685399454806,
      "learning_rate": 1.9940320923750796e-05,
      "loss": 0.3573,
      "step": 2082
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.3779191305340197,
      "learning_rate": 1.9940212672229652e-05,
      "loss": 0.2921,
      "step": 2083
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9596800327088947,
      "learning_rate": 1.994010432291343e-05,
      "loss": 0.1786,
      "step": 2084
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6839755267597689,
      "learning_rate": 1.9939995875803185e-05,
      "loss": 0.4323,
      "step": 2085
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.063665825703754,
      "learning_rate": 1.9939887330899994e-05,
      "loss": 0.5202,
      "step": 2086
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7260381738423739,
      "learning_rate": 1.9939778688204917e-05,
      "loss": 0.4181,
      "step": 2087
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6374913487093484,
      "learning_rate": 1.9939669947719032e-05,
      "loss": 0.2855,
      "step": 2088
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5102695671503644,
      "learning_rate": 1.99395611094434e-05,
      "loss": 0.3465,
      "step": 2089
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4052027799554896,
      "learning_rate": 1.9939452173379096e-05,
      "loss": 0.5123,
      "step": 2090
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.2766027994450506,
      "learning_rate": 1.993934313952719e-05,
      "loss": 0.5723,
      "step": 2091
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.1677669741078553,
      "learning_rate": 1.9939234007888757e-05,
      "loss": 0.2149,
      "step": 2092
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4104420527053422,
      "learning_rate": 1.993912477846487e-05,
      "loss": 0.2422,
      "step": 2093
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5332160997640081,
      "learning_rate": 1.99390154512566e-05,
      "loss": 0.3855,
      "step": 2094
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9043063843381408,
      "learning_rate": 1.993890602626503e-05,
      "loss": 0.4607,
      "step": 2095
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.8752093800794581,
      "learning_rate": 1.993879650349123e-05,
      "loss": 0.5808,
      "step": 2096
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.4843085734835342,
      "learning_rate": 1.9938686882936276e-05,
      "loss": 0.1946,
      "step": 2097
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5439232676296337,
      "learning_rate": 1.9938577164601256e-05,
      "loss": 0.3639,
      "step": 2098
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7071614456971645,
      "learning_rate": 1.9938467348487242e-05,
      "loss": 0.3179,
      "step": 2099
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6705604884643782,
      "learning_rate": 1.9938357434595313e-05,
      "loss": 0.4102,
      "step": 2100
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.41440838603856534,
      "learning_rate": 1.9938247422926556e-05,
      "loss": 0.1444,
      "step": 2101
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5737186546332997,
      "learning_rate": 1.993813731348205e-05,
      "loss": 0.3568,
      "step": 2102
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6917878269272112,
      "learning_rate": 1.9938027106262884e-05,
      "loss": 0.3313,
      "step": 2103
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0764789821047522,
      "learning_rate": 1.993791680127013e-05,
      "loss": 0.4986,
      "step": 2104
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6503021705013606,
      "learning_rate": 1.9937806398504885e-05,
      "loss": 0.35,
      "step": 2105
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9802217375196707,
      "learning_rate": 1.9937695897968227e-05,
      "loss": 0.396,
      "step": 2106
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.7097256271601825,
      "learning_rate": 1.993758529966125e-05,
      "loss": 0.3246,
      "step": 2107
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.6093799529532578,
      "learning_rate": 1.9937474603585035e-05,
      "loss": 0.3408,
      "step": 2108
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.7918735749637245,
      "learning_rate": 1.993736380974068e-05,
      "loss": 0.8251,
      "step": 2109
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.158744301423645,
      "learning_rate": 1.993725291812927e-05,
      "loss": 0.423,
      "step": 2110
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4474680571100877,
      "learning_rate": 1.9937141928751892e-05,
      "loss": 0.3369,
      "step": 2111
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4309174828115532,
      "learning_rate": 1.993703084160964e-05,
      "loss": 0.2959,
      "step": 2112
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.574117928853068,
      "learning_rate": 1.9936919656703615e-05,
      "loss": 0.3882,
      "step": 2113
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.555774320385182,
      "learning_rate": 1.9936808374034903e-05,
      "loss": 0.1023,
      "step": 2114
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.112571407620934,
      "learning_rate": 1.99366969936046e-05,
      "loss": 0.7937,
      "step": 2115
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.4229677095605159,
      "learning_rate": 1.9936585515413803e-05,
      "loss": 0.2347,
      "step": 2116
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.5113744475149686,
      "learning_rate": 1.9936473939463606e-05,
      "loss": 0.3103,
      "step": 2117
    },
    {
      "epoch": 0.06,
      "grad_norm": 2.054482135134421,
      "learning_rate": 1.993636226575511e-05,
      "loss": 0.8989,
      "step": 2118
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.9498182236015078,
      "learning_rate": 1.9936250494289414e-05,
      "loss": 0.4761,
      "step": 2119
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.127198120466241,
      "learning_rate": 1.9936138625067612e-05,
      "loss": 0.4319,
      "step": 2120
    },
    {
      "epoch": 0.06,
      "grad_norm": 0.45800241986626106,
      "learning_rate": 1.9936026658090815e-05,
      "loss": 0.3304,
      "step": 2121
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.0552723167807425,
      "learning_rate": 1.9935914593360113e-05,
      "loss": 0.3266,
      "step": 2122
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.39293186771902433,
      "learning_rate": 1.9935802430876616e-05,
      "loss": 0.2535,
      "step": 2123
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3683864557576833,
      "learning_rate": 1.9935690170641427e-05,
      "loss": 0.6292,
      "step": 2124
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4555919590286193,
      "learning_rate": 1.9935577812655644e-05,
      "loss": 0.2338,
      "step": 2125
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2606189800238874,
      "learning_rate": 1.9935465356920382e-05,
      "loss": 0.9703,
      "step": 2126
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4789571688823322,
      "learning_rate": 1.9935352803436742e-05,
      "loss": 0.62,
      "step": 2127
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.643352297343668,
      "learning_rate": 1.993524015220583e-05,
      "loss": 0.4277,
      "step": 2128
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4897138243509305,
      "learning_rate": 1.993512740322876e-05,
      "loss": 0.2546,
      "step": 2129
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5722037760413792,
      "learning_rate": 1.9935014556506633e-05,
      "loss": 0.3311,
      "step": 2130
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1283825395400635,
      "learning_rate": 1.9934901612040563e-05,
      "loss": 0.6224,
      "step": 2131
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6981041677760393,
      "learning_rate": 1.9934788569831666e-05,
      "loss": 0.1473,
      "step": 2132
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.79666569522215,
      "learning_rate": 1.9934675429881048e-05,
      "loss": 0.7387,
      "step": 2133
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4232514992513827,
      "learning_rate": 1.9934562192189824e-05,
      "loss": 0.2044,
      "step": 2134
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4732611975497927,
      "learning_rate": 1.993444885675911e-05,
      "loss": 0.3496,
      "step": 2135
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5247900342232665,
      "learning_rate": 1.993433542359002e-05,
      "loss": 0.3765,
      "step": 2136
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8638000480778925,
      "learning_rate": 1.9934221892683666e-05,
      "loss": 0.5637,
      "step": 2137
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7161102102472234,
      "learning_rate": 1.993410826404117e-05,
      "loss": 0.3421,
      "step": 2138
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5197146101100131,
      "learning_rate": 1.993399453766365e-05,
      "loss": 0.3182,
      "step": 2139
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.38796610673511206,
      "learning_rate": 1.9933880713552218e-05,
      "loss": 0.1999,
      "step": 2140
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4166575939205571,
      "learning_rate": 1.9933766791708003e-05,
      "loss": 0.2693,
      "step": 2141
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.276838901644802,
      "learning_rate": 1.9933652772132118e-05,
      "loss": 0.3588,
      "step": 2142
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43019546480753723,
      "learning_rate": 1.9933538654825693e-05,
      "loss": 0.2251,
      "step": 2143
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6931293717932216,
      "learning_rate": 1.9933424439789844e-05,
      "loss": 0.4528,
      "step": 2144
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.005285102412226,
      "learning_rate": 1.9933310127025694e-05,
      "loss": 0.4647,
      "step": 2145
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6714556512709938,
      "learning_rate": 1.9933195716534376e-05,
      "loss": 0.4552,
      "step": 2146
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4889171197446195,
      "learning_rate": 1.9933081208317008e-05,
      "loss": 0.2828,
      "step": 2147
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5230819965432931,
      "learning_rate": 1.9932966602374718e-05,
      "loss": 0.3545,
      "step": 2148
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5954951496473997,
      "learning_rate": 1.9932851898708635e-05,
      "loss": 0.3684,
      "step": 2149
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43834250376454315,
      "learning_rate": 1.9932737097319882e-05,
      "loss": 0.215,
      "step": 2150
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2131532141551276,
      "learning_rate": 1.99326221982096e-05,
      "loss": 0.507,
      "step": 2151
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5446660775023346,
      "learning_rate": 1.9932507201378907e-05,
      "loss": 0.3022,
      "step": 2152
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4851490972775182,
      "learning_rate": 1.9932392106828944e-05,
      "loss": 0.2822,
      "step": 2153
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6351256170356364,
      "learning_rate": 1.9932276914560836e-05,
      "loss": 0.4362,
      "step": 2154
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8655190398828324,
      "learning_rate": 1.9932161624575724e-05,
      "loss": 0.4728,
      "step": 2155
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1193377924511272,
      "learning_rate": 1.9932046236874733e-05,
      "loss": 0.281,
      "step": 2156
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6922246171868123,
      "learning_rate": 1.9931930751459005e-05,
      "loss": 0.3504,
      "step": 2157
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.41713456124457887,
      "learning_rate": 1.9931815168329674e-05,
      "loss": 0.2953,
      "step": 2158
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4728863270914019,
      "learning_rate": 1.993169948748788e-05,
      "loss": 0.2829,
      "step": 2159
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4767368219276049,
      "learning_rate": 1.9931583708934753e-05,
      "loss": 0.1642,
      "step": 2160
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.780404289939947,
      "learning_rate": 1.9931467832671443e-05,
      "loss": 0.2913,
      "step": 2161
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48165047391795174,
      "learning_rate": 1.993135185869908e-05,
      "loss": 0.312,
      "step": 2162
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9241061335613543,
      "learning_rate": 1.9931235787018815e-05,
      "loss": 0.5712,
      "step": 2163
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.42887936372309526,
      "learning_rate": 1.993111961763178e-05,
      "loss": 0.2941,
      "step": 2164
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1202859308631465,
      "learning_rate": 1.9931003350539125e-05,
      "loss": 0.9147,
      "step": 2165
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4524211729588168,
      "learning_rate": 1.993088698574199e-05,
      "loss": 0.2303,
      "step": 2166
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6021401002754665,
      "learning_rate": 1.9930770523241522e-05,
      "loss": 0.7357,
      "step": 2167
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7210307350318815,
      "learning_rate": 1.9930653963038863e-05,
      "loss": 0.0968,
      "step": 2168
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4393385289926787,
      "learning_rate": 1.9930537305135165e-05,
      "loss": 0.192,
      "step": 2169
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5363333212141284,
      "learning_rate": 1.9930420549531574e-05,
      "loss": 0.3278,
      "step": 2170
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3938725843656259,
      "learning_rate": 1.9930303696229237e-05,
      "loss": 0.2851,
      "step": 2171
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.056081913019101,
      "learning_rate": 1.9930186745229307e-05,
      "loss": 0.5786,
      "step": 2172
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7947549862393402,
      "learning_rate": 1.9930069696532932e-05,
      "loss": 0.4744,
      "step": 2173
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2146500845804282,
      "learning_rate": 1.992995255014126e-05,
      "loss": 0.6542,
      "step": 2174
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4489491155144778,
      "learning_rate": 1.9929835306055453e-05,
      "loss": 0.2234,
      "step": 2175
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.298701702907797,
      "learning_rate": 1.9929717964276652e-05,
      "loss": 0.8755,
      "step": 2176
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4951768703716881,
      "learning_rate": 1.992960052480602e-05,
      "loss": 0.2812,
      "step": 2177
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2120330675095632,
      "learning_rate": 1.9929482987644716e-05,
      "loss": 0.2125,
      "step": 2178
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5590659782102387,
      "learning_rate": 1.9929365352793888e-05,
      "loss": 0.2728,
      "step": 2179
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7265890500828796,
      "learning_rate": 1.99292476202547e-05,
      "loss": 0.3819,
      "step": 2180
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.034805853435492,
      "learning_rate": 1.9929129790028302e-05,
      "loss": 0.4735,
      "step": 2181
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4798398143867511,
      "learning_rate": 1.9929011862115858e-05,
      "loss": 0.3042,
      "step": 2182
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9157627200835761,
      "learning_rate": 1.992889383651853e-05,
      "loss": 0.4337,
      "step": 2183
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6052921079038361,
      "learning_rate": 1.9928775713237476e-05,
      "loss": 0.2155,
      "step": 2184
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8039174322386181,
      "learning_rate": 1.992865749227386e-05,
      "loss": 0.3936,
      "step": 2185
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9461439653385084,
      "learning_rate": 1.9928539173628846e-05,
      "loss": 0.0895,
      "step": 2186
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5735448836618573,
      "learning_rate": 1.9928420757303595e-05,
      "loss": 0.4412,
      "step": 2187
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5034873348630676,
      "learning_rate": 1.9928302243299276e-05,
      "loss": 0.2597,
      "step": 2188
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7804143408788076,
      "learning_rate": 1.992818363161705e-05,
      "loss": 0.3027,
      "step": 2189
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9079168666129114,
      "learning_rate": 1.9928064922258087e-05,
      "loss": 0.4911,
      "step": 2190
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3282916722401408,
      "learning_rate": 1.9927946115223554e-05,
      "loss": 0.7795,
      "step": 2191
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0854131985128024,
      "learning_rate": 1.9927827210514623e-05,
      "loss": 0.3455,
      "step": 2192
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5949996198444266,
      "learning_rate": 1.9927708208132462e-05,
      "loss": 0.3229,
      "step": 2193
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49027420343889166,
      "learning_rate": 1.9927589108078238e-05,
      "loss": 0.2727,
      "step": 2194
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5425869653014076,
      "learning_rate": 1.9927469910353126e-05,
      "loss": 0.2863,
      "step": 2195
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5396817285245231,
      "learning_rate": 1.99273506149583e-05,
      "loss": 0.2964,
      "step": 2196
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6187244109846918,
      "learning_rate": 1.992723122189493e-05,
      "loss": 0.3358,
      "step": 2197
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5300813922294082,
      "learning_rate": 1.9927111731164194e-05,
      "loss": 0.3882,
      "step": 2198
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4357421215900072,
      "learning_rate": 1.9926992142767267e-05,
      "loss": 0.5668,
      "step": 2199
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4887916001845073,
      "learning_rate": 1.9926872456705326e-05,
      "loss": 0.359,
      "step": 2200
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4749418476001237,
      "learning_rate": 1.9926752672979544e-05,
      "loss": 0.1137,
      "step": 2201
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5688460120718388,
      "learning_rate": 1.9926632791591104e-05,
      "loss": 0.3363,
      "step": 2202
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5807753596158184,
      "learning_rate": 1.9926512812541185e-05,
      "loss": 0.2724,
      "step": 2203
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2491917132674162,
      "learning_rate": 1.992639273583097e-05,
      "loss": 0.7128,
      "step": 2204
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5797717299438788,
      "learning_rate": 1.9926272561461633e-05,
      "loss": 0.3746,
      "step": 2205
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5110606268582706,
      "learning_rate": 1.992615228943436e-05,
      "loss": 0.3616,
      "step": 2206
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.504290919862146,
      "learning_rate": 1.9926031919750336e-05,
      "loss": 0.2559,
      "step": 2207
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.503766324771535,
      "learning_rate": 1.992591145241074e-05,
      "loss": 0.2402,
      "step": 2208
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6209294979098225,
      "learning_rate": 1.9925790887416765e-05,
      "loss": 0.3094,
      "step": 2209
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5384294745933907,
      "learning_rate": 1.9925670224769594e-05,
      "loss": 0.084,
      "step": 2210
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1220565046958604,
      "learning_rate": 1.9925549464470413e-05,
      "loss": 0.4946,
      "step": 2211
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4623244829583321,
      "learning_rate": 1.9925428606520406e-05,
      "loss": 0.2638,
      "step": 2212
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5413354820481084,
      "learning_rate": 1.9925307650920774e-05,
      "loss": 0.4097,
      "step": 2213
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7043590927305234,
      "learning_rate": 1.9925186597672694e-05,
      "loss": 0.4693,
      "step": 2214
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5443522634098255,
      "learning_rate": 1.9925065446777365e-05,
      "loss": 0.7552,
      "step": 2215
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40781038110680884,
      "learning_rate": 1.9924944198235976e-05,
      "loss": 0.2275,
      "step": 2216
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7056363900176785,
      "learning_rate": 1.9924822852049716e-05,
      "loss": 0.9321,
      "step": 2217
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.37236939309311445,
      "learning_rate": 1.992470140821979e-05,
      "loss": 0.2524,
      "step": 2218
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5969171235389027,
      "learning_rate": 1.992457986674738e-05,
      "loss": 0.2473,
      "step": 2219
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5890660936990006,
      "learning_rate": 1.992445822763369e-05,
      "loss": 0.1621,
      "step": 2220
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5069661765987807,
      "learning_rate": 1.9924336490879916e-05,
      "loss": 0.2657,
      "step": 2221
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9209813305870408,
      "learning_rate": 1.9924214656487254e-05,
      "loss": 0.5597,
      "step": 2222
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47176037934794085,
      "learning_rate": 1.99240927244569e-05,
      "loss": 0.3438,
      "step": 2223
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6429957436515642,
      "learning_rate": 1.9923970694790056e-05,
      "loss": 0.3501,
      "step": 2224
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4205766102125558,
      "learning_rate": 1.9923848567487926e-05,
      "loss": 0.2282,
      "step": 2225
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7035706370718695,
      "learning_rate": 1.992372634255171e-05,
      "loss": 0.8652,
      "step": 2226
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2090414783005081,
      "learning_rate": 1.9923604019982603e-05,
      "loss": 0.5425,
      "step": 2227
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.562358338212064,
      "learning_rate": 1.992348159978182e-05,
      "loss": 0.235,
      "step": 2228
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5199385202170156,
      "learning_rate": 1.9923359081950554e-05,
      "loss": 0.213,
      "step": 2229
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5112603393893287,
      "learning_rate": 1.992323646649002e-05,
      "loss": 0.3465,
      "step": 2230
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5439417580293584,
      "learning_rate": 1.992311375340142e-05,
      "loss": 0.3541,
      "step": 2231
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8159173560760881,
      "learning_rate": 1.9922990942685957e-05,
      "loss": 0.5597,
      "step": 2232
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.675768944442618,
      "learning_rate": 1.992286803434485e-05,
      "loss": 0.1655,
      "step": 2233
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5333089227623855,
      "learning_rate": 1.99227450283793e-05,
      "loss": 0.2199,
      "step": 2234
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6857517050492911,
      "learning_rate": 1.9922621924790523e-05,
      "loss": 0.4318,
      "step": 2235
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4176960734082557,
      "learning_rate": 1.992249872357972e-05,
      "loss": 0.3038,
      "step": 2236
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6447760206443526,
      "learning_rate": 1.9922375424748116e-05,
      "loss": 0.1932,
      "step": 2237
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6688707826249464,
      "learning_rate": 1.9922252028296914e-05,
      "loss": 0.0857,
      "step": 2238
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4897847833359655,
      "learning_rate": 1.9922128534227332e-05,
      "loss": 0.3509,
      "step": 2239
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.921052212829947,
      "learning_rate": 1.9922004942540585e-05,
      "loss": 0.4622,
      "step": 2240
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4264367693053563,
      "learning_rate": 1.992188125323789e-05,
      "loss": 0.3332,
      "step": 2241
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3734494552846517,
      "learning_rate": 1.992175746632046e-05,
      "loss": 0.1505,
      "step": 2242
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.578621244631372,
      "learning_rate": 1.9921633581789516e-05,
      "loss": 0.3072,
      "step": 2243
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6052902225693444,
      "learning_rate": 1.9921509599646276e-05,
      "loss": 0.3162,
      "step": 2244
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7285528494587228,
      "learning_rate": 1.992138551989196e-05,
      "loss": 0.7464,
      "step": 2245
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3710994268665916,
      "learning_rate": 1.992126134252779e-05,
      "loss": 0.08,
      "step": 2246
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4177266602913466,
      "learning_rate": 1.9921137067554987e-05,
      "loss": 0.2668,
      "step": 2247
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.43398983342700004,
      "learning_rate": 1.9921012694974772e-05,
      "loss": 0.3369,
      "step": 2248
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.864832420929456,
      "learning_rate": 1.992088822478837e-05,
      "loss": 0.4562,
      "step": 2249
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.748982855206908,
      "learning_rate": 1.9920763656997005e-05,
      "loss": 0.6771,
      "step": 2250
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1471445446163036,
      "learning_rate": 1.9920638991601902e-05,
      "loss": 0.5141,
      "step": 2251
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5497695886726438,
      "learning_rate": 1.9920514228604287e-05,
      "loss": 0.2989,
      "step": 2252
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6136201217277044,
      "learning_rate": 1.992038936800539e-05,
      "loss": 0.3076,
      "step": 2253
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4591229269939774,
      "learning_rate": 1.992026440980644e-05,
      "loss": 0.3391,
      "step": 2254
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7142625816540887,
      "learning_rate": 1.9920139354008665e-05,
      "loss": 0.251,
      "step": 2255
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9573935732988644,
      "learning_rate": 1.992001420061329e-05,
      "loss": 0.3867,
      "step": 2256
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.545672545229935,
      "learning_rate": 1.9919888949621555e-05,
      "loss": 0.2501,
      "step": 2257
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.584440313168011,
      "learning_rate": 1.991976360103469e-05,
      "loss": 0.3576,
      "step": 2258
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5997851620195095,
      "learning_rate": 1.9919638154853925e-05,
      "loss": 0.2974,
      "step": 2259
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6554085667447374,
      "learning_rate": 1.9919512611080492e-05,
      "loss": 0.4107,
      "step": 2260
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9181684976446769,
      "learning_rate": 1.9919386969715633e-05,
      "loss": 0.0918,
      "step": 2261
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4793700552060602,
      "learning_rate": 1.991926123076058e-05,
      "loss": 0.2893,
      "step": 2262
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2676850374322155,
      "learning_rate": 1.9919135394216577e-05,
      "loss": 0.6531,
      "step": 2263
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5352364054309509,
      "learning_rate": 1.991900946008485e-05,
      "loss": 0.3723,
      "step": 2264
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.657807036572158,
      "learning_rate": 1.9918883428366645e-05,
      "loss": 0.3868,
      "step": 2265
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49207241482308595,
      "learning_rate": 1.9918757299063202e-05,
      "loss": 0.2228,
      "step": 2266
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5465898606542452,
      "learning_rate": 1.991863107217576e-05,
      "loss": 0.3463,
      "step": 2267
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5714903233383996,
      "learning_rate": 1.991850474770556e-05,
      "loss": 0.1274,
      "step": 2268
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.279655883375166,
      "learning_rate": 1.991837832565385e-05,
      "loss": 0.9356,
      "step": 2269
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47508007532480895,
      "learning_rate": 1.991825180602187e-05,
      "loss": 0.2075,
      "step": 2270
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7202764662510102,
      "learning_rate": 1.9918125188810867e-05,
      "loss": 0.4355,
      "step": 2271
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4265536340142561,
      "learning_rate": 1.9917998474022083e-05,
      "loss": 0.264,
      "step": 2272
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1149344155494116,
      "learning_rate": 1.9917871661656766e-05,
      "loss": 0.5996,
      "step": 2273
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5809829302210976,
      "learning_rate": 1.9917744751716165e-05,
      "loss": 0.3816,
      "step": 2274
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.38071229951464064,
      "learning_rate": 1.9917617744201528e-05,
      "loss": 0.2402,
      "step": 2275
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6147473678268499,
      "learning_rate": 1.9917490639114104e-05,
      "loss": 0.2068,
      "step": 2276
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5973824050068126,
      "learning_rate": 1.9917363436455145e-05,
      "loss": 0.2933,
      "step": 2277
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6133933863949514,
      "learning_rate": 1.9917236136225898e-05,
      "loss": 0.419,
      "step": 2278
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40598617764470424,
      "learning_rate": 1.991710873842762e-05,
      "loss": 0.2105,
      "step": 2279
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6500476437971695,
      "learning_rate": 1.9916981243061563e-05,
      "loss": 0.392,
      "step": 2280
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.275411504666421,
      "learning_rate": 1.9916853650128982e-05,
      "loss": 0.4447,
      "step": 2281
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8702990327132093,
      "learning_rate": 1.991672595963113e-05,
      "loss": 0.5374,
      "step": 2282
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3843139766673315,
      "learning_rate": 1.9916598171569267e-05,
      "loss": 0.2653,
      "step": 2283
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6463221898224031,
      "learning_rate": 1.9916470285944648e-05,
      "loss": 0.3708,
      "step": 2284
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6275477644005868,
      "learning_rate": 1.991634230275853e-05,
      "loss": 0.292,
      "step": 2285
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3520699489486033,
      "learning_rate": 1.9916214222012176e-05,
      "loss": 0.1799,
      "step": 2286
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2626225340094437,
      "learning_rate": 1.9916086043706845e-05,
      "loss": 0.5765,
      "step": 2287
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.501903060812653,
      "learning_rate": 1.9915957767843792e-05,
      "loss": 0.0838,
      "step": 2288
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49774597000810983,
      "learning_rate": 1.9915829394424286e-05,
      "loss": 0.3395,
      "step": 2289
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4772156416834036,
      "learning_rate": 1.9915700923449586e-05,
      "loss": 0.3417,
      "step": 2290
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9721746503467019,
      "learning_rate": 1.991557235492096e-05,
      "loss": 0.5801,
      "step": 2291
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.5510894442056617,
      "learning_rate": 1.9915443688839668e-05,
      "loss": 0.2301,
      "step": 2292
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5957236042090355,
      "learning_rate": 1.9915314925206984e-05,
      "loss": 0.3345,
      "step": 2293
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.34765313307759743,
      "learning_rate": 1.9915186064024165e-05,
      "loss": 0.1587,
      "step": 2294
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3728504750691907,
      "learning_rate": 1.991505710529248e-05,
      "loss": 0.2712,
      "step": 2295
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6509445151659004,
      "learning_rate": 1.9914928049013204e-05,
      "loss": 0.474,
      "step": 2296
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.052437463443468,
      "learning_rate": 1.9914798895187603e-05,
      "loss": 0.5615,
      "step": 2297
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40663467891023647,
      "learning_rate": 1.991466964381695e-05,
      "loss": 0.2588,
      "step": 2298
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8456333912339026,
      "learning_rate": 1.9914540294902513e-05,
      "loss": 0.5568,
      "step": 2299
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.8448963614202915,
      "learning_rate": 1.9914410848445565e-05,
      "loss": 0.4659,
      "step": 2300
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3810617795195018,
      "learning_rate": 1.9914281304447383e-05,
      "loss": 0.2756,
      "step": 2301
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40034819361511537,
      "learning_rate": 1.9914151662909237e-05,
      "loss": 0.2922,
      "step": 2302
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4567151768835319,
      "learning_rate": 1.9914021923832405e-05,
      "loss": 0.1668,
      "step": 2303
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6231212112683545,
      "learning_rate": 1.9913892087218167e-05,
      "loss": 0.2482,
      "step": 2304
    },
    {
      "epoch": 0.07,
      "grad_norm": 3.3380133372900844,
      "learning_rate": 1.9913762153067793e-05,
      "loss": 0.5134,
      "step": 2305
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7018928847035943,
      "learning_rate": 1.9913632121382565e-05,
      "loss": 0.4224,
      "step": 2306
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3552818341197802,
      "learning_rate": 1.9913501992163763e-05,
      "loss": 0.2611,
      "step": 2307
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0629282773799467,
      "learning_rate": 1.9913371765412668e-05,
      "loss": 0.613,
      "step": 2308
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7607798125405769,
      "learning_rate": 1.9913241441130554e-05,
      "loss": 0.4874,
      "step": 2309
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.7544050330838366,
      "learning_rate": 1.9913111019318715e-05,
      "loss": 0.7055,
      "step": 2310
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4064119567352598,
      "learning_rate": 1.9912980499978425e-05,
      "loss": 0.1937,
      "step": 2311
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6814737483632154,
      "learning_rate": 1.9912849883110975e-05,
      "loss": 0.4134,
      "step": 2312
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.31895170412843027,
      "learning_rate": 1.991271916871764e-05,
      "loss": 0.2249,
      "step": 2313
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5708477310838259,
      "learning_rate": 1.9912588356799714e-05,
      "loss": 0.2491,
      "step": 2314
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.88415137652953,
      "learning_rate": 1.9912457447358485e-05,
      "loss": 0.472,
      "step": 2315
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4160805444506543,
      "learning_rate": 1.9912326440395235e-05,
      "loss": 0.2398,
      "step": 2316
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9995736584045192,
      "learning_rate": 1.9912195335911256e-05,
      "loss": 0.5618,
      "step": 2317
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5499799083355724,
      "learning_rate": 1.991206413390784e-05,
      "loss": 0.3226,
      "step": 2318
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6355854783830308,
      "learning_rate": 1.9911932834386273e-05,
      "loss": 0.3899,
      "step": 2319
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.38324412742531266,
      "learning_rate": 1.991180143734785e-05,
      "loss": 0.186,
      "step": 2320
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6650373851406701,
      "learning_rate": 1.9911669942793864e-05,
      "loss": 0.3911,
      "step": 2321
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2960901815581095,
      "learning_rate": 1.9911538350725608e-05,
      "loss": 0.0851,
      "step": 2322
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.347641192536215,
      "learning_rate": 1.9911406661144378e-05,
      "loss": 0.8797,
      "step": 2323
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47135303739589507,
      "learning_rate": 1.9911274874051464e-05,
      "loss": 0.3202,
      "step": 2324
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5330972338696812,
      "learning_rate": 1.9911142989448166e-05,
      "loss": 0.3353,
      "step": 2325
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5574269495529507,
      "learning_rate": 1.9911011007335786e-05,
      "loss": 0.3734,
      "step": 2326
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3347419647212395,
      "learning_rate": 1.9910878927715616e-05,
      "loss": 0.5299,
      "step": 2327
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5249780792184184,
      "learning_rate": 1.9910746750588957e-05,
      "loss": 0.6835,
      "step": 2328
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4987228178222628,
      "learning_rate": 1.9910614475957114e-05,
      "loss": 0.1957,
      "step": 2329
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5804237266242942,
      "learning_rate": 1.9910482103821383e-05,
      "loss": 0.3631,
      "step": 2330
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4598863327143161,
      "learning_rate": 1.9910349634183067e-05,
      "loss": 0.2765,
      "step": 2331
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4487262504052757,
      "learning_rate": 1.9910217067043472e-05,
      "loss": 0.222,
      "step": 2332
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6182013868618069,
      "learning_rate": 1.99100844024039e-05,
      "loss": 0.3112,
      "step": 2333
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4774620156847644,
      "learning_rate": 1.9909951640265656e-05,
      "loss": 0.3399,
      "step": 2334
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.2706507095508726,
      "learning_rate": 1.990981878063005e-05,
      "loss": 0.5765,
      "step": 2335
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4646271865366416,
      "learning_rate": 1.9909685823498383e-05,
      "loss": 0.7505,
      "step": 2336
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4271489261164477,
      "learning_rate": 1.9909552768871968e-05,
      "loss": 0.2987,
      "step": 2337
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5746005698523303,
      "learning_rate": 1.990941961675211e-05,
      "loss": 0.3044,
      "step": 2338
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6616113630000705,
      "learning_rate": 1.990928636714012e-05,
      "loss": 0.2575,
      "step": 2339
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9950927810930129,
      "learning_rate": 1.9909153020037314e-05,
      "loss": 0.4249,
      "step": 2340
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7267216051550978,
      "learning_rate": 1.9909019575445e-05,
      "loss": 0.497,
      "step": 2341
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.561546275938596,
      "learning_rate": 1.990888603336449e-05,
      "loss": 0.1734,
      "step": 2342
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6624136895752073,
      "learning_rate": 1.9908752393797096e-05,
      "loss": 0.3395,
      "step": 2343
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4370254897955004,
      "learning_rate": 1.990861865674414e-05,
      "loss": 0.3029,
      "step": 2344
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7147358804650152,
      "learning_rate": 1.990848482220693e-05,
      "loss": 0.3132,
      "step": 2345
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6364879480411174,
      "learning_rate": 1.9908350890186786e-05,
      "loss": 0.3553,
      "step": 2346
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.886059975857861,
      "learning_rate": 1.9908216860685025e-05,
      "loss": 0.4087,
      "step": 2347
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4912357401638878,
      "learning_rate": 1.9908082733702972e-05,
      "loss": 0.2848,
      "step": 2348
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46786218666190477,
      "learning_rate": 1.9907948509241938e-05,
      "loss": 0.331,
      "step": 2349
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1186803825004457,
      "learning_rate": 1.990781418730324e-05,
      "loss": 0.4726,
      "step": 2350
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5818002333621555,
      "learning_rate": 1.9907679767888217e-05,
      "loss": 0.376,
      "step": 2351
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4503540162363936,
      "learning_rate": 1.990754525099817e-05,
      "loss": 0.2267,
      "step": 2352
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4655750384213841,
      "learning_rate": 1.9907410636634438e-05,
      "loss": 0.7123,
      "step": 2353
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7149598152704313,
      "learning_rate": 1.990727592479834e-05,
      "loss": 0.2632,
      "step": 2354
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.396374515572363,
      "learning_rate": 1.99071411154912e-05,
      "loss": 0.2631,
      "step": 2355
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0556401196962268,
      "learning_rate": 1.9907006208714347e-05,
      "loss": 0.3283,
      "step": 2356
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4920794666415349,
      "learning_rate": 1.9906871204469106e-05,
      "loss": 0.2642,
      "step": 2357
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.392768629173656,
      "learning_rate": 1.9906736102756808e-05,
      "loss": 0.6075,
      "step": 2358
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9696635400372198,
      "learning_rate": 1.990660090357878e-05,
      "loss": 0.4671,
      "step": 2359
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4278889084584596,
      "learning_rate": 1.9906465606936353e-05,
      "loss": 0.3355,
      "step": 2360
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40975297365113356,
      "learning_rate": 1.9906330212830857e-05,
      "loss": 0.2211,
      "step": 2361
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.0809279687639006,
      "learning_rate": 1.9906194721263625e-05,
      "loss": 0.8465,
      "step": 2362
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.32403645279160337,
      "learning_rate": 1.9906059132235992e-05,
      "loss": 0.1712,
      "step": 2363
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5651838560323345,
      "learning_rate": 1.9905923445749288e-05,
      "loss": 0.2431,
      "step": 2364
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4179579858248022,
      "learning_rate": 1.9905787661804846e-05,
      "loss": 0.2258,
      "step": 2365
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.689466474862711,
      "learning_rate": 1.9905651780404008e-05,
      "loss": 0.3466,
      "step": 2366
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47899616377647275,
      "learning_rate": 1.990551580154811e-05,
      "loss": 0.3702,
      "step": 2367
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0164763707056603,
      "learning_rate": 1.9905379725238486e-05,
      "loss": 0.4811,
      "step": 2368
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.3773333966324046,
      "learning_rate": 1.990524355147648e-05,
      "loss": 0.5091,
      "step": 2369
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.48768476402652444,
      "learning_rate": 1.9905107280263425e-05,
      "loss": 0.2137,
      "step": 2370
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7102979584900528,
      "learning_rate": 1.9904970911600668e-05,
      "loss": 0.4717,
      "step": 2371
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4321859354681942,
      "learning_rate": 1.9904834445489548e-05,
      "loss": 0.3068,
      "step": 2372
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4632922685833403,
      "learning_rate": 1.9904697881931407e-05,
      "loss": 0.1694,
      "step": 2373
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.531228813134312,
      "learning_rate": 1.9904561220927588e-05,
      "loss": 0.0924,
      "step": 2374
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4944089774463863,
      "learning_rate": 1.9904424462479438e-05,
      "loss": 0.3319,
      "step": 2375
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0267525540796643,
      "learning_rate": 1.99042876065883e-05,
      "loss": 0.4972,
      "step": 2376
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1156571182223765,
      "learning_rate": 1.9904150653255524e-05,
      "loss": 0.7788,
      "step": 2377
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.45562999970386303,
      "learning_rate": 1.990401360248245e-05,
      "loss": 0.2922,
      "step": 2378
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3954755540000614,
      "learning_rate": 1.9903876454270436e-05,
      "loss": 0.2182,
      "step": 2379
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.6222720264464758,
      "learning_rate": 1.9903739208620827e-05,
      "loss": 0.8995,
      "step": 2380
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0620871801530163,
      "learning_rate": 1.990360186553497e-05,
      "loss": 0.4567,
      "step": 2381
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.38394553580932544,
      "learning_rate": 1.990346442501422e-05,
      "loss": 0.1836,
      "step": 2382
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6610858848011318,
      "learning_rate": 1.9903326887059927e-05,
      "loss": 0.3196,
      "step": 2383
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5507040560249284,
      "learning_rate": 1.990318925167345e-05,
      "loss": 0.3779,
      "step": 2384
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4846283185577167,
      "learning_rate": 1.9903051518856132e-05,
      "loss": 0.2844,
      "step": 2385
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9761454944687968,
      "learning_rate": 1.990291368860934e-05,
      "loss": 0.6162,
      "step": 2386
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0872590894126324,
      "learning_rate": 1.9902775760934423e-05,
      "loss": 0.4973,
      "step": 2387
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5030101558762593,
      "learning_rate": 1.990263773583274e-05,
      "loss": 0.2962,
      "step": 2388
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.579566373763756,
      "learning_rate": 1.990249961330565e-05,
      "loss": 0.2628,
      "step": 2389
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6078037775166755,
      "learning_rate": 1.990236139335451e-05,
      "loss": 0.4116,
      "step": 2390
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3566255314031026,
      "learning_rate": 1.9902223075980675e-05,
      "loss": 0.2156,
      "step": 2391
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.6366004757698804,
      "learning_rate": 1.9902084661185516e-05,
      "loss": 0.3139,
      "step": 2392
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1298446701264437,
      "learning_rate": 1.990194614897039e-05,
      "loss": 0.409,
      "step": 2393
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3591975763472116,
      "learning_rate": 1.9901807539336658e-05,
      "loss": 0.2194,
      "step": 2394
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4359939258815997,
      "learning_rate": 1.9901668832285686e-05,
      "loss": 0.5812,
      "step": 2395
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4564438080906683,
      "learning_rate": 1.9901530027818837e-05,
      "loss": 0.2965,
      "step": 2396
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.852391395998973,
      "learning_rate": 1.990139112593748e-05,
      "loss": 0.3759,
      "step": 2397
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.49067411021874185,
      "learning_rate": 1.9901252126642973e-05,
      "loss": 0.2832,
      "step": 2398
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5538728720683341,
      "learning_rate": 1.9901113029936696e-05,
      "loss": 0.7414,
      "step": 2399
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47282052753999865,
      "learning_rate": 1.990097383582001e-05,
      "loss": 0.2702,
      "step": 2400
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7907486042983629,
      "learning_rate": 1.9900834544294282e-05,
      "loss": 0.5316,
      "step": 2401
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4022929963740973,
      "learning_rate": 1.9900695155360893e-05,
      "loss": 0.2263,
      "step": 2402
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.37331626218533326,
      "learning_rate": 1.99005556690212e-05,
      "loss": 0.2884,
      "step": 2403
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0798424361761667,
      "learning_rate": 1.9900416085276587e-05,
      "loss": 0.3814,
      "step": 2404
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4596083927517276,
      "learning_rate": 1.990027640412842e-05,
      "loss": 0.5371,
      "step": 2405
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7864803111101997,
      "learning_rate": 1.990013662557808e-05,
      "loss": 0.3627,
      "step": 2406
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4681331870602761,
      "learning_rate": 1.9899996749626935e-05,
      "loss": 0.275,
      "step": 2407
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44506141114229025,
      "learning_rate": 1.9899856776276364e-05,
      "loss": 0.3242,
      "step": 2408
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9210230676592281,
      "learning_rate": 1.9899716705527744e-05,
      "loss": 0.4358,
      "step": 2409
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.977796765467007,
      "learning_rate": 1.9899576537382458e-05,
      "loss": 0.5072,
      "step": 2410
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46266612384340955,
      "learning_rate": 1.9899436271841875e-05,
      "loss": 0.2552,
      "step": 2411
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5623877713615943,
      "learning_rate": 1.9899295908907387e-05,
      "loss": 0.2392,
      "step": 2412
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.40928155275592365,
      "learning_rate": 1.9899155448580366e-05,
      "loss": 0.1201,
      "step": 2413
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44859010353900736,
      "learning_rate": 1.9899014890862195e-05,
      "loss": 0.3286,
      "step": 2414
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3790802789296572,
      "learning_rate": 1.989887423575426e-05,
      "loss": 0.2059,
      "step": 2415
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7108541616185341,
      "learning_rate": 1.9898733483257938e-05,
      "loss": 0.4018,
      "step": 2416
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.0844409703985178,
      "learning_rate": 1.9898592633374627e-05,
      "loss": 0.5042,
      "step": 2417
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.9660315032681883,
      "learning_rate": 1.9898451686105698e-05,
      "loss": 0.4681,
      "step": 2418
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4932376105563733,
      "learning_rate": 1.9898310641452548e-05,
      "loss": 0.3391,
      "step": 2419
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4504411479176759,
      "learning_rate": 1.989816949941656e-05,
      "loss": 0.2086,
      "step": 2420
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.47049361941026585,
      "learning_rate": 1.9898028259999123e-05,
      "loss": 0.3576,
      "step": 2421
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3413201903759992,
      "learning_rate": 1.989788692320163e-05,
      "loss": 0.0807,
      "step": 2422
    },
    {
      "epoch": 0.07,
      "grad_norm": 2.1345370843244704,
      "learning_rate": 1.9897745489025464e-05,
      "loss": 0.917,
      "step": 2423
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.4378562569343592,
      "learning_rate": 1.9897603957472022e-05,
      "loss": 0.1738,
      "step": 2424
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5045632608821237,
      "learning_rate": 1.9897462328542698e-05,
      "loss": 0.3293,
      "step": 2425
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46968015667493007,
      "learning_rate": 1.9897320602238882e-05,
      "loss": 0.3138,
      "step": 2426
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.1127413401837696,
      "learning_rate": 1.9897178778561968e-05,
      "loss": 0.5874,
      "step": 2427
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.4105374392924592,
      "learning_rate": 1.9897036857513355e-05,
      "loss": 0.1077,
      "step": 2428
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5077217792644466,
      "learning_rate": 1.9896894839094435e-05,
      "loss": 0.2982,
      "step": 2429
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.2609741845196946,
      "learning_rate": 1.9896752723306605e-05,
      "loss": 0.1329,
      "step": 2430
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.46975507828288005,
      "learning_rate": 1.9896610510151267e-05,
      "loss": 0.2901,
      "step": 2431
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5633813580529403,
      "learning_rate": 1.989646819962982e-05,
      "loss": 0.3879,
      "step": 2432
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.460098207785054,
      "learning_rate": 1.9896325791743664e-05,
      "loss": 0.1825,
      "step": 2433
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5062432347996024,
      "learning_rate": 1.9896183286494195e-05,
      "loss": 0.3502,
      "step": 2434
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7616569899322614,
      "learning_rate": 1.9896040683882818e-05,
      "loss": 0.4725,
      "step": 2435
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7501525387939654,
      "learning_rate": 1.989589798391094e-05,
      "loss": 0.5405,
      "step": 2436
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.44488625566597995,
      "learning_rate": 1.9895755186579956e-05,
      "loss": 0.2835,
      "step": 2437
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5538809074860503,
      "learning_rate": 1.9895612291891283e-05,
      "loss": 0.3139,
      "step": 2438
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5571536836567288,
      "learning_rate": 1.9895469299846314e-05,
      "loss": 0.2708,
      "step": 2439
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3871031592105126,
      "learning_rate": 1.9895326210446468e-05,
      "loss": 0.1552,
      "step": 2440
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.5287560617237288,
      "learning_rate": 1.9895183023693144e-05,
      "loss": 0.6371,
      "step": 2441
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.7306548388535495,
      "learning_rate": 1.989503973958775e-05,
      "loss": 0.4052,
      "step": 2442
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5178268767155664,
      "learning_rate": 1.9894896358131707e-05,
      "loss": 0.2484,
      "step": 2443
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5570069315300485,
      "learning_rate": 1.989475287932641e-05,
      "loss": 0.4213,
      "step": 2444
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.5031579053319065,
      "learning_rate": 1.9894609303173284e-05,
      "loss": 0.3415,
      "step": 2445
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.668312240152447,
      "learning_rate": 1.989446562967373e-05,
      "loss": 0.1951,
      "step": 2446
    },
    {
      "epoch": 0.07,
      "grad_norm": 1.2287721917728205,
      "learning_rate": 1.9894321858829173e-05,
      "loss": 0.504,
      "step": 2447
    },
    {
      "epoch": 0.07,
      "grad_norm": 0.3373056413009252,
      "learning_rate": 1.9894177990641018e-05,
      "loss": 0.1947,
      "step": 2448
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3865821498515026,
      "learning_rate": 1.9894034025110687e-05,
      "loss": 0.2515,
      "step": 2449
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5249304105900161,
      "learning_rate": 1.9893889962239595e-05,
      "loss": 0.3036,
      "step": 2450
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1819107711634647,
      "learning_rate": 1.9893745802029154e-05,
      "loss": 0.5796,
      "step": 2451
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6042865307563515,
      "learning_rate": 1.989360154448079e-05,
      "loss": 0.2365,
      "step": 2452
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7894374675714207,
      "learning_rate": 1.9893457189595915e-05,
      "loss": 0.5469,
      "step": 2453
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8889698726794625,
      "learning_rate": 1.9893312737375954e-05,
      "loss": 0.4497,
      "step": 2454
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6226255938218005,
      "learning_rate": 1.989316818782233e-05,
      "loss": 0.3686,
      "step": 2455
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.38306703299636063,
      "learning_rate": 1.9893023540936457e-05,
      "loss": 0.2004,
      "step": 2456
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4991443539861452,
      "learning_rate": 1.9892878796719767e-05,
      "loss": 0.3748,
      "step": 2457
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.24317096701972593,
      "learning_rate": 1.9892733955173678e-05,
      "loss": 0.0895,
      "step": 2458
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1883739380388376,
      "learning_rate": 1.989258901629962e-05,
      "loss": 0.4868,
      "step": 2459
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8546215602902226,
      "learning_rate": 1.9892443980099015e-05,
      "loss": 0.4639,
      "step": 2460
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4887100458852336,
      "learning_rate": 1.9892298846573287e-05,
      "loss": 0.2688,
      "step": 2461
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4710542750391097,
      "learning_rate": 1.9892153615723873e-05,
      "loss": 0.3256,
      "step": 2462
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3293081376302982,
      "learning_rate": 1.9892008287552196e-05,
      "loss": 0.5309,
      "step": 2463
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.353517323161283,
      "learning_rate": 1.9891862862059686e-05,
      "loss": 0.6463,
      "step": 2464
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41764071800149927,
      "learning_rate": 1.9891717339247772e-05,
      "loss": 0.2198,
      "step": 2465
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.642396302016468,
      "learning_rate": 1.989157171911789e-05,
      "loss": 0.3946,
      "step": 2466
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3610183608335325,
      "learning_rate": 1.989142600167147e-05,
      "loss": 0.2023,
      "step": 2467
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5304380860096839,
      "learning_rate": 1.989128018690995e-05,
      "loss": 0.3958,
      "step": 2468
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7569858322916249,
      "learning_rate": 1.9891134274834757e-05,
      "loss": 0.4157,
      "step": 2469
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5802422904933968,
      "learning_rate": 1.9890988265447332e-05,
      "loss": 0.3273,
      "step": 2470
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6093104075326058,
      "learning_rate": 1.9890842158749107e-05,
      "loss": 0.2946,
      "step": 2471
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3163516717845644,
      "learning_rate": 1.9890695954741527e-05,
      "loss": 0.1089,
      "step": 2472
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4827093015611848,
      "learning_rate": 1.9890549653426023e-05,
      "loss": 0.3489,
      "step": 2473
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3718294655520517,
      "learning_rate": 1.9890403254804035e-05,
      "loss": 0.0792,
      "step": 2474
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5500457107482118,
      "learning_rate": 1.989025675887701e-05,
      "loss": 0.3506,
      "step": 2475
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.34597666099736524,
      "learning_rate": 1.989011016564638e-05,
      "loss": 0.1435,
      "step": 2476
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7816698225181832,
      "learning_rate": 1.9889963475113594e-05,
      "loss": 0.529,
      "step": 2477
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8409839850276869,
      "learning_rate": 1.9889816687280095e-05,
      "loss": 0.4792,
      "step": 2478
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5392515652140164,
      "learning_rate": 1.9889669802147322e-05,
      "loss": 0.311,
      "step": 2479
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3888495262910305,
      "learning_rate": 1.9889522819716728e-05,
      "loss": 0.2735,
      "step": 2480
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5872221374825055,
      "learning_rate": 1.988937573998975e-05,
      "loss": 0.2783,
      "step": 2481
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2632380419255134,
      "learning_rate": 1.988922856296784e-05,
      "loss": 0.334,
      "step": 2482
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1257098632257794,
      "learning_rate": 1.9889081288652444e-05,
      "loss": 0.5521,
      "step": 2483
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3996615257615847,
      "learning_rate": 1.9888933917045013e-05,
      "loss": 0.2459,
      "step": 2484
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5240952116973452,
      "learning_rate": 1.9888786448147003e-05,
      "loss": 0.3162,
      "step": 2485
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6039655900578003,
      "learning_rate": 1.988863888195985e-05,
      "loss": 0.3907,
      "step": 2486
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7713671919686323,
      "learning_rate": 1.9888491218485017e-05,
      "loss": 0.3837,
      "step": 2487
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.452290139362517,
      "learning_rate": 1.9888343457723955e-05,
      "loss": 0.315,
      "step": 2488
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2941680556853339,
      "learning_rate": 1.9888195599678114e-05,
      "loss": 0.6205,
      "step": 2489
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4178354679827409,
      "learning_rate": 1.9888047644348953e-05,
      "loss": 0.1608,
      "step": 2490
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4114482554484251,
      "learning_rate": 1.9887899591737922e-05,
      "loss": 0.2917,
      "step": 2491
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.78481872556425,
      "learning_rate": 1.9887751441846482e-05,
      "loss": 0.3744,
      "step": 2492
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.45096163098953296,
      "learning_rate": 1.9887603194676094e-05,
      "loss": 0.2502,
      "step": 2493
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1406177223168261,
      "learning_rate": 1.988745485022821e-05,
      "loss": 0.5963,
      "step": 2494
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.80504909315514,
      "learning_rate": 1.9887306408504293e-05,
      "loss": 0.4612,
      "step": 2495
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6024452988618249,
      "learning_rate": 1.98871578695058e-05,
      "loss": 0.4035,
      "step": 2496
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3900664572752684,
      "learning_rate": 1.9887009233234197e-05,
      "loss": 0.1833,
      "step": 2497
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.43999625632652184,
      "learning_rate": 1.988686049969094e-05,
      "loss": 0.2772,
      "step": 2498
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.39650223126683287,
      "learning_rate": 1.98867116688775e-05,
      "loss": 0.1931,
      "step": 2499
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2679829269856377,
      "learning_rate": 1.9886562740795337e-05,
      "loss": 0.4021,
      "step": 2500
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3411272502723266,
      "learning_rate": 1.9886413715445917e-05,
      "loss": 0.8051,
      "step": 2501
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3429738980622655,
      "learning_rate": 1.98862645928307e-05,
      "loss": 0.2122,
      "step": 2502
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.572918785290539,
      "learning_rate": 1.9886115372951165e-05,
      "loss": 0.4136,
      "step": 2503
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5402046004626597,
      "learning_rate": 1.988596605580877e-05,
      "loss": 0.3771,
      "step": 2504
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.936239616126333,
      "learning_rate": 1.9885816641404992e-05,
      "loss": 0.7267,
      "step": 2505
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46774605015623555,
      "learning_rate": 1.98856671297413e-05,
      "loss": 0.213,
      "step": 2506
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6974634388269304,
      "learning_rate": 1.9885517520819154e-05,
      "loss": 0.4114,
      "step": 2507
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.27507410965728274,
      "learning_rate": 1.9885367814640036e-05,
      "loss": 0.1626,
      "step": 2508
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.682331268231879,
      "learning_rate": 1.988521801120542e-05,
      "loss": 0.4429,
      "step": 2509
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9719453992819844,
      "learning_rate": 1.9885068110516772e-05,
      "loss": 0.2931,
      "step": 2510
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3941525489863427,
      "learning_rate": 1.9884918112575574e-05,
      "loss": 0.2786,
      "step": 2511
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6698801840575369,
      "learning_rate": 1.98847680173833e-05,
      "loss": 0.5509,
      "step": 2512
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3726130481072951,
      "learning_rate": 1.9884617824941426e-05,
      "loss": 0.3571,
      "step": 2513
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7018036982807323,
      "learning_rate": 1.9884467535251424e-05,
      "loss": 0.4166,
      "step": 2514
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3812100421989752,
      "learning_rate": 1.9884317148314783e-05,
      "loss": 0.2307,
      "step": 2515
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7325107400720774,
      "learning_rate": 1.988416666413298e-05,
      "loss": 0.4625,
      "step": 2516
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2382793568527626,
      "learning_rate": 1.9884016082707487e-05,
      "loss": 0.4022,
      "step": 2517
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.34762257810777814,
      "learning_rate": 1.9883865404039793e-05,
      "loss": 0.1611,
      "step": 2518
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6287432914859435,
      "learning_rate": 1.9883714628131382e-05,
      "loss": 0.322,
      "step": 2519
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48217047381480577,
      "learning_rate": 1.988356375498373e-05,
      "loss": 0.3579,
      "step": 2520
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5656320890007368,
      "learning_rate": 1.9883412784598328e-05,
      "loss": 0.3103,
      "step": 2521
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6346033947655668,
      "learning_rate": 1.988326171697666e-05,
      "loss": 0.3643,
      "step": 2522
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6345806447326412,
      "learning_rate": 1.9883110552120212e-05,
      "loss": 0.612,
      "step": 2523
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42579068288883504,
      "learning_rate": 1.9882959290030465e-05,
      "loss": 0.077,
      "step": 2524
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4963087974012725,
      "learning_rate": 1.9882807930708918e-05,
      "loss": 0.3132,
      "step": 2525
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.27583042296569954,
      "learning_rate": 1.9882656474157055e-05,
      "loss": 0.127,
      "step": 2526
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41664207467854175,
      "learning_rate": 1.9882504920376364e-05,
      "loss": 0.3108,
      "step": 2527
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.167208317825821,
      "learning_rate": 1.9882353269368335e-05,
      "loss": 0.4397,
      "step": 2528
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5429600266479384,
      "learning_rate": 1.9882201521134466e-05,
      "loss": 0.3005,
      "step": 2529
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5737227367603034,
      "learning_rate": 1.9882049675676245e-05,
      "loss": 0.3366,
      "step": 2530
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0430436667408047,
      "learning_rate": 1.988189773299517e-05,
      "loss": 0.8697,
      "step": 2531
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4900564263613011,
      "learning_rate": 1.988174569309273e-05,
      "loss": 0.2699,
      "step": 2532
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48071685614932624,
      "learning_rate": 1.9881593555970428e-05,
      "loss": 0.3069,
      "step": 2533
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42936637355907564,
      "learning_rate": 1.9881441321629755e-05,
      "loss": 0.248,
      "step": 2534
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.392903075392397,
      "learning_rate": 1.988128899007221e-05,
      "loss": 0.7316,
      "step": 2535
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41393760387967093,
      "learning_rate": 1.9881136561299295e-05,
      "loss": 0.2118,
      "step": 2536
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.748549991044365,
      "learning_rate": 1.9880984035312507e-05,
      "loss": 0.4087,
      "step": 2537
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4610092716135914,
      "learning_rate": 1.9880831412113346e-05,
      "loss": 0.3041,
      "step": 2538
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.34222724434104085,
      "learning_rate": 1.9880678691703314e-05,
      "loss": 0.2094,
      "step": 2539
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.3856858299336237,
      "learning_rate": 1.9880525874083916e-05,
      "loss": 0.9291,
      "step": 2540
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.282738248858967,
      "learning_rate": 1.9880372959256648e-05,
      "loss": 0.5663,
      "step": 2541
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7578260893652423,
      "learning_rate": 1.9880219947223025e-05,
      "loss": 0.3885,
      "step": 2542
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5457899331953815,
      "learning_rate": 1.9880066837984545e-05,
      "loss": 0.2551,
      "step": 2543
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.6480979814634746,
      "learning_rate": 1.987991363154272e-05,
      "loss": 0.6419,
      "step": 2544
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4270763446552489,
      "learning_rate": 1.987976032789905e-05,
      "loss": 0.3264,
      "step": 2545
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.499963462015848,
      "learning_rate": 1.9879606927055045e-05,
      "loss": 0.3654,
      "step": 2546
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3828975467080931,
      "learning_rate": 1.987945342901222e-05,
      "loss": 0.2192,
      "step": 2547
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7259664896570491,
      "learning_rate": 1.9879299833772083e-05,
      "loss": 0.299,
      "step": 2548
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2168017519609717,
      "learning_rate": 1.9879146141336138e-05,
      "loss": 0.4672,
      "step": 2549
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42766614471323117,
      "learning_rate": 1.9878992351705903e-05,
      "loss": 0.2917,
      "step": 2550
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7703133526070699,
      "learning_rate": 1.9878838464882894e-05,
      "loss": 0.286,
      "step": 2551
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5190780436165613,
      "learning_rate": 1.9878684480868623e-05,
      "loss": 0.2729,
      "step": 2552
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.605455962058836,
      "learning_rate": 1.98785303996646e-05,
      "loss": 0.7004,
      "step": 2553
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6168936269558797,
      "learning_rate": 1.9878376221272345e-05,
      "loss": 0.4393,
      "step": 2554
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6079531014652053,
      "learning_rate": 1.9878221945693376e-05,
      "loss": 0.2575,
      "step": 2555
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.40327904633979655,
      "learning_rate": 1.9878067572929208e-05,
      "loss": 0.1909,
      "step": 2556
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.398453198237851,
      "learning_rate": 1.987791310298136e-05,
      "loss": 0.3239,
      "step": 2557
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42111691618027347,
      "learning_rate": 1.9877758535851355e-05,
      "loss": 0.1688,
      "step": 2558
    },
    {
      "epoch": 0.08,
      "grad_norm": 5.19214253019828,
      "learning_rate": 1.987760387154071e-05,
      "loss": 0.8672,
      "step": 2559
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6431502338601814,
      "learning_rate": 1.9877449110050947e-05,
      "loss": 0.2352,
      "step": 2560
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46329981033312884,
      "learning_rate": 1.987729425138359e-05,
      "loss": 0.3149,
      "step": 2561
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4861236961039765,
      "learning_rate": 1.9877139295540163e-05,
      "loss": 0.3447,
      "step": 2562
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5057260571830046,
      "learning_rate": 1.987698424252219e-05,
      "loss": 0.3124,
      "step": 2563
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9847091116697284,
      "learning_rate": 1.9876829092331194e-05,
      "loss": 0.623,
      "step": 2564
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.45204382799567006,
      "learning_rate": 1.9876673844968706e-05,
      "loss": 0.1963,
      "step": 2565
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4047857778051316,
      "learning_rate": 1.987651850043625e-05,
      "loss": 0.2186,
      "step": 2566
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1330657250646023,
      "learning_rate": 1.9876363058735352e-05,
      "loss": 0.3486,
      "step": 2567
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6191120414033356,
      "learning_rate": 1.987620751986755e-05,
      "loss": 0.4095,
      "step": 2568
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.35892800708738065,
      "learning_rate": 1.9876051883834364e-05,
      "loss": 0.2067,
      "step": 2569
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5056288776869012,
      "learning_rate": 1.987589615063733e-05,
      "loss": 0.3324,
      "step": 2570
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2132471644866263,
      "learning_rate": 1.9875740320277987e-05,
      "loss": 0.4325,
      "step": 2571
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.735294314383823,
      "learning_rate": 1.9875584392757854e-05,
      "loss": 0.5227,
      "step": 2572
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5719744239408906,
      "learning_rate": 1.9875428368078475e-05,
      "loss": 0.2906,
      "step": 2573
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5301898236673804,
      "learning_rate": 1.987527224624138e-05,
      "loss": 0.3143,
      "step": 2574
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4189514039050631,
      "learning_rate": 1.987511602724811e-05,
      "loss": 0.2704,
      "step": 2575
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2147795584424881,
      "learning_rate": 1.98749597111002e-05,
      "loss": 0.1071,
      "step": 2576
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8067445823799149,
      "learning_rate": 1.9874803297799183e-05,
      "loss": 0.9366,
      "step": 2577
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7880388774117284,
      "learning_rate": 1.98746467873466e-05,
      "loss": 0.3417,
      "step": 2578
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5852804806026695,
      "learning_rate": 1.9874490179744e-05,
      "loss": 0.3672,
      "step": 2579
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4137957403997683,
      "learning_rate": 1.9874333474992916e-05,
      "loss": 0.3242,
      "step": 2580
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5338000854541078,
      "learning_rate": 1.9874176673094885e-05,
      "loss": 0.4167,
      "step": 2581
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0630515576714767,
      "learning_rate": 1.987401977405146e-05,
      "loss": 0.3197,
      "step": 2582
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7421580179779756,
      "learning_rate": 1.9873862777864175e-05,
      "loss": 0.294,
      "step": 2583
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.32185637479843165,
      "learning_rate": 1.9873705684534582e-05,
      "loss": 0.2198,
      "step": 2584
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48228108027749017,
      "learning_rate": 1.987354849406422e-05,
      "loss": 0.2451,
      "step": 2585
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.463024517507618,
      "learning_rate": 1.9873391206454646e-05,
      "loss": 0.2894,
      "step": 2586
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.110799824474495,
      "learning_rate": 1.9873233821707392e-05,
      "loss": 0.6545,
      "step": 2587
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3996514584320875,
      "learning_rate": 1.987307633982402e-05,
      "loss": 0.2215,
      "step": 2588
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8209258952830693,
      "learning_rate": 1.9872918760806074e-05,
      "loss": 0.4625,
      "step": 2589
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.042008219168712,
      "learning_rate": 1.9872761084655102e-05,
      "loss": 0.8755,
      "step": 2590
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.526197540327772,
      "learning_rate": 1.9872603311372664e-05,
      "loss": 0.3076,
      "step": 2591
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4635848547913331,
      "learning_rate": 1.9872445440960297e-05,
      "loss": 0.3003,
      "step": 2592
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3370461824691647,
      "learning_rate": 1.987228747341957e-05,
      "loss": 0.2167,
      "step": 2593
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5809279928670591,
      "learning_rate": 1.9872129408752027e-05,
      "loss": 0.2885,
      "step": 2594
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.164839320745597,
      "learning_rate": 1.9871971246959228e-05,
      "loss": 0.5087,
      "step": 2595
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0541069678607038,
      "learning_rate": 1.9871812988042728e-05,
      "loss": 0.5462,
      "step": 2596
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.35940063531623706,
      "learning_rate": 1.9871654632004085e-05,
      "loss": 0.213,
      "step": 2597
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.883120809002606,
      "learning_rate": 1.9871496178844854e-05,
      "loss": 0.9244,
      "step": 2598
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4853545122468549,
      "learning_rate": 1.9871337628566594e-05,
      "loss": 0.3066,
      "step": 2599
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.691625960250169,
      "learning_rate": 1.9871178981170867e-05,
      "loss": 0.7418,
      "step": 2600
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42776230790167125,
      "learning_rate": 1.9871020236659235e-05,
      "loss": 0.1944,
      "step": 2601
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5958809916206221,
      "learning_rate": 1.9870861395033256e-05,
      "loss": 0.2489,
      "step": 2602
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.366349827716197,
      "learning_rate": 1.9870702456294495e-05,
      "loss": 0.2017,
      "step": 2603
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.45319531966523924,
      "learning_rate": 1.9870543420444515e-05,
      "loss": 0.3081,
      "step": 2604
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8986837090826182,
      "learning_rate": 1.987038428748488e-05,
      "loss": 0.5035,
      "step": 2605
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44541442208767906,
      "learning_rate": 1.9870225057417162e-05,
      "loss": 0.205,
      "step": 2606
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.615410576150907,
      "learning_rate": 1.987006573024292e-05,
      "loss": 0.4382,
      "step": 2607
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8948017125361588,
      "learning_rate": 1.986990630596372e-05,
      "loss": 0.4569,
      "step": 2608
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5846130488395066,
      "learning_rate": 1.9869746784581133e-05,
      "loss": 0.3746,
      "step": 2609
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42880546067194447,
      "learning_rate": 1.9869587166096735e-05,
      "loss": 0.2313,
      "step": 2610
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41396399322551536,
      "learning_rate": 1.9869427450512088e-05,
      "loss": 0.2936,
      "step": 2611
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.2621058667432115,
      "learning_rate": 1.9869267637828765e-05,
      "loss": 0.1044,
      "step": 2612
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8470765874993353,
      "learning_rate": 1.9869107728048344e-05,
      "loss": 0.5646,
      "step": 2613
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8370332528109596,
      "learning_rate": 1.986894772117239e-05,
      "loss": 0.459,
      "step": 2614
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.39397589342977174,
      "learning_rate": 1.9868787617202483e-05,
      "loss": 0.215,
      "step": 2615
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4506248899796773,
      "learning_rate": 1.9868627416140194e-05,
      "loss": 0.3458,
      "step": 2616
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.361431722658811,
      "learning_rate": 1.9868467117987103e-05,
      "loss": 0.1135,
      "step": 2617
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.513605106241938,
      "learning_rate": 1.9868306722744788e-05,
      "loss": 0.709,
      "step": 2618
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4676197663146046,
      "learning_rate": 1.986814623041482e-05,
      "loss": 0.0922,
      "step": 2619
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5643953021719603,
      "learning_rate": 1.9867985640998782e-05,
      "loss": 0.3591,
      "step": 2620
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.36637999892466305,
      "learning_rate": 1.986782495449826e-05,
      "loss": 0.1924,
      "step": 2621
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.40083050928856667,
      "learning_rate": 1.986766417091482e-05,
      "loss": 0.3595,
      "step": 2622
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8248135905783988,
      "learning_rate": 1.986750329025006e-05,
      "loss": 0.4894,
      "step": 2623
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49484712277164866,
      "learning_rate": 1.9867342312505556e-05,
      "loss": 0.3059,
      "step": 2624
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5029865518856217,
      "learning_rate": 1.986718123768289e-05,
      "loss": 0.2496,
      "step": 2625
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.539299109017532,
      "learning_rate": 1.9867020065783648e-05,
      "loss": 0.2616,
      "step": 2626
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4683085345113678,
      "learning_rate": 1.9866858796809412e-05,
      "loss": 0.2985,
      "step": 2627
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4709073280146877,
      "learning_rate": 1.986669743076178e-05,
      "loss": 0.2943,
      "step": 2628
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4711082500497439,
      "learning_rate": 1.9866535967642327e-05,
      "loss": 0.2629,
      "step": 2629
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8786329301220754,
      "learning_rate": 1.986637440745265e-05,
      "loss": 0.4454,
      "step": 2630
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8711436404145841,
      "learning_rate": 1.9866212750194333e-05,
      "loss": 0.5196,
      "step": 2631
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5028050101548731,
      "learning_rate": 1.9866050995868965e-05,
      "loss": 0.3012,
      "step": 2632
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.1564889283595707,
      "learning_rate": 1.9865889144478144e-05,
      "loss": 0.4286,
      "step": 2633
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3933735882528053,
      "learning_rate": 1.986572719602346e-05,
      "loss": 0.2763,
      "step": 2634
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.35628979705050995,
      "learning_rate": 1.986556515050651e-05,
      "loss": 0.1845,
      "step": 2635
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2602243739464483,
      "learning_rate": 1.9865403007928875e-05,
      "loss": 0.0927,
      "step": 2636
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8850014183614228,
      "learning_rate": 1.9865240768292165e-05,
      "loss": 0.9181,
      "step": 2637
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.39464831666338823,
      "learning_rate": 1.9865078431597968e-05,
      "loss": 0.2293,
      "step": 2638
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.204293130240816,
      "learning_rate": 1.9864915997847883e-05,
      "loss": 0.5611,
      "step": 2639
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48952285888121505,
      "learning_rate": 1.9864753467043513e-05,
      "loss": 0.3204,
      "step": 2640
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4363372780395336,
      "learning_rate": 1.9864590839186446e-05,
      "loss": 0.7605,
      "step": 2641
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5717124171427158,
      "learning_rate": 1.9864428114278293e-05,
      "loss": 0.1831,
      "step": 2642
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3120877941103709,
      "learning_rate": 1.9864265292320648e-05,
      "loss": 0.2033,
      "step": 2643
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49447783384493776,
      "learning_rate": 1.9864102373315115e-05,
      "loss": 0.2329,
      "step": 2644
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5195151437364919,
      "learning_rate": 1.98639393572633e-05,
      "loss": 0.2764,
      "step": 2645
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5749115719028882,
      "learning_rate": 1.98637762441668e-05,
      "loss": 0.3957,
      "step": 2646
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4561826659049693,
      "learning_rate": 1.9863613034027224e-05,
      "loss": 0.2802,
      "step": 2647
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.037994911022978,
      "learning_rate": 1.986344972684618e-05,
      "loss": 0.5828,
      "step": 2648
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9216753256431317,
      "learning_rate": 1.9863286322625273e-05,
      "loss": 0.435,
      "step": 2649
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7706392617248445,
      "learning_rate": 1.9863122821366104e-05,
      "loss": 0.8183,
      "step": 2650
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5150210974486834,
      "learning_rate": 1.986295922307029e-05,
      "loss": 0.2008,
      "step": 2651
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7205883999851901,
      "learning_rate": 1.986279552773944e-05,
      "loss": 0.4078,
      "step": 2652
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3473902838304371,
      "learning_rate": 1.9862631735375163e-05,
      "loss": 0.229,
      "step": 2653
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6035169908834597,
      "learning_rate": 1.9862467845979067e-05,
      "loss": 0.2378,
      "step": 2654
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0282380684701504,
      "learning_rate": 1.986230385955277e-05,
      "loss": 0.3966,
      "step": 2655
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5026064093544258,
      "learning_rate": 1.9862139776097883e-05,
      "loss": 0.2101,
      "step": 2656
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6882018387796499,
      "learning_rate": 1.9861975595616013e-05,
      "loss": 0.4452,
      "step": 2657
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.45270655360396833,
      "learning_rate": 1.986181131810879e-05,
      "loss": 0.2854,
      "step": 2658
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5057196397957908,
      "learning_rate": 1.9861646943577816e-05,
      "loss": 0.8037,
      "step": 2659
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5255609357366617,
      "learning_rate": 1.986148247202472e-05,
      "loss": 0.0819,
      "step": 2660
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5047605197639821,
      "learning_rate": 1.986131790345111e-05,
      "loss": 0.3291,
      "step": 2661
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.24127158472553395,
      "learning_rate": 1.9861153237858612e-05,
      "loss": 0.0745,
      "step": 2662
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.422657988719516,
      "learning_rate": 1.9860988475248844e-05,
      "loss": 0.3494,
      "step": 2663
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.050309757024405,
      "learning_rate": 1.9860823615623425e-05,
      "loss": 0.4525,
      "step": 2664
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5426416221227767,
      "learning_rate": 1.986065865898398e-05,
      "loss": 0.3078,
      "step": 2665
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6224898975570619,
      "learning_rate": 1.986049360533213e-05,
      "loss": 0.3617,
      "step": 2666
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8654653431071473,
      "learning_rate": 1.98603284546695e-05,
      "loss": 0.7726,
      "step": 2667
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3247406989699515,
      "learning_rate": 1.9860163206997714e-05,
      "loss": 0.5485,
      "step": 2668
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4121097020910126,
      "learning_rate": 1.98599978623184e-05,
      "loss": 0.2301,
      "step": 2669
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.39197408798420436,
      "learning_rate": 1.9859832420633177e-05,
      "loss": 0.3097,
      "step": 2670
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4021050873177243,
      "learning_rate": 1.9859666881943684e-05,
      "loss": 0.1285,
      "step": 2671
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5428127725071364,
      "learning_rate": 1.985950124625154e-05,
      "loss": 0.3531,
      "step": 2672
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8944913354951117,
      "learning_rate": 1.985933551355838e-05,
      "loss": 0.4521,
      "step": 2673
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5181161184244384,
      "learning_rate": 1.9859169683865833e-05,
      "loss": 0.2958,
      "step": 2674
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5732177303395601,
      "learning_rate": 1.9859003757175534e-05,
      "loss": 0.2396,
      "step": 2675
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.44266078060299935,
      "learning_rate": 1.9858837733489107e-05,
      "loss": 0.3387,
      "step": 2676
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0743383624045246,
      "learning_rate": 1.985867161280819e-05,
      "loss": 0.5582,
      "step": 2677
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8416084303088409,
      "learning_rate": 1.9858505395134422e-05,
      "loss": 0.3479,
      "step": 2678
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4051371000293946,
      "learning_rate": 1.985833908046943e-05,
      "loss": 0.25,
      "step": 2679
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.0606923299887188,
      "learning_rate": 1.9858172668814856e-05,
      "loss": 0.597,
      "step": 2680
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3229640008344072,
      "learning_rate": 1.9858006160172336e-05,
      "loss": 0.2521,
      "step": 2681
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.725843293055468,
      "learning_rate": 1.9857839554543507e-05,
      "loss": 0.3848,
      "step": 2682
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3834115496276986,
      "learning_rate": 1.985767285193001e-05,
      "loss": 0.4113,
      "step": 2683
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.32558343590311506,
      "learning_rate": 1.9857506052333484e-05,
      "loss": 0.1927,
      "step": 2684
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.271049406053352,
      "learning_rate": 1.985733915575557e-05,
      "loss": 0.9287,
      "step": 2685
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.497357348541388,
      "learning_rate": 1.9857172162197906e-05,
      "loss": 0.3379,
      "step": 2686
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6135371078836072,
      "learning_rate": 1.9857005071662142e-05,
      "loss": 0.4166,
      "step": 2687
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.33720865086669327,
      "learning_rate": 1.985683788414992e-05,
      "loss": 0.2162,
      "step": 2688
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.1341296445298523,
      "learning_rate": 1.985667059966288e-05,
      "loss": 0.8619,
      "step": 2689
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7337610171136344,
      "learning_rate": 1.9856503218202674e-05,
      "loss": 0.4402,
      "step": 2690
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5753455948635117,
      "learning_rate": 1.9856335739770943e-05,
      "loss": 0.245,
      "step": 2691
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3109616656754481,
      "learning_rate": 1.9856168164369343e-05,
      "loss": 0.1474,
      "step": 2692
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6264663848029998,
      "learning_rate": 1.9856000491999515e-05,
      "loss": 0.4117,
      "step": 2693
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41074780620650125,
      "learning_rate": 1.985583272266311e-05,
      "loss": 0.2902,
      "step": 2694
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2212805430626963,
      "learning_rate": 1.985566485636178e-05,
      "loss": 0.53,
      "step": 2695
    },
    {
      "epoch": 0.08,
      "grad_norm": 4.2547986929211765,
      "learning_rate": 1.9855496893097176e-05,
      "loss": 0.5221,
      "step": 2696
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3904756721438764,
      "learning_rate": 1.9855328832870952e-05,
      "loss": 0.2315,
      "step": 2697
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9646447567781676,
      "learning_rate": 1.985516067568476e-05,
      "loss": 0.514,
      "step": 2698
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4185386028068234,
      "learning_rate": 1.9854992421540255e-05,
      "loss": 0.2822,
      "step": 2699
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9365275190916157,
      "learning_rate": 1.985482407043909e-05,
      "loss": 0.5013,
      "step": 2700
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.31225948799414366,
      "learning_rate": 1.9854655622382925e-05,
      "loss": 0.1447,
      "step": 2701
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4219698673721544,
      "learning_rate": 1.9854487077373416e-05,
      "loss": 0.2474,
      "step": 2702
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5254692493988884,
      "learning_rate": 1.9854318435412218e-05,
      "loss": 0.3928,
      "step": 2703
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.5946525498683106,
      "learning_rate": 1.9854149696500994e-05,
      "loss": 0.8801,
      "step": 2704
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.37804161581461804,
      "learning_rate": 1.98539808606414e-05,
      "loss": 0.2603,
      "step": 2705
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5828571514661745,
      "learning_rate": 1.9853811927835106e-05,
      "loss": 0.3435,
      "step": 2706
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5219637027621965,
      "learning_rate": 1.9853642898083763e-05,
      "loss": 0.3211,
      "step": 2707
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8653190736966858,
      "learning_rate": 1.985347377138904e-05,
      "loss": 0.4538,
      "step": 2708
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.522971296467486,
      "learning_rate": 1.98533045477526e-05,
      "loss": 0.2996,
      "step": 2709
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5262694576401478,
      "learning_rate": 1.9853135227176107e-05,
      "loss": 0.0764,
      "step": 2710
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3670355034393411,
      "learning_rate": 1.985296580966123e-05,
      "loss": 0.2526,
      "step": 2711
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4395237750265028,
      "learning_rate": 1.985279629520963e-05,
      "loss": 0.2837,
      "step": 2712
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8100448819130814,
      "learning_rate": 1.985262668382298e-05,
      "loss": 0.9903,
      "step": 2713
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9117159713308935,
      "learning_rate": 1.9852456975502946e-05,
      "loss": 0.4419,
      "step": 2714
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.506813792274228,
      "learning_rate": 1.98522871702512e-05,
      "loss": 0.3004,
      "step": 2715
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4918844796590224,
      "learning_rate": 1.985211726806941e-05,
      "loss": 0.3753,
      "step": 2716
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.41434829672836365,
      "learning_rate": 1.985194726895925e-05,
      "loss": 0.3096,
      "step": 2717
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.8013968237710871,
      "learning_rate": 1.985177717292239e-05,
      "loss": 0.1144,
      "step": 2718
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4842170833561563,
      "learning_rate": 1.9851606979960502e-05,
      "loss": 0.21,
      "step": 2719
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.29575959772243554,
      "learning_rate": 1.985143669007526e-05,
      "loss": 0.1803,
      "step": 2720
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.3744349711674022,
      "learning_rate": 1.985126630326835e-05,
      "loss": 0.6363,
      "step": 2721
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.0810335960367277,
      "learning_rate": 1.985109581954144e-05,
      "loss": 0.811,
      "step": 2722
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.36046169235314895,
      "learning_rate": 1.9850925238896205e-05,
      "loss": 0.2748,
      "step": 2723
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4431429956023485,
      "learning_rate": 1.9850754561334325e-05,
      "loss": 0.308,
      "step": 2724
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.757036359105908,
      "learning_rate": 1.9850583786857483e-05,
      "loss": 0.4494,
      "step": 2725
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4757118675966163,
      "learning_rate": 1.9850412915467356e-05,
      "loss": 0.6897,
      "step": 2726
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7225145597816207,
      "learning_rate": 1.9850241947165625e-05,
      "loss": 0.4164,
      "step": 2727
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.59339355752286,
      "learning_rate": 1.9850070881953974e-05,
      "loss": 0.2265,
      "step": 2728
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3888616335401785,
      "learning_rate": 1.9849899719834082e-05,
      "loss": 0.2674,
      "step": 2729
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4253579192064456,
      "learning_rate": 1.984972846080764e-05,
      "loss": 0.2731,
      "step": 2730
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.7427505606973688,
      "learning_rate": 1.9849557104876322e-05,
      "loss": 0.7248,
      "step": 2731
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8753619231770652,
      "learning_rate": 1.9849385652041825e-05,
      "loss": 0.5184,
      "step": 2732
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3931677163176497,
      "learning_rate": 1.9849214102305826e-05,
      "loss": 0.1675,
      "step": 2733
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6006763800252166,
      "learning_rate": 1.984904245567002e-05,
      "loss": 0.3065,
      "step": 2734
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.47353117290762103,
      "learning_rate": 1.98488707121361e-05,
      "loss": 0.3362,
      "step": 2735
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4630895146798977,
      "learning_rate": 1.9848698871705743e-05,
      "loss": 0.364,
      "step": 2736
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6704927311521387,
      "learning_rate": 1.9848526934380644e-05,
      "loss": 0.1968,
      "step": 2737
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4642681786545334,
      "learning_rate": 1.98483549001625e-05,
      "loss": 0.2304,
      "step": 2738
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6267675561000589,
      "learning_rate": 1.9848182769053e-05,
      "loss": 0.3345,
      "step": 2739
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49623546075283176,
      "learning_rate": 1.9848010541053836e-05,
      "loss": 0.3172,
      "step": 2740
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2109358711958762,
      "learning_rate": 1.9847838216166705e-05,
      "loss": 0.6069,
      "step": 2741
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.43602836809957934,
      "learning_rate": 1.98476657943933e-05,
      "loss": 0.257,
      "step": 2742
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5843859298891526,
      "learning_rate": 1.984749327573532e-05,
      "loss": 0.4361,
      "step": 2743
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8972279953479751,
      "learning_rate": 1.9847320660194464e-05,
      "loss": 0.0229,
      "step": 2744
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.692682256726189,
      "learning_rate": 1.984714794777242e-05,
      "loss": 0.8139,
      "step": 2745
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4693529599106682,
      "learning_rate": 1.98469751384709e-05,
      "loss": 0.2873,
      "step": 2746
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3742415485510754,
      "learning_rate": 1.9846802232291593e-05,
      "loss": 0.2469,
      "step": 2747
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.570129473154517,
      "learning_rate": 1.984662922923621e-05,
      "loss": 0.2524,
      "step": 2748
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.9821147231936354,
      "learning_rate": 1.9846456129306447e-05,
      "loss": 0.4307,
      "step": 2749
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8677180108512729,
      "learning_rate": 1.984628293250401e-05,
      "loss": 0.5882,
      "step": 2750
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.42667301380239064,
      "learning_rate": 1.9846109638830603e-05,
      "loss": 0.2166,
      "step": 2751
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4510687885789769,
      "learning_rate": 1.9845936248287926e-05,
      "loss": 0.2728,
      "step": 2752
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.40020602469124106,
      "learning_rate": 1.984576276087769e-05,
      "loss": 0.2672,
      "step": 2753
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.560421477739964,
      "learning_rate": 1.98455891766016e-05,
      "loss": 0.7195,
      "step": 2754
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4408988343403528,
      "learning_rate": 1.9845415495461364e-05,
      "loss": 0.1516,
      "step": 2755
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.3584487444731583,
      "learning_rate": 1.9845241717458692e-05,
      "loss": 0.2592,
      "step": 2756
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2076693832908392,
      "learning_rate": 1.984506784259529e-05,
      "loss": 0.555,
      "step": 2757
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6391401577457677,
      "learning_rate": 1.9844893870872874e-05,
      "loss": 0.4264,
      "step": 2758
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5995384504549274,
      "learning_rate": 1.984471980229315e-05,
      "loss": 0.3677,
      "step": 2759
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.46204518691581015,
      "learning_rate": 1.9844545636857835e-05,
      "loss": 0.182,
      "step": 2760
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.6548916248912457,
      "learning_rate": 1.9844371374568642e-05,
      "loss": 0.3932,
      "step": 2761
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.36468054612534145,
      "learning_rate": 1.9844197015427283e-05,
      "loss": 0.1596,
      "step": 2762
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2833080031336186,
      "learning_rate": 1.9844022559435472e-05,
      "loss": 0.6342,
      "step": 2763
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4310563144532898,
      "learning_rate": 1.984384800659493e-05,
      "loss": 0.2694,
      "step": 2764
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.49304238472187284,
      "learning_rate": 1.9843673356907372e-05,
      "loss": 0.2888,
      "step": 2765
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.5617575439246695,
      "learning_rate": 1.984349861037452e-05,
      "loss": 0.3262,
      "step": 2766
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.7279290429072082,
      "learning_rate": 1.9843323766998085e-05,
      "loss": 0.5478,
      "step": 2767
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.2475366839995514,
      "learning_rate": 1.984314882677979e-05,
      "loss": 0.5481,
      "step": 2768
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4565562395539627,
      "learning_rate": 1.9842973789721366e-05,
      "loss": 0.2132,
      "step": 2769
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.35482148705804456,
      "learning_rate": 1.984279865582452e-05,
      "loss": 0.1823,
      "step": 2770
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.48788593738651487,
      "learning_rate": 1.9842623425090988e-05,
      "loss": 0.3273,
      "step": 2771
    },
    {
      "epoch": 0.08,
      "grad_norm": 1.4181819732019127,
      "learning_rate": 1.9842448097522484e-05,
      "loss": 0.3769,
      "step": 2772
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8094506510913,
      "learning_rate": 1.9842272673120737e-05,
      "loss": 0.2646,
      "step": 2773
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.4782253060202552,
      "learning_rate": 1.9842097151887474e-05,
      "loss": 0.2943,
      "step": 2774
    },
    {
      "epoch": 0.08,
      "grad_norm": 0.8086901378436665,
      "learning_rate": 1.984192153382442e-05,
      "loss": 0.4452,
      "step": 2775
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0459478374478115,
      "learning_rate": 1.9841745818933303e-05,
      "loss": 0.8093,
      "step": 2776
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4101829606651121,
      "learning_rate": 1.9841570007215855e-05,
      "loss": 0.2995,
      "step": 2777
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7484631338992115,
      "learning_rate": 1.9841394098673802e-05,
      "loss": 0.391,
      "step": 2778
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3534209411823509,
      "learning_rate": 1.9841218093308873e-05,
      "loss": 0.212,
      "step": 2779
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5364103234974321,
      "learning_rate": 1.9841041991122804e-05,
      "loss": 0.2807,
      "step": 2780
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.455857342807894,
      "learning_rate": 1.9840865792117324e-05,
      "loss": 0.381,
      "step": 2781
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4457490803118789,
      "learning_rate": 1.9840689496294172e-05,
      "loss": 0.3288,
      "step": 2782
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3669212781312033,
      "learning_rate": 1.9840513103655077e-05,
      "loss": 0.2166,
      "step": 2783
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0611859961969128,
      "learning_rate": 1.9840336614201776e-05,
      "loss": 0.6196,
      "step": 2784
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.832778605820326,
      "learning_rate": 1.9840160027936005e-05,
      "loss": 0.4554,
      "step": 2785
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7970976362368376,
      "learning_rate": 1.9839983344859503e-05,
      "loss": 0.5724,
      "step": 2786
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4607730326326234,
      "learning_rate": 1.9839806564974007e-05,
      "loss": 0.279,
      "step": 2787
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3298448870339918,
      "learning_rate": 1.9839629688281255e-05,
      "loss": 0.1626,
      "step": 2788
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.32054391651324987,
      "learning_rate": 1.983945271478299e-05,
      "loss": 0.2579,
      "step": 2789
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8571310078061602,
      "learning_rate": 1.9839275644480952e-05,
      "loss": 0.3276,
      "step": 2790
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2086329543884098,
      "learning_rate": 1.983909847737688e-05,
      "loss": 0.5794,
      "step": 2791
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3781532235121487,
      "learning_rate": 1.9838921213472518e-05,
      "loss": 0.1923,
      "step": 2792
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5748061170575857,
      "learning_rate": 1.9838743852769618e-05,
      "loss": 0.417,
      "step": 2793
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.39907760433690864,
      "learning_rate": 1.9838566395269914e-05,
      "loss": 0.2685,
      "step": 2794
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.708589126026211,
      "learning_rate": 1.9838388840975158e-05,
      "loss": 0.8099,
      "step": 2795
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5137734135735051,
      "learning_rate": 1.9838211189887097e-05,
      "loss": 0.2599,
      "step": 2796
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4341399789253827,
      "learning_rate": 1.9838033442007473e-05,
      "loss": 0.2973,
      "step": 2797
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23215705085520216,
      "learning_rate": 1.983785559733804e-05,
      "loss": 0.1288,
      "step": 2798
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3024596379351725,
      "learning_rate": 1.983767765588055e-05,
      "loss": 0.4378,
      "step": 2799
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4179535117239235,
      "learning_rate": 1.9837499617636745e-05,
      "loss": 0.3495,
      "step": 2800
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.33195901215187895,
      "learning_rate": 1.9837321482608387e-05,
      "loss": 0.2102,
      "step": 2801
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1417780540606401,
      "learning_rate": 1.983714325079722e-05,
      "loss": 0.5222,
      "step": 2802
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1429503724081433,
      "learning_rate": 1.9836964922205002e-05,
      "loss": 0.3503,
      "step": 2803
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6825908872937705,
      "learning_rate": 1.9836786496833485e-05,
      "loss": 0.8588,
      "step": 2804
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44689127651863936,
      "learning_rate": 1.983660797468443e-05,
      "loss": 0.1469,
      "step": 2805
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.39765174016707094,
      "learning_rate": 1.9836429355759582e-05,
      "loss": 0.3387,
      "step": 2806
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.311416489775637,
      "learning_rate": 1.9836250640060713e-05,
      "loss": 0.1701,
      "step": 2807
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5124078124757144,
      "learning_rate": 1.9836071827589566e-05,
      "loss": 0.3122,
      "step": 2808
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8122885626727,
      "learning_rate": 1.9835892918347913e-05,
      "loss": 0.4637,
      "step": 2809
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4431294726561914,
      "learning_rate": 1.983571391233751e-05,
      "loss": 0.268,
      "step": 2810
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5553115239939933,
      "learning_rate": 1.9835534809560115e-05,
      "loss": 0.233,
      "step": 2811
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3955792299969899,
      "learning_rate": 1.983535561001749e-05,
      "loss": 0.3066,
      "step": 2812
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6537461901452186,
      "learning_rate": 1.9835176313711405e-05,
      "loss": 0.8377,
      "step": 2813
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4808360127760597,
      "learning_rate": 1.9834996920643615e-05,
      "loss": 0.132,
      "step": 2814
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.715976246092527,
      "learning_rate": 1.983481743081589e-05,
      "loss": 0.375,
      "step": 2815
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.26282129730555126,
      "learning_rate": 1.9834637844229996e-05,
      "loss": 0.1302,
      "step": 2816
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7911955915935208,
      "learning_rate": 1.98344581608877e-05,
      "loss": 0.5314,
      "step": 2817
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3955211287172157,
      "learning_rate": 1.9834278380790767e-05,
      "loss": 0.2926,
      "step": 2818
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8554283397594763,
      "learning_rate": 1.9834098503940966e-05,
      "loss": 0.3777,
      "step": 2819
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45871418186760915,
      "learning_rate": 1.9833918530340074e-05,
      "loss": 0.2479,
      "step": 2820
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.125382443998886,
      "learning_rate": 1.9833738459989848e-05,
      "loss": 0.6714,
      "step": 2821
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1715559188108686,
      "learning_rate": 1.983355829289207e-05,
      "loss": 0.6197,
      "step": 2822
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5952530148510679,
      "learning_rate": 1.9833378029048512e-05,
      "loss": 0.3426,
      "step": 2823
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.35249172566676434,
      "learning_rate": 1.9833197668460947e-05,
      "loss": 0.2052,
      "step": 2824
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5602633946409628,
      "learning_rate": 1.9833017211131142e-05,
      "loss": 0.3872,
      "step": 2825
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0400343014363862,
      "learning_rate": 1.9832836657060882e-05,
      "loss": 0.4452,
      "step": 2826
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.33390560704366173,
      "learning_rate": 1.9832656006251937e-05,
      "loss": 0.1442,
      "step": 2827
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8536969243587964,
      "learning_rate": 1.983247525870609e-05,
      "loss": 0.3959,
      "step": 2828
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4550994237127207,
      "learning_rate": 1.9832294414425113e-05,
      "loss": 0.2773,
      "step": 2829
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.621281456169803,
      "learning_rate": 1.983211347341079e-05,
      "loss": 0.3486,
      "step": 2830
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5077032098199927,
      "learning_rate": 1.98319324356649e-05,
      "loss": 0.3065,
      "step": 2831
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.225954500290524,
      "learning_rate": 1.983175130118922e-05,
      "loss": 0.4524,
      "step": 2832
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.39665743351576205,
      "learning_rate": 1.983157006998554e-05,
      "loss": 0.222,
      "step": 2833
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.3780039696579114,
      "learning_rate": 1.983138874205564e-05,
      "loss": 0.7426,
      "step": 2834
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9481555288631136,
      "learning_rate": 1.9831207317401297e-05,
      "loss": 0.4763,
      "step": 2835
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3762412755393053,
      "learning_rate": 1.9831025796024303e-05,
      "loss": 0.28,
      "step": 2836
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.327879792428348,
      "learning_rate": 1.9830844177926444e-05,
      "loss": 0.1389,
      "step": 2837
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6596441477284959,
      "learning_rate": 1.9830662463109507e-05,
      "loss": 0.4004,
      "step": 2838
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0632403554981475,
      "learning_rate": 1.9830480651575278e-05,
      "loss": 0.4243,
      "step": 2839
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0028152421623908,
      "learning_rate": 1.983029874332554e-05,
      "loss": 0.5568,
      "step": 2840
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4052640472057631,
      "learning_rate": 1.9830116738362092e-05,
      "loss": 0.3203,
      "step": 2841
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49382206968131187,
      "learning_rate": 1.9829934636686722e-05,
      "loss": 0.2212,
      "step": 2842
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6010818125250447,
      "learning_rate": 1.9829752438301217e-05,
      "loss": 0.4255,
      "step": 2843
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8286578629105262,
      "learning_rate": 1.9829570143207378e-05,
      "loss": 0.4502,
      "step": 2844
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3580317406144618,
      "learning_rate": 1.982938775140699e-05,
      "loss": 0.1849,
      "step": 2845
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6307346503633489,
      "learning_rate": 1.9829205262901853e-05,
      "loss": 0.2031,
      "step": 2846
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4308055638127976,
      "learning_rate": 1.9829022677693757e-05,
      "loss": 0.2931,
      "step": 2847
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4179042636819441,
      "learning_rate": 1.9828839995784505e-05,
      "loss": 0.291,
      "step": 2848
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.674244511481382,
      "learning_rate": 1.982865721717589e-05,
      "loss": 0.813,
      "step": 2849
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8539840946528918,
      "learning_rate": 1.982847434186971e-05,
      "loss": 0.4314,
      "step": 2850
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5259217449043307,
      "learning_rate": 1.9828291369867768e-05,
      "loss": 0.3165,
      "step": 2851
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.44749670602248687,
      "learning_rate": 1.982810830117186e-05,
      "loss": 0.3729,
      "step": 2852
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0979417371108782,
      "learning_rate": 1.982792513578379e-05,
      "loss": 0.091,
      "step": 2853
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4093307595199676,
      "learning_rate": 1.9827741873705356e-05,
      "loss": 0.3252,
      "step": 2854
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.340536566777189,
      "learning_rate": 1.9827558514938367e-05,
      "loss": 0.0745,
      "step": 2855
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47598751538855905,
      "learning_rate": 1.982737505948462e-05,
      "loss": 0.3622,
      "step": 2856
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.877317568045397,
      "learning_rate": 1.982719150734593e-05,
      "loss": 0.4469,
      "step": 2857
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5984753129278952,
      "learning_rate": 1.982700785852409e-05,
      "loss": 0.8565,
      "step": 2858
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.36556385368809136,
      "learning_rate": 1.9826824113020914e-05,
      "loss": 0.3108,
      "step": 2859
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5673377757895016,
      "learning_rate": 1.9826640270838213e-05,
      "loss": 0.3153,
      "step": 2860
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0624950838272604,
      "learning_rate": 1.9826456331977786e-05,
      "loss": 0.3112,
      "step": 2861
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8923557246166488,
      "learning_rate": 1.982627229644145e-05,
      "loss": 0.5343,
      "step": 2862
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2681528425967638,
      "learning_rate": 1.9826088164231015e-05,
      "loss": 0.0685,
      "step": 2863
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48333570732271225,
      "learning_rate": 1.9825903935348293e-05,
      "loss": 0.2217,
      "step": 2864
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4530339048153765,
      "learning_rate": 1.9825719609795093e-05,
      "loss": 0.2518,
      "step": 2865
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3953065514582752,
      "learning_rate": 1.982553518757323e-05,
      "loss": 0.2917,
      "step": 2866
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6458146354347778,
      "learning_rate": 1.982535066868452e-05,
      "loss": 0.896,
      "step": 2867
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.65139122597186,
      "learning_rate": 1.9825166053130777e-05,
      "loss": 0.3947,
      "step": 2868
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0378460570723413,
      "learning_rate": 1.982498134091382e-05,
      "loss": 0.5031,
      "step": 2869
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.40581473190330974,
      "learning_rate": 1.9824796532035458e-05,
      "loss": 0.294,
      "step": 2870
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6073794966763707,
      "learning_rate": 1.9824611626497518e-05,
      "loss": 0.4179,
      "step": 2871
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5068054594170543,
      "learning_rate": 1.9824426624301816e-05,
      "loss": 0.3126,
      "step": 2872
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.57645836873245,
      "learning_rate": 1.982424152545017e-05,
      "loss": 0.2332,
      "step": 2873
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.34414614430516616,
      "learning_rate": 1.9824056329944407e-05,
      "loss": 0.2144,
      "step": 2874
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5246380637507121,
      "learning_rate": 1.9823871037786345e-05,
      "loss": 0.258,
      "step": 2875
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7857825020981849,
      "learning_rate": 1.9823685648977803e-05,
      "loss": 0.4719,
      "step": 2876
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6142536949869878,
      "learning_rate": 1.9823500163520613e-05,
      "loss": 0.374,
      "step": 2877
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.37591172031577885,
      "learning_rate": 1.9823314581416594e-05,
      "loss": 0.2493,
      "step": 2878
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4341484114543308,
      "learning_rate": 1.9823128902667575e-05,
      "loss": 0.283,
      "step": 2879
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7025119000697007,
      "learning_rate": 1.982294312727538e-05,
      "loss": 0.6685,
      "step": 2880
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1742979927819852,
      "learning_rate": 1.9822757255241844e-05,
      "loss": 0.3573,
      "step": 2881
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4010038991712438,
      "learning_rate": 1.9822571286568783e-05,
      "loss": 0.1696,
      "step": 2882
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3466182634991714,
      "learning_rate": 1.9822385221258037e-05,
      "loss": 0.2438,
      "step": 2883
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6542858005593588,
      "learning_rate": 1.9822199059311434e-05,
      "loss": 0.4002,
      "step": 2884
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0890261961983034,
      "learning_rate": 1.9822012800730803e-05,
      "loss": 0.5039,
      "step": 2885
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7524620740956798,
      "learning_rate": 1.9821826445517977e-05,
      "loss": 0.5282,
      "step": 2886
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3584896848944665,
      "learning_rate": 1.982163999367479e-05,
      "loss": 0.1903,
      "step": 2887
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6720592655900046,
      "learning_rate": 1.982145344520308e-05,
      "loss": 0.3648,
      "step": 2888
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5351740610293426,
      "learning_rate": 1.9821266800104682e-05,
      "loss": 0.2856,
      "step": 2889
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5677777056636526,
      "learning_rate": 1.9821080058381425e-05,
      "loss": 0.3956,
      "step": 2890
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.31590966467067927,
      "learning_rate": 1.982089322003515e-05,
      "loss": 0.1225,
      "step": 2891
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4363693183087573,
      "learning_rate": 1.98207062850677e-05,
      "loss": 0.2165,
      "step": 2892
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6825213972036596,
      "learning_rate": 1.9820519253480908e-05,
      "loss": 0.4588,
      "step": 2893
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1731771327734224,
      "learning_rate": 1.9820332125276615e-05,
      "loss": 0.5045,
      "step": 2894
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4423087523118776,
      "learning_rate": 1.9820144900456664e-05,
      "loss": 0.3366,
      "step": 2895
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3127650879741985,
      "learning_rate": 1.98199575790229e-05,
      "loss": 0.0787,
      "step": 2896
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4727085127492885,
      "learning_rate": 1.9819770160977156e-05,
      "loss": 0.3393,
      "step": 2897
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.37619059984512415,
      "learning_rate": 1.9819582646321284e-05,
      "loss": 0.0732,
      "step": 2898
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.465896672122633,
      "learning_rate": 1.981939503505713e-05,
      "loss": 0.6917,
      "step": 2899
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3207354174745262,
      "learning_rate": 1.9819207327186534e-05,
      "loss": 0.1133,
      "step": 2900
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.496046986895135,
      "learning_rate": 1.9819019522711347e-05,
      "loss": 0.3312,
      "step": 2901
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.331639508557245,
      "learning_rate": 1.9818831621633413e-05,
      "loss": 0.2783,
      "step": 2902
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7125667576922063,
      "learning_rate": 1.981864362395458e-05,
      "loss": 0.5529,
      "step": 2903
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0249541267821944,
      "learning_rate": 1.981845552967671e-05,
      "loss": 0.5285,
      "step": 2904
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3553510295798955,
      "learning_rate": 1.9818267338801635e-05,
      "loss": 0.1002,
      "step": 2905
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4677778804341947,
      "learning_rate": 1.9818079051331218e-05,
      "loss": 0.3358,
      "step": 2906
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3982077372804894,
      "learning_rate": 1.9817890667267306e-05,
      "loss": 0.2843,
      "step": 2907
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.440293321974253,
      "learning_rate": 1.981770218661176e-05,
      "loss": 0.6492,
      "step": 2908
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9953050030914121,
      "learning_rate": 1.9817513609366426e-05,
      "loss": 0.4223,
      "step": 2909
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49012167373521975,
      "learning_rate": 1.9817324935533162e-05,
      "loss": 0.2924,
      "step": 2910
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.861303024816207,
      "learning_rate": 1.981713616511383e-05,
      "loss": 0.3449,
      "step": 2911
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.93496196938769,
      "learning_rate": 1.981694729811028e-05,
      "loss": 0.5511,
      "step": 2912
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4006389436216019,
      "learning_rate": 1.9816758334524368e-05,
      "loss": 0.2874,
      "step": 2913
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7455943151693911,
      "learning_rate": 1.981656927435796e-05,
      "loss": 0.2794,
      "step": 2914
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7095938625285556,
      "learning_rate": 1.9816380117612914e-05,
      "loss": 0.1942,
      "step": 2915
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6947353121262925,
      "learning_rate": 1.9816190864291092e-05,
      "loss": 0.3343,
      "step": 2916
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.823363150071661,
      "learning_rate": 1.981600151439435e-05,
      "loss": 0.4142,
      "step": 2917
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.45316801560199194,
      "learning_rate": 1.9815812067924562e-05,
      "loss": 0.3032,
      "step": 2918
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1107859603736057,
      "learning_rate": 1.9815622524883577e-05,
      "loss": 0.4186,
      "step": 2919
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3470371117314479,
      "learning_rate": 1.9815432885273274e-05,
      "loss": 0.2872,
      "step": 2920
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.950884872596661,
      "learning_rate": 1.981524314909551e-05,
      "loss": 0.5519,
      "step": 2921
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1577769081616316,
      "learning_rate": 1.9815053316352153e-05,
      "loss": 0.6047,
      "step": 2922
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6188931499086273,
      "learning_rate": 1.9814863387045076e-05,
      "loss": 0.2802,
      "step": 2923
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3732654099580661,
      "learning_rate": 1.981467336117614e-05,
      "loss": 0.2164,
      "step": 2924
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.35632018951330185,
      "learning_rate": 1.981448323874722e-05,
      "loss": 0.2724,
      "step": 2925
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1586322347807803,
      "learning_rate": 1.9814293019760183e-05,
      "loss": 0.4194,
      "step": 2926
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.145771631379044,
      "learning_rate": 1.98141027042169e-05,
      "loss": 0.6208,
      "step": 2927
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.38847774520322986,
      "learning_rate": 1.981391229211925e-05,
      "loss": 0.2608,
      "step": 2928
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5649916927593236,
      "learning_rate": 1.98137217834691e-05,
      "loss": 0.4188,
      "step": 2929
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5239314168557508,
      "learning_rate": 1.9813531178268323e-05,
      "loss": 0.3096,
      "step": 2930
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4896561286645591,
      "learning_rate": 1.98133404765188e-05,
      "loss": 0.2923,
      "step": 2931
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1979308999536493,
      "learning_rate": 1.9813149678222404e-05,
      "loss": 0.6292,
      "step": 2932
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3248455057715817,
      "learning_rate": 1.9812958783381013e-05,
      "loss": 0.1915,
      "step": 2933
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5640090635390766,
      "learning_rate": 1.9812767791996505e-05,
      "loss": 0.2716,
      "step": 2934
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.124783954395501,
      "learning_rate": 1.9812576704070755e-05,
      "loss": 0.3865,
      "step": 2935
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.564099521322284,
      "learning_rate": 1.981238551960565e-05,
      "loss": 0.3982,
      "step": 2936
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.31546524285179034,
      "learning_rate": 1.981219423860307e-05,
      "loss": 0.2165,
      "step": 2937
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5840448294935122,
      "learning_rate": 1.9812002861064886e-05,
      "loss": 0.4004,
      "step": 2938
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5547162437621254,
      "learning_rate": 1.9811811386992995e-05,
      "loss": 0.4347,
      "step": 2939
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.6284225422722947,
      "learning_rate": 1.9811619816389278e-05,
      "loss": 0.9019,
      "step": 2940
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.48517394876813813,
      "learning_rate": 1.9811428149255613e-05,
      "loss": 0.1985,
      "step": 2941
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47401854310703784,
      "learning_rate": 1.981123638559389e-05,
      "loss": 0.3267,
      "step": 2942
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.257701096538524,
      "learning_rate": 1.9811044525405997e-05,
      "loss": 0.1544,
      "step": 2943
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.527635304628084,
      "learning_rate": 1.9810852568693813e-05,
      "loss": 0.2985,
      "step": 2944
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8484264128883036,
      "learning_rate": 1.9810660515459237e-05,
      "loss": 0.5201,
      "step": 2945
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4913541266678322,
      "learning_rate": 1.981046836570416e-05,
      "loss": 0.0816,
      "step": 2946
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4271160070373856,
      "learning_rate": 1.9810276119430463e-05,
      "loss": 0.3139,
      "step": 2947
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5048354256211024,
      "learning_rate": 1.981008377664004e-05,
      "loss": 0.2996,
      "step": 2948
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6181443439412069,
      "learning_rate": 1.9809891337334785e-05,
      "loss": 0.4179,
      "step": 2949
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.9191952631101405,
      "learning_rate": 1.9809698801516594e-05,
      "loss": 0.1451,
      "step": 2950
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4745028040911917,
      "learning_rate": 1.9809506169187357e-05,
      "loss": 0.2735,
      "step": 2951
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2963571811879334,
      "learning_rate": 1.980931344034897e-05,
      "loss": 0.158,
      "step": 2952
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3884133567698802,
      "learning_rate": 1.980912061500333e-05,
      "loss": 0.5562,
      "step": 2953
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.52167886301913,
      "learning_rate": 1.9808927693152333e-05,
      "loss": 0.3224,
      "step": 2954
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5603072299018138,
      "learning_rate": 1.980873467479788e-05,
      "loss": 0.3846,
      "step": 2955
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4258391927656811,
      "learning_rate": 1.9808541559941864e-05,
      "loss": 0.2398,
      "step": 2956
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2524405118655615,
      "learning_rate": 1.980834834858619e-05,
      "loss": 0.6245,
      "step": 2957
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7964343414132333,
      "learning_rate": 1.980815504073276e-05,
      "loss": 0.964,
      "step": 2958
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7761763344524475,
      "learning_rate": 1.9807961636383474e-05,
      "loss": 0.2636,
      "step": 2959
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.504259015674766,
      "learning_rate": 1.980776813554023e-05,
      "loss": 0.3241,
      "step": 2960
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.27521497553481794,
      "learning_rate": 1.980757453820494e-05,
      "loss": 0.1962,
      "step": 2961
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3365249103920651,
      "learning_rate": 1.9807380844379503e-05,
      "loss": 0.6113,
      "step": 2962
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9136340349288881,
      "learning_rate": 1.9807187054065825e-05,
      "loss": 0.4686,
      "step": 2963
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8639249616549302,
      "learning_rate": 1.9806993167265815e-05,
      "loss": 0.3918,
      "step": 2964
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5741241346657474,
      "learning_rate": 1.9806799183981376e-05,
      "loss": 0.2828,
      "step": 2965
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7704392754078626,
      "learning_rate": 1.9806605104214426e-05,
      "loss": 0.964,
      "step": 2966
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5527733294452354,
      "learning_rate": 1.9806410927966865e-05,
      "loss": 0.3052,
      "step": 2967
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.0079586337339888,
      "learning_rate": 1.9806216655240606e-05,
      "loss": 0.8343,
      "step": 2968
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.43931054624080157,
      "learning_rate": 1.9806022286037558e-05,
      "loss": 0.1871,
      "step": 2969
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6371547455814559,
      "learning_rate": 1.9805827820359642e-05,
      "loss": 0.3261,
      "step": 2970
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9352447521165401,
      "learning_rate": 1.9805633258208764e-05,
      "loss": 0.5715,
      "step": 2971
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.31081851551049533,
      "learning_rate": 1.9805438599586836e-05,
      "loss": 0.1907,
      "step": 2972
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8482418217770463,
      "learning_rate": 1.9805243844495778e-05,
      "loss": 0.3168,
      "step": 2973
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46300729423561093,
      "learning_rate": 1.9805048992937505e-05,
      "loss": 0.2794,
      "step": 2974
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.3505635817976425,
      "learning_rate": 1.9804854044913934e-05,
      "loss": 0.7505,
      "step": 2975
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9739646090277272,
      "learning_rate": 1.9804659000426984e-05,
      "loss": 0.5056,
      "step": 2976
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1203137190180577,
      "learning_rate": 1.980446385947857e-05,
      "loss": 0.6199,
      "step": 2977
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42340486870078237,
      "learning_rate": 1.980426862207062e-05,
      "loss": 0.1851,
      "step": 2978
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4642416509913408,
      "learning_rate": 1.9804073288205042e-05,
      "loss": 0.2953,
      "step": 2979
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8666224071172786,
      "learning_rate": 1.980387785788377e-05,
      "loss": 0.4375,
      "step": 2980
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6711088988034064,
      "learning_rate": 1.980368233110872e-05,
      "loss": 0.2506,
      "step": 2981
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2778163230061274,
      "learning_rate": 1.980348670788182e-05,
      "loss": 0.0797,
      "step": 2982
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4486196575372541,
      "learning_rate": 1.9803290988204988e-05,
      "loss": 0.2437,
      "step": 2983
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5801694426070012,
      "learning_rate": 1.9803095172080154e-05,
      "loss": 0.411,
      "step": 2984
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4673944246207907,
      "learning_rate": 1.9802899259509245e-05,
      "loss": 0.2779,
      "step": 2985
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.7573696193652764,
      "learning_rate": 1.9802703250494186e-05,
      "loss": 0.8073,
      "step": 2986
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.46426784096467055,
      "learning_rate": 1.9802507145036907e-05,
      "loss": 0.2586,
      "step": 2987
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6058974320082227,
      "learning_rate": 1.980231094313934e-05,
      "loss": 0.4215,
      "step": 2988
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9699056059000533,
      "learning_rate": 1.980211464480341e-05,
      "loss": 0.4026,
      "step": 2989
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4711732398641032,
      "learning_rate": 1.9801918250031055e-05,
      "loss": 0.3181,
      "step": 2990
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4045798616385958,
      "learning_rate": 1.9801721758824198e-05,
      "loss": 0.0795,
      "step": 2991
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.500565201061359,
      "learning_rate": 1.980152517118478e-05,
      "loss": 0.3511,
      "step": 2992
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0889476478762024,
      "learning_rate": 1.9801328487114732e-05,
      "loss": 0.5365,
      "step": 2993
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.706950108679738,
      "learning_rate": 1.980113170661599e-05,
      "loss": 0.8738,
      "step": 2994
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5520815635179459,
      "learning_rate": 1.9800934829690488e-05,
      "loss": 0.3729,
      "step": 2995
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.40928942071771285,
      "learning_rate": 1.9800737856340165e-05,
      "loss": 0.1965,
      "step": 2996
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4230532641149488,
      "learning_rate": 1.9800540786566955e-05,
      "loss": 0.3206,
      "step": 2997
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9030433076744755,
      "learning_rate": 1.9800343620372804e-05,
      "loss": 0.4614,
      "step": 2998
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.47119753204464426,
      "learning_rate": 1.9800146357759645e-05,
      "loss": 0.2144,
      "step": 2999
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6769872273194791,
      "learning_rate": 1.979994899872942e-05,
      "loss": 0.1724,
      "step": 3000
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5870725919153206,
      "learning_rate": 1.9799751543284075e-05,
      "loss": 0.3358,
      "step": 3001
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42490972907372027,
      "learning_rate": 1.979955399142555e-05,
      "loss": 0.2797,
      "step": 3002
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5995256598442252,
      "learning_rate": 1.979935634315579e-05,
      "loss": 0.3769,
      "step": 3003
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7985851571939874,
      "learning_rate": 1.9799158598476736e-05,
      "loss": 0.4355,
      "step": 3004
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8563459388241157,
      "learning_rate": 1.9798960757390334e-05,
      "loss": 0.3693,
      "step": 3005
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.429015303912622,
      "learning_rate": 1.9798762819898535e-05,
      "loss": 0.2877,
      "step": 3006
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2303475201810872,
      "learning_rate": 1.9798564786003277e-05,
      "loss": 0.6478,
      "step": 3007
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4109217440957952,
      "learning_rate": 1.979836665570652e-05,
      "loss": 0.2696,
      "step": 3008
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3832518141094461,
      "learning_rate": 1.9798168429010206e-05,
      "loss": 0.2007,
      "step": 3009
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3924450095487536,
      "learning_rate": 1.9797970105916287e-05,
      "loss": 0.2129,
      "step": 3010
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.224113062540262,
      "learning_rate": 1.9797771686426714e-05,
      "loss": 0.6738,
      "step": 3011
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.5018051659844978,
      "learning_rate": 1.979757317054344e-05,
      "loss": 0.7641,
      "step": 3012
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5510311259347134,
      "learning_rate": 1.9797374558268418e-05,
      "loss": 0.3531,
      "step": 3013
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.511977140631845,
      "learning_rate": 1.97971758496036e-05,
      "loss": 0.2834,
      "step": 3014
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4532136599396729,
      "learning_rate": 1.9796977044550942e-05,
      "loss": 0.3075,
      "step": 3015
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.5523522666294247,
      "learning_rate": 1.97967781431124e-05,
      "loss": 0.7226,
      "step": 3016
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1413933446996298,
      "learning_rate": 1.979657914528993e-05,
      "loss": 0.3358,
      "step": 3017
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6433430533968464,
      "learning_rate": 1.9796380051085493e-05,
      "loss": 0.3439,
      "step": 3018
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3425117907494366,
      "learning_rate": 1.9796180860501046e-05,
      "loss": 0.2058,
      "step": 3019
    },
    {
      "epoch": 0.09,
      "grad_norm": 2.071310920611504,
      "learning_rate": 1.9795981573538548e-05,
      "loss": 0.8894,
      "step": 3020
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4427147214518008,
      "learning_rate": 1.9795782190199963e-05,
      "loss": 0.3259,
      "step": 3021
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8289063135403839,
      "learning_rate": 1.9795582710487243e-05,
      "loss": 0.521,
      "step": 3022
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2907095990739268,
      "learning_rate": 1.979538313440236e-05,
      "loss": 0.1435,
      "step": 3023
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4906187787333198,
      "learning_rate": 1.9795183461947273e-05,
      "loss": 0.2204,
      "step": 3024
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.180291035292142,
      "learning_rate": 1.979498369312395e-05,
      "loss": 0.6562,
      "step": 3025
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3940709840998934,
      "learning_rate": 1.9794783827934356e-05,
      "loss": 0.2568,
      "step": 3026
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4714219006033104,
      "learning_rate": 1.9794583866380456e-05,
      "loss": 0.2498,
      "step": 3027
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4501448024576677,
      "learning_rate": 1.979438380846421e-05,
      "loss": 0.2141,
      "step": 3028
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6488035857123272,
      "learning_rate": 1.9794183654187602e-05,
      "loss": 0.3975,
      "step": 3029
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.123465699471245,
      "learning_rate": 1.979398340355259e-05,
      "loss": 0.4561,
      "step": 3030
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5415410236917368,
      "learning_rate": 1.9793783056561144e-05,
      "loss": 0.3429,
      "step": 3031
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.31567336841400784,
      "learning_rate": 1.9793582613215242e-05,
      "loss": 0.1697,
      "step": 3032
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4497420635729705,
      "learning_rate": 1.9793382073516846e-05,
      "loss": 0.3019,
      "step": 3033
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1123700724073189,
      "learning_rate": 1.979318143746794e-05,
      "loss": 0.353,
      "step": 3034
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2214646145997694,
      "learning_rate": 1.9792980705070487e-05,
      "loss": 0.7214,
      "step": 3035
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9468701564558897,
      "learning_rate": 1.9792779876326474e-05,
      "loss": 0.5067,
      "step": 3036
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.381496677641494,
      "learning_rate": 1.9792578951237865e-05,
      "loss": 0.2219,
      "step": 3037
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42312530315817215,
      "learning_rate": 1.9792377929806644e-05,
      "loss": 0.3501,
      "step": 3038
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0028041329633162,
      "learning_rate": 1.979217681203479e-05,
      "loss": 0.4472,
      "step": 3039
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8391621206088762,
      "learning_rate": 1.9791975597924272e-05,
      "loss": 0.5662,
      "step": 3040
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.2306125239047556,
      "learning_rate": 1.979177428747708e-05,
      "loss": 0.0774,
      "step": 3041
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5109157755460906,
      "learning_rate": 1.9791572880695194e-05,
      "loss": 0.3293,
      "step": 3042
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4735468493020334,
      "learning_rate": 1.9791371377580585e-05,
      "loss": 0.385,
      "step": 3043
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.452635111026272,
      "learning_rate": 1.979116977813525e-05,
      "loss": 0.3456,
      "step": 3044
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.4951750359437113,
      "learning_rate": 1.979096808236116e-05,
      "loss": 0.5395,
      "step": 3045
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4775694166034198,
      "learning_rate": 1.9790766290260306e-05,
      "loss": 0.3278,
      "step": 3046
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4969842701178249,
      "learning_rate": 1.979056440183467e-05,
      "loss": 0.326,
      "step": 3047
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7326840505027824,
      "learning_rate": 1.9790362417086242e-05,
      "loss": 0.5493,
      "step": 3048
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.39247398527965593,
      "learning_rate": 1.9790160336017004e-05,
      "loss": 0.2882,
      "step": 3049
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.24069025069368055,
      "learning_rate": 1.978995815862895e-05,
      "loss": 0.0779,
      "step": 3050
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6771490846811827,
      "learning_rate": 1.9789755884924067e-05,
      "loss": 0.3637,
      "step": 3051
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5617584023711366,
      "learning_rate": 1.978955351490434e-05,
      "loss": 0.2807,
      "step": 3052
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.573513782719794,
      "learning_rate": 1.978935104857177e-05,
      "loss": 0.8142,
      "step": 3053
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6876867398035936,
      "learning_rate": 1.978914848592834e-05,
      "loss": 0.3961,
      "step": 3054
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0189892478569615,
      "learning_rate": 1.9788945826976047e-05,
      "loss": 0.4223,
      "step": 3055
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3338915061847251,
      "learning_rate": 1.9788743071716884e-05,
      "loss": 0.2636,
      "step": 3056
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0703748909257713,
      "learning_rate": 1.9788540220152846e-05,
      "loss": 0.5877,
      "step": 3057
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9834766174382952,
      "learning_rate": 1.9788337272285927e-05,
      "loss": 0.5134,
      "step": 3058
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.37035045502457237,
      "learning_rate": 1.9788134228118127e-05,
      "loss": 0.15,
      "step": 3059
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.40931099592424597,
      "learning_rate": 1.978793108765144e-05,
      "loss": 0.2243,
      "step": 3060
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2958067510407907,
      "learning_rate": 1.9787727850887866e-05,
      "loss": 0.6891,
      "step": 3061
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42600324628167624,
      "learning_rate": 1.9787524517829407e-05,
      "loss": 0.2675,
      "step": 3062
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.074538875378146,
      "learning_rate": 1.978732108847806e-05,
      "loss": 0.4124,
      "step": 3063
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6750033954337032,
      "learning_rate": 1.9787117562835828e-05,
      "loss": 0.3614,
      "step": 3064
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4459961306555261,
      "learning_rate": 1.978691394090471e-05,
      "loss": 0.3097,
      "step": 3065
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.785909639760455,
      "learning_rate": 1.978671022268672e-05,
      "loss": 0.8687,
      "step": 3066
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5089898434182076,
      "learning_rate": 1.9786506408183847e-05,
      "loss": 0.3079,
      "step": 3067
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4810065098630965,
      "learning_rate": 1.9786302497398105e-05,
      "loss": 0.3626,
      "step": 3068
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.28208033662594845,
      "learning_rate": 1.97860984903315e-05,
      "loss": 0.1895,
      "step": 3069
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5586907513828926,
      "learning_rate": 1.9785894386986043e-05,
      "loss": 0.2511,
      "step": 3070
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1288388023816656,
      "learning_rate": 1.9785690187363727e-05,
      "loss": 0.41,
      "step": 3071
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1209170467610237,
      "learning_rate": 1.978548589146658e-05,
      "loss": 0.5784,
      "step": 3072
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3463518253180831,
      "learning_rate": 1.9785281499296598e-05,
      "loss": 0.2287,
      "step": 3073
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.6505836329446324,
      "learning_rate": 1.97850770108558e-05,
      "loss": 0.4018,
      "step": 3074
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.438242017619427,
      "learning_rate": 1.978487242614619e-05,
      "loss": 0.3327,
      "step": 3075
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9097318496824717,
      "learning_rate": 1.9784667745169792e-05,
      "loss": 0.6061,
      "step": 3076
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3568544400814545,
      "learning_rate": 1.9784462967928607e-05,
      "loss": 0.1875,
      "step": 3077
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4404719936769004,
      "learning_rate": 1.9784258094424656e-05,
      "loss": 0.209,
      "step": 3078
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5830947427109604,
      "learning_rate": 1.9784053124659957e-05,
      "loss": 0.3634,
      "step": 3079
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.39972231191155133,
      "learning_rate": 1.9783848058636526e-05,
      "loss": 0.2969,
      "step": 3080
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9477850852270007,
      "learning_rate": 1.9783642896356377e-05,
      "loss": 0.5097,
      "step": 3081
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3029338024888685,
      "learning_rate": 1.978343763782153e-05,
      "loss": 0.0766,
      "step": 3082
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4504619614974874,
      "learning_rate": 1.9783232283034004e-05,
      "loss": 0.3306,
      "step": 3083
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.1225891537564558,
      "learning_rate": 1.9783026831995817e-05,
      "loss": 0.5607,
      "step": 3084
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4176889172969015,
      "learning_rate": 1.9782821284708994e-05,
      "loss": 0.3402,
      "step": 3085
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.0422715142661152,
      "learning_rate": 1.978261564117556e-05,
      "loss": 0.3316,
      "step": 3086
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49810540115229057,
      "learning_rate": 1.978240990139753e-05,
      "loss": 0.2906,
      "step": 3087
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3035933553975485,
      "learning_rate": 1.9782204065376936e-05,
      "loss": 0.1453,
      "step": 3088
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9447842542794987,
      "learning_rate": 1.9781998133115797e-05,
      "loss": 0.4493,
      "step": 3089
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8451841443617557,
      "learning_rate": 1.9781792104616144e-05,
      "loss": 0.5645,
      "step": 3090
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.3591238728424981,
      "learning_rate": 1.978158597988e-05,
      "loss": 0.2458,
      "step": 3091
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.49547446196685707,
      "learning_rate": 1.9781379758909394e-05,
      "loss": 0.3511,
      "step": 3092
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.2356872288815,
      "learning_rate": 1.9781173441706356e-05,
      "loss": 0.5222,
      "step": 3093
    },
    {
      "epoch": 0.09,
      "grad_norm": 1.832758388458753,
      "learning_rate": 1.9780967028272915e-05,
      "loss": 0.8993,
      "step": 3094
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.23383310680599445,
      "learning_rate": 1.9780760518611103e-05,
      "loss": 0.1028,
      "step": 3095
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4691786869037803,
      "learning_rate": 1.978055391272295e-05,
      "loss": 0.2979,
      "step": 3096
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.5390739726875066,
      "learning_rate": 1.978034721061049e-05,
      "loss": 0.2517,
      "step": 3097
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.42238502683014634,
      "learning_rate": 1.9780140412275757e-05,
      "loss": 0.3546,
      "step": 3098
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.8935165330903219,
      "learning_rate": 1.977993351772078e-05,
      "loss": 0.4559,
      "step": 3099
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.7758600130860067,
      "learning_rate": 1.9779726526947603e-05,
      "loss": 0.3522,
      "step": 3100
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.4585277484660794,
      "learning_rate": 1.977951943995826e-05,
      "loss": 0.2582,
      "step": 3101
    },
    {
      "epoch": 0.09,
      "grad_norm": 0.9867537518190859,
      "learning_rate": 1.9779312256754784e-05,
      "loss": 0.5635,
      "step": 3102
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4056248207905426,
      "learning_rate": 1.9779104977339218e-05,
      "loss": 0.3296,
      "step": 3103
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47146369561819534,
      "learning_rate": 1.9778897601713594e-05,
      "loss": 0.074,
      "step": 3104
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.042637590101159,
      "learning_rate": 1.9778690129879963e-05,
      "loss": 0.3988,
      "step": 3105
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2877157950728075,
      "learning_rate": 1.9778482561840364e-05,
      "loss": 0.2334,
      "step": 3106
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6432719931600057,
      "learning_rate": 1.977827489759683e-05,
      "loss": 0.5105,
      "step": 3107
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3294449800273349,
      "learning_rate": 1.977806713715141e-05,
      "loss": 0.2141,
      "step": 3108
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.62776592193145,
      "learning_rate": 1.9777859280506152e-05,
      "loss": 0.3885,
      "step": 3109
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.324604305275625,
      "learning_rate": 1.9777651327663096e-05,
      "loss": 0.2048,
      "step": 3110
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5509032342511795,
      "learning_rate": 1.9777443278624284e-05,
      "loss": 0.8718,
      "step": 3111
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.18743546918241,
      "learning_rate": 1.9777235133391772e-05,
      "loss": 0.5148,
      "step": 3112
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.253835431077211,
      "learning_rate": 1.9777026891967606e-05,
      "loss": 0.7672,
      "step": 3113
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4007776039949197,
      "learning_rate": 1.9776818554353827e-05,
      "loss": 0.2124,
      "step": 3114
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3378297999308525,
      "learning_rate": 1.9776610120552494e-05,
      "loss": 0.2938,
      "step": 3115
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8127263028255006,
      "learning_rate": 1.9776401590565654e-05,
      "loss": 0.5711,
      "step": 3116
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2918261124947684,
      "learning_rate": 1.9776192964395356e-05,
      "loss": 0.0735,
      "step": 3117
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9555238719279151,
      "learning_rate": 1.9775984242043654e-05,
      "loss": 0.5434,
      "step": 3118
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3864980128381281,
      "learning_rate": 1.9775775423512607e-05,
      "loss": 0.236,
      "step": 3119
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7432117063210375,
      "learning_rate": 1.977556650880426e-05,
      "loss": 0.8466,
      "step": 3120
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4194613006626841,
      "learning_rate": 1.9775357497920674e-05,
      "loss": 0.2742,
      "step": 3121
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5887267214714627,
      "learning_rate": 1.9775148390863907e-05,
      "loss": 0.7947,
      "step": 3122
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46975801574030146,
      "learning_rate": 1.9774939187636012e-05,
      "loss": 0.2427,
      "step": 3123
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6346982162364542,
      "learning_rate": 1.977472988823905e-05,
      "loss": 0.3763,
      "step": 3124
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8065493389657518,
      "learning_rate": 1.9774520492675077e-05,
      "loss": 0.4311,
      "step": 3125
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3829216637160689,
      "learning_rate": 1.9774311000946154e-05,
      "loss": 0.2267,
      "step": 3126
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.34596781659307274,
      "learning_rate": 1.9774101413054347e-05,
      "loss": 0.2647,
      "step": 3127
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45117456228322156,
      "learning_rate": 1.9773891729001714e-05,
      "loss": 0.2038,
      "step": 3128
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6369735349630409,
      "learning_rate": 1.9773681948790317e-05,
      "loss": 0.4301,
      "step": 3129
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.263577967829006,
      "learning_rate": 1.9773472072422218e-05,
      "loss": 0.5365,
      "step": 3130
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0941437132851806,
      "learning_rate": 1.9773262099899488e-05,
      "loss": 0.5565,
      "step": 3131
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.38486489130539603,
      "learning_rate": 1.977305203122419e-05,
      "loss": 0.0794,
      "step": 3132
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4135911411213359,
      "learning_rate": 1.9772841866398387e-05,
      "loss": 0.3069,
      "step": 3133
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48146725832515685,
      "learning_rate": 1.9772631605424154e-05,
      "loss": 0.3351,
      "step": 3134
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.38253220910338204,
      "learning_rate": 1.9772421248303553e-05,
      "loss": 0.187,
      "step": 3135
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.031552403443217,
      "learning_rate": 1.9772210795038657e-05,
      "loss": 0.4396,
      "step": 3136
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.705540563846146,
      "learning_rate": 1.9772000245631538e-05,
      "loss": 0.2744,
      "step": 3137
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5501530336941991,
      "learning_rate": 1.9771789600084262e-05,
      "loss": 0.3049,
      "step": 3138
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41532646397516343,
      "learning_rate": 1.9771578858398905e-05,
      "loss": 0.3166,
      "step": 3139
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8319955745522588,
      "learning_rate": 1.9771368020577538e-05,
      "loss": 0.4262,
      "step": 3140
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5114130822404538,
      "learning_rate": 1.977115708662224e-05,
      "loss": 0.3016,
      "step": 3141
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42512276821264605,
      "learning_rate": 1.9770946056535085e-05,
      "loss": 0.3321,
      "step": 3142
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4126794634079477,
      "learning_rate": 1.9770734930318146e-05,
      "loss": 0.1708,
      "step": 3143
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6437538720385443,
      "learning_rate": 1.9770523707973505e-05,
      "loss": 0.3185,
      "step": 3144
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4057196073951426,
      "learning_rate": 1.977031238950323e-05,
      "loss": 0.2556,
      "step": 3145
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.484742809174835,
      "learning_rate": 1.9770100974909412e-05,
      "loss": 0.289,
      "step": 3146
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3926559523161006,
      "learning_rate": 1.976988946419412e-05,
      "loss": 0.6429,
      "step": 3147
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7957437270925267,
      "learning_rate": 1.976967785735945e-05,
      "loss": 0.8076,
      "step": 3148
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9155492695711616,
      "learning_rate": 1.9769466154407467e-05,
      "loss": 0.4602,
      "step": 3149
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8190162190599362,
      "learning_rate": 1.9769254355340264e-05,
      "loss": 0.3783,
      "step": 3150
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.36973826919945035,
      "learning_rate": 1.9769042460159923e-05,
      "loss": 0.2437,
      "step": 3151
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.602012310430608,
      "learning_rate": 1.9768830468868527e-05,
      "loss": 0.4289,
      "step": 3152
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41669788237397504,
      "learning_rate": 1.9768618381468164e-05,
      "loss": 0.1644,
      "step": 3153
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5866198889052917,
      "learning_rate": 1.9768406197960916e-05,
      "loss": 0.1278,
      "step": 3154
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6627344458941661,
      "learning_rate": 1.9768193918348875e-05,
      "loss": 0.3592,
      "step": 3155
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5953912087885329,
      "learning_rate": 1.976798154263413e-05,
      "loss": 0.3014,
      "step": 3156
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.496782841378338,
      "learning_rate": 1.9767769070818764e-05,
      "loss": 0.3547,
      "step": 3157
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.260930138806961,
      "learning_rate": 1.9767556502904877e-05,
      "loss": 0.4799,
      "step": 3158
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2225708125055204,
      "learning_rate": 1.976734383889455e-05,
      "loss": 0.8983,
      "step": 3159
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.28450771456386975,
      "learning_rate": 1.9767131078789884e-05,
      "loss": 0.2024,
      "step": 3160
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3989134003324855,
      "learning_rate": 1.976691822259297e-05,
      "loss": 0.6585,
      "step": 3161
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.35439542625171044,
      "learning_rate": 1.9766705270305895e-05,
      "loss": 0.2758,
      "step": 3162
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7757719559514802,
      "learning_rate": 1.9766492221930763e-05,
      "loss": 0.3079,
      "step": 3163
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3814420299985198,
      "learning_rate": 1.9766279077469667e-05,
      "loss": 0.2026,
      "step": 3164
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6792041906849627,
      "learning_rate": 1.9766065836924705e-05,
      "loss": 0.4187,
      "step": 3165
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2769175370158732,
      "learning_rate": 1.9765852500297976e-05,
      "loss": 0.498,
      "step": 3166
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4836843170018655,
      "learning_rate": 1.976563906759157e-05,
      "loss": 0.2105,
      "step": 3167
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49799737375622644,
      "learning_rate": 1.97654255388076e-05,
      "loss": 0.36,
      "step": 3168
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3380227355612195,
      "learning_rate": 1.9765211913948156e-05,
      "loss": 0.2472,
      "step": 3169
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3375252327369647,
      "learning_rate": 1.976499819301535e-05,
      "loss": 0.6366,
      "step": 3170
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9947488452015848,
      "learning_rate": 1.976478437601127e-05,
      "loss": 0.3222,
      "step": 3171
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5062149477337861,
      "learning_rate": 1.9764570462938032e-05,
      "loss": 0.2683,
      "step": 3172
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4216021497302388,
      "learning_rate": 1.976435645379774e-05,
      "loss": 0.2131,
      "step": 3173
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4481419825422221,
      "learning_rate": 1.9764142348592494e-05,
      "loss": 0.3589,
      "step": 3174
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4621517111886709,
      "learning_rate": 1.9763928147324402e-05,
      "loss": 0.3136,
      "step": 3175
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8792384204173725,
      "learning_rate": 1.976371384999557e-05,
      "loss": 0.5769,
      "step": 3176
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2419921724937964,
      "learning_rate": 1.976349945660811e-05,
      "loss": 0.0781,
      "step": 3177
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5090266753244845,
      "learning_rate": 1.976328496716413e-05,
      "loss": 0.314,
      "step": 3178
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9857324291296725,
      "learning_rate": 1.9763070381665743e-05,
      "loss": 0.1553,
      "step": 3179
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40159605218625194,
      "learning_rate": 1.9762855700115056e-05,
      "loss": 0.2687,
      "step": 3180
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9403888750039813,
      "learning_rate": 1.976264092251418e-05,
      "loss": 0.8941,
      "step": 3181
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3053841366820049,
      "learning_rate": 1.9762426048865233e-05,
      "loss": 0.0805,
      "step": 3182
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41579841204952395,
      "learning_rate": 1.9762211079170322e-05,
      "loss": 0.3409,
      "step": 3183
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7007893568338335,
      "learning_rate": 1.976199601343157e-05,
      "loss": 0.4502,
      "step": 3184
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44991546587875203,
      "learning_rate": 1.976178085165109e-05,
      "loss": 0.2004,
      "step": 3185
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3918316680597214,
      "learning_rate": 1.9761565593830993e-05,
      "loss": 0.2797,
      "step": 3186
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5025468801514436,
      "learning_rate": 1.9761350239973406e-05,
      "loss": 0.3252,
      "step": 3187
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.594174533101341,
      "learning_rate": 1.976113479008044e-05,
      "loss": 0.3085,
      "step": 3188
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.333951097208183,
      "learning_rate": 1.976091924415422e-05,
      "loss": 0.6911,
      "step": 3189
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4382807970373654,
      "learning_rate": 1.9760703602196868e-05,
      "loss": 0.5245,
      "step": 3190
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6602513113968667,
      "learning_rate": 1.9760487864210497e-05,
      "loss": 0.4271,
      "step": 3191
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3447267564952889,
      "learning_rate": 1.9760272030197237e-05,
      "loss": 0.2665,
      "step": 3192
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4149737044438138,
      "learning_rate": 1.9760056100159207e-05,
      "loss": 0.3629,
      "step": 3193
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.208966635142069,
      "learning_rate": 1.9759840074098537e-05,
      "loss": 0.5405,
      "step": 3194
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.34150255852115297,
      "learning_rate": 1.9759623952017347e-05,
      "loss": 0.075,
      "step": 3195
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41340318780278734,
      "learning_rate": 1.9759407733917767e-05,
      "loss": 0.3018,
      "step": 3196
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.031088535610706,
      "learning_rate": 1.975919141980192e-05,
      "loss": 0.3915,
      "step": 3197
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6072008083472545,
      "learning_rate": 1.975897500967194e-05,
      "loss": 0.4123,
      "step": 3198
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47943190549693954,
      "learning_rate": 1.975875850352995e-05,
      "loss": 0.3433,
      "step": 3199
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.676188043430951,
      "learning_rate": 1.9758541901378083e-05,
      "loss": 0.4031,
      "step": 3200
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47877577742067556,
      "learning_rate": 1.975832520321847e-05,
      "loss": 0.3136,
      "step": 3201
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9913265309724086,
      "learning_rate": 1.9758108409053242e-05,
      "loss": 0.9997,
      "step": 3202
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.32189833982194704,
      "learning_rate": 1.9757891518884535e-05,
      "loss": 0.1774,
      "step": 3203
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3189489455809828,
      "learning_rate": 1.975767453271448e-05,
      "loss": 0.278,
      "step": 3204
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3300067342558879,
      "learning_rate": 1.9757457450545208e-05,
      "loss": 0.2079,
      "step": 3205
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2487297455793858,
      "learning_rate": 1.9757240272378864e-05,
      "loss": 0.7695,
      "step": 3206
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0641596323924185,
      "learning_rate": 1.9757022998217575e-05,
      "loss": 0.3314,
      "step": 3207
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5951549118605364,
      "learning_rate": 1.9756805628063486e-05,
      "loss": 0.4399,
      "step": 3208
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8157669281731191,
      "learning_rate": 1.9756588161918734e-05,
      "loss": 0.3539,
      "step": 3209
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44729325762796224,
      "learning_rate": 1.9756370599785457e-05,
      "loss": 0.2643,
      "step": 3210
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47376200390481227,
      "learning_rate": 1.9756152941665797e-05,
      "loss": 0.3512,
      "step": 3211
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6543057521244444,
      "learning_rate": 1.975593518756189e-05,
      "loss": 0.6009,
      "step": 3212
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4438423571233257,
      "learning_rate": 1.9755717337475884e-05,
      "loss": 0.2082,
      "step": 3213
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.34651133720461064,
      "learning_rate": 1.975549939140992e-05,
      "loss": 0.1823,
      "step": 3214
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6557492130229463,
      "learning_rate": 1.9755281349366147e-05,
      "loss": 0.4322,
      "step": 3215
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.47914575930978026,
      "learning_rate": 1.9755063211346703e-05,
      "loss": 0.2671,
      "step": 3216
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9624658341323001,
      "learning_rate": 1.9754844977353736e-05,
      "loss": 0.5473,
      "step": 3217
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7005043399448793,
      "learning_rate": 1.97546266473894e-05,
      "loss": 0.3389,
      "step": 3218
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4300762188831888,
      "learning_rate": 1.9754408221455833e-05,
      "loss": 0.3374,
      "step": 3219
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3177871113839963,
      "learning_rate": 1.9754189699555185e-05,
      "loss": 0.6984,
      "step": 3220
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9800292239378987,
      "learning_rate": 1.9753971081689614e-05,
      "loss": 0.4068,
      "step": 3221
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3317894410167143,
      "learning_rate": 1.9753752367861265e-05,
      "loss": 0.2638,
      "step": 3222
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3979957888292622,
      "learning_rate": 1.9753533558072292e-05,
      "loss": 0.2176,
      "step": 3223
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5800069222293336,
      "learning_rate": 1.9753314652324844e-05,
      "loss": 0.3216,
      "step": 3224
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.286421604646045,
      "learning_rate": 1.9753095650621083e-05,
      "loss": 0.4514,
      "step": 3225
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8354716211090308,
      "learning_rate": 1.975287655296315e-05,
      "loss": 0.53,
      "step": 3226
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44253937017390027,
      "learning_rate": 1.975265735935321e-05,
      "loss": 0.1415,
      "step": 3227
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3974392031130655,
      "learning_rate": 1.975243806979342e-05,
      "loss": 0.3,
      "step": 3228
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5372763784702392,
      "learning_rate": 1.9752218684285934e-05,
      "loss": 0.2841,
      "step": 3229
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7125847573807502,
      "learning_rate": 1.9751999202832916e-05,
      "loss": 0.8532,
      "step": 3230
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3462472882903044,
      "learning_rate": 1.9751779625436516e-05,
      "loss": 0.0751,
      "step": 3231
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5129930876556698,
      "learning_rate": 1.97515599520989e-05,
      "loss": 0.3327,
      "step": 3232
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5417340199336893,
      "learning_rate": 1.975134018282223e-05,
      "loss": 0.2902,
      "step": 3233
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44163313527115944,
      "learning_rate": 1.9751120317608667e-05,
      "loss": 0.3428,
      "step": 3234
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.783504240906327,
      "learning_rate": 1.9750900356460373e-05,
      "loss": 0.5763,
      "step": 3235
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5343889158079861,
      "learning_rate": 1.975068029937951e-05,
      "loss": 0.2638,
      "step": 3236
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.542781961469345,
      "learning_rate": 1.975046014636825e-05,
      "loss": 0.3298,
      "step": 3237
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.537351699625222,
      "learning_rate": 1.9750239897428754e-05,
      "loss": 0.3267,
      "step": 3238
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5728400936891782,
      "learning_rate": 1.9750019552563188e-05,
      "loss": 0.3844,
      "step": 3239
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.32191508859630663,
      "learning_rate": 1.974979911177372e-05,
      "loss": 0.1859,
      "step": 3240
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8176866135004761,
      "learning_rate": 1.9749578575062524e-05,
      "loss": 0.394,
      "step": 3241
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3524281588485298,
      "learning_rate": 1.9749357942431768e-05,
      "loss": 0.1966,
      "step": 3242
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7421027762845457,
      "learning_rate": 1.9749137213883617e-05,
      "loss": 0.5499,
      "step": 3243
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8940289358505655,
      "learning_rate": 1.9748916389420244e-05,
      "loss": 0.4437,
      "step": 3244
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6027758329215058,
      "learning_rate": 1.9748695469043823e-05,
      "loss": 0.4268,
      "step": 3245
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3820029867280337,
      "learning_rate": 1.9748474452756533e-05,
      "loss": 0.209,
      "step": 3246
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5103191346350411,
      "learning_rate": 1.974825334056054e-05,
      "loss": 0.3446,
      "step": 3247
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1457221419256465,
      "learning_rate": 1.9748032132458025e-05,
      "loss": 0.9111,
      "step": 3248
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40063876277727056,
      "learning_rate": 1.9747810828451164e-05,
      "loss": 0.1909,
      "step": 3249
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6347559071905122,
      "learning_rate": 1.974758942854213e-05,
      "loss": 0.3676,
      "step": 3250
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49075191482607394,
      "learning_rate": 1.9747367932733105e-05,
      "loss": 0.3324,
      "step": 3251
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6467145703906202,
      "learning_rate": 1.9747146341026268e-05,
      "loss": 0.4071,
      "step": 3252
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3880591448924327,
      "learning_rate": 1.9746924653423795e-05,
      "loss": 0.2319,
      "step": 3253
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0760262721887446,
      "learning_rate": 1.9746702869927872e-05,
      "loss": 0.4884,
      "step": 3254
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3984726550833471,
      "learning_rate": 1.974648099054068e-05,
      "loss": 0.2347,
      "step": 3255
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0128612877812873,
      "learning_rate": 1.97462590152644e-05,
      "loss": 0.8725,
      "step": 3256
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45473880281754364,
      "learning_rate": 1.9746036944101218e-05,
      "loss": 0.3045,
      "step": 3257
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5466957367051873,
      "learning_rate": 1.9745814777053316e-05,
      "loss": 0.3569,
      "step": 3258
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4070066940694174,
      "learning_rate": 1.974559251412288e-05,
      "loss": 0.2627,
      "step": 3259
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.569139102758117,
      "learning_rate": 1.97453701553121e-05,
      "loss": 0.3054,
      "step": 3260
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.823866130116206,
      "learning_rate": 1.9745147700623164e-05,
      "loss": 0.5343,
      "step": 3261
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.24898211145157392,
      "learning_rate": 1.9744925150058253e-05,
      "loss": 0.0965,
      "step": 3262
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4084338162313876,
      "learning_rate": 1.974470250361957e-05,
      "loss": 0.3267,
      "step": 3263
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4663145766249361,
      "learning_rate": 1.974447976130929e-05,
      "loss": 0.1881,
      "step": 3264
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7051227935496217,
      "learning_rate": 1.9744256923129614e-05,
      "loss": 0.5029,
      "step": 3265
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8979470351942007,
      "learning_rate": 1.974403398908273e-05,
      "loss": 0.3924,
      "step": 3266
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7417541327160941,
      "learning_rate": 1.9743810959170836e-05,
      "loss": 0.5185,
      "step": 3267
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5892237493120499,
      "learning_rate": 1.9743587833396126e-05,
      "loss": 0.2917,
      "step": 3268
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.45712321026470576,
      "learning_rate": 1.974336461176079e-05,
      "loss": 0.2992,
      "step": 3269
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4561355597811826,
      "learning_rate": 1.9743141294267025e-05,
      "loss": 0.3078,
      "step": 3270
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3840593820260666,
      "learning_rate": 1.9742917880917033e-05,
      "loss": 0.1923,
      "step": 3271
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1622259345077397,
      "learning_rate": 1.9742694371713006e-05,
      "loss": 0.5042,
      "step": 3272
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3863414052168958,
      "learning_rate": 1.9742470766657148e-05,
      "loss": 0.2187,
      "step": 3273
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.5103215551758598,
      "learning_rate": 1.974224706575166e-05,
      "loss": 0.7845,
      "step": 3274
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.49063683249846746,
      "learning_rate": 1.9742023268998736e-05,
      "loss": 0.2742,
      "step": 3275
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5171753022938992,
      "learning_rate": 1.974179937640058e-05,
      "loss": 0.4088,
      "step": 3276
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5631016391804641,
      "learning_rate": 1.97415753879594e-05,
      "loss": 0.2938,
      "step": 3277
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.451062995796852,
      "learning_rate": 1.9741351303677395e-05,
      "loss": 0.363,
      "step": 3278
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2032717475760184,
      "learning_rate": 1.974112712355677e-05,
      "loss": 0.0919,
      "step": 3279
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.39055794389583093,
      "learning_rate": 1.974090284759973e-05,
      "loss": 0.2246,
      "step": 3280
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.36709314490491385,
      "learning_rate": 1.9740678475808485e-05,
      "loss": 0.2874,
      "step": 3281
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4961723634246517,
      "learning_rate": 1.9740454008185238e-05,
      "loss": 0.3273,
      "step": 3282
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4986512793350479,
      "learning_rate": 1.97402294447322e-05,
      "loss": 0.2554,
      "step": 3283
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.4301171190756907,
      "learning_rate": 1.974000478545158e-05,
      "loss": 0.6721,
      "step": 3284
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8288746211920681,
      "learning_rate": 1.9739780030345588e-05,
      "loss": 0.4599,
      "step": 3285
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.35660196030810987,
      "learning_rate": 1.9739555179416435e-05,
      "loss": 0.198,
      "step": 3286
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6312085021144727,
      "learning_rate": 1.9739330232666335e-05,
      "loss": 0.3374,
      "step": 3287
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3999524397630294,
      "learning_rate": 1.9739105190097496e-05,
      "loss": 0.2714,
      "step": 3288
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.36121004642731913,
      "learning_rate": 1.9738880051712138e-05,
      "loss": 0.1954,
      "step": 3289
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6186244905918428,
      "learning_rate": 1.9738654817512476e-05,
      "loss": 0.3019,
      "step": 3290
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0975229508331221,
      "learning_rate": 1.973842948750072e-05,
      "loss": 0.5265,
      "step": 3291
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.39072512851834307,
      "learning_rate": 1.973820406167909e-05,
      "loss": 0.2646,
      "step": 3292
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5350610447988682,
      "learning_rate": 1.9737978540049806e-05,
      "loss": 0.4217,
      "step": 3293
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.536688640930329,
      "learning_rate": 1.9737752922615084e-05,
      "loss": 0.386,
      "step": 3294
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.730223707275207,
      "learning_rate": 1.973752720937714e-05,
      "loss": 0.9267,
      "step": 3295
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.377246300274689,
      "learning_rate": 1.9737301400338207e-05,
      "loss": 0.2143,
      "step": 3296
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1932900270474331,
      "learning_rate": 1.9737075495500494e-05,
      "loss": 0.6431,
      "step": 3297
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3883857850063387,
      "learning_rate": 1.9736849494866234e-05,
      "loss": 0.1618,
      "step": 3298
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4334536338754083,
      "learning_rate": 1.973662339843764e-05,
      "loss": 0.2776,
      "step": 3299
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.780550729220879,
      "learning_rate": 1.973639720621694e-05,
      "loss": 0.3597,
      "step": 3300
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.32322748208987995,
      "learning_rate": 1.973617091820636e-05,
      "loss": 0.2031,
      "step": 3301
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8151616492738143,
      "learning_rate": 1.9735944534408132e-05,
      "loss": 0.5156,
      "step": 3302
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7258737830292604,
      "learning_rate": 1.9735718054824476e-05,
      "loss": 0.4193,
      "step": 3303
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0210287996900036,
      "learning_rate": 1.9735491479457623e-05,
      "loss": 0.4963,
      "step": 3304
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3638886761061808,
      "learning_rate": 1.97352648083098e-05,
      "loss": 0.2447,
      "step": 3305
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5341742427845061,
      "learning_rate": 1.973503804138324e-05,
      "loss": 0.379,
      "step": 3306
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40387025232809465,
      "learning_rate": 1.973481117868017e-05,
      "loss": 0.1217,
      "step": 3307
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1029237358232944,
      "learning_rate": 1.973458422020283e-05,
      "loss": 0.8468,
      "step": 3308
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4328496788795056,
      "learning_rate": 1.973435716595344e-05,
      "loss": 0.1957,
      "step": 3309
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7196732609821253,
      "learning_rate": 1.9734130015934245e-05,
      "loss": 0.375,
      "step": 3310
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7528423318970938,
      "learning_rate": 1.9733902770147475e-05,
      "loss": 0.3904,
      "step": 3311
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.37055680770811616,
      "learning_rate": 1.9733675428595367e-05,
      "loss": 0.2398,
      "step": 3312
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1786978447211363,
      "learning_rate": 1.9733447991280156e-05,
      "loss": 0.4891,
      "step": 3313
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44203880197543355,
      "learning_rate": 1.9733220458204083e-05,
      "loss": 0.2112,
      "step": 3314
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6383273590324312,
      "learning_rate": 1.9732992829369383e-05,
      "loss": 0.3908,
      "step": 3315
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5377193082177073,
      "learning_rate": 1.9732765104778297e-05,
      "loss": 0.3391,
      "step": 3316
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5391242327264196,
      "learning_rate": 1.973253728443307e-05,
      "loss": 0.3765,
      "step": 3317
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5136898915291167,
      "learning_rate": 1.9732309368335935e-05,
      "loss": 0.075,
      "step": 3318
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4859103013058434,
      "learning_rate": 1.9732081356489135e-05,
      "loss": 0.3392,
      "step": 3319
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.868439685693736,
      "learning_rate": 1.9731853248894917e-05,
      "loss": 0.4557,
      "step": 3320
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.410811701006169,
      "learning_rate": 1.973162504555553e-05,
      "loss": 0.1937,
      "step": 3321
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4773795198155746,
      "learning_rate": 1.973139674647321e-05,
      "loss": 0.2811,
      "step": 3322
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48500024507661765,
      "learning_rate": 1.9731168351650208e-05,
      "loss": 0.2877,
      "step": 3323
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.38713822438380485,
      "learning_rate": 1.973093986108877e-05,
      "loss": 0.2714,
      "step": 3324
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.0962669854933766,
      "learning_rate": 1.973071127479114e-05,
      "loss": 0.1416,
      "step": 3325
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8169475195190758,
      "learning_rate": 1.9730482592759577e-05,
      "loss": 0.8558,
      "step": 3326
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6694674821267088,
      "learning_rate": 1.973025381499632e-05,
      "loss": 0.3242,
      "step": 3327
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.48510680273441115,
      "learning_rate": 1.9730024941503624e-05,
      "loss": 0.345,
      "step": 3328
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4356758104464746,
      "learning_rate": 1.972979597228374e-05,
      "loss": 0.2863,
      "step": 3329
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.611917459539103,
      "learning_rate": 1.9729566907338927e-05,
      "loss": 0.7626,
      "step": 3330
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3272649391223014,
      "learning_rate": 1.9729337746671427e-05,
      "loss": 0.0989,
      "step": 3331
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4651904867213246,
      "learning_rate": 1.9729108490283506e-05,
      "loss": 0.2462,
      "step": 3332
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5619933941765786,
      "learning_rate": 1.9728879138177412e-05,
      "loss": 0.2759,
      "step": 3333
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1386412934443135,
      "learning_rate": 1.9728649690355406e-05,
      "loss": 0.7788,
      "step": 3334
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40870560893504526,
      "learning_rate": 1.972842014681974e-05,
      "loss": 0.2989,
      "step": 3335
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.9321839433934748,
      "learning_rate": 1.9728190507572677e-05,
      "loss": 0.4005,
      "step": 3336
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4698497609622882,
      "learning_rate": 1.9727960772616477e-05,
      "loss": 0.2472,
      "step": 3337
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8111668750717693,
      "learning_rate": 1.9727730941953397e-05,
      "loss": 0.4353,
      "step": 3338
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43636583073054247,
      "learning_rate": 1.9727501015585697e-05,
      "loss": 0.2193,
      "step": 3339
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3756255742204158,
      "learning_rate": 1.9727270993515645e-05,
      "loss": 0.254,
      "step": 3340
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3020880571750333,
      "learning_rate": 1.9727040875745493e-05,
      "loss": 0.4898,
      "step": 3341
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4951846273035573,
      "learning_rate": 1.9726810662277517e-05,
      "loss": 0.2585,
      "step": 3342
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.2661547031461744,
      "learning_rate": 1.972658035311398e-05,
      "loss": 0.6997,
      "step": 3343
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8267516830019656,
      "learning_rate": 1.972634994825714e-05,
      "loss": 0.4491,
      "step": 3344
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1697520269611414,
      "learning_rate": 1.9726119447709272e-05,
      "loss": 0.5837,
      "step": 3345
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3384011092875309,
      "learning_rate": 1.9725888851472637e-05,
      "loss": 0.2099,
      "step": 3346
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4309535919727237,
      "learning_rate": 1.9725658159549513e-05,
      "loss": 0.3444,
      "step": 3347
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3634366695429794,
      "learning_rate": 1.972542737194216e-05,
      "loss": 0.1283,
      "step": 3348
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5745202908390933,
      "learning_rate": 1.972519648865285e-05,
      "loss": 0.2487,
      "step": 3349
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4101419099450087,
      "learning_rate": 1.972496550968386e-05,
      "loss": 0.1849,
      "step": 3350
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.618489420026972,
      "learning_rate": 1.972473443503746e-05,
      "loss": 0.3055,
      "step": 3351
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.64271707853532,
      "learning_rate": 1.972450326471592e-05,
      "loss": 0.4419,
      "step": 3352
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5042828031777327,
      "learning_rate": 1.972427199872152e-05,
      "loss": 0.3104,
      "step": 3353
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7789084232405223,
      "learning_rate": 1.9724040637056535e-05,
      "loss": 0.4803,
      "step": 3354
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3841016519150726,
      "learning_rate": 1.972380917972323e-05,
      "loss": 0.2422,
      "step": 3355
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.7141989658378347,
      "learning_rate": 1.97235776267239e-05,
      "loss": 0.8807,
      "step": 3356
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.39868309977223665,
      "learning_rate": 1.9723345978060807e-05,
      "loss": 0.1294,
      "step": 3357
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3439795794205533,
      "learning_rate": 1.972311423373624e-05,
      "loss": 0.2906,
      "step": 3358
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3601648652425431,
      "learning_rate": 1.9722882393752476e-05,
      "loss": 0.1682,
      "step": 3359
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7239407058220272,
      "learning_rate": 1.9722650458111793e-05,
      "loss": 0.3997,
      "step": 3360
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.0943391065226873,
      "learning_rate": 1.972241842681648e-05,
      "loss": 0.4304,
      "step": 3361
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7399171624784473,
      "learning_rate": 1.9722186299868814e-05,
      "loss": 0.5254,
      "step": 3362
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4703176018007111,
      "learning_rate": 1.9721954077271078e-05,
      "loss": 0.1901,
      "step": 3363
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.44311965250432356,
      "learning_rate": 1.972172175902556e-05,
      "loss": 0.2497,
      "step": 3364
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.41113246396042347,
      "learning_rate": 1.9721489345134542e-05,
      "loss": 0.3242,
      "step": 3365
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.8959790890863584,
      "learning_rate": 1.9721256835600317e-05,
      "loss": 0.4924,
      "step": 3366
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3709154113721197,
      "learning_rate": 1.9721024230425168e-05,
      "loss": 0.1954,
      "step": 3367
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.46144642215293497,
      "learning_rate": 1.9720791529611384e-05,
      "loss": 0.0805,
      "step": 3368
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5082363163609421,
      "learning_rate": 1.972055873316125e-05,
      "loss": 0.3726,
      "step": 3369
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5771163047900945,
      "learning_rate": 1.9720325841077066e-05,
      "loss": 0.348,
      "step": 3370
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5408021834565164,
      "learning_rate": 1.9720092853361116e-05,
      "loss": 0.3932,
      "step": 3371
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.2854013263217996,
      "learning_rate": 1.9719859770015696e-05,
      "loss": 0.3812,
      "step": 3372
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4843472449359799,
      "learning_rate": 1.9719626591043095e-05,
      "loss": 0.3081,
      "step": 3373
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5382244355838187,
      "learning_rate": 1.9719393316445607e-05,
      "loss": 0.3209,
      "step": 3374
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.566307058326034,
      "learning_rate": 1.9719159946225534e-05,
      "loss": 0.7523,
      "step": 3375
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3636336092041782,
      "learning_rate": 1.9718926480385168e-05,
      "loss": 0.195,
      "step": 3376
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3067553890376749,
      "learning_rate": 1.9718692918926803e-05,
      "loss": 0.1658,
      "step": 3377
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43370969531309383,
      "learning_rate": 1.9718459261852737e-05,
      "loss": 0.3133,
      "step": 3378
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6924479272942751,
      "learning_rate": 1.9718225509165273e-05,
      "loss": 0.4476,
      "step": 3379
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.465048681529723,
      "learning_rate": 1.9717991660866712e-05,
      "loss": 0.6781,
      "step": 3380
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6351458428337726,
      "learning_rate": 1.9717757716959348e-05,
      "loss": 0.3012,
      "step": 3381
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4163084283244735,
      "learning_rate": 1.9717523677445484e-05,
      "loss": 0.3129,
      "step": 3382
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.42008970955723995,
      "learning_rate": 1.971728954232743e-05,
      "loss": 0.3022,
      "step": 3383
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.1865774931053332,
      "learning_rate": 1.9717055311607482e-05,
      "loss": 0.8391,
      "step": 3384
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.2877049753410503,
      "learning_rate": 1.9716820985287944e-05,
      "loss": 0.0739,
      "step": 3385
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6481713303003924,
      "learning_rate": 1.9716586563371126e-05,
      "loss": 0.399,
      "step": 3386
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.40723601516205465,
      "learning_rate": 1.971635204585933e-05,
      "loss": 0.2813,
      "step": 3387
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7674271456659147,
      "learning_rate": 1.9716117432754868e-05,
      "loss": 0.5433,
      "step": 3388
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.35306171577530093,
      "learning_rate": 1.9715882724060046e-05,
      "loss": 0.2643,
      "step": 3389
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.3839142347371618,
      "learning_rate": 1.9715647919777174e-05,
      "loss": 0.7059,
      "step": 3390
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4402032002369694,
      "learning_rate": 1.9715413019908554e-05,
      "loss": 0.1859,
      "step": 3391
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5186785401360696,
      "learning_rate": 1.9715178024456508e-05,
      "loss": 0.2945,
      "step": 3392
    },
    {
      "epoch": 0.1,
      "grad_norm": 2.030807011363542,
      "learning_rate": 1.9714942933423343e-05,
      "loss": 0.9045,
      "step": 3393
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.31101411892597947,
      "learning_rate": 1.9714707746811377e-05,
      "loss": 0.2063,
      "step": 3394
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8338846359852157,
      "learning_rate": 1.9714472464622917e-05,
      "loss": 0.2797,
      "step": 3395
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.43790074081332386,
      "learning_rate": 1.971423708686028e-05,
      "loss": 0.2863,
      "step": 3396
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8549564652462044,
      "learning_rate": 1.971400161352578e-05,
      "loss": 0.5049,
      "step": 3397
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3293358567437919,
      "learning_rate": 1.971376604462174e-05,
      "loss": 0.1507,
      "step": 3398
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.863805137288676,
      "learning_rate": 1.9713530380150473e-05,
      "loss": 0.8927,
      "step": 3399
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.36985545849411333,
      "learning_rate": 1.9713294620114295e-05,
      "loss": 0.2213,
      "step": 3400
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4669650504745605,
      "learning_rate": 1.971305876451553e-05,
      "loss": 0.3526,
      "step": 3401
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.8733462539579183,
      "learning_rate": 1.97128228133565e-05,
      "loss": 0.4457,
      "step": 3402
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.554601823813819,
      "learning_rate": 1.971258676663952e-05,
      "loss": 0.7432,
      "step": 3403
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.649653744995658,
      "learning_rate": 1.9712350624366913e-05,
      "loss": 0.3822,
      "step": 3404
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3527979762017107,
      "learning_rate": 1.9712114386541008e-05,
      "loss": 0.2554,
      "step": 3405
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.32784829997577986,
      "learning_rate": 1.9711878053164125e-05,
      "loss": 0.257,
      "step": 3406
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4208739674483356,
      "learning_rate": 1.9711641624238588e-05,
      "loss": 0.076,
      "step": 3407
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.6617768098008399,
      "learning_rate": 1.9711405099766728e-05,
      "loss": 0.9088,
      "step": 3408
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4394540049479102,
      "learning_rate": 1.971116847975087e-05,
      "loss": 0.2048,
      "step": 3409
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.6760516627184867,
      "learning_rate": 1.971093176419334e-05,
      "loss": 0.4123,
      "step": 3410
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.9658621602790511,
      "learning_rate": 1.9710694953096466e-05,
      "loss": 0.3867,
      "step": 3411
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3836949421900617,
      "learning_rate": 1.9710458046462585e-05,
      "loss": 0.3443,
      "step": 3412
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4752454230081485,
      "learning_rate": 1.9710221044294016e-05,
      "loss": 0.3043,
      "step": 3413
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4261569881156143,
      "learning_rate": 1.97099839465931e-05,
      "loss": 0.3821,
      "step": 3414
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1265047365512404,
      "learning_rate": 1.970974675336217e-05,
      "loss": 0.3508,
      "step": 3415
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.33627903855862157,
      "learning_rate": 1.9709509464603555e-05,
      "loss": 0.1699,
      "step": 3416
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4954565270774085,
      "learning_rate": 1.970927208031959e-05,
      "loss": 0.3136,
      "step": 3417
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4016901445117461,
      "learning_rate": 1.970903460051261e-05,
      "loss": 0.199,
      "step": 3418
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4255468400758043,
      "learning_rate": 1.9708797025184955e-05,
      "loss": 0.318,
      "step": 3419
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1079219201709571,
      "learning_rate": 1.9708559354338958e-05,
      "loss": 0.439,
      "step": 3420
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7194995803078934,
      "learning_rate": 1.9708321587976963e-05,
      "loss": 0.5164,
      "step": 3421
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4836329210219121,
      "learning_rate": 1.9708083726101303e-05,
      "loss": 0.3156,
      "step": 3422
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.5534068603395104,
      "learning_rate": 1.9707845768714324e-05,
      "loss": 0.3799,
      "step": 3423
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.38415590565991814,
      "learning_rate": 1.970760771581836e-05,
      "loss": 0.2731,
      "step": 3424
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.3460858705044697,
      "learning_rate": 1.970736956741576e-05,
      "loss": 0.1796,
      "step": 3425
    },
    {
      "epoch": 0.1,
      "grad_norm": 1.1291646163558442,
      "learning_rate": 1.970713132350886e-05,
      "loss": 0.5312,
      "step": 3426
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.7268433235819058,
      "learning_rate": 1.970689298410001e-05,
      "loss": 0.349,
      "step": 3427
    },
    {
      "epoch": 0.1,
      "grad_norm": 0.4427110616218943,
      "learning_rate": 1.9706654549191555e-05,
      "loss": 0.2634,
      "step": 3428
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9411204493479061,
      "learning_rate": 1.9706416018785838e-05,
      "loss": 0.5731,
      "step": 3429
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3684552984814075,
      "learning_rate": 1.9706177392885206e-05,
      "loss": 0.2646,
      "step": 3430
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7382726264529268,
      "learning_rate": 1.9705938671492007e-05,
      "loss": 0.3755,
      "step": 3431
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3963615862292563,
      "learning_rate": 1.970569985460859e-05,
      "loss": 0.2877,
      "step": 3432
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.37289974965759926,
      "learning_rate": 1.9705460942237303e-05,
      "loss": 0.1238,
      "step": 3433
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5664197216204874,
      "learning_rate": 1.9705221934380496e-05,
      "loss": 0.2152,
      "step": 3434
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5254776554438767,
      "learning_rate": 1.9704982831040527e-05,
      "loss": 0.3366,
      "step": 3435
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46383689551677504,
      "learning_rate": 1.970474363221974e-05,
      "loss": 0.3016,
      "step": 3436
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3920883673447425,
      "learning_rate": 1.9704504337920493e-05,
      "loss": 0.2601,
      "step": 3437
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1319695733960704,
      "learning_rate": 1.970426494814514e-05,
      "loss": 0.6177,
      "step": 3438
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8728948227257699,
      "learning_rate": 1.9704025462896036e-05,
      "loss": 0.4604,
      "step": 3439
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5295907002848514,
      "learning_rate": 1.9703785882175533e-05,
      "loss": 0.746,
      "step": 3440
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.33550587990929875,
      "learning_rate": 1.9703546205985993e-05,
      "loss": 0.2087,
      "step": 3441
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5247948192038066,
      "learning_rate": 1.9703306434329775e-05,
      "loss": 0.3518,
      "step": 3442
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.30613294523664397,
      "learning_rate": 1.970306656720923e-05,
      "loss": 0.1751,
      "step": 3443
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1603992043975475,
      "learning_rate": 1.970282660462673e-05,
      "loss": 0.6776,
      "step": 3444
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.444746528735391,
      "learning_rate": 1.9702586546584627e-05,
      "loss": 0.2513,
      "step": 3445
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2980039987457932,
      "learning_rate": 1.9702346393085282e-05,
      "loss": 0.1571,
      "step": 3446
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8610315983381991,
      "learning_rate": 1.970210614413106e-05,
      "loss": 0.5647,
      "step": 3447
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.38170348733152853,
      "learning_rate": 1.970186579972433e-05,
      "loss": 0.3002,
      "step": 3448
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1486015880550282,
      "learning_rate": 1.970162535986745e-05,
      "loss": 0.6504,
      "step": 3449
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.39718425057376805,
      "learning_rate": 1.9701384824562786e-05,
      "loss": 0.191,
      "step": 3450
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.577120948856292,
      "learning_rate": 1.970114419381271e-05,
      "loss": 0.3947,
      "step": 3451
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3061538469949408,
      "learning_rate": 1.970090346761958e-05,
      "loss": 0.1275,
      "step": 3452
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5369300709957167,
      "learning_rate": 1.9700662645985773e-05,
      "loss": 0.369,
      "step": 3453
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3833387204905759,
      "learning_rate": 1.9700421728913653e-05,
      "loss": 0.2402,
      "step": 3454
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42086509807599604,
      "learning_rate": 1.9700180716405593e-05,
      "loss": 0.3174,
      "step": 3455
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6298129770085584,
      "learning_rate": 1.9699939608463964e-05,
      "loss": 0.4377,
      "step": 3456
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.27726996922847746,
      "learning_rate": 1.969969840509114e-05,
      "loss": 0.0746,
      "step": 3457
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0881966301001573,
      "learning_rate": 1.9699457106289488e-05,
      "loss": 0.7973,
      "step": 3458
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3540539968478035,
      "learning_rate": 1.9699215712061385e-05,
      "loss": 0.2103,
      "step": 3459
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.41860526323790637,
      "learning_rate": 1.969897422240921e-05,
      "loss": 0.332,
      "step": 3460
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9982821526647279,
      "learning_rate": 1.969873263733533e-05,
      "loss": 0.3931,
      "step": 3461
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8140484925751472,
      "learning_rate": 1.969849095684213e-05,
      "loss": 0.9087,
      "step": 3462
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4703203448584776,
      "learning_rate": 1.9698249180931985e-05,
      "loss": 0.2975,
      "step": 3463
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4278383249466331,
      "learning_rate": 1.9698007309607275e-05,
      "loss": 0.3441,
      "step": 3464
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9563470792304015,
      "learning_rate": 1.9697765342870372e-05,
      "loss": 0.4407,
      "step": 3465
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3632852714550674,
      "learning_rate": 1.9697523280723668e-05,
      "loss": 0.2766,
      "step": 3466
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42444729334060516,
      "learning_rate": 1.9697281123169538e-05,
      "loss": 0.1293,
      "step": 3467
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4996253165486201,
      "learning_rate": 1.9697038870210363e-05,
      "loss": 0.3005,
      "step": 3468
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.564640481498744,
      "learning_rate": 1.9696796521848532e-05,
      "loss": 0.2375,
      "step": 3469
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0358219226445151,
      "learning_rate": 1.9696554078086424e-05,
      "loss": 0.4897,
      "step": 3470
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40956444790849245,
      "learning_rate": 1.969631153892643e-05,
      "loss": 0.3299,
      "step": 3471
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.671124994324767,
      "learning_rate": 1.969606890437093e-05,
      "loss": 0.3132,
      "step": 3472
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7528223950888313,
      "learning_rate": 1.9695826174422312e-05,
      "loss": 0.3751,
      "step": 3473
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5200840748333859,
      "learning_rate": 1.969558334908297e-05,
      "loss": 0.3491,
      "step": 3474
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4037511717097504,
      "learning_rate": 1.9695340428355285e-05,
      "loss": 0.1619,
      "step": 3475
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2279020279165918,
      "learning_rate": 1.969509741224165e-05,
      "loss": 0.3327,
      "step": 3476
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.809485264083097,
      "learning_rate": 1.969485430074446e-05,
      "loss": 0.3928,
      "step": 3477
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3280995853481177,
      "learning_rate": 1.96946110938661e-05,
      "loss": 0.2614,
      "step": 3478
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.4883956400586218,
      "learning_rate": 1.9694367791608968e-05,
      "loss": 0.8332,
      "step": 3479
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8423849206208973,
      "learning_rate": 1.9694124393975458e-05,
      "loss": 0.4405,
      "step": 3480
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.865558106663981,
      "learning_rate": 1.969388090096796e-05,
      "loss": 0.6023,
      "step": 3481
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42506527165934577,
      "learning_rate": 1.9693637312588868e-05,
      "loss": 0.2139,
      "step": 3482
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3841452121957171,
      "learning_rate": 1.9693393628840588e-05,
      "loss": 0.2885,
      "step": 3483
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5662149536793906,
      "learning_rate": 1.969314984972551e-05,
      "loss": 0.2647,
      "step": 3484
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.38604400257365246,
      "learning_rate": 1.9692905975246037e-05,
      "loss": 0.1362,
      "step": 3485
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8293856628865219,
      "learning_rate": 1.9692662005404563e-05,
      "loss": 0.3536,
      "step": 3486
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4112564496352325,
      "learning_rate": 1.969241794020349e-05,
      "loss": 0.2455,
      "step": 3487
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.780078734747479,
      "learning_rate": 1.9692173779645222e-05,
      "loss": 0.8226,
      "step": 3488
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4456739989210251,
      "learning_rate": 1.969192952373216e-05,
      "loss": 0.3369,
      "step": 3489
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.803205958390472,
      "learning_rate": 1.9691685172466705e-05,
      "loss": 0.4796,
      "step": 3490
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4225265722207492,
      "learning_rate": 1.969144072585126e-05,
      "loss": 0.2365,
      "step": 3491
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0221449046245854,
      "learning_rate": 1.9691196183888235e-05,
      "loss": 0.9314,
      "step": 3492
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40509197335586594,
      "learning_rate": 1.969095154658003e-05,
      "loss": 0.1072,
      "step": 3493
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6003597169516106,
      "learning_rate": 1.969070681392906e-05,
      "loss": 0.2962,
      "step": 3494
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3445446318295137,
      "learning_rate": 1.969046198593772e-05,
      "loss": 0.2344,
      "step": 3495
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3953230131586065,
      "learning_rate": 1.9690217062608437e-05,
      "loss": 0.2606,
      "step": 3496
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0483044972570608,
      "learning_rate": 1.9689972043943605e-05,
      "loss": 0.6366,
      "step": 3497
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9850276599862928,
      "learning_rate": 1.9689726929945638e-05,
      "loss": 0.4591,
      "step": 3498
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0807742355314798,
      "learning_rate": 1.968948172061695e-05,
      "loss": 0.5823,
      "step": 3499
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4398347176425235,
      "learning_rate": 1.9689236415959953e-05,
      "loss": 0.1957,
      "step": 3500
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6962103360981055,
      "learning_rate": 1.968899101597706e-05,
      "loss": 0.4253,
      "step": 3501
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3217129672530627,
      "learning_rate": 1.968874552067069e-05,
      "loss": 0.2244,
      "step": 3502
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5721917430239251,
      "learning_rate": 1.9688499930043244e-05,
      "loss": 0.2775,
      "step": 3503
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43645708648067855,
      "learning_rate": 1.9688254244097155e-05,
      "loss": 0.0815,
      "step": 3504
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.46300839066634736,
      "learning_rate": 1.968800846283483e-05,
      "loss": 0.3423,
      "step": 3505
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.124463242802941,
      "learning_rate": 1.9687762586258687e-05,
      "loss": 0.4628,
      "step": 3506
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40511179555029636,
      "learning_rate": 1.968751661437115e-05,
      "loss": 0.3651,
      "step": 3507
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1078437724028172,
      "learning_rate": 1.968727054717464e-05,
      "loss": 0.5304,
      "step": 3508
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.385120829343743,
      "learning_rate": 1.9687024384671575e-05,
      "loss": 0.2103,
      "step": 3509
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7143955707072678,
      "learning_rate": 1.9686778126864376e-05,
      "loss": 0.8372,
      "step": 3510
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.35873333140427255,
      "learning_rate": 1.9686531773755463e-05,
      "loss": 0.1554,
      "step": 3511
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6332333569381147,
      "learning_rate": 1.9686285325347266e-05,
      "loss": 0.332,
      "step": 3512
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3814326797223837,
      "learning_rate": 1.9686038781642205e-05,
      "loss": 0.228,
      "step": 3513
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4889948144912823,
      "learning_rate": 1.9685792142642705e-05,
      "loss": 0.3351,
      "step": 3514
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.233399981644852,
      "learning_rate": 1.9685545408351202e-05,
      "loss": 0.476,
      "step": 3515
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8627784605592406,
      "learning_rate": 1.9685298578770112e-05,
      "loss": 0.5289,
      "step": 3516
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.312301548996366,
      "learning_rate": 1.9685051653901864e-05,
      "loss": 0.6902,
      "step": 3517
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4748251194809778,
      "learning_rate": 1.96848046337489e-05,
      "loss": 0.278,
      "step": 3518
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5251113589893223,
      "learning_rate": 1.9684557518313633e-05,
      "loss": 0.2681,
      "step": 3519
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5016590982047854,
      "learning_rate": 1.9684310307598505e-05,
      "loss": 0.3515,
      "step": 3520
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42174999074714353,
      "learning_rate": 1.968406300160595e-05,
      "loss": 0.0809,
      "step": 3521
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3777912851360465,
      "learning_rate": 1.9683815600338393e-05,
      "loss": 0.1975,
      "step": 3522
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7137908383176655,
      "learning_rate": 1.9683568103798272e-05,
      "loss": 0.4332,
      "step": 3523
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9763350134068807,
      "learning_rate": 1.968332051198802e-05,
      "loss": 0.4986,
      "step": 3524
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42959125266745857,
      "learning_rate": 1.9683072824910078e-05,
      "loss": 0.3209,
      "step": 3525
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6783573176819886,
      "learning_rate": 1.9682825042566874e-05,
      "loss": 0.3307,
      "step": 3526
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9983156137703691,
      "learning_rate": 1.9682577164960856e-05,
      "loss": 0.4108,
      "step": 3527
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4196924134471987,
      "learning_rate": 1.9682329192094457e-05,
      "loss": 0.2589,
      "step": 3528
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9420857389928808,
      "learning_rate": 1.968208112397012e-05,
      "loss": 0.5623,
      "step": 3529
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2903570080973607,
      "learning_rate": 1.968183296059028e-05,
      "loss": 0.1944,
      "step": 3530
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9434820405803831,
      "learning_rate": 1.968158470195738e-05,
      "loss": 0.5806,
      "step": 3531
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3604090131328366,
      "learning_rate": 1.9681336348073867e-05,
      "loss": 0.2116,
      "step": 3532
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5897270590643848,
      "learning_rate": 1.968108789894218e-05,
      "loss": 0.5234,
      "step": 3533
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8159288022651602,
      "learning_rate": 1.9680839354564766e-05,
      "loss": 0.4691,
      "step": 3534
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0633538385285777,
      "learning_rate": 1.9680590714944067e-05,
      "loss": 0.5401,
      "step": 3535
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49631398281301614,
      "learning_rate": 1.9680341980082532e-05,
      "loss": 0.3036,
      "step": 3536
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.36967886635960584,
      "learning_rate": 1.968009314998261e-05,
      "loss": 0.2833,
      "step": 3537
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.5311190452804513,
      "learning_rate": 1.9679844224646742e-05,
      "loss": 0.8738,
      "step": 3538
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0375053860141292,
      "learning_rate": 1.9679595204077383e-05,
      "loss": 0.4269,
      "step": 3539
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7065717210152465,
      "learning_rate": 1.9679346088276983e-05,
      "loss": 0.4398,
      "step": 3540
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3433451984393618,
      "learning_rate": 1.967909687724799e-05,
      "loss": 0.275,
      "step": 3541
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44815981232453267,
      "learning_rate": 1.967884757099286e-05,
      "loss": 0.1916,
      "step": 3542
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.36889711631979283,
      "learning_rate": 1.9678598169514037e-05,
      "loss": 0.2657,
      "step": 3543
    },
    {
      "epoch": 0.11,
      "grad_norm": 3.4161654244557025,
      "learning_rate": 1.9678348672813983e-05,
      "loss": 0.8481,
      "step": 3544
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.387553045369222,
      "learning_rate": 1.967809908089515e-05,
      "loss": 0.2086,
      "step": 3545
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6625639598398697,
      "learning_rate": 1.9677849393759997e-05,
      "loss": 0.3614,
      "step": 3546
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9475030631615613,
      "learning_rate": 1.9677599611410973e-05,
      "loss": 0.3727,
      "step": 3547
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4505740244401815,
      "learning_rate": 1.967734973385054e-05,
      "loss": 0.3176,
      "step": 3548
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0349043067222812,
      "learning_rate": 1.967709976108116e-05,
      "loss": 0.5059,
      "step": 3549
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4601713284250646,
      "learning_rate": 1.9676849693105284e-05,
      "loss": 0.1997,
      "step": 3550
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40533644630483723,
      "learning_rate": 1.967659952992538e-05,
      "loss": 0.2948,
      "step": 3551
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3577584365967452,
      "learning_rate": 1.9676349271543903e-05,
      "loss": 0.1814,
      "step": 3552
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7404724897412684,
      "learning_rate": 1.967609891796332e-05,
      "loss": 0.8163,
      "step": 3553
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43035978345710907,
      "learning_rate": 1.967584846918609e-05,
      "loss": 0.0805,
      "step": 3554
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42966216617521186,
      "learning_rate": 1.967559792521468e-05,
      "loss": 0.3027,
      "step": 3555
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5089245020082539,
      "learning_rate": 1.9675347286051557e-05,
      "loss": 0.297,
      "step": 3556
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7450470102925686,
      "learning_rate": 1.967509655169918e-05,
      "loss": 0.5088,
      "step": 3557
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8799461120681218,
      "learning_rate": 1.967484572216002e-05,
      "loss": 0.4116,
      "step": 3558
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4845972255285936,
      "learning_rate": 1.9674594797436542e-05,
      "loss": 0.2809,
      "step": 3559
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.27858114650551813,
      "learning_rate": 1.967434377753122e-05,
      "loss": 0.134,
      "step": 3560
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.37945455827140456,
      "learning_rate": 1.967409266244652e-05,
      "loss": 0.281,
      "step": 3561
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0413750209096797,
      "learning_rate": 1.9673841452184917e-05,
      "loss": 0.8169,
      "step": 3562
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45033486254990773,
      "learning_rate": 1.967359014674887e-05,
      "loss": 0.1237,
      "step": 3563
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5419141358461274,
      "learning_rate": 1.9673338746140868e-05,
      "loss": 0.356,
      "step": 3564
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2142395838271645,
      "learning_rate": 1.9673087250363374e-05,
      "loss": 0.4617,
      "step": 3565
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6438616055220631,
      "learning_rate": 1.9672835659418864e-05,
      "loss": 0.4371,
      "step": 3566
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4389797825361181,
      "learning_rate": 1.9672583973309817e-05,
      "loss": 0.2933,
      "step": 3567
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44879856957701814,
      "learning_rate": 1.9672332192038702e-05,
      "loss": 0.2879,
      "step": 3568
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2070062935238857,
      "learning_rate": 1.9672080315608003e-05,
      "loss": 0.3829,
      "step": 3569
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6727443121274178,
      "learning_rate": 1.9671828344020196e-05,
      "loss": 0.2584,
      "step": 3570
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3713313892280532,
      "learning_rate": 1.9671576277277754e-05,
      "loss": 0.1682,
      "step": 3571
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5294955422474348,
      "learning_rate": 1.967132411538317e-05,
      "loss": 0.3515,
      "step": 3572
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.39526814815864625,
      "learning_rate": 1.9671071858338912e-05,
      "loss": 0.2606,
      "step": 3573
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.2147664083426597,
      "learning_rate": 1.9670819506147465e-05,
      "loss": 0.8553,
      "step": 3574
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2847148402812905,
      "learning_rate": 1.9670567058811318e-05,
      "loss": 0.4885,
      "step": 3575
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7355993361968892,
      "learning_rate": 1.967031451633295e-05,
      "loss": 0.2018,
      "step": 3576
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0644775068237504,
      "learning_rate": 1.9670061878714842e-05,
      "loss": 0.4445,
      "step": 3577
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3310938813446195,
      "learning_rate": 1.9669809145959485e-05,
      "loss": 0.216,
      "step": 3578
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43699763016980914,
      "learning_rate": 1.9669556318069368e-05,
      "loss": 0.2674,
      "step": 3579
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.330111188312197,
      "learning_rate": 1.966930339504697e-05,
      "loss": 0.6826,
      "step": 3580
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.5983333426690507,
      "learning_rate": 1.9669050376894785e-05,
      "loss": 0.8162,
      "step": 3581
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3368627584024841,
      "learning_rate": 1.9668797263615297e-05,
      "loss": 0.2077,
      "step": 3582
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7040473887002037,
      "learning_rate": 1.9668544055211004e-05,
      "loss": 0.5196,
      "step": 3583
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.35903762772245695,
      "learning_rate": 1.9668290751684392e-05,
      "loss": 0.313,
      "step": 3584
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7544220944130033,
      "learning_rate": 1.9668037353037955e-05,
      "loss": 0.7553,
      "step": 3585
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.37969349714036427,
      "learning_rate": 1.9667783859274188e-05,
      "loss": 0.1841,
      "step": 3586
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7328214983014727,
      "learning_rate": 1.966753027039558e-05,
      "loss": 0.4171,
      "step": 3587
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.38404681870638163,
      "learning_rate": 1.9667276586404624e-05,
      "loss": 0.1377,
      "step": 3588
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1831950378249863,
      "learning_rate": 1.9667022807303826e-05,
      "loss": 0.5732,
      "step": 3589
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.412678802583059,
      "learning_rate": 1.9666768933095674e-05,
      "loss": 0.2916,
      "step": 3590
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.33871498881901163,
      "learning_rate": 1.9666514963782667e-05,
      "loss": 0.261,
      "step": 3591
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7336740902536896,
      "learning_rate": 1.9666260899367307e-05,
      "loss": 0.539,
      "step": 3592
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.316849295635626,
      "learning_rate": 1.9666006739852093e-05,
      "loss": 0.0738,
      "step": 3593
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.551075360095804,
      "learning_rate": 1.9665752485239523e-05,
      "loss": 0.6597,
      "step": 3594
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3924059423941724,
      "learning_rate": 1.96654981355321e-05,
      "loss": 0.1841,
      "step": 3595
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.47657587920998784,
      "learning_rate": 1.9665243690732323e-05,
      "loss": 0.3221,
      "step": 3596
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.36009350620174185,
      "learning_rate": 1.9664989150842702e-05,
      "loss": 0.1824,
      "step": 3597
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.065040036023923,
      "learning_rate": 1.9664734515865736e-05,
      "loss": 0.9581,
      "step": 3598
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5214964874765103,
      "learning_rate": 1.966447978580393e-05,
      "loss": 0.263,
      "step": 3599
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4574545865235964,
      "learning_rate": 1.9664224960659798e-05,
      "loss": 0.3484,
      "step": 3600
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8529681871543567,
      "learning_rate": 1.9663970040435836e-05,
      "loss": 0.4463,
      "step": 3601
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.29059000673422275,
      "learning_rate": 1.9663715025134555e-05,
      "loss": 0.2175,
      "step": 3602
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1626407445756266,
      "learning_rate": 1.966345991475847e-05,
      "loss": 0.6842,
      "step": 3603
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3625864982951864,
      "learning_rate": 1.9663204709310085e-05,
      "loss": 0.0749,
      "step": 3604
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5056349268045602,
      "learning_rate": 1.9662949408791912e-05,
      "loss": 0.3177,
      "step": 3605
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2964767390419136,
      "learning_rate": 1.966269401320646e-05,
      "loss": 0.5241,
      "step": 3606
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.631881423182667,
      "learning_rate": 1.9662438522556252e-05,
      "loss": 0.4136,
      "step": 3607
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4924265448192713,
      "learning_rate": 1.966218293684379e-05,
      "loss": 0.2961,
      "step": 3608
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5068762066379493,
      "learning_rate": 1.9661927256071595e-05,
      "loss": 0.3148,
      "step": 3609
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5475712509261986,
      "learning_rate": 1.9661671480242178e-05,
      "loss": 0.3511,
      "step": 3610
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.355044988644434,
      "learning_rate": 1.966141560935806e-05,
      "loss": 0.1447,
      "step": 3611
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.03650736173417,
      "learning_rate": 1.9661159643421753e-05,
      "loss": 0.3871,
      "step": 3612
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8846679384469022,
      "learning_rate": 1.966090358243578e-05,
      "loss": 0.4105,
      "step": 3613
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.31134461172527905,
      "learning_rate": 1.9660647426402662e-05,
      "loss": 0.2338,
      "step": 3614
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4560411457139377,
      "learning_rate": 1.9660391175324915e-05,
      "loss": 0.2978,
      "step": 3615
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6561885250122144,
      "learning_rate": 1.9660134829205057e-05,
      "loss": 0.6042,
      "step": 3616
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6189704044497163,
      "learning_rate": 1.9659878388045618e-05,
      "loss": 0.3643,
      "step": 3617
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4054433563125435,
      "learning_rate": 1.9659621851849114e-05,
      "loss": 0.315,
      "step": 3618
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3468427434559217,
      "learning_rate": 1.9659365220618074e-05,
      "loss": 0.1687,
      "step": 3619
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6085216731802086,
      "learning_rate": 1.9659108494355016e-05,
      "loss": 0.3992,
      "step": 3620
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3227371325667059,
      "learning_rate": 1.9658851673062475e-05,
      "loss": 0.1815,
      "step": 3621
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8316034305335196,
      "learning_rate": 1.9658594756742974e-05,
      "loss": 0.384,
      "step": 3622
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4237611505918197,
      "learning_rate": 1.9658337745399035e-05,
      "loss": 0.2597,
      "step": 3623
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7803267798499058,
      "learning_rate": 1.9658080639033198e-05,
      "loss": 0.8995,
      "step": 3624
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2150202667877763,
      "learning_rate": 1.965782343764798e-05,
      "loss": 0.4877,
      "step": 3625
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40991518480152955,
      "learning_rate": 1.965756614124592e-05,
      "loss": 0.3635,
      "step": 3626
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3882485761828329,
      "learning_rate": 1.9657308749829544e-05,
      "loss": 0.1852,
      "step": 3627
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.555438697778699,
      "learning_rate": 1.9657051263401386e-05,
      "loss": 0.3016,
      "step": 3628
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3238136211854268,
      "learning_rate": 1.9656793681963985e-05,
      "loss": 0.1663,
      "step": 3629
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9673474029472895,
      "learning_rate": 1.9656536005519865e-05,
      "loss": 0.3754,
      "step": 3630
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8430165247872016,
      "learning_rate": 1.965627823407157e-05,
      "loss": 0.3566,
      "step": 3631
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3568279234249118,
      "learning_rate": 1.9656020367621634e-05,
      "loss": 0.2498,
      "step": 3632
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5964952450665594,
      "learning_rate": 1.9655762406172588e-05,
      "loss": 0.3909,
      "step": 3633
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1838797994604546,
      "learning_rate": 1.9655504349726974e-05,
      "loss": 0.4946,
      "step": 3634
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.7848974595298064,
      "learning_rate": 1.9655246198287333e-05,
      "loss": 0.9094,
      "step": 3635
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40708497000620303,
      "learning_rate": 1.9654987951856204e-05,
      "loss": 0.1878,
      "step": 3636
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6414187265130639,
      "learning_rate": 1.9654729610436126e-05,
      "loss": 0.3787,
      "step": 3637
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.283849698190462,
      "learning_rate": 1.965447117402964e-05,
      "loss": 0.2126,
      "step": 3638
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5971160897241224,
      "learning_rate": 1.965421264263929e-05,
      "loss": 0.2616,
      "step": 3639
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5746435105748213,
      "learning_rate": 1.9653954016267623e-05,
      "loss": 0.2637,
      "step": 3640
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40909955747446364,
      "learning_rate": 1.9653695294917178e-05,
      "loss": 0.2322,
      "step": 3641
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8261664310822158,
      "learning_rate": 1.96534364785905e-05,
      "loss": 0.5227,
      "step": 3642
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8040278866877786,
      "learning_rate": 1.9653177567290136e-05,
      "loss": 0.4272,
      "step": 3643
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40936824113032233,
      "learning_rate": 1.9652918561018637e-05,
      "loss": 0.344,
      "step": 3644
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3586919651424627,
      "learning_rate": 1.965265945977855e-05,
      "loss": 0.1956,
      "step": 3645
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6047720952259049,
      "learning_rate": 1.9652400263572423e-05,
      "loss": 0.361,
      "step": 3646
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3580484253719978,
      "learning_rate": 1.9652140972402806e-05,
      "loss": 0.0755,
      "step": 3647
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5533773224266357,
      "learning_rate": 1.965188158627225e-05,
      "loss": 0.2507,
      "step": 3648
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.39651513442414343,
      "learning_rate": 1.9651622105183308e-05,
      "loss": 0.2279,
      "step": 3649
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.388467546955183,
      "learning_rate": 1.965136252913853e-05,
      "loss": 0.3008,
      "step": 3650
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9356247644193911,
      "learning_rate": 1.965110285814047e-05,
      "loss": 0.4084,
      "step": 3651
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8400365590574732,
      "learning_rate": 1.965084309219169e-05,
      "loss": 0.5162,
      "step": 3652
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2208719787638804,
      "learning_rate": 1.9650583231294735e-05,
      "loss": 0.6803,
      "step": 3653
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4201132614852247,
      "learning_rate": 1.965032327545217e-05,
      "loss": 0.1846,
      "step": 3654
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.697005804860223,
      "learning_rate": 1.9650063224666547e-05,
      "loss": 0.4109,
      "step": 3655
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40990620217541135,
      "learning_rate": 1.964980307894043e-05,
      "loss": 0.2707,
      "step": 3656
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5237912889071623,
      "learning_rate": 1.9649542838276372e-05,
      "loss": 0.2145,
      "step": 3657
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7200172519298285,
      "learning_rate": 1.9649282502676936e-05,
      "loss": 0.3114,
      "step": 3658
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5269361610477882,
      "learning_rate": 1.9649022072144684e-05,
      "loss": 0.331,
      "step": 3659
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5606396461911692,
      "learning_rate": 1.964876154668218e-05,
      "loss": 0.3708,
      "step": 3660
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6459302220611193,
      "learning_rate": 1.9648500926291985e-05,
      "loss": 0.4833,
      "step": 3661
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5096699273324564,
      "learning_rate": 1.964824021097666e-05,
      "loss": 0.2982,
      "step": 3662
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8608833190677918,
      "learning_rate": 1.9647979400738773e-05,
      "loss": 0.3991,
      "step": 3663
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42506794950204435,
      "learning_rate": 1.9647718495580896e-05,
      "loss": 0.2731,
      "step": 3664
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2284166286087612,
      "learning_rate": 1.9647457495505584e-05,
      "loss": 0.649,
      "step": 3665
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.32806299360377233,
      "learning_rate": 1.9647196400515412e-05,
      "loss": 0.0766,
      "step": 3666
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43516758589227134,
      "learning_rate": 1.9646935210612952e-05,
      "loss": 0.2942,
      "step": 3667
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45615435153544576,
      "learning_rate": 1.9646673925800763e-05,
      "loss": 0.3217,
      "step": 3668
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44444741208403965,
      "learning_rate": 1.9646412546081424e-05,
      "loss": 0.3449,
      "step": 3669
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.441380636173214,
      "learning_rate": 1.9646151071457503e-05,
      "loss": 0.7767,
      "step": 3670
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0730395826144572,
      "learning_rate": 1.9645889501931576e-05,
      "loss": 0.6826,
      "step": 3671
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8421743162788707,
      "learning_rate": 1.9645627837506213e-05,
      "loss": 0.3072,
      "step": 3672
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.37830384711275755,
      "learning_rate": 1.964536607818399e-05,
      "loss": 0.2593,
      "step": 3673
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6072743074637879,
      "learning_rate": 1.9645104223967482e-05,
      "loss": 0.3721,
      "step": 3674
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2729094269506829,
      "learning_rate": 1.9644842274859264e-05,
      "loss": 0.0988,
      "step": 3675
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7149842612590915,
      "learning_rate": 1.9644580230861917e-05,
      "loss": 0.4314,
      "step": 3676
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40594534829420637,
      "learning_rate": 1.964431809197801e-05,
      "loss": 0.2668,
      "step": 3677
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1065164613694034,
      "learning_rate": 1.9644055858210137e-05,
      "loss": 0.6437,
      "step": 3678
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5020885944640873,
      "learning_rate": 1.9643793529560862e-05,
      "loss": 0.2694,
      "step": 3679
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4678057365640286,
      "learning_rate": 1.9643531106032778e-05,
      "loss": 0.2874,
      "step": 3680
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.493349532516944,
      "learning_rate": 1.9643268587628458e-05,
      "loss": 0.3085,
      "step": 3681
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5551447133817841,
      "learning_rate": 1.964300597435049e-05,
      "loss": 0.2449,
      "step": 3682
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.0266517078681046,
      "learning_rate": 1.9642743266201454e-05,
      "loss": 0.6199,
      "step": 3683
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9100984877681141,
      "learning_rate": 1.964248046318394e-05,
      "loss": 0.421,
      "step": 3684
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.41541664293008135,
      "learning_rate": 1.964221756530053e-05,
      "loss": 0.2979,
      "step": 3685
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4427503202374864,
      "learning_rate": 1.9641954572553808e-05,
      "loss": 0.1949,
      "step": 3686
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.39044448732545456,
      "learning_rate": 1.9641691484946367e-05,
      "loss": 0.2981,
      "step": 3687
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.29793575177567216,
      "learning_rate": 1.9641428302480793e-05,
      "loss": 0.1391,
      "step": 3688
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.8961638537559171,
      "learning_rate": 1.9641165025159676e-05,
      "loss": 0.888,
      "step": 3689
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42744823345720356,
      "learning_rate": 1.96409016529856e-05,
      "loss": 0.0802,
      "step": 3690
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42595196626410514,
      "learning_rate": 1.9640638185961164e-05,
      "loss": 0.2773,
      "step": 3691
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3923255592220609,
      "learning_rate": 1.964037462408896e-05,
      "loss": 0.2669,
      "step": 3692
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9753670744739139,
      "learning_rate": 1.9640110967371575e-05,
      "loss": 0.4215,
      "step": 3693
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9310826091427906,
      "learning_rate": 1.9639847215811608e-05,
      "loss": 0.5401,
      "step": 3694
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3733452201182855,
      "learning_rate": 1.9639583369411652e-05,
      "loss": 0.2149,
      "step": 3695
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4094411804995846,
      "learning_rate": 1.9639319428174303e-05,
      "loss": 0.1777,
      "step": 3696
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5083797372143694,
      "learning_rate": 1.9639055392102156e-05,
      "loss": 0.2847,
      "step": 3697
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5587682673025572,
      "learning_rate": 1.9638791261197813e-05,
      "loss": 0.3766,
      "step": 3698
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.37947744061727956,
      "learning_rate": 1.963852703546387e-05,
      "loss": 0.1298,
      "step": 3699
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49718828929519676,
      "learning_rate": 1.9638262714902923e-05,
      "loss": 0.3366,
      "step": 3700
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2252652513027913,
      "learning_rate": 1.9637998299517578e-05,
      "loss": 0.3134,
      "step": 3701
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9962332759340391,
      "learning_rate": 1.9637733789310434e-05,
      "loss": 0.5505,
      "step": 3702
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4192145179675772,
      "learning_rate": 1.9637469184284094e-05,
      "loss": 0.3006,
      "step": 3703
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4860636533169467,
      "learning_rate": 1.963720448444116e-05,
      "loss": 0.3037,
      "step": 3704
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5397502444775804,
      "learning_rate": 1.963693968978424e-05,
      "loss": 0.2335,
      "step": 3705
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.2729699434511242,
      "learning_rate": 1.9636674800315933e-05,
      "loss": 0.1109,
      "step": 3706
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.6713530463030946,
      "learning_rate": 1.963640981603885e-05,
      "loss": 0.8683,
      "step": 3707
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.44922126668424817,
      "learning_rate": 1.96361447369556e-05,
      "loss": 0.2231,
      "step": 3708
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6419609706057593,
      "learning_rate": 1.9635879563068782e-05,
      "loss": 0.3865,
      "step": 3709
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.40610525295195865,
      "learning_rate": 1.9635614294381012e-05,
      "loss": 0.3082,
      "step": 3710
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.9118597101854219,
      "learning_rate": 1.96353489308949e-05,
      "loss": 0.5135,
      "step": 3711
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.653619221405037,
      "learning_rate": 1.9635083472613052e-05,
      "loss": 0.2664,
      "step": 3712
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1775822177062079,
      "learning_rate": 1.9634817919538084e-05,
      "loss": 0.4379,
      "step": 3713
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.33518460455457116,
      "learning_rate": 1.9634552271672606e-05,
      "loss": 0.1949,
      "step": 3714
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.448300293352258,
      "learning_rate": 1.9634286529019233e-05,
      "loss": 0.343,
      "step": 3715
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.1909942761330368,
      "learning_rate": 1.963402069158058e-05,
      "loss": 0.6282,
      "step": 3716
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.62978660892272,
      "learning_rate": 1.963375475935926e-05,
      "loss": 0.3047,
      "step": 3717
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.33991705575458203,
      "learning_rate": 1.963348873235789e-05,
      "loss": 0.2071,
      "step": 3718
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8142499148829295,
      "learning_rate": 1.963322261057909e-05,
      "loss": 0.4592,
      "step": 3719
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.2386596868031,
      "learning_rate": 1.9632956394025476e-05,
      "loss": 0.6707,
      "step": 3720
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.41383407115992416,
      "learning_rate": 1.9632690082699667e-05,
      "loss": 0.2709,
      "step": 3721
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7144685227821954,
      "learning_rate": 1.9632423676604285e-05,
      "loss": 0.2585,
      "step": 3722
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3071645838655207,
      "learning_rate": 1.963215717574195e-05,
      "loss": 0.1904,
      "step": 3723
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.007498861409469,
      "learning_rate": 1.9631890580115278e-05,
      "loss": 0.9568,
      "step": 3724
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4301697003420852,
      "learning_rate": 1.9631623889726897e-05,
      "loss": 0.1513,
      "step": 3725
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8179972759919613,
      "learning_rate": 1.9631357104579437e-05,
      "loss": 0.4657,
      "step": 3726
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.30280350457027727,
      "learning_rate": 1.963109022467551e-05,
      "loss": 0.2358,
      "step": 3727
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5941051895540331,
      "learning_rate": 1.9630823250017758e-05,
      "loss": 0.3838,
      "step": 3728
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.6407522293861586,
      "learning_rate": 1.9630556180608788e-05,
      "loss": 0.4376,
      "step": 3729
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0196364637393884,
      "learning_rate": 1.9630289016451244e-05,
      "loss": 0.6295,
      "step": 3730
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3715841613271716,
      "learning_rate": 1.9630021757547742e-05,
      "loss": 0.2076,
      "step": 3731
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.49107045338194955,
      "learning_rate": 1.9629754403900922e-05,
      "loss": 0.2626,
      "step": 3732
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.43675340595749307,
      "learning_rate": 1.962948695551341e-05,
      "loss": 0.3489,
      "step": 3733
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3298688985783819,
      "learning_rate": 1.9629219412387833e-05,
      "loss": 0.1683,
      "step": 3734
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.7699526254505461,
      "learning_rate": 1.962895177452683e-05,
      "loss": 0.4426,
      "step": 3735
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4307221594390144,
      "learning_rate": 1.962868404193303e-05,
      "loss": 0.2012,
      "step": 3736
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5277259715934066,
      "learning_rate": 1.962841621460907e-05,
      "loss": 0.4421,
      "step": 3737
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.317077254235611,
      "learning_rate": 1.9628148292557578e-05,
      "loss": 0.1717,
      "step": 3738
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5858571856540546,
      "learning_rate": 1.9627880275781203e-05,
      "loss": 0.4123,
      "step": 3739
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5072280167692044,
      "learning_rate": 1.9627612164282566e-05,
      "loss": 0.0766,
      "step": 3740
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.45577471018553795,
      "learning_rate": 1.9627343958064315e-05,
      "loss": 0.3491,
      "step": 3741
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8858859568787107,
      "learning_rate": 1.9627075657129088e-05,
      "loss": 0.5391,
      "step": 3742
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8292531788231897,
      "learning_rate": 1.9626807261479524e-05,
      "loss": 0.5423,
      "step": 3743
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.447491379453378,
      "learning_rate": 1.962653877111826e-05,
      "loss": 0.3121,
      "step": 3744
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.35300372183154616,
      "learning_rate": 1.962627018604794e-05,
      "loss": 0.2073,
      "step": 3745
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.3786776948470594,
      "learning_rate": 1.9626001506271207e-05,
      "loss": 0.3002,
      "step": 3746
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.36415380282861326,
      "learning_rate": 1.9625732731790705e-05,
      "loss": 0.1559,
      "step": 3747
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.1373086317193217,
      "learning_rate": 1.9625463862609076e-05,
      "loss": 0.7854,
      "step": 3748
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4609083898319341,
      "learning_rate": 1.9625194898728964e-05,
      "loss": 0.0718,
      "step": 3749
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.4883162620123983,
      "learning_rate": 1.962492584015302e-05,
      "loss": 0.3599,
      "step": 3750
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.42668710569360074,
      "learning_rate": 1.9624656686883887e-05,
      "loss": 0.2867,
      "step": 3751
    },
    {
      "epoch": 0.11,
      "grad_norm": 2.0340544238818477,
      "learning_rate": 1.962438743892422e-05,
      "loss": 0.9307,
      "step": 3752
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.8131109500942434,
      "learning_rate": 1.9624118096276655e-05,
      "loss": 0.4372,
      "step": 3753
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.483697619098292,
      "learning_rate": 1.9623848658943855e-05,
      "loss": 0.2614,
      "step": 3754
    },
    {
      "epoch": 0.11,
      "grad_norm": 0.5052969933829391,
      "learning_rate": 1.9623579126928462e-05,
      "loss": 0.2157,
      "step": 3755
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40633999299697454,
      "learning_rate": 1.9623309500233132e-05,
      "loss": 0.1519,
      "step": 3756
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3803048581270422,
      "learning_rate": 1.962303977886052e-05,
      "loss": 0.3,
      "step": 3757
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7933431291471925,
      "learning_rate": 1.962276996281327e-05,
      "loss": 0.3287,
      "step": 3758
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.41919571159518354,
      "learning_rate": 1.9622500052094048e-05,
      "loss": 0.2563,
      "step": 3759
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0369364622110657,
      "learning_rate": 1.96222300467055e-05,
      "loss": 0.5444,
      "step": 3760
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8323062396889348,
      "learning_rate": 1.9621959946650287e-05,
      "loss": 0.523,
      "step": 3761
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.32217495911837024,
      "learning_rate": 1.962168975193107e-05,
      "loss": 0.2744,
      "step": 3762
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.212105181158275,
      "learning_rate": 1.96214194625505e-05,
      "loss": 0.4615,
      "step": 3763
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42470465677728964,
      "learning_rate": 1.9621149078511238e-05,
      "loss": 0.2553,
      "step": 3764
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.35813804493121215,
      "learning_rate": 1.962087859981595e-05,
      "loss": 0.1743,
      "step": 3765
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0746033499338485,
      "learning_rate": 1.9620608026467288e-05,
      "loss": 0.4264,
      "step": 3766
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2683530330586854,
      "learning_rate": 1.9620337358467925e-05,
      "loss": 0.8461,
      "step": 3767
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3618777274821402,
      "learning_rate": 1.962006659582051e-05,
      "loss": 0.2105,
      "step": 3768
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39874699128703467,
      "learning_rate": 1.961979573852772e-05,
      "loss": 0.3211,
      "step": 3769
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9115863635342798,
      "learning_rate": 1.9619524786592213e-05,
      "loss": 0.4584,
      "step": 3770
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3652753091349243,
      "learning_rate": 1.961925374001666e-05,
      "loss": 0.9106,
      "step": 3771
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42358306728365597,
      "learning_rate": 1.9618982598803715e-05,
      "loss": 0.207,
      "step": 3772
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3214437121125839,
      "learning_rate": 1.9618711362956056e-05,
      "loss": 0.187,
      "step": 3773
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.32882945162996813,
      "learning_rate": 1.9618440032476352e-05,
      "loss": 0.2709,
      "step": 3774
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6566382342272915,
      "learning_rate": 1.9618168607367272e-05,
      "loss": 0.0336,
      "step": 3775
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8340274743679861,
      "learning_rate": 1.9617897087631483e-05,
      "loss": 0.4302,
      "step": 3776
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.35663361904827884,
      "learning_rate": 1.9617625473271655e-05,
      "loss": 0.2419,
      "step": 3777
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2354467893207448,
      "learning_rate": 1.9617353764290467e-05,
      "loss": 0.6904,
      "step": 3778
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.052929659258199,
      "learning_rate": 1.9617081960690586e-05,
      "loss": 0.4525,
      "step": 3779
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40498378852438977,
      "learning_rate": 1.9616810062474686e-05,
      "loss": 0.3328,
      "step": 3780
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.37780122309352837,
      "learning_rate": 1.961653806964545e-05,
      "loss": 0.2037,
      "step": 3781
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6693745265244472,
      "learning_rate": 1.9616265982205543e-05,
      "loss": 0.3978,
      "step": 3782
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3063959165727487,
      "learning_rate": 1.961599380015765e-05,
      "loss": 0.126,
      "step": 3783
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6471155239412162,
      "learning_rate": 1.9615721523504444e-05,
      "loss": 0.3049,
      "step": 3784
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9835634354330318,
      "learning_rate": 1.9615449152248605e-05,
      "loss": 0.3638,
      "step": 3785
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4948358342251519,
      "learning_rate": 1.9615176686392816e-05,
      "loss": 0.2166,
      "step": 3786
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.37737523487700053,
      "learning_rate": 1.9614904125939754e-05,
      "loss": 0.3189,
      "step": 3787
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0019759363371135,
      "learning_rate": 1.96146314708921e-05,
      "loss": 0.4692,
      "step": 3788
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7758988042411192,
      "learning_rate": 1.961435872125254e-05,
      "loss": 0.8778,
      "step": 3789
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5854309778661086,
      "learning_rate": 1.9614085877023753e-05,
      "loss": 0.0778,
      "step": 3790
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42388098547228326,
      "learning_rate": 1.961381293820843e-05,
      "loss": 0.3364,
      "step": 3791
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2707978609106062,
      "learning_rate": 1.961353990480925e-05,
      "loss": 0.15,
      "step": 3792
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5448439434624848,
      "learning_rate": 1.9613266776828896e-05,
      "loss": 0.4361,
      "step": 3793
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0725196957243124,
      "learning_rate": 1.9612993554270064e-05,
      "loss": 0.4292,
      "step": 3794
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5573429073967979,
      "learning_rate": 1.961272023713544e-05,
      "loss": 0.322,
      "step": 3795
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.618360862462875,
      "learning_rate": 1.961244682542771e-05,
      "loss": 0.3609,
      "step": 3796
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7609244210804984,
      "learning_rate": 1.9612173319149564e-05,
      "loss": 0.9485,
      "step": 3797
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3874663785609166,
      "learning_rate": 1.9611899718303697e-05,
      "loss": 0.2592,
      "step": 3798
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5819638606481046,
      "learning_rate": 1.9611626022892798e-05,
      "loss": 0.1268,
      "step": 3799
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4620612067278116,
      "learning_rate": 1.9611352232919558e-05,
      "loss": 0.3301,
      "step": 3800
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25951332215750555,
      "learning_rate": 1.9611078348386673e-05,
      "loss": 0.0783,
      "step": 3801
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2173368733781795,
      "learning_rate": 1.9610804369296834e-05,
      "loss": 0.8646,
      "step": 3802
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8100664598891696,
      "learning_rate": 1.961053029565274e-05,
      "loss": 0.366,
      "step": 3803
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4450556007101596,
      "learning_rate": 1.9610256127457087e-05,
      "loss": 0.3219,
      "step": 3804
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4246963769751536,
      "learning_rate": 1.9609981864712576e-05,
      "loss": 0.297,
      "step": 3805
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9878675596846538,
      "learning_rate": 1.9609707507421896e-05,
      "loss": 0.6245,
      "step": 3806
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0006554835363193,
      "learning_rate": 1.9609433055587753e-05,
      "loss": 0.5467,
      "step": 3807
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8333993605643042,
      "learning_rate": 1.9609158509212847e-05,
      "loss": 0.4503,
      "step": 3808
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39421355284529197,
      "learning_rate": 1.960888386829988e-05,
      "loss": 0.2512,
      "step": 3809
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4697874144730292,
      "learning_rate": 1.9608609132851546e-05,
      "loss": 0.7313,
      "step": 3810
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.26483448375069224,
      "learning_rate": 1.960833430287056e-05,
      "loss": 0.1864,
      "step": 3811
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0541377368314293,
      "learning_rate": 1.9608059378359613e-05,
      "loss": 0.4,
      "step": 3812
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6065137002616295,
      "learning_rate": 1.960778435932142e-05,
      "loss": 0.3707,
      "step": 3813
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.506799639401769,
      "learning_rate": 1.960750924575868e-05,
      "loss": 0.3278,
      "step": 3814
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.8463781019071375,
      "learning_rate": 1.960723403767411e-05,
      "loss": 0.8909,
      "step": 3815
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3568580951126002,
      "learning_rate": 1.9606958735070403e-05,
      "loss": 0.2716,
      "step": 3816
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0427641757089532,
      "learning_rate": 1.960668333795028e-05,
      "loss": 0.4152,
      "step": 3817
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3456946280323653,
      "learning_rate": 1.960640784631644e-05,
      "loss": 0.2351,
      "step": 3818
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.134957028475314,
      "learning_rate": 1.96061322601716e-05,
      "loss": 0.8286,
      "step": 3819
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6324576358199199,
      "learning_rate": 1.9605856579518473e-05,
      "loss": 0.3198,
      "step": 3820
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1976740558236814,
      "learning_rate": 1.960558080435977e-05,
      "loss": 0.6025,
      "step": 3821
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.36751232963798225,
      "learning_rate": 1.96053049346982e-05,
      "loss": 0.2233,
      "step": 3822
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3072411113001017,
      "learning_rate": 1.9605028970536476e-05,
      "loss": 0.247,
      "step": 3823
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0310797265546452,
      "learning_rate": 1.9604752911877315e-05,
      "loss": 0.5876,
      "step": 3824
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2566474239631884,
      "learning_rate": 1.960447675872344e-05,
      "loss": 0.3798,
      "step": 3825
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.779665667323783,
      "learning_rate": 1.960420051107756e-05,
      "loss": 0.4,
      "step": 3826
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.36663153617572547,
      "learning_rate": 1.9603924168942398e-05,
      "loss": 0.2338,
      "step": 3827
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4225546962263599,
      "learning_rate": 1.9603647732320666e-05,
      "loss": 0.3534,
      "step": 3828
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1180754819204015,
      "learning_rate": 1.9603371201215092e-05,
      "loss": 0.427,
      "step": 3829
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7740697633390523,
      "learning_rate": 1.960309457562839e-05,
      "loss": 0.5066,
      "step": 3830
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25712572384129134,
      "learning_rate": 1.9602817855563282e-05,
      "loss": 0.1387,
      "step": 3831
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40812315680337424,
      "learning_rate": 1.9602541041022496e-05,
      "loss": 0.2692,
      "step": 3832
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2138564866991475,
      "learning_rate": 1.960226413200875e-05,
      "loss": 0.5346,
      "step": 3833
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39289560556677033,
      "learning_rate": 1.9601987128524765e-05,
      "loss": 0.3302,
      "step": 3834
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.361166390444441,
      "learning_rate": 1.9601710030573278e-05,
      "loss": 0.1306,
      "step": 3835
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5152219567391908,
      "learning_rate": 1.9601432838157003e-05,
      "loss": 0.3591,
      "step": 3836
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.783219803072587,
      "learning_rate": 1.9601155551278676e-05,
      "loss": 0.1312,
      "step": 3837
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8544617470288641,
      "learning_rate": 1.960087816994102e-05,
      "loss": 0.4925,
      "step": 3838
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4047433513406247,
      "learning_rate": 1.9600600694146766e-05,
      "loss": 0.3427,
      "step": 3839
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4710710551768312,
      "learning_rate": 1.9600323123898642e-05,
      "loss": 0.1849,
      "step": 3840
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4307815635299165,
      "learning_rate": 1.9600045459199385e-05,
      "loss": 0.281,
      "step": 3841
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4276242565594175,
      "learning_rate": 1.959976770005172e-05,
      "loss": 0.1898,
      "step": 3842
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.302581849210801,
      "learning_rate": 1.959948984645838e-05,
      "loss": 0.792,
      "step": 3843
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1588322517867442,
      "learning_rate": 1.9599211898422098e-05,
      "loss": 0.3217,
      "step": 3844
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49970895033692353,
      "learning_rate": 1.9598933855945614e-05,
      "loss": 0.2729,
      "step": 3845
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.32042656773753825,
      "learning_rate": 1.9598655719031662e-05,
      "loss": 0.2925,
      "step": 3846
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7185332956025553,
      "learning_rate": 1.9598377487682972e-05,
      "loss": 0.5299,
      "step": 3847
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1876094375730422,
      "learning_rate": 1.9598099161902286e-05,
      "loss": 0.3293,
      "step": 3848
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7700865513039526,
      "learning_rate": 1.9597820741692347e-05,
      "loss": 0.2893,
      "step": 3849
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2902112057531311,
      "learning_rate": 1.9597542227055887e-05,
      "loss": 0.1595,
      "step": 3850
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4750352362865415,
      "learning_rate": 1.9597263617995647e-05,
      "loss": 0.3079,
      "step": 3851
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5748401821650149,
      "learning_rate": 1.959698491451437e-05,
      "loss": 0.3834,
      "step": 3852
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.866268367585774,
      "learning_rate": 1.9596706116614798e-05,
      "loss": 0.1993,
      "step": 3853
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40978684434829193,
      "learning_rate": 1.9596427224299676e-05,
      "loss": 0.2999,
      "step": 3854
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.590309882007751,
      "learning_rate": 1.9596148237571744e-05,
      "loss": 0.4356,
      "step": 3855
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0919162200712287,
      "learning_rate": 1.9595869156433746e-05,
      "loss": 0.6787,
      "step": 3856
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38714113061976707,
      "learning_rate": 1.959558998088843e-05,
      "loss": 0.2689,
      "step": 3857
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8757999956118468,
      "learning_rate": 1.9595310710938547e-05,
      "loss": 0.3799,
      "step": 3858
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.29623960454666154,
      "learning_rate": 1.9595031346586837e-05,
      "loss": 0.159,
      "step": 3859
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.734835541921478,
      "learning_rate": 1.959475188783605e-05,
      "loss": 0.8747,
      "step": 3860
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.088539832331522,
      "learning_rate": 1.9594472334688935e-05,
      "loss": 0.5651,
      "step": 3861
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7594888368342986,
      "learning_rate": 1.959419268714825e-05,
      "loss": 0.5577,
      "step": 3862
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3683489369716779,
      "learning_rate": 1.9593912945216736e-05,
      "loss": 0.2111,
      "step": 3863
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3872952949449496,
      "learning_rate": 1.9593633108897153e-05,
      "loss": 0.2379,
      "step": 3864
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9404293725633701,
      "learning_rate": 1.9593353178192248e-05,
      "loss": 0.573,
      "step": 3865
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.2045519015065873,
      "learning_rate": 1.959307315310478e-05,
      "loss": 0.3377,
      "step": 3866
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.164708439492218,
      "learning_rate": 1.95927930336375e-05,
      "loss": 0.4193,
      "step": 3867
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.29085881745656567,
      "learning_rate": 1.9592512819793166e-05,
      "loss": 0.1851,
      "step": 3868
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5609071514225863,
      "learning_rate": 1.9592232511574533e-05,
      "loss": 0.9336,
      "step": 3869
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.295093851619692,
      "learning_rate": 1.959195210898436e-05,
      "loss": 0.229,
      "step": 3870
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8333487773133379,
      "learning_rate": 1.9591671612025413e-05,
      "loss": 0.5159,
      "step": 3871
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38994058587097713,
      "learning_rate": 1.9591391020700437e-05,
      "loss": 0.1737,
      "step": 3872
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.48351904457066297,
      "learning_rate": 1.9591110335012203e-05,
      "loss": 0.4118,
      "step": 3873
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9531863166223195,
      "learning_rate": 1.959082955496347e-05,
      "loss": 0.3974,
      "step": 3874
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5541302495483051,
      "learning_rate": 1.9590548680556996e-05,
      "loss": 0.3919,
      "step": 3875
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4654724936862979,
      "learning_rate": 1.9590267711795553e-05,
      "loss": 0.266,
      "step": 3876
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.330936352944438,
      "learning_rate": 1.95899866486819e-05,
      "loss": 0.2373,
      "step": 3877
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.592603466725591,
      "learning_rate": 1.95897054912188e-05,
      "loss": 0.933,
      "step": 3878
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3391112801561564,
      "learning_rate": 1.9589424239409026e-05,
      "loss": 0.1082,
      "step": 3879
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7539757405016713,
      "learning_rate": 1.958914289325534e-05,
      "loss": 0.5084,
      "step": 3880
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3837308757382986,
      "learning_rate": 1.958886145276051e-05,
      "loss": 0.2084,
      "step": 3881
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4906057875387233,
      "learning_rate": 1.9588579917927308e-05,
      "loss": 0.3553,
      "step": 3882
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2639265454586145,
      "learning_rate": 1.9588298288758497e-05,
      "loss": 0.0744,
      "step": 3883
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.140393527122893,
      "learning_rate": 1.9588016565256856e-05,
      "loss": 0.922,
      "step": 3884
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.31403677984476763,
      "learning_rate": 1.958773474742515e-05,
      "loss": 0.0819,
      "step": 3885
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47085867961410377,
      "learning_rate": 1.958745283526616e-05,
      "loss": 0.3479,
      "step": 3886
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4418483124142656,
      "learning_rate": 1.9587170828782652e-05,
      "loss": 0.5394,
      "step": 3887
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.41575595847452684,
      "learning_rate": 1.95868887279774e-05,
      "loss": 0.3329,
      "step": 3888
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8185654631885318,
      "learning_rate": 1.9586606532853183e-05,
      "loss": 0.4434,
      "step": 3889
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3889227022608875,
      "learning_rate": 1.958632424341278e-05,
      "loss": 0.212,
      "step": 3890
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6680958987354916,
      "learning_rate": 1.9586041859658963e-05,
      "loss": 0.318,
      "step": 3891
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3957895908633809,
      "learning_rate": 1.9585759381594512e-05,
      "loss": 0.1742,
      "step": 3892
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4456727048282335,
      "learning_rate": 1.9585476809222205e-05,
      "loss": 0.3311,
      "step": 3893
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6157149129636978,
      "learning_rate": 1.9585194142544824e-05,
      "loss": 0.2502,
      "step": 3894
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5338723710436465,
      "learning_rate": 1.9584911381565147e-05,
      "loss": 0.3277,
      "step": 3895
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.57884721940116,
      "learning_rate": 1.9584628526285963e-05,
      "loss": 0.3116,
      "step": 3896
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1800340407484917,
      "learning_rate": 1.9584345576710047e-05,
      "loss": 0.5178,
      "step": 3897
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4949563079639354,
      "learning_rate": 1.9584062532840184e-05,
      "loss": 0.343,
      "step": 3898
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8642625195293073,
      "learning_rate": 1.9583779394679162e-05,
      "loss": 0.2696,
      "step": 3899
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39333894476036013,
      "learning_rate": 1.9583496162229766e-05,
      "loss": 0.2739,
      "step": 3900
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3745246304709466,
      "learning_rate": 1.9583212835494775e-05,
      "loss": 0.1903,
      "step": 3901
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.444224615938925,
      "learning_rate": 1.958292941447699e-05,
      "loss": 0.3537,
      "step": 3902
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5605214781728844,
      "learning_rate": 1.958264589917919e-05,
      "loss": 0.1679,
      "step": 3903
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5130911283464499,
      "learning_rate": 1.9582362289604163e-05,
      "loss": 0.303,
      "step": 3904
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4353948934090876,
      "learning_rate": 1.9582078585754708e-05,
      "loss": 0.286,
      "step": 3905
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1218528912513932,
      "learning_rate": 1.958179478763361e-05,
      "loss": 0.6482,
      "step": 3906
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8853951249370672,
      "learning_rate": 1.958151089524366e-05,
      "loss": 0.4688,
      "step": 3907
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8130329239787574,
      "learning_rate": 1.9581226908587654e-05,
      "loss": 0.3686,
      "step": 3908
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2502247148467696,
      "learning_rate": 1.9580942827668384e-05,
      "loss": 0.1616,
      "step": 3909
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.259186886376815,
      "learning_rate": 1.9580658652488648e-05,
      "loss": 0.9588,
      "step": 3910
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49768098217283513,
      "learning_rate": 1.958037438305124e-05,
      "loss": 0.286,
      "step": 3911
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.02005075290249,
      "learning_rate": 1.9580090019358953e-05,
      "loss": 0.5537,
      "step": 3912
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39609911913927065,
      "learning_rate": 1.957980556141459e-05,
      "loss": 0.2257,
      "step": 3913
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.081655411003621,
      "learning_rate": 1.957952100922095e-05,
      "loss": 0.9125,
      "step": 3914
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0907470532127856,
      "learning_rate": 1.957923636278083e-05,
      "loss": 0.4608,
      "step": 3915
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46942881877740966,
      "learning_rate": 1.957895162209703e-05,
      "loss": 0.2979,
      "step": 3916
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45707043893055793,
      "learning_rate": 1.957866678717235e-05,
      "loss": 0.3009,
      "step": 3917
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3650670127352482,
      "learning_rate": 1.9578381858009593e-05,
      "loss": 0.2565,
      "step": 3918
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4334218039565727,
      "learning_rate": 1.957809683461157e-05,
      "loss": 0.1914,
      "step": 3919
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1076159138550463,
      "learning_rate": 1.957781171698107e-05,
      "loss": 0.43,
      "step": 3920
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7393072739000355,
      "learning_rate": 1.9577526505120913e-05,
      "loss": 0.4701,
      "step": 3921
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46010420743132446,
      "learning_rate": 1.95772411990339e-05,
      "loss": 0.1991,
      "step": 3922
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43600921181014946,
      "learning_rate": 1.9576955798722832e-05,
      "loss": 0.3286,
      "step": 3923
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5382938962017335,
      "learning_rate": 1.9576670304190524e-05,
      "loss": 0.3624,
      "step": 3924
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5543641141198883,
      "learning_rate": 1.957638471543978e-05,
      "loss": 0.8697,
      "step": 3925
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5468196160696456,
      "learning_rate": 1.9576099032473418e-05,
      "loss": 0.1879,
      "step": 3926
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45373622584611567,
      "learning_rate": 1.9575813255294238e-05,
      "loss": 0.2717,
      "step": 3927
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.1801792939515128,
      "learning_rate": 1.9575527383905055e-05,
      "loss": 0.0727,
      "step": 3928
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40723879626290077,
      "learning_rate": 1.9575241418308685e-05,
      "loss": 0.282,
      "step": 3929
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.412577309741295,
      "learning_rate": 1.957495535850794e-05,
      "loss": 0.7245,
      "step": 3930
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3372972043614093,
      "learning_rate": 1.9574669204505635e-05,
      "loss": 0.2041,
      "step": 3931
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.746761698570121,
      "learning_rate": 1.957438295630458e-05,
      "loss": 0.5642,
      "step": 3932
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0348406416041256,
      "learning_rate": 1.9574096613907598e-05,
      "loss": 0.4332,
      "step": 3933
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.9340278869338285,
      "learning_rate": 1.9573810177317504e-05,
      "loss": 0.9305,
      "step": 3934
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.29457637362117495,
      "learning_rate": 1.9573523646537112e-05,
      "loss": 0.2284,
      "step": 3935
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.47680305214338264,
      "learning_rate": 1.9573237021569246e-05,
      "loss": 0.3563,
      "step": 3936
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.27150545520035724,
      "learning_rate": 1.9572950302416722e-05,
      "loss": 0.1171,
      "step": 3937
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.511291299450359,
      "learning_rate": 1.9572663489082365e-05,
      "loss": 0.2596,
      "step": 3938
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8515478174369918,
      "learning_rate": 1.9572376581568993e-05,
      "loss": 0.4498,
      "step": 3939
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.464251002150469,
      "learning_rate": 1.9572089579879435e-05,
      "loss": 0.3007,
      "step": 3940
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.35722109221115794,
      "learning_rate": 1.9571802484016507e-05,
      "loss": 0.2503,
      "step": 3941
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4800401377257534,
      "learning_rate": 1.9571515293983037e-05,
      "loss": 0.4005,
      "step": 3942
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8185578706608446,
      "learning_rate": 1.9571228009781848e-05,
      "loss": 0.5463,
      "step": 3943
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38271530455380465,
      "learning_rate": 1.9570940631415773e-05,
      "loss": 0.2169,
      "step": 3944
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6124421029758744,
      "learning_rate": 1.957065315888763e-05,
      "loss": 0.3282,
      "step": 3945
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.28010891895313766,
      "learning_rate": 1.9570365592200255e-05,
      "loss": 0.1235,
      "step": 3946
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3913999978770391,
      "learning_rate": 1.9570077931356474e-05,
      "loss": 0.3399,
      "step": 3947
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7374668053300562,
      "learning_rate": 1.9569790176359117e-05,
      "loss": 0.4034,
      "step": 3948
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0313119601532337,
      "learning_rate": 1.9569502327211015e-05,
      "loss": 0.4512,
      "step": 3949
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3871372100507086,
      "learning_rate": 1.9569214383915e-05,
      "loss": 0.2797,
      "step": 3950
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.671523886988457,
      "learning_rate": 1.9568926346473907e-05,
      "loss": 0.8467,
      "step": 3951
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4044658866429065,
      "learning_rate": 1.9568638214890566e-05,
      "loss": 0.2691,
      "step": 3952
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.0027992821780294,
      "learning_rate": 1.9568349989167818e-05,
      "loss": 0.9052,
      "step": 3953
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3824514097511866,
      "learning_rate": 1.9568061669308487e-05,
      "loss": 0.202,
      "step": 3954
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.507848269383036,
      "learning_rate": 1.956777325531542e-05,
      "loss": 0.2957,
      "step": 3955
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4148313932849212,
      "learning_rate": 1.9567484747191458e-05,
      "loss": 0.2015,
      "step": 3956
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7878697487408418,
      "learning_rate": 1.9567196144939423e-05,
      "loss": 0.4316,
      "step": 3957
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.48127638749770424,
      "learning_rate": 1.956690744856217e-05,
      "loss": 0.2893,
      "step": 3958
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3794460721263731,
      "learning_rate": 1.9566618658062532e-05,
      "loss": 0.2592,
      "step": 3959
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.6143391677458367,
      "learning_rate": 1.9566329773443354e-05,
      "loss": 0.844,
      "step": 3960
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9884231937148853,
      "learning_rate": 1.956604079470747e-05,
      "loss": 0.4977,
      "step": 3961
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9344527035839739,
      "learning_rate": 1.9565751721857735e-05,
      "loss": 0.5333,
      "step": 3962
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.37119835730342754,
      "learning_rate": 1.9565462554896986e-05,
      "loss": 0.2189,
      "step": 3963
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5753255817893125,
      "learning_rate": 1.956517329382807e-05,
      "loss": 0.2397,
      "step": 3964
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3628865136371346,
      "learning_rate": 1.9564883938653828e-05,
      "loss": 0.287,
      "step": 3965
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8873039289073857,
      "learning_rate": 1.956459448937711e-05,
      "loss": 0.5593,
      "step": 3966
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.25307819901542583,
      "learning_rate": 1.9564304946000767e-05,
      "loss": 0.1396,
      "step": 3967
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6117470402185189,
      "learning_rate": 1.9564015308527647e-05,
      "loss": 0.4165,
      "step": 3968
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.351130543952461,
      "learning_rate": 1.9563725576960594e-05,
      "loss": 0.636,
      "step": 3969
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3580938487932815,
      "learning_rate": 1.956343575130246e-05,
      "loss": 0.2728,
      "step": 3970
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8501215608308406,
      "learning_rate": 1.9563145831556102e-05,
      "loss": 0.4792,
      "step": 3971
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43968886027674836,
      "learning_rate": 1.956285581772437e-05,
      "loss": 0.2091,
      "step": 3972
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5897211010750988,
      "learning_rate": 1.9562565709810112e-05,
      "loss": 0.4042,
      "step": 3973
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6735881665750015,
      "learning_rate": 1.956227550781619e-05,
      "loss": 0.4521,
      "step": 3974
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38082490406519204,
      "learning_rate": 1.9561985211745452e-05,
      "loss": 0.2021,
      "step": 3975
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5588557605562879,
      "learning_rate": 1.9561694821600756e-05,
      "loss": 0.0802,
      "step": 3976
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3942705909427179,
      "learning_rate": 1.9561404337384962e-05,
      "loss": 0.2848,
      "step": 3977
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5799671716268796,
      "learning_rate": 1.9561113759100926e-05,
      "loss": 0.308,
      "step": 3978
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3610493982667302,
      "learning_rate": 1.956082308675151e-05,
      "loss": 0.7631,
      "step": 3979
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8495573023656036,
      "learning_rate": 1.9560532320339564e-05,
      "loss": 0.3816,
      "step": 3980
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5432678951275043,
      "learning_rate": 1.9560241459867958e-05,
      "loss": 0.2684,
      "step": 3981
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.45344308429559765,
      "learning_rate": 1.9559950505339554e-05,
      "loss": 0.2934,
      "step": 3982
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3847792087279534,
      "learning_rate": 1.9559659456757207e-05,
      "loss": 0.3328,
      "step": 3983
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.257515155797347,
      "learning_rate": 1.9559368314123785e-05,
      "loss": 0.6458,
      "step": 3984
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43142408850473696,
      "learning_rate": 1.9559077077442157e-05,
      "loss": 0.076,
      "step": 3985
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46818931514236967,
      "learning_rate": 1.9558785746715178e-05,
      "loss": 0.3635,
      "step": 3986
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.4088528273469576,
      "learning_rate": 1.955849432194572e-05,
      "loss": 0.6238,
      "step": 3987
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5333212018208007,
      "learning_rate": 1.9558202803136655e-05,
      "loss": 0.3812,
      "step": 3988
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.37555185666574775,
      "learning_rate": 1.955791119029084e-05,
      "loss": 0.2091,
      "step": 3989
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43432362038299094,
      "learning_rate": 1.955761948341115e-05,
      "loss": 0.3215,
      "step": 3990
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0671495322977969,
      "learning_rate": 1.9557327682500457e-05,
      "loss": 0.4379,
      "step": 3991
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9371169581583392,
      "learning_rate": 1.9557035787561627e-05,
      "loss": 0.5474,
      "step": 3992
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39758584401943525,
      "learning_rate": 1.9556743798597536e-05,
      "loss": 0.1609,
      "step": 3993
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4588890180975591,
      "learning_rate": 1.9556451715611057e-05,
      "loss": 0.2957,
      "step": 3994
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40955729887547326,
      "learning_rate": 1.9556159538605058e-05,
      "loss": 0.2515,
      "step": 3995
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.49689501415781817,
      "learning_rate": 1.955586726758242e-05,
      "loss": 0.332,
      "step": 3996
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5279007884448303,
      "learning_rate": 1.955557490254601e-05,
      "loss": 0.3137,
      "step": 3997
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9917873595882684,
      "learning_rate": 1.955528244349872e-05,
      "loss": 0.4013,
      "step": 3998
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.069430477506503,
      "learning_rate": 1.955498989044341e-05,
      "loss": 0.4784,
      "step": 3999
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4135237961813551,
      "learning_rate": 1.9554697243382964e-05,
      "loss": 0.3159,
      "step": 4000
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40172714353590583,
      "learning_rate": 1.9554404502320266e-05,
      "loss": 0.3275,
      "step": 4001
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3700832292059282,
      "learning_rate": 1.9554111667258193e-05,
      "loss": 0.13,
      "step": 4002
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7632863339837543,
      "learning_rate": 1.9553818738199624e-05,
      "loss": 0.9003,
      "step": 4003
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38943805176457413,
      "learning_rate": 1.9553525715147444e-05,
      "loss": 0.2077,
      "step": 4004
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5117348744833268,
      "learning_rate": 1.955323259810453e-05,
      "loss": 0.8916,
      "step": 4005
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2821306051406962,
      "learning_rate": 1.9552939387073776e-05,
      "loss": 0.226,
      "step": 4006
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5911970918894074,
      "learning_rate": 1.955264608205806e-05,
      "loss": 0.7349,
      "step": 4007
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38026145356752156,
      "learning_rate": 1.955235268306027e-05,
      "loss": 0.2543,
      "step": 4008
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5141038750596498,
      "learning_rate": 1.9552059190083287e-05,
      "loss": 0.3689,
      "step": 4009
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.2036544689410262,
      "learning_rate": 1.9551765603130006e-05,
      "loss": 0.8866,
      "step": 4010
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0308720276440073,
      "learning_rate": 1.9551471922203308e-05,
      "loss": 0.3099,
      "step": 4011
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39165794409373056,
      "learning_rate": 1.955117814730609e-05,
      "loss": 0.2962,
      "step": 4012
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3447561198611963,
      "learning_rate": 1.9550884278441237e-05,
      "loss": 0.2417,
      "step": 4013
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5485780320472886,
      "learning_rate": 1.9550590315611644e-05,
      "loss": 0.8892,
      "step": 4014
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2570921829943094,
      "learning_rate": 1.9550296258820198e-05,
      "loss": 0.0772,
      "step": 4015
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6877536781830955,
      "learning_rate": 1.9550002108069796e-05,
      "loss": 0.5416,
      "step": 4016
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4100583517312407,
      "learning_rate": 1.954970786336333e-05,
      "loss": 0.2504,
      "step": 4017
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.43506926363352216,
      "learning_rate": 1.95494135247037e-05,
      "loss": 0.3336,
      "step": 4018
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4900899689188894,
      "learning_rate": 1.954911909209379e-05,
      "loss": 0.2469,
      "step": 4019
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3686777057382862,
      "learning_rate": 1.954882456553651e-05,
      "loss": 0.6172,
      "step": 4020
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3629826141568152,
      "learning_rate": 1.954852994503475e-05,
      "loss": 0.1253,
      "step": 4021
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44305887895249874,
      "learning_rate": 1.9548235230591414e-05,
      "loss": 0.2585,
      "step": 4022
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.806187171421492,
      "learning_rate": 1.9547940422209396e-05,
      "loss": 0.9558,
      "step": 4023
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.36472395134653546,
      "learning_rate": 1.9547645519891593e-05,
      "loss": 0.2872,
      "step": 4024
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9775244253316107,
      "learning_rate": 1.9547350523640916e-05,
      "loss": 0.5086,
      "step": 4025
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3140168819019414,
      "learning_rate": 1.9547055433460266e-05,
      "loss": 0.0725,
      "step": 4026
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3526407697300695,
      "learning_rate": 1.9546760249352542e-05,
      "loss": 0.2598,
      "step": 4027
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.36421115479288946,
      "learning_rate": 1.9546464971320648e-05,
      "loss": 0.1576,
      "step": 4028
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5604227054088795,
      "learning_rate": 1.954616959936749e-05,
      "loss": 0.4265,
      "step": 4029
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.46138857980543774,
      "learning_rate": 1.9545874133495975e-05,
      "loss": 0.264,
      "step": 4030
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4714062676358204,
      "learning_rate": 1.9545578573709006e-05,
      "loss": 0.3142,
      "step": 4031
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4911359547360168,
      "learning_rate": 1.95452829200095e-05,
      "loss": 0.3193,
      "step": 4032
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.9960156906623122,
      "learning_rate": 1.9544987172400357e-05,
      "loss": 0.5689,
      "step": 4033
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0652322717679625,
      "learning_rate": 1.9544691330884488e-05,
      "loss": 0.4213,
      "step": 4034
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.2500688018771123,
      "learning_rate": 1.9544395395464808e-05,
      "loss": 0.077,
      "step": 4035
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4215855660335697,
      "learning_rate": 1.9544099366144224e-05,
      "loss": 0.323,
      "step": 4036
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5551896812356616,
      "learning_rate": 1.954380324292565e-05,
      "loss": 0.302,
      "step": 4037
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.7764518175017912,
      "learning_rate": 1.9543507025812e-05,
      "loss": 0.7176,
      "step": 4038
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.508249617353247,
      "learning_rate": 1.9543210714806188e-05,
      "loss": 0.2945,
      "step": 4039
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.42962718921472676,
      "learning_rate": 1.9542914309911128e-05,
      "loss": 0.3154,
      "step": 4040
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.1781784634519856,
      "learning_rate": 1.9542617811129733e-05,
      "loss": 0.5214,
      "step": 4041
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5641184591919156,
      "learning_rate": 1.954232121846493e-05,
      "loss": 0.3341,
      "step": 4042
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5044625812822632,
      "learning_rate": 1.9542024531919627e-05,
      "loss": 0.3522,
      "step": 4043
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.7344850946912492,
      "learning_rate": 1.9541727751496752e-05,
      "loss": 0.4251,
      "step": 4044
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.33146343356916086,
      "learning_rate": 1.9541430877199214e-05,
      "loss": 0.1924,
      "step": 4045
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5345502411510884,
      "learning_rate": 1.9541133909029944e-05,
      "loss": 0.3446,
      "step": 4046
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.44593882886869496,
      "learning_rate": 1.954083684699186e-05,
      "loss": 0.2526,
      "step": 4047
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.40206672953955963,
      "learning_rate": 1.954053969108788e-05,
      "loss": 0.2761,
      "step": 4048
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5394774194521859,
      "learning_rate": 1.9540242441320928e-05,
      "loss": 0.3466,
      "step": 4049
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.471991930039426,
      "learning_rate": 1.9539945097693936e-05,
      "loss": 0.3416,
      "step": 4050
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8464281303862896,
      "learning_rate": 1.953964766020983e-05,
      "loss": 0.564,
      "step": 4051
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.38165027396441525,
      "learning_rate": 1.9539350128871524e-05,
      "loss": 0.1652,
      "step": 4052
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.753755966524246,
      "learning_rate": 1.9539052503681955e-05,
      "loss": 0.3006,
      "step": 4053
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39890971504345507,
      "learning_rate": 1.953875478464405e-05,
      "loss": 0.2639,
      "step": 4054
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3374468366625659,
      "learning_rate": 1.9538456971760733e-05,
      "loss": 0.2529,
      "step": 4055
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8709443731208732,
      "learning_rate": 1.9538159065034942e-05,
      "loss": 0.4355,
      "step": 4056
    },
    {
      "epoch": 0.12,
      "grad_norm": 2.3354390219947545,
      "learning_rate": 1.9537861064469603e-05,
      "loss": 0.9848,
      "step": 4057
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.37145192867831733,
      "learning_rate": 1.9537562970067644e-05,
      "loss": 0.2053,
      "step": 4058
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5279847803683899,
      "learning_rate": 1.953726478183201e-05,
      "loss": 0.4279,
      "step": 4059
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.35970996273189093,
      "learning_rate": 1.953696649976562e-05,
      "loss": 0.3128,
      "step": 4060
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.0009673008737952,
      "learning_rate": 1.9536668123871417e-05,
      "loss": 0.3528,
      "step": 4061
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.249661900471496,
      "learning_rate": 1.9536369654152337e-05,
      "loss": 0.4969,
      "step": 4062
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.31355439528538764,
      "learning_rate": 1.9536071090611314e-05,
      "loss": 0.1806,
      "step": 4063
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6375274733304737,
      "learning_rate": 1.9535772433251285e-05,
      "loss": 0.2766,
      "step": 4064
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5167462573533956,
      "learning_rate": 1.953547368207519e-05,
      "loss": 0.2607,
      "step": 4065
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5671134943731729,
      "learning_rate": 1.9535174837085966e-05,
      "loss": 0.4017,
      "step": 4066
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4487663193714869,
      "learning_rate": 1.9534875898286555e-05,
      "loss": 0.2449,
      "step": 4067
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6523737596777692,
      "learning_rate": 1.95345768656799e-05,
      "loss": 0.421,
      "step": 4068
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.8037168991519745,
      "learning_rate": 1.9534277739268937e-05,
      "loss": 0.4328,
      "step": 4069
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.5344077954363569,
      "learning_rate": 1.9533978519056616e-05,
      "loss": 0.769,
      "step": 4070
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.39661749563715926,
      "learning_rate": 1.953367920504587e-05,
      "loss": 0.2394,
      "step": 4071
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4137772617600597,
      "learning_rate": 1.953337979723966e-05,
      "loss": 0.3106,
      "step": 4072
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.36608356114227003,
      "learning_rate": 1.9533080295640916e-05,
      "loss": 0.1287,
      "step": 4073
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.398254676802367,
      "learning_rate": 1.9532780700252596e-05,
      "loss": 0.1936,
      "step": 4074
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.74877162522224,
      "learning_rate": 1.953248101107764e-05,
      "loss": 0.5107,
      "step": 4075
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3232952574916896,
      "learning_rate": 1.9532181228119e-05,
      "loss": 0.1736,
      "step": 4076
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.6055778187767188,
      "learning_rate": 1.9531881351379624e-05,
      "loss": 0.3893,
      "step": 4077
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.3747637525058865,
      "learning_rate": 1.9531581380862462e-05,
      "loss": 0.2858,
      "step": 4078
    },
    {
      "epoch": 0.12,
      "grad_norm": 1.3693502980834567,
      "learning_rate": 1.9531281316570468e-05,
      "loss": 0.7552,
      "step": 4079
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.4823563435057522,
      "learning_rate": 1.953098115850659e-05,
      "loss": 0.0788,
      "step": 4080
    },
    {
      "epoch": 0.12,
      "grad_norm": 0.5489654716912306,
      "learning_rate": 1.953068090667379e-05,
      "loss": 0.3831,
      "step": 4081
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.31393326966457885,
      "learning_rate": 1.9530380561075008e-05,
      "loss": 0.1335,
      "step": 4082
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5477386997092898,
      "learning_rate": 1.9530080121713207e-05,
      "loss": 0.4172,
      "step": 4083
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45676271198739093,
      "learning_rate": 1.9529779588591343e-05,
      "loss": 0.3048,
      "step": 4084
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7906903672684091,
      "learning_rate": 1.9529478961712372e-05,
      "loss": 0.3665,
      "step": 4085
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.39024042020852345,
      "learning_rate": 1.952917824107925e-05,
      "loss": 0.3024,
      "step": 4086
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9857336213762168,
      "learning_rate": 1.952887742669494e-05,
      "loss": 0.657,
      "step": 4087
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.483997215386739,
      "learning_rate": 1.9528576518562395e-05,
      "loss": 0.692,
      "step": 4088
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3520849140327104,
      "learning_rate": 1.952827551668458e-05,
      "loss": 0.2589,
      "step": 4089
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41663135055044087,
      "learning_rate": 1.9527974421064458e-05,
      "loss": 0.3226,
      "step": 4090
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2703580570901048,
      "learning_rate": 1.9527673231704987e-05,
      "loss": 0.0971,
      "step": 4091
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2208178617376424,
      "learning_rate": 1.9527371948609128e-05,
      "loss": 0.7423,
      "step": 4092
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8447851472801242,
      "learning_rate": 1.9527070571779853e-05,
      "loss": 0.4132,
      "step": 4093
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7176029719573979,
      "learning_rate": 1.952676910122012e-05,
      "loss": 0.3425,
      "step": 4094
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.35051786353426384,
      "learning_rate": 1.95264675369329e-05,
      "loss": 0.2573,
      "step": 4095
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6425985571196959,
      "learning_rate": 1.952616587892116e-05,
      "loss": 0.3757,
      "step": 4096
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2920299208837338,
      "learning_rate": 1.9525864127187862e-05,
      "loss": 0.6032,
      "step": 4097
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9253095331655723,
      "learning_rate": 1.952556228173598e-05,
      "loss": 0.4369,
      "step": 4098
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3675737098187821,
      "learning_rate": 1.952526034256848e-05,
      "loss": 0.2193,
      "step": 4099
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2701167720977435,
      "learning_rate": 1.9524958309688336e-05,
      "loss": 0.0748,
      "step": 4100
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5867335198203567,
      "learning_rate": 1.9524656183098517e-05,
      "loss": 0.4194,
      "step": 4101
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4736801760241682,
      "learning_rate": 1.9524353962802e-05,
      "loss": 0.3492,
      "step": 4102
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4533374529096282,
      "learning_rate": 1.9524051648801746e-05,
      "loss": 0.2629,
      "step": 4103
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48500449567765225,
      "learning_rate": 1.9523749241100747e-05,
      "loss": 0.2547,
      "step": 4104
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6516349054879786,
      "learning_rate": 1.9523446739701966e-05,
      "loss": 0.9202,
      "step": 4105
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.46772234938918567,
      "learning_rate": 1.9523144144608382e-05,
      "loss": 0.2903,
      "step": 4106
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4663151199848451,
      "learning_rate": 1.9522841455822972e-05,
      "loss": 0.3229,
      "step": 4107
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45879231533102294,
      "learning_rate": 1.9522538673348718e-05,
      "loss": 0.2054,
      "step": 4108
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5664893986797243,
      "learning_rate": 1.9522235797188592e-05,
      "loss": 0.3998,
      "step": 4109
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5329405457703145,
      "learning_rate": 1.9521932827345575e-05,
      "loss": 0.2975,
      "step": 4110
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48812056693935274,
      "learning_rate": 1.9521629763822655e-05,
      "loss": 0.3034,
      "step": 4111
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2316328664685496,
      "learning_rate": 1.9521326606622805e-05,
      "loss": 0.1768,
      "step": 4112
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3446302564995967,
      "learning_rate": 1.9521023355749008e-05,
      "loss": 0.24,
      "step": 4113
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5843749067961278,
      "learning_rate": 1.9520720011204256e-05,
      "loss": 0.3954,
      "step": 4114
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1937560203233362,
      "learning_rate": 1.9520416572991525e-05,
      "loss": 0.4008,
      "step": 4115
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0772830810177725,
      "learning_rate": 1.9520113041113806e-05,
      "loss": 0.8599,
      "step": 4116
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3553075829383505,
      "learning_rate": 1.951980941557408e-05,
      "loss": 0.217,
      "step": 4117
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.939529127725896,
      "learning_rate": 1.9519505696375336e-05,
      "loss": 0.5729,
      "step": 4118
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45726041766532244,
      "learning_rate": 1.9519201883520562e-05,
      "loss": 0.299,
      "step": 4119
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49110635690732396,
      "learning_rate": 1.951889797701275e-05,
      "loss": 0.4112,
      "step": 4120
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3511311760526776,
      "learning_rate": 1.9518593976854886e-05,
      "loss": 0.0776,
      "step": 4121
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4494257881971608,
      "learning_rate": 1.9518289883049963e-05,
      "loss": 0.3497,
      "step": 4122
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3017174542302181,
      "learning_rate": 1.951798569560097e-05,
      "loss": 0.5619,
      "step": 4123
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1702238162293346,
      "learning_rate": 1.9517681414510907e-05,
      "loss": 0.7105,
      "step": 4124
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.35705018688421136,
      "learning_rate": 1.951737703978276e-05,
      "loss": 0.2784,
      "step": 4125
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3252116641992362,
      "learning_rate": 1.9517072571419526e-05,
      "loss": 0.2059,
      "step": 4126
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1247776204337676,
      "learning_rate": 1.95167680094242e-05,
      "loss": 0.5956,
      "step": 4127
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7767625823498513,
      "learning_rate": 1.9516463353799778e-05,
      "loss": 0.4239,
      "step": 4128
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3337435867360441,
      "learning_rate": 1.9516158604549257e-05,
      "loss": 0.1866,
      "step": 4129
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3737109222862293,
      "learning_rate": 1.951585376167564e-05,
      "loss": 0.2027,
      "step": 4130
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6693214051045081,
      "learning_rate": 1.9515548825181924e-05,
      "loss": 0.4059,
      "step": 4131
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3612809019674024,
      "learning_rate": 1.95152437950711e-05,
      "loss": 0.2928,
      "step": 4132
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.695496514684055,
      "learning_rate": 1.9514938671346184e-05,
      "loss": 0.8392,
      "step": 4133
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9711766833646915,
      "learning_rate": 1.951463345401017e-05,
      "loss": 0.4518,
      "step": 4134
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7187782837423469,
      "learning_rate": 1.9514328143066057e-05,
      "loss": 0.3586,
      "step": 4135
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.403374619367682,
      "learning_rate": 1.9514022738516853e-05,
      "loss": 0.3133,
      "step": 4136
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40144025777472014,
      "learning_rate": 1.9513717240365567e-05,
      "loss": 0.3384,
      "step": 4137
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.6011963515386416,
      "learning_rate": 1.95134116486152e-05,
      "loss": 0.3769,
      "step": 4138
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5177636509547269,
      "learning_rate": 1.9513105963268756e-05,
      "loss": 0.2495,
      "step": 4139
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2884823147516792,
      "learning_rate": 1.9512800184329247e-05,
      "loss": 0.201,
      "step": 4140
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3693066271421794,
      "learning_rate": 1.951249431179968e-05,
      "loss": 0.1794,
      "step": 4141
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.161827768700656,
      "learning_rate": 1.951218834568306e-05,
      "loss": 0.8612,
      "step": 4142
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.39203892324130135,
      "learning_rate": 1.9511882285982405e-05,
      "loss": 0.2967,
      "step": 4143
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6551390546770415,
      "learning_rate": 1.9511576132700724e-05,
      "loss": 0.4176,
      "step": 4144
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44928359510131655,
      "learning_rate": 1.9511269885841025e-05,
      "loss": 0.3114,
      "step": 4145
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.970365985488066,
      "learning_rate": 1.9510963545406323e-05,
      "loss": 0.7787,
      "step": 4146
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4708396616697003,
      "learning_rate": 1.951065711139963e-05,
      "loss": 0.0784,
      "step": 4147
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6691188884884904,
      "learning_rate": 1.9510350583823966e-05,
      "loss": 0.498,
      "step": 4148
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7538742841456422,
      "learning_rate": 1.9510043962682342e-05,
      "loss": 0.2542,
      "step": 4149
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.015786602386307,
      "learning_rate": 1.9509737247977777e-05,
      "loss": 1.0158,
      "step": 4150
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.122798444158538,
      "learning_rate": 1.950943043971329e-05,
      "loss": 0.2909,
      "step": 4151
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2413148524437654,
      "learning_rate": 1.9509123537891894e-05,
      "loss": 0.572,
      "step": 4152
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.1335940875583996,
      "learning_rate": 1.9508816542516613e-05,
      "loss": 0.271,
      "step": 4153
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.175403624991403,
      "learning_rate": 1.9508509453590466e-05,
      "loss": 0.3938,
      "step": 4154
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5749916711402273,
      "learning_rate": 1.9508202271116476e-05,
      "loss": 0.3503,
      "step": 4155
    },
    {
      "epoch": 0.13,
      "grad_norm": 35.89715573280314,
      "learning_rate": 1.950789499509766e-05,
      "loss": 1.1332,
      "step": 4156
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.711221397229549,
      "learning_rate": 1.9507587625537047e-05,
      "loss": 0.599,
      "step": 4157
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.430752863000864,
      "learning_rate": 1.9507280162437657e-05,
      "loss": 0.2455,
      "step": 4158
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8276230973838892,
      "learning_rate": 1.950697260580252e-05,
      "loss": 0.5366,
      "step": 4159
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7756706326943773,
      "learning_rate": 1.9506664955634656e-05,
      "loss": 0.2176,
      "step": 4160
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.125004535141411,
      "learning_rate": 1.9506357211937092e-05,
      "loss": 0.4737,
      "step": 4161
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.879930263358028,
      "learning_rate": 1.9506049374712858e-05,
      "loss": 0.0844,
      "step": 4162
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.820085653977227,
      "learning_rate": 1.9505741443964985e-05,
      "loss": 0.3845,
      "step": 4163
    },
    {
      "epoch": 0.13,
      "grad_norm": 11.175812791693241,
      "learning_rate": 1.95054334196965e-05,
      "loss": 0.723,
      "step": 4164
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.648739679964551,
      "learning_rate": 1.950512530191043e-05,
      "loss": 1.0643,
      "step": 4165
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.094534022693212,
      "learning_rate": 1.950481709060981e-05,
      "loss": 0.3426,
      "step": 4166
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9497571718382989,
      "learning_rate": 1.9504508785797674e-05,
      "loss": 0.2373,
      "step": 4167
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.662745797679474,
      "learning_rate": 1.9504200387477053e-05,
      "loss": 1.0364,
      "step": 4168
    },
    {
      "epoch": 0.13,
      "grad_norm": 49.37464325819019,
      "learning_rate": 1.9503891895650985e-05,
      "loss": 0.9715,
      "step": 4169
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4407771909267337,
      "learning_rate": 1.9503583310322495e-05,
      "loss": 0.6577,
      "step": 4170
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5398729698031524,
      "learning_rate": 1.9503274631494625e-05,
      "loss": 0.1182,
      "step": 4171
    },
    {
      "epoch": 0.13,
      "grad_norm": 10.77187744465479,
      "learning_rate": 1.9502965859170418e-05,
      "loss": 0.7163,
      "step": 4172
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6500706848055826,
      "learning_rate": 1.9502656993352903e-05,
      "loss": 0.3795,
      "step": 4173
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.654657140740927,
      "learning_rate": 1.950234803404512e-05,
      "loss": 1.3108,
      "step": 4174
    },
    {
      "epoch": 0.13,
      "grad_norm": 7.860499979799707,
      "learning_rate": 1.9502038981250113e-05,
      "loss": 1.1294,
      "step": 4175
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2702115866797032,
      "learning_rate": 1.9501729834970916e-05,
      "loss": 0.39,
      "step": 4176
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.219561353863381,
      "learning_rate": 1.950142059521058e-05,
      "loss": 0.3509,
      "step": 4177
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.8569259138458536,
      "learning_rate": 1.9501111261972138e-05,
      "loss": 0.7349,
      "step": 4178
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.3599185736168833,
      "learning_rate": 1.9500801835258638e-05,
      "loss": 0.415,
      "step": 4179
    },
    {
      "epoch": 0.13,
      "grad_norm": 5.617497727389722,
      "learning_rate": 1.9500492315073124e-05,
      "loss": 0.3003,
      "step": 4180
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0474199906558357,
      "learning_rate": 1.9500182701418643e-05,
      "loss": 0.4661,
      "step": 4181
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8666456831027206,
      "learning_rate": 1.9499872994298234e-05,
      "loss": 0.2097,
      "step": 4182
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2821524212561406,
      "learning_rate": 1.949956319371495e-05,
      "loss": 0.9323,
      "step": 4183
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9354869913478543,
      "learning_rate": 1.949925329967184e-05,
      "loss": 0.3155,
      "step": 4184
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.409543038054492,
      "learning_rate": 1.9498943312171944e-05,
      "loss": 0.5897,
      "step": 4185
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9251530549450017,
      "learning_rate": 1.949863323121832e-05,
      "loss": 0.3176,
      "step": 4186
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.6969740859236113,
      "learning_rate": 1.9498323056814022e-05,
      "loss": 0.72,
      "step": 4187
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.069413997795415,
      "learning_rate": 1.9498012788962094e-05,
      "loss": 0.2803,
      "step": 4188
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8161491302883568,
      "learning_rate": 1.949770242766559e-05,
      "loss": 0.3603,
      "step": 4189
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6819277811006091,
      "learning_rate": 1.9497391972927565e-05,
      "loss": 0.2447,
      "step": 4190
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0917468243096504,
      "learning_rate": 1.9497081424751073e-05,
      "loss": 0.3981,
      "step": 4191
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.620238477363167,
      "learning_rate": 1.9496770783139166e-05,
      "loss": 0.793,
      "step": 4192
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8444853656276377,
      "learning_rate": 1.9496460048094906e-05,
      "loss": 0.5732,
      "step": 4193
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5671879027858304,
      "learning_rate": 1.949614921962135e-05,
      "loss": 0.325,
      "step": 4194
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.900654214184514,
      "learning_rate": 1.949583829772155e-05,
      "loss": 0.4789,
      "step": 4195
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8022689827874081,
      "learning_rate": 1.9495527282398566e-05,
      "loss": 0.4788,
      "step": 4196
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6935048123637749,
      "learning_rate": 1.9495216173655466e-05,
      "loss": 0.3402,
      "step": 4197
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8126314207484937,
      "learning_rate": 1.94949049714953e-05,
      "loss": 0.2077,
      "step": 4198
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.38404880808779274,
      "learning_rate": 1.9494593675921137e-05,
      "loss": 0.198,
      "step": 4199
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.882863171932184,
      "learning_rate": 1.9494282286936038e-05,
      "loss": 0.7665,
      "step": 4200
    },
    {
      "epoch": 0.13,
      "grad_norm": 8.67263681101445,
      "learning_rate": 1.9493970804543062e-05,
      "loss": 0.78,
      "step": 4201
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5301474264616409,
      "learning_rate": 1.949365922874528e-05,
      "loss": 0.3731,
      "step": 4202
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.663323338902421,
      "learning_rate": 1.9493347559545754e-05,
      "loss": 0.2875,
      "step": 4203
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9814140883265808,
      "learning_rate": 1.9493035796947554e-05,
      "loss": 0.4586,
      "step": 4204
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2807270016711865,
      "learning_rate": 1.949272394095374e-05,
      "loss": 0.519,
      "step": 4205
    },
    {
      "epoch": 0.13,
      "grad_norm": 18.446229378645707,
      "learning_rate": 1.9492411991567387e-05,
      "loss": 1.2035,
      "step": 4206
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1347619767276975,
      "learning_rate": 1.949209994879156e-05,
      "loss": 0.2925,
      "step": 4207
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2472231230588946,
      "learning_rate": 1.949178781262933e-05,
      "loss": 0.2346,
      "step": 4208
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3801915198020254,
      "learning_rate": 1.949147558308377e-05,
      "loss": 0.3274,
      "step": 4209
    },
    {
      "epoch": 0.13,
      "grad_norm": 22.44702833323849,
      "learning_rate": 1.949116326015795e-05,
      "loss": 1.8554,
      "step": 4210
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8241863099904623,
      "learning_rate": 1.949085084385494e-05,
      "loss": 0.6993,
      "step": 4211
    },
    {
      "epoch": 0.13,
      "grad_norm": 4.940050787270754,
      "learning_rate": 1.949053833417782e-05,
      "loss": 0.1794,
      "step": 4212
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.2189772741870795,
      "learning_rate": 1.949022573112966e-05,
      "loss": 0.4443,
      "step": 4213
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.7624985272353997,
      "learning_rate": 1.9489913034713535e-05,
      "loss": 0.4124,
      "step": 4214
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.536343726883222,
      "learning_rate": 1.9489600244932523e-05,
      "loss": 1.0353,
      "step": 4215
    },
    {
      "epoch": 0.13,
      "grad_norm": 3.2606100480960234,
      "learning_rate": 1.9489287361789704e-05,
      "loss": 0.1232,
      "step": 4216
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7955935413311159,
      "learning_rate": 1.9488974385288154e-05,
      "loss": 0.3143,
      "step": 4217
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6100953375559655,
      "learning_rate": 1.9488661315430948e-05,
      "loss": 0.1475,
      "step": 4218
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0105901696006798,
      "learning_rate": 1.9488348152221174e-05,
      "loss": 0.4471,
      "step": 4219
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7824038523870017,
      "learning_rate": 1.94880348956619e-05,
      "loss": 0.5073,
      "step": 4220
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6144721880569897,
      "learning_rate": 1.9487721545756227e-05,
      "loss": 0.3706,
      "step": 4221
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.613778038387364,
      "learning_rate": 1.948740810250722e-05,
      "loss": 0.4159,
      "step": 4222
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3580672733102377,
      "learning_rate": 1.9487094565917978e-05,
      "loss": 0.6716,
      "step": 4223
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0182176934197096,
      "learning_rate": 1.9486780935991573e-05,
      "loss": 0.9668,
      "step": 4224
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0155439426507156,
      "learning_rate": 1.9486467212731095e-05,
      "loss": 0.0974,
      "step": 4225
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5521585276964742,
      "learning_rate": 1.9486153396139635e-05,
      "loss": 0.3456,
      "step": 4226
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44862627687074,
      "learning_rate": 1.9485839486220272e-05,
      "loss": 0.2473,
      "step": 4227
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0735577105301233,
      "learning_rate": 1.9485525482976103e-05,
      "loss": 0.9752,
      "step": 4228
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.391843640741438,
      "learning_rate": 1.948521138641021e-05,
      "loss": 0.5222,
      "step": 4229
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1932011892100223,
      "learning_rate": 1.9484897196525686e-05,
      "loss": 0.4128,
      "step": 4230
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5795960116652522,
      "learning_rate": 1.9484582913325626e-05,
      "loss": 0.2949,
      "step": 4231
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.495638028452694,
      "learning_rate": 1.9484268536813115e-05,
      "loss": 0.3139,
      "step": 4232
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.5157545063562012,
      "learning_rate": 1.9483954066991252e-05,
      "loss": 0.9121,
      "step": 4233
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.278426390170203,
      "learning_rate": 1.9483639503863127e-05,
      "loss": 0.3853,
      "step": 4234
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5253355670726407,
      "learning_rate": 1.9483324847431835e-05,
      "loss": 0.5109,
      "step": 4235
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3820044926498998,
      "learning_rate": 1.9483010097700474e-05,
      "loss": 0.1755,
      "step": 4236
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2426005724827238,
      "learning_rate": 1.948269525467214e-05,
      "loss": 0.627,
      "step": 4237
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4762539776186355,
      "learning_rate": 1.948238031834993e-05,
      "loss": 0.3312,
      "step": 4238
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2132505297424123,
      "learning_rate": 1.948206528873694e-05,
      "loss": 0.5368,
      "step": 4239
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41907576256937173,
      "learning_rate": 1.948175016583627e-05,
      "loss": 0.247,
      "step": 4240
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.134009617154343,
      "learning_rate": 1.948143494965103e-05,
      "loss": 0.9163,
      "step": 4241
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.1114386772184277,
      "learning_rate": 1.9481119640184304e-05,
      "loss": 0.7029,
      "step": 4242
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9765842791520891,
      "learning_rate": 1.9480804237439208e-05,
      "loss": 0.8942,
      "step": 4243
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3907365102873694,
      "learning_rate": 1.948048874141884e-05,
      "loss": 0.2425,
      "step": 4244
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2843411741672761,
      "learning_rate": 1.9480173152126304e-05,
      "loss": 0.1882,
      "step": 4245
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8922247526010109,
      "learning_rate": 1.9479857469564704e-05,
      "loss": 0.5294,
      "step": 4246
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.192206541756111,
      "learning_rate": 1.947954169373715e-05,
      "loss": 0.4307,
      "step": 4247
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5791900943339507,
      "learning_rate": 1.947922582464674e-05,
      "loss": 0.575,
      "step": 4248
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4331583493395854,
      "learning_rate": 1.9478909862296595e-05,
      "loss": 0.2579,
      "step": 4249
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4847451775536445,
      "learning_rate": 1.9478593806689813e-05,
      "loss": 0.3929,
      "step": 4250
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.3159542659954,
      "learning_rate": 1.9478277657829504e-05,
      "loss": 0.0658,
      "step": 4251
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9719290673824068,
      "learning_rate": 1.947796141571878e-05,
      "loss": 0.8267,
      "step": 4252
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5201175413324859,
      "learning_rate": 1.9477645080360754e-05,
      "loss": 0.2518,
      "step": 4253
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6386302901215516,
      "learning_rate": 1.9477328651758538e-05,
      "loss": 0.4279,
      "step": 4254
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1044451623035207,
      "learning_rate": 1.9477012129915243e-05,
      "loss": 0.4671,
      "step": 4255
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4003464348721224,
      "learning_rate": 1.9476695514833987e-05,
      "loss": 0.2781,
      "step": 4256
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8001815440411912,
      "learning_rate": 1.947637880651788e-05,
      "loss": 0.1122,
      "step": 4257
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4851382193217318,
      "learning_rate": 1.947606200497004e-05,
      "loss": 0.28,
      "step": 4258
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0416817617012604,
      "learning_rate": 1.9475745110193586e-05,
      "loss": 1.0,
      "step": 4259
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5298791818245603,
      "learning_rate": 1.947542812219163e-05,
      "loss": 0.5242,
      "step": 4260
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45804797902819294,
      "learning_rate": 1.9475111040967297e-05,
      "loss": 0.3503,
      "step": 4261
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2648821547933404,
      "learning_rate": 1.9474793866523703e-05,
      "loss": 0.0962,
      "step": 4262
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5244021725241017,
      "learning_rate": 1.947447659886397e-05,
      "loss": 0.3477,
      "step": 4263
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0689488838665755,
      "learning_rate": 1.9474159237991222e-05,
      "loss": 0.4677,
      "step": 4264
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5710274517668728,
      "learning_rate": 1.9473841783908574e-05,
      "loss": 0.3063,
      "step": 4265
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5416072565162602,
      "learning_rate": 1.9473524236619153e-05,
      "loss": 0.554,
      "step": 4266
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5429806273717023,
      "learning_rate": 1.9473206596126083e-05,
      "loss": 0.3095,
      "step": 4267
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40073762233636606,
      "learning_rate": 1.947288886243249e-05,
      "loss": 0.2722,
      "step": 4268
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5836701000217496,
      "learning_rate": 1.94725710355415e-05,
      "loss": 0.3201,
      "step": 4269
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5590654065409442,
      "learning_rate": 1.947225311545624e-05,
      "loss": 0.0466,
      "step": 4270
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4788828609268632,
      "learning_rate": 1.9471935102179836e-05,
      "loss": 0.2605,
      "step": 4271
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5801460553487524,
      "learning_rate": 1.9471616995715418e-05,
      "loss": 0.4391,
      "step": 4272
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.399399068924746,
      "learning_rate": 1.9471298796066117e-05,
      "loss": 0.3015,
      "step": 4273
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5177781168365891,
      "learning_rate": 1.947098050323506e-05,
      "loss": 0.8673,
      "step": 4274
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.26858227589399036,
      "learning_rate": 1.9470662117225378e-05,
      "loss": 0.1302,
      "step": 4275
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4316090690099432,
      "learning_rate": 1.9470343638040212e-05,
      "loss": 0.301,
      "step": 4276
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3425113898267944,
      "learning_rate": 1.9470025065682682e-05,
      "loss": 0.1827,
      "step": 4277
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.565038961469726,
      "learning_rate": 1.9469706400155936e-05,
      "loss": 0.8364,
      "step": 4278
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4137702971012709,
      "learning_rate": 1.9469387641463097e-05,
      "loss": 0.3055,
      "step": 4279
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6838216238472333,
      "learning_rate": 1.9469068789607314e-05,
      "loss": 0.421,
      "step": 4280
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44026247037813654,
      "learning_rate": 1.946874984459171e-05,
      "loss": 0.3109,
      "step": 4281
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.048883115296053,
      "learning_rate": 1.946843080641943e-05,
      "loss": 0.9355,
      "step": 4282
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.9375544574895436,
      "learning_rate": 1.9468111675093615e-05,
      "loss": 0.4499,
      "step": 4283
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5046658685589709,
      "learning_rate": 1.9467792450617396e-05,
      "loss": 0.3012,
      "step": 4284
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5317998020233493,
      "learning_rate": 1.9467473132993926e-05,
      "loss": 0.3176,
      "step": 4285
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3371871714520758,
      "learning_rate": 1.946715372222634e-05,
      "loss": 0.2857,
      "step": 4286
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3775672443659015,
      "learning_rate": 1.946683421831778e-05,
      "loss": 0.195,
      "step": 4287
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7468979520429584,
      "learning_rate": 1.9466514621271385e-05,
      "loss": 0.4474,
      "step": 4288
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8020210441844275,
      "learning_rate": 1.946619493109031e-05,
      "loss": 0.3497,
      "step": 4289
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.43411597084066106,
      "learning_rate": 1.9465875147777692e-05,
      "loss": 0.3035,
      "step": 4290
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6497149311883988,
      "learning_rate": 1.9465555271336677e-05,
      "loss": 0.4436,
      "step": 4291
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4880982017631553,
      "learning_rate": 1.946523530177042e-05,
      "loss": 0.2959,
      "step": 4292
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.8011247024765786,
      "learning_rate": 1.9464915239082058e-05,
      "loss": 0.8948,
      "step": 4293
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.39414703455646594,
      "learning_rate": 1.946459508327475e-05,
      "loss": 0.2032,
      "step": 4294
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4232324037639147,
      "learning_rate": 1.946427483435164e-05,
      "loss": 0.2836,
      "step": 4295
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4267571233511559,
      "learning_rate": 1.946395449231588e-05,
      "loss": 0.2475,
      "step": 4296
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.363438219260039,
      "learning_rate": 1.9463634057170622e-05,
      "loss": 0.2736,
      "step": 4297
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8058632363559833,
      "learning_rate": 1.946331352891902e-05,
      "loss": 0.4492,
      "step": 4298
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3704989558681432,
      "learning_rate": 1.946299290756422e-05,
      "loss": 0.2382,
      "step": 4299
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3801245545667573,
      "learning_rate": 1.9462672193109386e-05,
      "loss": 0.7122,
      "step": 4300
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2747218927863948,
      "learning_rate": 1.946235138555767e-05,
      "loss": 0.3422,
      "step": 4301
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4398267454201323,
      "learning_rate": 1.9462030484912223e-05,
      "loss": 0.9104,
      "step": 4302
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.36458666691259806,
      "learning_rate": 1.9461709491176214e-05,
      "loss": 0.2107,
      "step": 4303
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4175336501549536,
      "learning_rate": 1.9461388404352787e-05,
      "loss": 0.3293,
      "step": 4304
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4856117759876905,
      "learning_rate": 1.9461067224445112e-05,
      "loss": 0.2009,
      "step": 4305
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6715511693678116,
      "learning_rate": 1.9460745951456343e-05,
      "loss": 0.5016,
      "step": 4306
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2397371166621541,
      "learning_rate": 1.946042458538964e-05,
      "loss": 0.1087,
      "step": 4307
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5032956045629163,
      "learning_rate": 1.9460103126248172e-05,
      "loss": 0.3603,
      "step": 4308
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2004918761576946,
      "learning_rate": 1.9459781574035093e-05,
      "loss": 0.3369,
      "step": 4309
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3802492449643304,
      "learning_rate": 1.945945992875357e-05,
      "loss": 0.2888,
      "step": 4310
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.631008610105019,
      "learning_rate": 1.945913819040677e-05,
      "loss": 0.9284,
      "step": 4311
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.35790345922783634,
      "learning_rate": 1.9458816358997853e-05,
      "loss": 0.1773,
      "step": 4312
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6781190845584056,
      "learning_rate": 1.9458494434529993e-05,
      "loss": 0.4454,
      "step": 4313
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8650227686433015,
      "learning_rate": 1.9458172417006347e-05,
      "loss": 0.4406,
      "step": 4314
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3785744497881036,
      "learning_rate": 1.945785030643009e-05,
      "loss": 0.347,
      "step": 4315
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.32998002484853955,
      "learning_rate": 1.9457528102804393e-05,
      "loss": 0.0798,
      "step": 4316
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5692706970673054,
      "learning_rate": 1.945720580613242e-05,
      "loss": 0.3509,
      "step": 4317
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7336734515649206,
      "learning_rate": 1.945688341641735e-05,
      "loss": 0.2838,
      "step": 4318
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6689806800967875,
      "learning_rate": 1.945656093366234e-05,
      "loss": 0.8256,
      "step": 4319
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4616649427896028,
      "learning_rate": 1.9456238357870576e-05,
      "loss": 0.3062,
      "step": 4320
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8014948076885988,
      "learning_rate": 1.9455915689045225e-05,
      "loss": 0.3692,
      "step": 4321
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3519196374487653,
      "learning_rate": 1.945559292718947e-05,
      "loss": 0.2668,
      "step": 4322
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8731856794975368,
      "learning_rate": 1.9455270072306477e-05,
      "loss": 0.5264,
      "step": 4323
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.078917927780516,
      "learning_rate": 1.9454947124399424e-05,
      "loss": 0.386,
      "step": 4324
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.28243048814384275,
      "learning_rate": 1.9454624083471493e-05,
      "loss": 0.0995,
      "step": 4325
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4285595626968762,
      "learning_rate": 1.945430094952586e-05,
      "loss": 0.3067,
      "step": 4326
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4076593868691795,
      "learning_rate": 1.94539777225657e-05,
      "loss": 0.2979,
      "step": 4327
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.092306954433205,
      "learning_rate": 1.94536544025942e-05,
      "loss": 0.9268,
      "step": 4328
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3445157695995131,
      "learning_rate": 1.9453330989614533e-05,
      "loss": 0.6888,
      "step": 4329
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7719672389424935,
      "learning_rate": 1.9453007483629888e-05,
      "loss": 0.4225,
      "step": 4330
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.529244339726201,
      "learning_rate": 1.9452683884643444e-05,
      "loss": 0.3331,
      "step": 4331
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1493365886032416,
      "learning_rate": 1.9452360192658387e-05,
      "loss": 0.6461,
      "step": 4332
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4218250719156446,
      "learning_rate": 1.9452036407677897e-05,
      "loss": 0.27,
      "step": 4333
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4705107791680091,
      "learning_rate": 1.9451712529705167e-05,
      "loss": 0.2452,
      "step": 4334
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.30469025320984716,
      "learning_rate": 1.9451388558743378e-05,
      "loss": 0.1983,
      "step": 4335
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.503776089522112,
      "learning_rate": 1.9451064494795718e-05,
      "loss": 0.6298,
      "step": 4336
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2790647103708763,
      "learning_rate": 1.9450740337865372e-05,
      "loss": 0.5117,
      "step": 4337
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40675804384406644,
      "learning_rate": 1.945041608795554e-05,
      "loss": 0.284,
      "step": 4338
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4572530227022848,
      "learning_rate": 1.94500917450694e-05,
      "loss": 0.3037,
      "step": 4339
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5976217181128897,
      "learning_rate": 1.9449767309210148e-05,
      "loss": 0.3207,
      "step": 4340
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.839190960767282,
      "learning_rate": 1.9449442780380977e-05,
      "loss": 0.5121,
      "step": 4341
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4385078079776346,
      "learning_rate": 1.944911815858508e-05,
      "loss": 0.1282,
      "step": 4342
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49453509747590074,
      "learning_rate": 1.9448793443825647e-05,
      "loss": 0.2452,
      "step": 4343
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.44201479430260376,
      "learning_rate": 1.9448468636105874e-05,
      "loss": 0.203,
      "step": 4344
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4018945424830133,
      "learning_rate": 1.944814373542896e-05,
      "loss": 0.3175,
      "step": 4345
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5738214826976881,
      "learning_rate": 1.94478187417981e-05,
      "loss": 0.3181,
      "step": 4346
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8965790633488763,
      "learning_rate": 1.9447493655216486e-05,
      "loss": 0.5154,
      "step": 4347
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.5184158833187673,
      "learning_rate": 1.9447168475687325e-05,
      "loss": 0.2827,
      "step": 4348
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.45987520085875083,
      "learning_rate": 1.944684320321381e-05,
      "loss": 0.3221,
      "step": 4349
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9185855748223348,
      "learning_rate": 1.9446517837799144e-05,
      "loss": 0.4566,
      "step": 4350
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.40290779463084375,
      "learning_rate": 1.944619237944653e-05,
      "loss": 0.2827,
      "step": 4351
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7577207226084596,
      "learning_rate": 1.9445866828159163e-05,
      "loss": 0.3012,
      "step": 4352
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.339095887369641,
      "learning_rate": 1.944554118394025e-05,
      "loss": 0.2192,
      "step": 4353
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5485260424815952,
      "learning_rate": 1.9445215446792996e-05,
      "loss": 0.2649,
      "step": 4354
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.553004270736274,
      "learning_rate": 1.9444889616720607e-05,
      "loss": 0.388,
      "step": 4355
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6711507829692369,
      "learning_rate": 1.9444563693726285e-05,
      "loss": 0.3918,
      "step": 4356
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.57222938143088,
      "learning_rate": 1.9444237677813234e-05,
      "loss": 0.2576,
      "step": 4357
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4445416097007528,
      "learning_rate": 1.9443911568984667e-05,
      "loss": 0.3301,
      "step": 4358
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.3642301157145418,
      "learning_rate": 1.9443585367243793e-05,
      "loss": 0.5472,
      "step": 4359
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1298547244378185,
      "learning_rate": 1.9443259072593815e-05,
      "loss": 0.6721,
      "step": 4360
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.275528463608183,
      "learning_rate": 1.944293268503795e-05,
      "loss": 0.0804,
      "step": 4361
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4464791406957934,
      "learning_rate": 1.9442606204579408e-05,
      "loss": 0.3134,
      "step": 4362
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3890047377998632,
      "learning_rate": 1.944227963122139e-05,
      "loss": 0.2596,
      "step": 4363
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.49948103702766594,
      "learning_rate": 1.9441952964967127e-05,
      "loss": 0.3249,
      "step": 4364
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.8491242751209145,
      "learning_rate": 1.9441626205819822e-05,
      "loss": 0.5216,
      "step": 4365
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3752799435007773,
      "learning_rate": 1.944129935378269e-05,
      "loss": 0.2305,
      "step": 4366
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.7438202394550946,
      "learning_rate": 1.944097240885895e-05,
      "loss": 0.3949,
      "step": 4367
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.947875102869952,
      "learning_rate": 1.944064537105182e-05,
      "loss": 0.619,
      "step": 4368
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4011079183246064,
      "learning_rate": 1.9440318240364505e-05,
      "loss": 0.3196,
      "step": 4369
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3058441514874082,
      "learning_rate": 1.943999101680024e-05,
      "loss": 0.1426,
      "step": 4370
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0686836700764073,
      "learning_rate": 1.943966370036224e-05,
      "loss": 0.4223,
      "step": 4371
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.30406929367333824,
      "learning_rate": 1.9439336291053715e-05,
      "loss": 0.1967,
      "step": 4372
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.2273139461197495,
      "learning_rate": 1.9439008788877898e-05,
      "loss": 0.5836,
      "step": 4373
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41554692095662416,
      "learning_rate": 1.9438681193838003e-05,
      "loss": 0.284,
      "step": 4374
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1852083521615508,
      "learning_rate": 1.9438353505937263e-05,
      "loss": 0.626,
      "step": 4375
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3533998893836437,
      "learning_rate": 1.943802572517889e-05,
      "loss": 0.206,
      "step": 4376
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4981421236774912,
      "learning_rate": 1.9437697851566116e-05,
      "loss": 0.6246,
      "step": 4377
    },
    {
      "epoch": 0.13,
      "grad_norm": 2.0583464643669407,
      "learning_rate": 1.9437369885102165e-05,
      "loss": 0.8594,
      "step": 4378
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3116294209399421,
      "learning_rate": 1.9437041825790265e-05,
      "loss": 0.114,
      "step": 4379
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48462986499522925,
      "learning_rate": 1.943671367363364e-05,
      "loss": 0.3132,
      "step": 4380
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2801089719353992,
      "learning_rate": 1.9436385428635525e-05,
      "loss": 0.2199,
      "step": 4381
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1526457679383308,
      "learning_rate": 1.943605709079914e-05,
      "loss": 0.6211,
      "step": 4382
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0594231030700385,
      "learning_rate": 1.9435728660127725e-05,
      "loss": 0.4242,
      "step": 4383
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1217775469781006,
      "learning_rate": 1.9435400136624504e-05,
      "loss": 0.5766,
      "step": 4384
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4616647057684564,
      "learning_rate": 1.9435071520292712e-05,
      "loss": 0.2443,
      "step": 4385
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4247851503843374,
      "learning_rate": 1.943474281113558e-05,
      "loss": 0.7838,
      "step": 4386
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4148178958568173,
      "learning_rate": 1.9434414009156348e-05,
      "loss": 0.2796,
      "step": 4387
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.4006366102868397,
      "learning_rate": 1.9434085114358243e-05,
      "loss": 0.7287,
      "step": 4388
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.2838745551915719,
      "learning_rate": 1.9433756126744508e-05,
      "loss": 0.1563,
      "step": 4389
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.34000289472542694,
      "learning_rate": 1.9433427046318374e-05,
      "loss": 0.1654,
      "step": 4390
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.0454419363727987,
      "learning_rate": 1.943309787308308e-05,
      "loss": 0.5597,
      "step": 4391
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3591317304428514,
      "learning_rate": 1.9432768607041866e-05,
      "loss": 0.3068,
      "step": 4392
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.9689946920763549,
      "learning_rate": 1.943243924819797e-05,
      "loss": 0.4994,
      "step": 4393
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4476447123557399,
      "learning_rate": 1.9432109796554633e-05,
      "loss": 0.2135,
      "step": 4394
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6676096890421254,
      "learning_rate": 1.94317802521151e-05,
      "loss": 0.4205,
      "step": 4395
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1316777994502962,
      "learning_rate": 1.9431450614882604e-05,
      "loss": 0.3936,
      "step": 4396
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5374449827646911,
      "learning_rate": 1.9431120884860398e-05,
      "loss": 0.3823,
      "step": 4397
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3837896671346948,
      "learning_rate": 1.943079106205172e-05,
      "loss": 0.0781,
      "step": 4398
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.3504787498247627,
      "learning_rate": 1.9430461146459816e-05,
      "loss": 0.3122,
      "step": 4399
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6633153067014762,
      "learning_rate": 1.943013113808793e-05,
      "loss": 0.4433,
      "step": 4400
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.4253773686576433,
      "learning_rate": 1.9429801036939314e-05,
      "loss": 0.1912,
      "step": 4401
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.1329406683673087,
      "learning_rate": 1.942947084301721e-05,
      "loss": 0.5261,
      "step": 4402
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.41975051941388614,
      "learning_rate": 1.9429140556324873e-05,
      "loss": 0.2171,
      "step": 4403
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.6336300937657903,
      "learning_rate": 1.9428810176865545e-05,
      "loss": 0.4153,
      "step": 4404
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.48963534685750165,
      "learning_rate": 1.9428479704642482e-05,
      "loss": 0.2698,
      "step": 4405
    },
    {
      "epoch": 0.13,
      "grad_norm": 1.6732601651623775,
      "learning_rate": 1.9428149139658933e-05,
      "loss": 0.6958,
      "step": 4406
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.587223716646387,
      "learning_rate": 1.942781848191815e-05,
      "loss": 0.2954,
      "step": 4407
    },
    {
      "epoch": 0.13,
      "grad_norm": 0.5452586415889928,
      "learning_rate": 1.9427487731423383e-05,
      "loss": 0.3751,
      "step": 4408
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6511913741400841,
      "learning_rate": 1.9427156888177893e-05,
      "loss": 0.2042,
      "step": 4409
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39257458308968385,
      "learning_rate": 1.9426825952184932e-05,
      "loss": 0.253,
      "step": 4410
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43096425272026784,
      "learning_rate": 1.9426494923447755e-05,
      "loss": 0.2744,
      "step": 4411
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45166522707850837,
      "learning_rate": 1.942616380196962e-05,
      "loss": 0.2969,
      "step": 4412
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3767498838944143,
      "learning_rate": 1.9425832587753784e-05,
      "loss": 0.5233,
      "step": 4413
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5697248840644447,
      "learning_rate": 1.9425501280803506e-05,
      "loss": 0.7345,
      "step": 4414
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47989708830142475,
      "learning_rate": 1.9425169881122046e-05,
      "loss": 0.297,
      "step": 4415
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.31638867207676563,
      "learning_rate": 1.942483838871266e-05,
      "loss": 0.2192,
      "step": 4416
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6676380635670066,
      "learning_rate": 1.9424506803578612e-05,
      "loss": 0.4267,
      "step": 4417
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8530953458198735,
      "learning_rate": 1.942417512572317e-05,
      "loss": 0.4235,
      "step": 4418
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6158919875486325,
      "learning_rate": 1.9423843355149588e-05,
      "loss": 0.3047,
      "step": 4419
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6718827001768082,
      "learning_rate": 1.9423511491861134e-05,
      "loss": 0.3095,
      "step": 4420
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.168873964286269,
      "learning_rate": 1.9423179535861076e-05,
      "loss": 0.5366,
      "step": 4421
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2670347737829968,
      "learning_rate": 1.9422847487152675e-05,
      "loss": 0.2116,
      "step": 4422
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5741129819406794,
      "learning_rate": 1.94225153457392e-05,
      "loss": 0.3856,
      "step": 4423
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8838819425145641,
      "learning_rate": 1.942218311162392e-05,
      "loss": 0.4273,
      "step": 4424
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1227054773673797,
      "learning_rate": 1.94218507848101e-05,
      "loss": 0.5339,
      "step": 4425
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.36745813366997676,
      "learning_rate": 1.9421518365301017e-05,
      "loss": 0.1996,
      "step": 4426
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.6086028112609023,
      "learning_rate": 1.942118585309993e-05,
      "loss": 0.8934,
      "step": 4427
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37548438080179686,
      "learning_rate": 1.942085324821012e-05,
      "loss": 0.2902,
      "step": 4428
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1701054803236135,
      "learning_rate": 1.9420520550634855e-05,
      "loss": 0.5268,
      "step": 4429
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5685403820557665,
      "learning_rate": 1.942018776037741e-05,
      "loss": 0.3051,
      "step": 4430
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37342425037013627,
      "learning_rate": 1.9419854877441055e-05,
      "loss": 0.1403,
      "step": 4431
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.94167694376089,
      "learning_rate": 1.941952190182907e-05,
      "loss": 0.5319,
      "step": 4432
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9650613097230752,
      "learning_rate": 1.941918883354473e-05,
      "loss": 0.4199,
      "step": 4433
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35363996198278297,
      "learning_rate": 1.941885567259131e-05,
      "loss": 0.2828,
      "step": 4434
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35968301871525066,
      "learning_rate": 1.941852241897209e-05,
      "loss": 0.2429,
      "step": 4435
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3994547890208022,
      "learning_rate": 1.9418189072690348e-05,
      "loss": 0.6927,
      "step": 4436
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2029305097782994,
      "learning_rate": 1.941785563374936e-05,
      "loss": 0.3597,
      "step": 4437
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5064334977200957,
      "learning_rate": 1.941752210215241e-05,
      "loss": 0.8882,
      "step": 4438
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2780552497400836,
      "learning_rate": 1.941718847790278e-05,
      "loss": 0.1403,
      "step": 4439
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4388925193409742,
      "learning_rate": 1.941685476100375e-05,
      "loss": 0.3383,
      "step": 4440
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3473163075805807,
      "learning_rate": 1.9416520951458605e-05,
      "loss": 0.2109,
      "step": 4441
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1453654282844996,
      "learning_rate": 1.941618704927063e-05,
      "loss": 0.4638,
      "step": 4442
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7391438935928513,
      "learning_rate": 1.9415853054443107e-05,
      "loss": 0.4151,
      "step": 4443
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38466156664172685,
      "learning_rate": 1.941551896697932e-05,
      "loss": 0.2455,
      "step": 4444
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.2847167927643373,
      "learning_rate": 1.9415184786882566e-05,
      "loss": 0.7877,
      "step": 4445
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37024096422258873,
      "learning_rate": 1.941485051415612e-05,
      "loss": 0.3009,
      "step": 4446
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.742657442091801,
      "learning_rate": 1.941451614880328e-05,
      "loss": 0.863,
      "step": 4447
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3862059844621594,
      "learning_rate": 1.9414181690827335e-05,
      "loss": 0.0824,
      "step": 4448
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48988848994956496,
      "learning_rate": 1.9413847140231565e-05,
      "loss": 0.3569,
      "step": 4449
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5366971090477679,
      "learning_rate": 1.9413512497019275e-05,
      "loss": 0.3007,
      "step": 4450
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3684778700956605,
      "learning_rate": 1.941317776119375e-05,
      "loss": 0.3148,
      "step": 4451
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5069678370715079,
      "learning_rate": 1.9412842932758285e-05,
      "loss": 0.2682,
      "step": 4452
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.47797172842863067,
      "learning_rate": 1.9412508011716176e-05,
      "loss": 0.2968,
      "step": 4453
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5215991698080075,
      "learning_rate": 1.9412172998070713e-05,
      "loss": 0.2982,
      "step": 4454
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6752357128122333,
      "learning_rate": 1.9411837891825195e-05,
      "loss": 0.6121,
      "step": 4455
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7985963968702101,
      "learning_rate": 1.941150269298292e-05,
      "loss": 0.8191,
      "step": 4456
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37855316046512844,
      "learning_rate": 1.9411167401547188e-05,
      "loss": 0.1979,
      "step": 4457
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.36461567137161643,
      "learning_rate": 1.941083201752129e-05,
      "loss": 0.3145,
      "step": 4458
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6884003327491215,
      "learning_rate": 1.941049654090853e-05,
      "loss": 0.4381,
      "step": 4459
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5786216865104697,
      "learning_rate": 1.9410160971712212e-05,
      "loss": 0.2532,
      "step": 4460
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.29067401363153994,
      "learning_rate": 1.940982530993563e-05,
      "loss": 0.0752,
      "step": 4461
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4748453946610672,
      "learning_rate": 1.9409489555582094e-05,
      "loss": 0.3179,
      "step": 4462
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1184591487467033,
      "learning_rate": 1.94091537086549e-05,
      "loss": 0.3578,
      "step": 4463
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44530704329559045,
      "learning_rate": 1.940881776915736e-05,
      "loss": 0.3367,
      "step": 4464
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2403306763249533,
      "learning_rate": 1.9408481737092774e-05,
      "loss": 0.6818,
      "step": 4465
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8244938565920898,
      "learning_rate": 1.9408145612464446e-05,
      "loss": 0.3667,
      "step": 4466
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4396503828739678,
      "learning_rate": 1.9407809395275686e-05,
      "loss": 0.2911,
      "step": 4467
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.875288459466907,
      "learning_rate": 1.9407473085529805e-05,
      "loss": 0.418,
      "step": 4468
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39654425922781233,
      "learning_rate": 1.9407136683230105e-05,
      "loss": 0.2942,
      "step": 4469
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3011646386341651,
      "learning_rate": 1.94068001883799e-05,
      "loss": 0.176,
      "step": 4470
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6338613615784958,
      "learning_rate": 1.94064636009825e-05,
      "loss": 0.339,
      "step": 4471
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4816583051324002,
      "learning_rate": 1.9406126921041216e-05,
      "loss": 0.2471,
      "step": 4472
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8013110200635347,
      "learning_rate": 1.940579014855936e-05,
      "loss": 0.9004,
      "step": 4473
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9127805439426525,
      "learning_rate": 1.9405453283540244e-05,
      "loss": 0.4506,
      "step": 4474
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8538439552729473,
      "learning_rate": 1.9405116325987188e-05,
      "loss": 0.3557,
      "step": 4475
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.33329272637611806,
      "learning_rate": 1.94047792759035e-05,
      "loss": 0.2594,
      "step": 4476
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5148509829814488,
      "learning_rate": 1.9404442133292498e-05,
      "loss": 0.4046,
      "step": 4477
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.27539588165890283,
      "learning_rate": 1.94041048981575e-05,
      "loss": 0.0738,
      "step": 4478
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5663769106352922,
      "learning_rate": 1.9403767570501825e-05,
      "loss": 0.2589,
      "step": 4479
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3725776525806671,
      "learning_rate": 1.9403430150328792e-05,
      "loss": 0.1757,
      "step": 4480
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35761286532711034,
      "learning_rate": 1.9403092637641716e-05,
      "loss": 0.2815,
      "step": 4481
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5398224023639357,
      "learning_rate": 1.940275503244392e-05,
      "loss": 0.4004,
      "step": 4482
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9732361573176133,
      "learning_rate": 1.940241733473873e-05,
      "loss": 0.44,
      "step": 4483
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6959659618913995,
      "learning_rate": 1.9402079544529463e-05,
      "loss": 0.476,
      "step": 4484
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38314493788982834,
      "learning_rate": 1.9401741661819442e-05,
      "loss": 0.2555,
      "step": 4485
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8938886720846067,
      "learning_rate": 1.9401403686611993e-05,
      "loss": 0.7328,
      "step": 4486
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4651302813878306,
      "learning_rate": 1.940106561891044e-05,
      "loss": 0.2672,
      "step": 4487
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2874058867952468,
      "learning_rate": 1.9400727458718115e-05,
      "loss": 0.2022,
      "step": 4488
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4340722315246684,
      "learning_rate": 1.9400389206038337e-05,
      "loss": 0.2036,
      "step": 4489
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5913096490147711,
      "learning_rate": 1.9400050860874435e-05,
      "loss": 0.3588,
      "step": 4490
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6188784835558396,
      "learning_rate": 1.9399712423229742e-05,
      "loss": 0.5488,
      "step": 4491
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0379435938513,
      "learning_rate": 1.9399373893107584e-05,
      "loss": 0.5315,
      "step": 4492
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3922120335643027,
      "learning_rate": 1.9399035270511292e-05,
      "loss": 0.2688,
      "step": 4493
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3235392670143237,
      "learning_rate": 1.93986965554442e-05,
      "loss": 0.2448,
      "step": 4494
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2239309325957801,
      "learning_rate": 1.939835774790964e-05,
      "loss": 0.6069,
      "step": 4495
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1203840663904683,
      "learning_rate": 1.9398018847910943e-05,
      "loss": 0.3728,
      "step": 4496
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39734384504420517,
      "learning_rate": 1.9397679855451443e-05,
      "loss": 0.1496,
      "step": 4497
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41449540820178477,
      "learning_rate": 1.939734077053448e-05,
      "loss": 0.1766,
      "step": 4498
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5946895263391149,
      "learning_rate": 1.9397001593163383e-05,
      "loss": 0.3605,
      "step": 4499
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35231116725577355,
      "learning_rate": 1.939666232334149e-05,
      "loss": 0.2847,
      "step": 4500
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9553004637263025,
      "learning_rate": 1.9396322961072143e-05,
      "loss": 0.5121,
      "step": 4501
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9843306223706957,
      "learning_rate": 1.9395983506358683e-05,
      "loss": 0.4354,
      "step": 4502
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45695287962080156,
      "learning_rate": 1.9395643959204442e-05,
      "loss": 0.2936,
      "step": 4503
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5389085390175592,
      "learning_rate": 1.9395304319612763e-05,
      "loss": 0.275,
      "step": 4504
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41574342599002667,
      "learning_rate": 1.9394964587586994e-05,
      "loss": 0.3184,
      "step": 4505
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40213397061190226,
      "learning_rate": 1.9394624763130466e-05,
      "loss": 0.1251,
      "step": 4506
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43538445506347123,
      "learning_rate": 1.939428484624653e-05,
      "loss": 0.202,
      "step": 4507
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4012503181240332,
      "learning_rate": 1.9393944836938528e-05,
      "loss": 0.2447,
      "step": 4508
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1038757688826655,
      "learning_rate": 1.939360473520981e-05,
      "loss": 0.463,
      "step": 4509
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1011519802724294,
      "learning_rate": 1.9393264541063712e-05,
      "loss": 0.6103,
      "step": 4510
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.31763954417917734,
      "learning_rate": 1.9392924254503593e-05,
      "loss": 0.2498,
      "step": 4511
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3971712772950084,
      "learning_rate": 1.9392583875532785e-05,
      "loss": 0.2957,
      "step": 4512
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1022947289376384,
      "learning_rate": 1.9392243404154657e-05,
      "loss": 0.4976,
      "step": 4513
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.388988250557099,
      "learning_rate": 1.939190284037254e-05,
      "loss": 0.8878,
      "step": 4514
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2922427196423614,
      "learning_rate": 1.9391562184189796e-05,
      "loss": 0.1265,
      "step": 4515
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7110589474764236,
      "learning_rate": 1.9391221435609772e-05,
      "loss": 0.3561,
      "step": 4516
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3292707555094415,
      "learning_rate": 1.939088059463582e-05,
      "loss": 0.2662,
      "step": 4517
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6233927905108789,
      "learning_rate": 1.93905396612713e-05,
      "loss": 0.3869,
      "step": 4518
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.055968062695148,
      "learning_rate": 1.9390198635519558e-05,
      "loss": 0.4915,
      "step": 4519
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9256629496435512,
      "learning_rate": 1.938985751738395e-05,
      "loss": 0.4286,
      "step": 4520
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.34593758110841355,
      "learning_rate": 1.9389516306867838e-05,
      "loss": 0.2415,
      "step": 4521
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.3015210491111704,
      "learning_rate": 1.938917500397457e-05,
      "loss": 0.3259,
      "step": 4522
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4005492765397159,
      "learning_rate": 1.9388833608707512e-05,
      "loss": 0.325,
      "step": 4523
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3854875333692061,
      "learning_rate": 1.9388492121070018e-05,
      "loss": 0.1247,
      "step": 4524
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45648561029376833,
      "learning_rate": 1.938815054106545e-05,
      "loss": 0.2924,
      "step": 4525
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.30034770520195425,
      "learning_rate": 1.938780886869717e-05,
      "loss": 0.1673,
      "step": 4526
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7358230522561412,
      "learning_rate": 1.9387467103968537e-05,
      "loss": 0.5208,
      "step": 4527
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4704571937314413,
      "learning_rate": 1.938712524688291e-05,
      "loss": 0.3244,
      "step": 4528
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5690476087180297,
      "learning_rate": 1.938678329744366e-05,
      "loss": 0.3853,
      "step": 4529
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40628882097044206,
      "learning_rate": 1.9386441255654144e-05,
      "loss": 0.1826,
      "step": 4530
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5628825410505355,
      "learning_rate": 1.9386099121517733e-05,
      "loss": 0.3893,
      "step": 4531
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6342900761414987,
      "learning_rate": 1.9385756895037787e-05,
      "loss": 0.0248,
      "step": 4532
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5302751304820961,
      "learning_rate": 1.938541457621768e-05,
      "loss": 0.2874,
      "step": 4533
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5517510372791498,
      "learning_rate": 1.9385072165060777e-05,
      "loss": 0.3437,
      "step": 4534
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3109916881017781,
      "learning_rate": 1.9384729661570442e-05,
      "loss": 0.2368,
      "step": 4535
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4709875064702547,
      "learning_rate": 1.938438706575005e-05,
      "loss": 0.4159,
      "step": 4536
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2607832815967417,
      "learning_rate": 1.9384044377602975e-05,
      "loss": 0.1047,
      "step": 4537
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7279017786838402,
      "learning_rate": 1.9383701597132574e-05,
      "loss": 0.9046,
      "step": 4538
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3464710469457841,
      "learning_rate": 1.9383358724342237e-05,
      "loss": 0.2111,
      "step": 4539
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.125094460495259,
      "learning_rate": 1.9383015759235328e-05,
      "loss": 0.6254,
      "step": 4540
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40315794438120595,
      "learning_rate": 1.938267270181522e-05,
      "loss": 0.2651,
      "step": 4541
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0351703540405035,
      "learning_rate": 1.938232955208529e-05,
      "loss": 0.8362,
      "step": 4542
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6824950130332903,
      "learning_rate": 1.938198631004892e-05,
      "loss": 0.2236,
      "step": 4543
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4345062641143386,
      "learning_rate": 1.9381642975709475e-05,
      "loss": 0.3407,
      "step": 4544
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7885318160775074,
      "learning_rate": 1.9381299549070345e-05,
      "loss": 0.3946,
      "step": 4545
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5089655236639673,
      "learning_rate": 1.93809560301349e-05,
      "loss": 0.2739,
      "step": 4546
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37922312006902625,
      "learning_rate": 1.9380612418906522e-05,
      "loss": 0.2945,
      "step": 4547
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.356366815251337,
      "learning_rate": 1.9380268715388595e-05,
      "loss": 0.2115,
      "step": 4548
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5510777606799322,
      "learning_rate": 1.9379924919584497e-05,
      "loss": 0.2529,
      "step": 4549
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0520059520363794,
      "learning_rate": 1.937958103149761e-05,
      "loss": 0.399,
      "step": 4550
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9804199991320818,
      "learning_rate": 1.937923705113132e-05,
      "loss": 0.4726,
      "step": 4551
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5370683549013022,
      "learning_rate": 1.937889297848901e-05,
      "loss": 0.2234,
      "step": 4552
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45903337371529007,
      "learning_rate": 1.9378548813574064e-05,
      "loss": 0.3348,
      "step": 4553
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3638400164942651,
      "learning_rate": 1.937820455638987e-05,
      "loss": 0.2901,
      "step": 4554
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5376850219825879,
      "learning_rate": 1.9377860206939812e-05,
      "loss": 0.306,
      "step": 4555
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.28162695209420674,
      "learning_rate": 1.937751576522728e-05,
      "loss": 0.1379,
      "step": 4556
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7114813349041892,
      "learning_rate": 1.9377171231255663e-05,
      "loss": 0.2964,
      "step": 4557
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.432691917844076,
      "learning_rate": 1.9376826605028348e-05,
      "loss": 0.2418,
      "step": 4558
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4304315904202547,
      "learning_rate": 1.937648188654873e-05,
      "loss": 0.3299,
      "step": 4559
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.128322398355075,
      "learning_rate": 1.9376137075820196e-05,
      "loss": 0.4233,
      "step": 4560
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7926515919443697,
      "learning_rate": 1.9375792172846143e-05,
      "loss": 0.3182,
      "step": 4561
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44924377715193303,
      "learning_rate": 1.9375447177629957e-05,
      "loss": 0.3353,
      "step": 4562
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1870534851522785,
      "learning_rate": 1.9375102090175042e-05,
      "loss": 0.4601,
      "step": 4563
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.810493878028183,
      "learning_rate": 1.9374756910484788e-05,
      "loss": 0.8962,
      "step": 4564
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35791986171900625,
      "learning_rate": 1.9374411638562586e-05,
      "loss": 0.2841,
      "step": 4565
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3880925359794621,
      "learning_rate": 1.937406627441184e-05,
      "loss": 0.1771,
      "step": 4566
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.32376501098104793,
      "learning_rate": 1.9373720818035945e-05,
      "loss": 0.2005,
      "step": 4567
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.1556286954865893,
      "learning_rate": 1.93733752694383e-05,
      "loss": 0.8798,
      "step": 4568
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.449434486920248,
      "learning_rate": 1.9373029628622305e-05,
      "loss": 0.4766,
      "step": 4569
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.515280063859492,
      "learning_rate": 1.9372683895591363e-05,
      "loss": 0.3743,
      "step": 4570
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.29772413997449765,
      "learning_rate": 1.937233807034887e-05,
      "loss": 0.242,
      "step": 4571
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5618186763325919,
      "learning_rate": 1.937199215289823e-05,
      "loss": 0.6831,
      "step": 4572
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.971563242117797,
      "learning_rate": 1.937164614324285e-05,
      "loss": 0.498,
      "step": 4573
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.34103018795960016,
      "learning_rate": 1.937130004138613e-05,
      "loss": 0.1388,
      "step": 4574
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6549439380127312,
      "learning_rate": 1.937095384733148e-05,
      "loss": 0.3232,
      "step": 4575
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.31029118519156196,
      "learning_rate": 1.9370607561082298e-05,
      "loss": 0.1572,
      "step": 4576
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41433631859166703,
      "learning_rate": 1.9370261182641997e-05,
      "loss": 0.3114,
      "step": 4577
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0960496638722765,
      "learning_rate": 1.9369914712013985e-05,
      "loss": 0.4462,
      "step": 4578
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1392477866656123,
      "learning_rate": 1.9369568149201666e-05,
      "loss": 0.4325,
      "step": 4579
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44381279054082284,
      "learning_rate": 1.9369221494208456e-05,
      "loss": 0.206,
      "step": 4580
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6074270287892426,
      "learning_rate": 1.936887474703776e-05,
      "loss": 0.3598,
      "step": 4581
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3757337837092695,
      "learning_rate": 1.936852790769299e-05,
      "loss": 0.2953,
      "step": 4582
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5858400354197622,
      "learning_rate": 1.9368180976177564e-05,
      "loss": 0.9291,
      "step": 4583
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3885846129635755,
      "learning_rate": 1.9367833952494886e-05,
      "loss": 0.0782,
      "step": 4584
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35780837532603466,
      "learning_rate": 1.936748683664838e-05,
      "loss": 0.2396,
      "step": 4585
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5711842585309557,
      "learning_rate": 1.9367139628641453e-05,
      "loss": 0.2893,
      "step": 4586
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8824765354712343,
      "learning_rate": 1.9366792328477527e-05,
      "loss": 0.4215,
      "step": 4587
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3867683818315083,
      "learning_rate": 1.9366444936160012e-05,
      "loss": 0.3099,
      "step": 4588
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.33327773961840956,
      "learning_rate": 1.9366097451692333e-05,
      "loss": 0.2117,
      "step": 4589
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5738417160119789,
      "learning_rate": 1.93657498750779e-05,
      "loss": 0.6967,
      "step": 4590
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2186825037907547,
      "learning_rate": 1.9365402206320144e-05,
      "loss": 0.5274,
      "step": 4591
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9072462929046763,
      "learning_rate": 1.936505444542248e-05,
      "loss": 0.8713,
      "step": 4592
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.28274788729388106,
      "learning_rate": 1.9364706592388325e-05,
      "loss": 0.0752,
      "step": 4593
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4664417666703843,
      "learning_rate": 1.936435864722111e-05,
      "loss": 0.3404,
      "step": 4594
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.37872194532704506,
      "learning_rate": 1.9364010609924248e-05,
      "loss": 0.2777,
      "step": 4595
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8829623675208775,
      "learning_rate": 1.936366248050117e-05,
      "loss": 0.5314,
      "step": 4596
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35934214940752335,
      "learning_rate": 1.93633142589553e-05,
      "loss": 0.1964,
      "step": 4597
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4773272707706888,
      "learning_rate": 1.9362965945290062e-05,
      "loss": 0.2867,
      "step": 4598
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5192544102215014,
      "learning_rate": 1.9362617539508884e-05,
      "loss": 0.2404,
      "step": 4599
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48115194401908046,
      "learning_rate": 1.9362269041615197e-05,
      "loss": 0.3047,
      "step": 4600
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5608382008652409,
      "learning_rate": 1.9361920451612428e-05,
      "loss": 0.3884,
      "step": 4601
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38555298452944103,
      "learning_rate": 1.9361571769504e-05,
      "loss": 0.1775,
      "step": 4602
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6530004373262187,
      "learning_rate": 1.936122299529335e-05,
      "loss": 0.3711,
      "step": 4603
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7973535504042361,
      "learning_rate": 1.936087412898391e-05,
      "loss": 0.4404,
      "step": 4604
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3831855882358246,
      "learning_rate": 1.936052517057911e-05,
      "loss": 0.2069,
      "step": 4605
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.32134768878847575,
      "learning_rate": 1.9360176120082383e-05,
      "loss": 0.2556,
      "step": 4606
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2651540908192063,
      "learning_rate": 1.935982697749716e-05,
      "loss": 0.5314,
      "step": 4607
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4620036339913462,
      "learning_rate": 1.9359477742826882e-05,
      "loss": 0.2442,
      "step": 4608
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7885534175202342,
      "learning_rate": 1.9359128416074983e-05,
      "loss": 0.9322,
      "step": 4609
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7704736170260802,
      "learning_rate": 1.9358778997244898e-05,
      "loss": 0.464,
      "step": 4610
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9842445179310295,
      "learning_rate": 1.9358429486340067e-05,
      "loss": 0.5915,
      "step": 4611
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.33433538175676564,
      "learning_rate": 1.9358079883363925e-05,
      "loss": 0.2102,
      "step": 4612
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35976448585722837,
      "learning_rate": 1.9357730188319915e-05,
      "loss": 0.2907,
      "step": 4613
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38156967986610607,
      "learning_rate": 1.9357380401211478e-05,
      "loss": 0.1482,
      "step": 4614
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2264232910581183,
      "learning_rate": 1.935703052204205e-05,
      "loss": 0.0427,
      "step": 4615
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4900405343825373,
      "learning_rate": 1.9356680550815083e-05,
      "loss": 0.2804,
      "step": 4616
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.512831032328289,
      "learning_rate": 1.9356330487534006e-05,
      "loss": 0.2832,
      "step": 4617
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5890909789179245,
      "learning_rate": 1.9355980332202277e-05,
      "loss": 0.4261,
      "step": 4618
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44767882431335754,
      "learning_rate": 1.9355630084823333e-05,
      "loss": 0.2959,
      "step": 4619
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7942726786705857,
      "learning_rate": 1.9355279745400622e-05,
      "loss": 0.4654,
      "step": 4620
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40111356344866833,
      "learning_rate": 1.9354929313937593e-05,
      "loss": 0.2725,
      "step": 4621
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9216899276355579,
      "learning_rate": 1.9354578790437684e-05,
      "loss": 0.5349,
      "step": 4622
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43965918016607125,
      "learning_rate": 1.935422817490436e-05,
      "loss": 0.1255,
      "step": 4623
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3191191357876307,
      "learning_rate": 1.9353877467341055e-05,
      "loss": 0.2654,
      "step": 4624
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4124792664690246,
      "learning_rate": 1.935352666775123e-05,
      "loss": 0.1877,
      "step": 4625
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5224255161986009,
      "learning_rate": 1.9353175776138327e-05,
      "loss": 0.2614,
      "step": 4626
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5800766245111182,
      "learning_rate": 1.9352824792505805e-05,
      "loss": 0.9497,
      "step": 4627
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0003252494923145,
      "learning_rate": 1.9352473716857117e-05,
      "loss": 0.4428,
      "step": 4628
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46089854092310606,
      "learning_rate": 1.9352122549195713e-05,
      "loss": 0.3289,
      "step": 4629
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4370660025363028,
      "learning_rate": 1.935177128952505e-05,
      "loss": 0.2094,
      "step": 4630
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.42553887364035886,
      "learning_rate": 1.935141993784859e-05,
      "loss": 0.3604,
      "step": 4631
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3528690673075765,
      "learning_rate": 1.9351068494169776e-05,
      "loss": 0.1606,
      "step": 4632
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5756182592254396,
      "learning_rate": 1.9350716958492076e-05,
      "loss": 0.2243,
      "step": 4633
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39639652454086316,
      "learning_rate": 1.9350365330818944e-05,
      "loss": 0.08,
      "step": 4634
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46134781757543153,
      "learning_rate": 1.9350013611153846e-05,
      "loss": 0.3278,
      "step": 4635
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.367660700377696,
      "learning_rate": 1.9349661799500232e-05,
      "loss": 0.2896,
      "step": 4636
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0160750951280197,
      "learning_rate": 1.9349309895861574e-05,
      "loss": 0.546,
      "step": 4637
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1755383473886647,
      "learning_rate": 1.9348957900241324e-05,
      "loss": 0.4256,
      "step": 4638
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3144267787244012,
      "learning_rate": 1.9348605812642952e-05,
      "loss": 0.2005,
      "step": 4639
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7702149155702602,
      "learning_rate": 1.934825363306992e-05,
      "loss": 0.8078,
      "step": 4640
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43572554095761973,
      "learning_rate": 1.9347901361525694e-05,
      "loss": 0.0763,
      "step": 4641
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43115315197113463,
      "learning_rate": 1.9347548998013736e-05,
      "loss": 0.3511,
      "step": 4642
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3209990465748396,
      "learning_rate": 1.9347196542537517e-05,
      "loss": 0.1052,
      "step": 4643
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.46459993675585354,
      "learning_rate": 1.93468439951005e-05,
      "loss": 0.3607,
      "step": 4644
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2045581616471615,
      "learning_rate": 1.934649135570616e-05,
      "loss": 0.5312,
      "step": 4645
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9371046712528132,
      "learning_rate": 1.9346138624357962e-05,
      "loss": 0.6064,
      "step": 4646
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3747890577890413,
      "learning_rate": 1.9345785801059376e-05,
      "loss": 0.2959,
      "step": 4647
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3678235531625247,
      "learning_rate": 1.9345432885813877e-05,
      "loss": 0.2862,
      "step": 4648
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3866942500767057,
      "learning_rate": 1.934507987862493e-05,
      "loss": 0.3033,
      "step": 4649
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1734776492320236,
      "learning_rate": 1.934472677949601e-05,
      "loss": 0.6988,
      "step": 4650
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3543815600931937,
      "learning_rate": 1.9344373588430597e-05,
      "loss": 0.0757,
      "step": 4651
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.383955582714573,
      "learning_rate": 1.9344020305432164e-05,
      "loss": 0.2012,
      "step": 4652
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6336808327724608,
      "learning_rate": 1.934366693050418e-05,
      "loss": 0.3834,
      "step": 4653
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3630566405597737,
      "learning_rate": 1.9343313463650132e-05,
      "loss": 0.3007,
      "step": 4654
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.695021486597671,
      "learning_rate": 1.9342959904873485e-05,
      "loss": 0.5212,
      "step": 4655
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5571007169228024,
      "learning_rate": 1.9342606254177728e-05,
      "loss": 0.3075,
      "step": 4656
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0516491129214331,
      "learning_rate": 1.9342252511566336e-05,
      "loss": 0.4893,
      "step": 4657
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4220267651800838,
      "learning_rate": 1.9341898677042793e-05,
      "loss": 0.2711,
      "step": 4658
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4056933067206647,
      "learning_rate": 1.9341544750610574e-05,
      "loss": 0.6959,
      "step": 4659
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2507447145620427,
      "learning_rate": 1.9341190732273162e-05,
      "loss": 0.1924,
      "step": 4660
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.944443899397076,
      "learning_rate": 1.9340836622034045e-05,
      "loss": 0.7656,
      "step": 4661
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3953698806006635,
      "learning_rate": 1.9340482419896702e-05,
      "loss": 0.2191,
      "step": 4662
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8936223219213371,
      "learning_rate": 1.9340128125864618e-05,
      "loss": 0.5326,
      "step": 4663
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.870425910890698,
      "learning_rate": 1.9339773739941284e-05,
      "loss": 0.491,
      "step": 4664
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.39593743604687476,
      "learning_rate": 1.933941926213018e-05,
      "loss": 0.2664,
      "step": 4665
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4989274694030455,
      "learning_rate": 1.93390646924348e-05,
      "loss": 0.2817,
      "step": 4666
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5372821499094251,
      "learning_rate": 1.9338710030858622e-05,
      "loss": 0.2905,
      "step": 4667
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.2410336518685858,
      "learning_rate": 1.9338355277405148e-05,
      "loss": 0.6312,
      "step": 4668
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2649376791619273,
      "learning_rate": 1.933800043207786e-05,
      "loss": 0.078,
      "step": 4669
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41329856983790664,
      "learning_rate": 1.9337645494880252e-05,
      "loss": 0.2456,
      "step": 4670
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4032028040341725,
      "learning_rate": 1.933729046581581e-05,
      "loss": 0.2689,
      "step": 4671
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38379261513638535,
      "learning_rate": 1.933693534488804e-05,
      "loss": 0.3373,
      "step": 4672
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8947291892908021,
      "learning_rate": 1.933658013210042e-05,
      "loss": 0.4558,
      "step": 4673
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6303917335544902,
      "learning_rate": 1.933622482745646e-05,
      "loss": 0.7946,
      "step": 4674
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.352614680073502,
      "learning_rate": 1.933586943095964e-05,
      "loss": 0.1852,
      "step": 4675
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6702160446179332,
      "learning_rate": 1.9335513942613474e-05,
      "loss": 0.3908,
      "step": 4676
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1151945047778002,
      "learning_rate": 1.9335158362421443e-05,
      "loss": 0.375,
      "step": 4677
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.2677822882741263,
      "learning_rate": 1.9334802690387057e-05,
      "loss": 0.2165,
      "step": 4678
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.589430461666161,
      "learning_rate": 1.9334446926513804e-05,
      "loss": 0.3779,
      "step": 4679
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.38874151010570684,
      "learning_rate": 1.9334091070805196e-05,
      "loss": 0.234,
      "step": 4680
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6360102597870778,
      "learning_rate": 1.9333735123264727e-05,
      "loss": 0.4991,
      "step": 4681
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35828215107437705,
      "learning_rate": 1.9333379083895902e-05,
      "loss": 0.1694,
      "step": 4682
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4005559945251265,
      "learning_rate": 1.933302295270222e-05,
      "loss": 0.3266,
      "step": 4683
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3453647052529382,
      "learning_rate": 1.933266672968719e-05,
      "loss": 0.0797,
      "step": 4684
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.45846025680629326,
      "learning_rate": 1.9332310414854313e-05,
      "loss": 0.3414,
      "step": 4685
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.0669935131238166,
      "learning_rate": 1.9331954008207097e-05,
      "loss": 0.0202,
      "step": 4686
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.769907907489017,
      "learning_rate": 1.9331597509749044e-05,
      "loss": 0.4949,
      "step": 4687
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8226865766854072,
      "learning_rate": 1.933124091948366e-05,
      "loss": 0.3998,
      "step": 4688
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.49563149711358434,
      "learning_rate": 1.9330884237414466e-05,
      "loss": 0.2786,
      "step": 4689
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.31581799923725973,
      "learning_rate": 1.933052746354496e-05,
      "loss": 0.2645,
      "step": 4690
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.32138078763670425,
      "learning_rate": 1.9330170597878654e-05,
      "loss": 0.1682,
      "step": 4691
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8161932089116037,
      "learning_rate": 1.9329813640419058e-05,
      "loss": 0.8462,
      "step": 4692
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.31508207369167257,
      "learning_rate": 1.932945659116969e-05,
      "loss": 0.1424,
      "step": 4693
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6422530327972771,
      "learning_rate": 1.9329099450134055e-05,
      "loss": 0.4212,
      "step": 4694
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35765106678147024,
      "learning_rate": 1.932874221731567e-05,
      "loss": 0.2817,
      "step": 4695
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.0841350155350176,
      "learning_rate": 1.9328384892718048e-05,
      "loss": 0.8288,
      "step": 4696
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.971250968036065,
      "learning_rate": 1.9328027476344708e-05,
      "loss": 0.3776,
      "step": 4697
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3970324800091083,
      "learning_rate": 1.9327669968199164e-05,
      "loss": 0.2999,
      "step": 4698
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.44680211095669525,
      "learning_rate": 1.9327312368284935e-05,
      "loss": 0.2181,
      "step": 4699
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.9068222129811483,
      "learning_rate": 1.9326954676605535e-05,
      "loss": 0.8209,
      "step": 4700
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.36539738935866867,
      "learning_rate": 1.9326596893164487e-05,
      "loss": 0.2995,
      "step": 4701
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.48607510520997704,
      "learning_rate": 1.9326239017965313e-05,
      "loss": 0.228,
      "step": 4702
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3770407015616102,
      "learning_rate": 1.9325881051011528e-05,
      "loss": 0.2489,
      "step": 4703
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6255088198992165,
      "learning_rate": 1.9325522992306657e-05,
      "loss": 0.754,
      "step": 4704
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.9495943592527324,
      "learning_rate": 1.932516484185422e-05,
      "loss": 0.4362,
      "step": 4705
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.423475826291647,
      "learning_rate": 1.9324806599657746e-05,
      "loss": 0.28,
      "step": 4706
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.5260663333187058,
      "learning_rate": 1.9324448265720753e-05,
      "loss": 0.3134,
      "step": 4707
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3697121091435745,
      "learning_rate": 1.9324089840046772e-05,
      "loss": 0.2622,
      "step": 4708
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.8361738916093038,
      "learning_rate": 1.932373132263933e-05,
      "loss": 0.8374,
      "step": 4709
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.25340393425400837,
      "learning_rate": 1.932337271350195e-05,
      "loss": 0.1067,
      "step": 4710
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4417852518438132,
      "learning_rate": 1.9323014012638158e-05,
      "loss": 0.3211,
      "step": 4711
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.40636861143495123,
      "learning_rate": 1.9322655220051486e-05,
      "loss": 0.2431,
      "step": 4712
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6231737850627934,
      "learning_rate": 1.9322296335745468e-05,
      "loss": 0.423,
      "step": 4713
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.6617331562410057,
      "learning_rate": 1.932193735972363e-05,
      "loss": 0.3625,
      "step": 4714
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.6863959175016554,
      "learning_rate": 1.9321578291989507e-05,
      "loss": 0.8063,
      "step": 4715
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3811408198829914,
      "learning_rate": 1.932121913254663e-05,
      "loss": 0.1886,
      "step": 4716
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.623417706435917,
      "learning_rate": 1.9320859881398525e-05,
      "loss": 0.4323,
      "step": 4717
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.1205715480522276,
      "learning_rate": 1.9320500538548738e-05,
      "loss": 0.6428,
      "step": 4718
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.35745296167519214,
      "learning_rate": 1.9320141104000803e-05,
      "loss": 0.2663,
      "step": 4719
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4757443166279797,
      "learning_rate": 1.9319781577758248e-05,
      "loss": 0.2095,
      "step": 4720
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.26533386937618697,
      "learning_rate": 1.9319421959824618e-05,
      "loss": 0.183,
      "step": 4721
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.8284641193135909,
      "learning_rate": 1.931906225020345e-05,
      "loss": 0.566,
      "step": 4722
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.7804955508901568,
      "learning_rate": 1.9318702448898277e-05,
      "loss": 0.4168,
      "step": 4723
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.362373942248984,
      "learning_rate": 1.9318342555912645e-05,
      "loss": 0.8473,
      "step": 4724
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3284186091745241,
      "learning_rate": 1.9317982571250093e-05,
      "loss": 0.209,
      "step": 4725
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.41652118790547665,
      "learning_rate": 1.931762249491416e-05,
      "loss": 0.3289,
      "step": 4726
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.7690881458272785,
      "learning_rate": 1.9317262326908393e-05,
      "loss": 0.5402,
      "step": 4727
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.4173730384890444,
      "learning_rate": 1.9316902067236337e-05,
      "loss": 0.7396,
      "step": 4728
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.16370552096637658,
      "learning_rate": 1.931654171590153e-05,
      "loss": 0.0759,
      "step": 4729
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.43400387944504093,
      "learning_rate": 1.931618127290752e-05,
      "loss": 0.3433,
      "step": 4730
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4547641221879374,
      "learning_rate": 1.931582073825785e-05,
      "loss": 0.2956,
      "step": 4731
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.4619887118683139,
      "learning_rate": 1.9315460111956074e-05,
      "loss": 0.3462,
      "step": 4732
    },
    {
      "epoch": 0.14,
      "grad_norm": 1.5542147660263599,
      "learning_rate": 1.9315099394005735e-05,
      "loss": 0.7555,
      "step": 4733
    },
    {
      "epoch": 0.14,
      "grad_norm": 0.3616878846398058,
      "learning_rate": 1.9314738584410382e-05,
      "loss": 0.1851,
      "step": 4734
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.57881337092648,
      "learning_rate": 1.931437768317357e-05,
      "loss": 0.3563,
      "step": 4735
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7855917505150087,
      "learning_rate": 1.931401669029884e-05,
      "loss": 0.7018,
      "step": 4736
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3828426292419453,
      "learning_rate": 1.9313655605789752e-05,
      "loss": 0.3189,
      "step": 4737
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9561716782935747,
      "learning_rate": 1.9313294429649857e-05,
      "loss": 0.281,
      "step": 4738
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5102175346490213,
      "learning_rate": 1.9312933161882704e-05,
      "loss": 0.3074,
      "step": 4739
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6046910659882606,
      "learning_rate": 1.931257180249185e-05,
      "loss": 0.3465,
      "step": 4740
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3181148064921457,
      "learning_rate": 1.9312210351480853e-05,
      "loss": 0.1638,
      "step": 4741
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2167469487188278,
      "learning_rate": 1.9311848808853267e-05,
      "loss": 0.3877,
      "step": 4742
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4132623871123325,
      "learning_rate": 1.9311487174612644e-05,
      "loss": 0.2805,
      "step": 4743
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3340374085198278,
      "learning_rate": 1.9311125448762552e-05,
      "loss": 0.2461,
      "step": 4744
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2016590518417,
      "learning_rate": 1.9310763631306538e-05,
      "loss": 0.6856,
      "step": 4745
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.2618168196704467,
      "learning_rate": 1.9310401722248172e-05,
      "loss": 0.7547,
      "step": 4746
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6522717780108738,
      "learning_rate": 1.931003972159101e-05,
      "loss": 0.3283,
      "step": 4747
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4440070622646401,
      "learning_rate": 1.930967762933861e-05,
      "loss": 0.3133,
      "step": 4748
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45498983149536903,
      "learning_rate": 1.9309315445494542e-05,
      "loss": 0.3092,
      "step": 4749
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6941729689437482,
      "learning_rate": 1.9308953170062365e-05,
      "loss": 0.3395,
      "step": 4750
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5372034932878252,
      "learning_rate": 1.9308590803045645e-05,
      "loss": 0.1933,
      "step": 4751
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7705372401542324,
      "learning_rate": 1.9308228344447943e-05,
      "loss": 0.3635,
      "step": 4752
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43916690011688486,
      "learning_rate": 1.9307865794272826e-05,
      "loss": 0.2756,
      "step": 4753
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9811105703898173,
      "learning_rate": 1.9307503152523867e-05,
      "loss": 1.0818,
      "step": 4754
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43795069760629446,
      "learning_rate": 1.9307140419204628e-05,
      "loss": 0.2786,
      "step": 4755
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7094779257640393,
      "learning_rate": 1.930677759431868e-05,
      "loss": 0.4579,
      "step": 4756
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3660761845499223,
      "learning_rate": 1.930641467786959e-05,
      "loss": 0.2597,
      "step": 4757
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9617411999834107,
      "learning_rate": 1.930605166986093e-05,
      "loss": 0.4002,
      "step": 4758
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43075630953918354,
      "learning_rate": 1.930568857029627e-05,
      "loss": 0.1615,
      "step": 4759
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.338695115503057,
      "learning_rate": 1.9305325379179186e-05,
      "loss": 0.2683,
      "step": 4760
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41486253984014104,
      "learning_rate": 1.930496209651325e-05,
      "loss": 0.2556,
      "step": 4761
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5606412062541541,
      "learning_rate": 1.9304598722302036e-05,
      "loss": 0.2664,
      "step": 4762
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6277230099976943,
      "learning_rate": 1.9304235256549115e-05,
      "loss": 0.8798,
      "step": 4763
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.888982819546261,
      "learning_rate": 1.930387169925807e-05,
      "loss": 0.4292,
      "step": 4764
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.876016421661972,
      "learning_rate": 1.9303508050432468e-05,
      "loss": 0.496,
      "step": 4765
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41342242331340395,
      "learning_rate": 1.9303144310075897e-05,
      "loss": 0.2004,
      "step": 4766
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41356221000441584,
      "learning_rate": 1.930278047819193e-05,
      "loss": 0.3322,
      "step": 4767
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3236755710097475,
      "learning_rate": 1.9302416554784148e-05,
      "loss": 0.196,
      "step": 4768
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5048706103470336,
      "learning_rate": 1.9302052539856132e-05,
      "loss": 0.249,
      "step": 4769
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4279976754402639,
      "learning_rate": 1.930168843341146e-05,
      "loss": 0.1607,
      "step": 4770
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.34164549778445724,
      "learning_rate": 1.930132423545372e-05,
      "loss": 0.2263,
      "step": 4771
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5098631157142046,
      "learning_rate": 1.9300959945986485e-05,
      "loss": 0.3804,
      "step": 4772
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4921675520016433,
      "learning_rate": 1.930059556501335e-05,
      "loss": 0.2994,
      "step": 4773
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8508241623474317,
      "learning_rate": 1.9300231092537896e-05,
      "loss": 0.5177,
      "step": 4774
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3462879788407259,
      "learning_rate": 1.929986652856371e-05,
      "loss": 0.216,
      "step": 4775
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.569046263770133,
      "learning_rate": 1.9299501873094372e-05,
      "loss": 0.7093,
      "step": 4776
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.412175037896325,
      "learning_rate": 1.929913712613348e-05,
      "loss": 0.1216,
      "step": 4777
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.35836044503734527,
      "learning_rate": 1.9298772287684612e-05,
      "loss": 0.2716,
      "step": 4778
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.34916549364342514,
      "learning_rate": 1.9298407357751366e-05,
      "loss": 0.2148,
      "step": 4779
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.97125719536261,
      "learning_rate": 1.9298042336337327e-05,
      "loss": 0.3371,
      "step": 4780
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0945400654505244,
      "learning_rate": 1.929767722344609e-05,
      "loss": 0.5425,
      "step": 4781
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8240338592254969,
      "learning_rate": 1.9297312019081245e-05,
      "loss": 0.5375,
      "step": 4782
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8398263023908328,
      "learning_rate": 1.9296946723246384e-05,
      "loss": 0.4124,
      "step": 4783
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.31740624726564987,
      "learning_rate": 1.92965813359451e-05,
      "loss": 0.2083,
      "step": 4784
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5573033170719983,
      "learning_rate": 1.9296215857180993e-05,
      "loss": 0.3676,
      "step": 4785
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4710326295954545,
      "learning_rate": 1.9295850286957655e-05,
      "loss": 0.2937,
      "step": 4786
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.37955381289976675,
      "learning_rate": 1.9295484625278683e-05,
      "loss": 0.1571,
      "step": 4787
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3685487493659983,
      "learning_rate": 1.9295118872147674e-05,
      "loss": 0.1986,
      "step": 4788
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7272126890708731,
      "learning_rate": 1.9294753027568227e-05,
      "loss": 0.4169,
      "step": 4789
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5501769473901426,
      "learning_rate": 1.9294387091543944e-05,
      "loss": 0.3686,
      "step": 4790
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.595986889443241,
      "learning_rate": 1.9294021064078418e-05,
      "loss": 0.4188,
      "step": 4791
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1542853740465282,
      "learning_rate": 1.929365494517526e-05,
      "loss": 0.51,
      "step": 4792
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.740019013614694,
      "learning_rate": 1.9293288734838064e-05,
      "loss": 0.3262,
      "step": 4793
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.46852333401600993,
      "learning_rate": 1.9292922433070432e-05,
      "loss": 0.2513,
      "step": 4794
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.042586674754215,
      "learning_rate": 1.9292556039875975e-05,
      "loss": 0.2958,
      "step": 4795
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.29769411728975415,
      "learning_rate": 1.9292189555258294e-05,
      "loss": 0.2056,
      "step": 4796
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6325909434475172,
      "learning_rate": 1.9291822979220994e-05,
      "loss": 0.2415,
      "step": 4797
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4090009033577601,
      "learning_rate": 1.929145631176768e-05,
      "loss": 0.3114,
      "step": 4798
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8710614918562232,
      "learning_rate": 1.9291089552901965e-05,
      "loss": 0.4397,
      "step": 4799
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9646540486717539,
      "learning_rate": 1.9290722702627455e-05,
      "loss": 0.5969,
      "step": 4800
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.130518986245823,
      "learning_rate": 1.9290355760947753e-05,
      "loss": 0.4344,
      "step": 4801
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45780748593637033,
      "learning_rate": 1.9289988727866474e-05,
      "loss": 0.2933,
      "step": 4802
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3581551423622533,
      "learning_rate": 1.9289621603387233e-05,
      "loss": 0.2712,
      "step": 4803
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8126835706248015,
      "learning_rate": 1.928925438751364e-05,
      "loss": 0.7529,
      "step": 4804
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.32455537130931644,
      "learning_rate": 1.9288887080249303e-05,
      "loss": 0.0706,
      "step": 4805
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8110993034230277,
      "learning_rate": 1.9288519681597836e-05,
      "loss": 0.5063,
      "step": 4806
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38518163677186806,
      "learning_rate": 1.9288152191562855e-05,
      "loss": 0.2561,
      "step": 4807
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.150204077634296,
      "learning_rate": 1.928778461014798e-05,
      "loss": 0.6411,
      "step": 4808
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3449350970788579,
      "learning_rate": 1.9287416937356825e-05,
      "loss": 0.2975,
      "step": 4809
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.174557138930483,
      "learning_rate": 1.9287049173193004e-05,
      "loss": 0.4947,
      "step": 4810
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4381925171716212,
      "learning_rate": 1.9286681317660138e-05,
      "loss": 0.2679,
      "step": 4811
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5128487171602198,
      "learning_rate": 1.9286313370761846e-05,
      "loss": 0.2383,
      "step": 4812
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2960733985292112,
      "learning_rate": 1.9285945332501748e-05,
      "loss": 0.601,
      "step": 4813
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.272216988270403,
      "learning_rate": 1.928557720288346e-05,
      "loss": 0.2127,
      "step": 4814
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.012891634715137,
      "learning_rate": 1.9285208981910616e-05,
      "loss": 0.4463,
      "step": 4815
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.464761400288994,
      "learning_rate": 1.9284840669586824e-05,
      "loss": 0.2091,
      "step": 4816
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5263584321307209,
      "learning_rate": 1.9284472265915716e-05,
      "loss": 0.4004,
      "step": 4817
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0182400560305602,
      "learning_rate": 1.9284103770900917e-05,
      "loss": 0.6483,
      "step": 4818
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.36853131074861145,
      "learning_rate": 1.928373518454605e-05,
      "loss": 0.239,
      "step": 4819
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2744503893094495,
      "learning_rate": 1.9283366506854737e-05,
      "loss": 0.0783,
      "step": 4820
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4237469672717667,
      "learning_rate": 1.9282997737830614e-05,
      "loss": 0.29,
      "step": 4821
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8909773674163606,
      "learning_rate": 1.9282628877477304e-05,
      "loss": 0.3757,
      "step": 4822
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45375502563438436,
      "learning_rate": 1.9282259925798434e-05,
      "loss": 0.1999,
      "step": 4823
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9436743266548532,
      "learning_rate": 1.9281890882797638e-05,
      "loss": 0.4927,
      "step": 4824
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3347974580388408,
      "learning_rate": 1.928152174847855e-05,
      "loss": 0.2138,
      "step": 4825
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6043504422456704,
      "learning_rate": 1.928115252284479e-05,
      "loss": 0.3753,
      "step": 4826
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45578342700206154,
      "learning_rate": 1.92807832059e-05,
      "loss": 0.2807,
      "step": 4827
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.429907185525814,
      "learning_rate": 1.9280413797647814e-05,
      "loss": 0.222,
      "step": 4828
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5957660876672654,
      "learning_rate": 1.9280044298091862e-05,
      "loss": 0.1586,
      "step": 4829
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47697845258540533,
      "learning_rate": 1.9279674707235777e-05,
      "loss": 0.3114,
      "step": 4830
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3384183129183835,
      "learning_rate": 1.9279305025083204e-05,
      "loss": 0.4226,
      "step": 4831
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42555367215093326,
      "learning_rate": 1.9278935251637773e-05,
      "loss": 0.3544,
      "step": 4832
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8071519968800566,
      "learning_rate": 1.9278565386903122e-05,
      "loss": 0.3727,
      "step": 4833
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4113615657700145,
      "learning_rate": 1.9278195430882893e-05,
      "loss": 0.2802,
      "step": 4834
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9034944231961959,
      "learning_rate": 1.9277825383580726e-05,
      "loss": 0.3898,
      "step": 4835
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2878679826302433,
      "learning_rate": 1.9277455245000255e-05,
      "loss": 0.6709,
      "step": 4836
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.29468831458375455,
      "learning_rate": 1.9277085015145133e-05,
      "loss": 0.2052,
      "step": 4837
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.33703309826241445,
      "learning_rate": 1.9276714694018993e-05,
      "loss": 0.2217,
      "step": 4838
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.656811373821644,
      "learning_rate": 1.927634428162548e-05,
      "loss": 0.3815,
      "step": 4839
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7527379640734813,
      "learning_rate": 1.927597377796824e-05,
      "loss": 0.0811,
      "step": 4840
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9141704123723897,
      "learning_rate": 1.9275603183050922e-05,
      "loss": 0.4931,
      "step": 4841
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7652920879309114,
      "learning_rate": 1.9275232496877163e-05,
      "loss": 0.2793,
      "step": 4842
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2024588148619026,
      "learning_rate": 1.9274861719450616e-05,
      "loss": 0.4527,
      "step": 4843
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3704959230037366,
      "learning_rate": 1.9274490850774928e-05,
      "loss": 0.2555,
      "step": 4844
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5982181141174613,
      "learning_rate": 1.9274119890853747e-05,
      "loss": 0.4184,
      "step": 4845
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2248714291151847,
      "learning_rate": 1.9273748839690726e-05,
      "loss": 0.1207,
      "step": 4846
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.602909447235969,
      "learning_rate": 1.927337769728951e-05,
      "loss": 0.813,
      "step": 4847
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3328642563014097,
      "learning_rate": 1.9273006463653753e-05,
      "loss": 0.2089,
      "step": 4848
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0688072001019817,
      "learning_rate": 1.9272635138787105e-05,
      "loss": 0.4322,
      "step": 4849
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3843062908296132,
      "learning_rate": 1.9272263722693224e-05,
      "loss": 0.3345,
      "step": 4850
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8691151490169277,
      "learning_rate": 1.927189221537576e-05,
      "loss": 0.376,
      "step": 4851
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45198976276550534,
      "learning_rate": 1.927152061683837e-05,
      "loss": 0.3079,
      "step": 4852
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5059570297296889,
      "learning_rate": 1.927114892708471e-05,
      "loss": 0.2572,
      "step": 4853
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5245990452371712,
      "learning_rate": 1.9270777146118437e-05,
      "loss": 0.9497,
      "step": 4854
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3263018255410115,
      "learning_rate": 1.9270405273943206e-05,
      "loss": 0.1965,
      "step": 4855
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.36809339131982755,
      "learning_rate": 1.927003331056268e-05,
      "loss": 0.2536,
      "step": 4856
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.37115786223266006,
      "learning_rate": 1.9269661255980515e-05,
      "loss": 0.2302,
      "step": 4857
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.127647101470919,
      "learning_rate": 1.926928911020037e-05,
      "loss": 0.6633,
      "step": 4858
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0624182769448376,
      "learning_rate": 1.926891687322591e-05,
      "loss": 0.4557,
      "step": 4859
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.220964402877201,
      "learning_rate": 1.9268544545060796e-05,
      "loss": 0.5726,
      "step": 4860
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3830599142356298,
      "learning_rate": 1.926817212570869e-05,
      "loss": 0.1873,
      "step": 4861
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4167312160611465,
      "learning_rate": 1.9267799615173263e-05,
      "loss": 0.2669,
      "step": 4862
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5209656528094412,
      "learning_rate": 1.926742701345817e-05,
      "loss": 0.3713,
      "step": 4863
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3620261276496452,
      "learning_rate": 1.926705432056708e-05,
      "loss": 0.1378,
      "step": 4864
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3781823889590063,
      "learning_rate": 1.926668153650366e-05,
      "loss": 0.1807,
      "step": 4865
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0869499258931494,
      "learning_rate": 1.926630866127158e-05,
      "loss": 0.2422,
      "step": 4866
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0634854091181507,
      "learning_rate": 1.9265935694874504e-05,
      "loss": 0.5499,
      "step": 4867
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.422808845998821,
      "learning_rate": 1.92655626373161e-05,
      "loss": 0.3072,
      "step": 4868
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4283500515452234,
      "learning_rate": 1.9265189488600052e-05,
      "loss": 0.6474,
      "step": 4869
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3839222388751769,
      "learning_rate": 1.9264816248730015e-05,
      "loss": 0.0805,
      "step": 4870
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5055040517928738,
      "learning_rate": 1.9264442917709667e-05,
      "loss": 0.3219,
      "step": 4871
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.780898024878889,
      "learning_rate": 1.926406949554268e-05,
      "loss": 0.6553,
      "step": 4872
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38676495173943404,
      "learning_rate": 1.9263695982232734e-05,
      "loss": 0.3294,
      "step": 4873
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.24801727387774153,
      "learning_rate": 1.9263322377783497e-05,
      "loss": 0.072,
      "step": 4874
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.36352123538877357,
      "learning_rate": 1.9262948682198645e-05,
      "loss": 0.2068,
      "step": 4875
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8255073608187087,
      "learning_rate": 1.9262574895481853e-05,
      "loss": 0.501,
      "step": 4876
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8180473282209957,
      "learning_rate": 1.9262201017636804e-05,
      "loss": 0.3882,
      "step": 4877
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6159244179784669,
      "learning_rate": 1.9261827048667175e-05,
      "loss": 0.7433,
      "step": 4878
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.357657186697144,
      "learning_rate": 1.9261452988576646e-05,
      "loss": 0.2186,
      "step": 4879
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38979643502400657,
      "learning_rate": 1.926107883736889e-05,
      "loss": 0.296,
      "step": 4880
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.359412533879823,
      "learning_rate": 1.9260704595047596e-05,
      "loss": 0.5898,
      "step": 4881
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3653577727841206,
      "learning_rate": 1.9260330261616444e-05,
      "loss": 0.697,
      "step": 4882
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6769506060462461,
      "learning_rate": 1.9259955837079113e-05,
      "loss": 0.3519,
      "step": 4883
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4486532127387991,
      "learning_rate": 1.9259581321439292e-05,
      "loss": 0.2949,
      "step": 4884
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.43366210911535896,
      "learning_rate": 1.925920671470066e-05,
      "loss": 0.2443,
      "step": 4885
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42047722680654764,
      "learning_rate": 1.925883201686691e-05,
      "loss": 0.3207,
      "step": 4886
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3169924240527282,
      "learning_rate": 1.925845722794172e-05,
      "loss": 0.0769,
      "step": 4887
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7558678650306947,
      "learning_rate": 1.9258082347928783e-05,
      "loss": 0.3364,
      "step": 4888
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44508488396802365,
      "learning_rate": 1.9257707376831788e-05,
      "loss": 0.2643,
      "step": 4889
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4323714618478192,
      "learning_rate": 1.9257332314654417e-05,
      "loss": 0.6269,
      "step": 4890
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5558469533462292,
      "learning_rate": 1.9256957161400367e-05,
      "loss": 0.3588,
      "step": 4891
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4375724314290666,
      "learning_rate": 1.925658191707333e-05,
      "loss": 0.3079,
      "step": 4892
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5697164996048574,
      "learning_rate": 1.9256206581676986e-05,
      "loss": 0.3401,
      "step": 4893
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5605131461621131,
      "learning_rate": 1.925583115521504e-05,
      "loss": 0.316,
      "step": 4894
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4142090958595202,
      "learning_rate": 1.9255455637691183e-05,
      "loss": 0.2032,
      "step": 4895
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.44802362619811004,
      "learning_rate": 1.9255080029109107e-05,
      "loss": 0.2935,
      "step": 4896
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1667302186406803,
      "learning_rate": 1.9254704329472508e-05,
      "loss": 0.4377,
      "step": 4897
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3278973993159169,
      "learning_rate": 1.9254328538785084e-05,
      "loss": 0.2365,
      "step": 4898
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5845752914613018,
      "learning_rate": 1.9253952657050533e-05,
      "loss": 0.9437,
      "step": 4899
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0029623170878788,
      "learning_rate": 1.9253576684272547e-05,
      "loss": 0.4308,
      "step": 4900
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0568932595917062,
      "learning_rate": 1.925320062045483e-05,
      "loss": 0.5365,
      "step": 4901
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39462754668111144,
      "learning_rate": 1.9252824465601077e-05,
      "loss": 0.2394,
      "step": 4902
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.347922747264585,
      "learning_rate": 1.9252448219714997e-05,
      "loss": 0.2511,
      "step": 4903
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3055606667111376,
      "learning_rate": 1.9252071882800285e-05,
      "loss": 0.2295,
      "step": 4904
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1709815483800472,
      "learning_rate": 1.9251695454860644e-05,
      "loss": 0.3368,
      "step": 4905
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9037595097443668,
      "learning_rate": 1.925131893589978e-05,
      "loss": 0.5385,
      "step": 4906
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.38144151364468787,
      "learning_rate": 1.92509423259214e-05,
      "loss": 0.2355,
      "step": 4907
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3554187560017545,
      "learning_rate": 1.9250565624929202e-05,
      "loss": 0.586,
      "step": 4908
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5165694197282255,
      "learning_rate": 1.9250188832926896e-05,
      "loss": 0.2981,
      "step": 4909
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6777621657612579,
      "learning_rate": 1.9249811949918187e-05,
      "loss": 0.403,
      "step": 4910
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.379579411523097,
      "learning_rate": 1.9249434975906786e-05,
      "loss": 0.1984,
      "step": 4911
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6296632659512628,
      "learning_rate": 1.92490579108964e-05,
      "loss": 0.432,
      "step": 4912
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.21796963003693862,
      "learning_rate": 1.9248680754890738e-05,
      "loss": 0.0707,
      "step": 4913
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1918071367909342,
      "learning_rate": 1.9248303507893517e-05,
      "loss": 0.6063,
      "step": 4914
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3211037231513025,
      "learning_rate": 1.9247926169908436e-05,
      "loss": 0.234,
      "step": 4915
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4154880144507991,
      "learning_rate": 1.9247548740939217e-05,
      "loss": 0.2545,
      "step": 4916
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.8440880376023634,
      "learning_rate": 1.924717122098957e-05,
      "loss": 0.8677,
      "step": 4917
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8991962692009369,
      "learning_rate": 1.924679361006321e-05,
      "loss": 0.435,
      "step": 4918
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6818974284103945,
      "learning_rate": 1.924641590816385e-05,
      "loss": 0.8455,
      "step": 4919
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3276364319676001,
      "learning_rate": 1.9246038115295213e-05,
      "loss": 0.0769,
      "step": 4920
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4203842340449677,
      "learning_rate": 1.9245660231461007e-05,
      "loss": 0.2966,
      "step": 4921
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.27935334042062687,
      "learning_rate": 1.9245282256664953e-05,
      "loss": 0.2381,
      "step": 4922
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5212923942097341,
      "learning_rate": 1.924490419091077e-05,
      "loss": 0.2363,
      "step": 4923
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0050170702449335,
      "learning_rate": 1.9244526034202184e-05,
      "loss": 0.5197,
      "step": 4924
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.47594029313082076,
      "learning_rate": 1.9244147786542902e-05,
      "loss": 0.3209,
      "step": 4925
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5386749259840562,
      "learning_rate": 1.9243769447936653e-05,
      "loss": 0.3652,
      "step": 4926
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3865600909695862,
      "learning_rate": 1.9243391018387162e-05,
      "loss": 0.3559,
      "step": 4927
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0771700541820866,
      "learning_rate": 1.9243012497898145e-05,
      "loss": 0.4482,
      "step": 4928
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3901438053036767,
      "learning_rate": 1.924263388647333e-05,
      "loss": 0.2005,
      "step": 4929
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5805084757459316,
      "learning_rate": 1.9242255184116442e-05,
      "loss": 0.3975,
      "step": 4930
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2905241959929946,
      "learning_rate": 1.9241876390831207e-05,
      "loss": 0.1239,
      "step": 4931
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.746234030857666,
      "learning_rate": 1.9241497506621353e-05,
      "loss": 0.8085,
      "step": 4932
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3432301824259858,
      "learning_rate": 1.9241118531490603e-05,
      "loss": 0.2503,
      "step": 4933
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.39387359515446174,
      "learning_rate": 1.924073946544269e-05,
      "loss": 0.2913,
      "step": 4934
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1140157537561939,
      "learning_rate": 1.924036030848134e-05,
      "loss": 0.4364,
      "step": 4935
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0298806581368667,
      "learning_rate": 1.9239981060610286e-05,
      "loss": 0.5679,
      "step": 4936
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2824278041447827,
      "learning_rate": 1.9239601721833255e-05,
      "loss": 0.4754,
      "step": 4937
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7346838267216995,
      "learning_rate": 1.9239222292153987e-05,
      "loss": 0.3524,
      "step": 4938
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.29750535331686384,
      "learning_rate": 1.9238842771576207e-05,
      "loss": 0.247,
      "step": 4939
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4648561645179234,
      "learning_rate": 1.9238463160103655e-05,
      "loss": 0.738,
      "step": 4940
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.23687237350556106,
      "learning_rate": 1.923808345774006e-05,
      "loss": 0.1239,
      "step": 4941
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8492854028372572,
      "learning_rate": 1.9237703664489157e-05,
      "loss": 0.3681,
      "step": 4942
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4354642752363457,
      "learning_rate": 1.923732378035469e-05,
      "loss": 0.337,
      "step": 4943
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9697386751061325,
      "learning_rate": 1.923694380534039e-05,
      "loss": 0.4198,
      "step": 4944
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40429720552391624,
      "learning_rate": 1.923656373945e-05,
      "loss": 0.3375,
      "step": 4945
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2973804300812575,
      "learning_rate": 1.9236183582687253e-05,
      "loss": 0.4722,
      "step": 4946
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7466135853566493,
      "learning_rate": 1.9235803335055894e-05,
      "loss": 0.3275,
      "step": 4947
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4142223883178654,
      "learning_rate": 1.9235422996559666e-05,
      "loss": 0.2452,
      "step": 4948
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3669113183968296,
      "learning_rate": 1.9235042567202304e-05,
      "loss": 0.6616,
      "step": 4949
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.244348524129615,
      "learning_rate": 1.9234662046987557e-05,
      "loss": 0.1623,
      "step": 4950
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5182924426664014,
      "learning_rate": 1.9234281435919165e-05,
      "loss": 0.3843,
      "step": 4951
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.336173709197215,
      "learning_rate": 1.9233900734000875e-05,
      "loss": 0.183,
      "step": 4952
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6085544293283149,
      "learning_rate": 1.9233519941236428e-05,
      "loss": 0.4132,
      "step": 4953
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0164395937341038,
      "learning_rate": 1.9233139057629574e-05,
      "loss": 0.4863,
      "step": 4954
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2787652230367568,
      "learning_rate": 1.9232758083184065e-05,
      "loss": 0.115,
      "step": 4955
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4482217236072808,
      "learning_rate": 1.923237701790364e-05,
      "loss": 0.3436,
      "step": 4956
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.31141653059678454,
      "learning_rate": 1.9231995861792054e-05,
      "loss": 0.2389,
      "step": 4957
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.3738459928229327,
      "learning_rate": 1.923161461485305e-05,
      "loss": 0.6533,
      "step": 4958
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2155438427007625,
      "learning_rate": 1.9231233277090386e-05,
      "loss": 0.075,
      "step": 4959
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8430133084317011,
      "learning_rate": 1.9230851848507813e-05,
      "loss": 0.4986,
      "step": 4960
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3247430930716634,
      "learning_rate": 1.9230470329109084e-05,
      "loss": 0.2092,
      "step": 4961
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9075185250840565,
      "learning_rate": 1.923008871889795e-05,
      "loss": 0.6256,
      "step": 4962
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.35176765120908415,
      "learning_rate": 1.9229707017878164e-05,
      "loss": 0.2643,
      "step": 4963
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.319745448914644,
      "learning_rate": 1.9229325226053484e-05,
      "loss": 0.7404,
      "step": 4964
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40729050021719443,
      "learning_rate": 1.9228943343427664e-05,
      "loss": 0.149,
      "step": 4965
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4304735116165624,
      "learning_rate": 1.9228561370004467e-05,
      "loss": 0.3096,
      "step": 4966
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.60027459866289,
      "learning_rate": 1.922817930578764e-05,
      "loss": 0.1665,
      "step": 4967
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5017231679999063,
      "learning_rate": 1.9227797150780956e-05,
      "loss": 0.3369,
      "step": 4968
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5030108188690171,
      "learning_rate": 1.922741490498816e-05,
      "loss": 0.414,
      "step": 4969
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3561304755601514,
      "learning_rate": 1.9227032568413028e-05,
      "loss": 0.1777,
      "step": 4970
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4255661932704321,
      "learning_rate": 1.9226650141059307e-05,
      "loss": 0.2546,
      "step": 4971
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1164754952408675,
      "learning_rate": 1.922626762293077e-05,
      "loss": 0.6838,
      "step": 4972
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4815287820393978,
      "learning_rate": 1.9225885014031176e-05,
      "loss": 0.2905,
      "step": 4973
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40954907817203556,
      "learning_rate": 1.9225502314364287e-05,
      "loss": 0.2001,
      "step": 4974
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45353798106113574,
      "learning_rate": 1.922511952393387e-05,
      "loss": 0.3199,
      "step": 4975
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42047074704437887,
      "learning_rate": 1.9224736642743695e-05,
      "loss": 0.2869,
      "step": 4976
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7992508680231115,
      "learning_rate": 1.922435367079752e-05,
      "loss": 0.4919,
      "step": 4977
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.605504067227892,
      "learning_rate": 1.9223970608099124e-05,
      "loss": 0.3951,
      "step": 4978
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6663849075471429,
      "learning_rate": 1.922358745465227e-05,
      "loss": 0.2933,
      "step": 4979
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3757749622843054,
      "learning_rate": 1.9223204210460725e-05,
      "loss": 0.2659,
      "step": 4980
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3816042061674259,
      "learning_rate": 1.9222820875528267e-05,
      "loss": 0.2828,
      "step": 4981
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.34539559336372644,
      "learning_rate": 1.9222437449858655e-05,
      "loss": 0.1838,
      "step": 4982
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.35626916575512485,
      "learning_rate": 1.9222053933455673e-05,
      "loss": 0.073,
      "step": 4983
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3995233829253778,
      "learning_rate": 1.922167032632309e-05,
      "loss": 0.3075,
      "step": 4984
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4527703553042377,
      "learning_rate": 1.922128662846468e-05,
      "loss": 0.4289,
      "step": 4985
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.690117818507904,
      "learning_rate": 1.9220902839884216e-05,
      "loss": 0.6048,
      "step": 4986
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3989930544005337,
      "learning_rate": 1.9220518960585475e-05,
      "loss": 0.2903,
      "step": 4987
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6840500475761911,
      "learning_rate": 1.922013499057224e-05,
      "loss": 0.3384,
      "step": 4988
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4215758790122503,
      "learning_rate": 1.921975092984828e-05,
      "loss": 0.2627,
      "step": 4989
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.7894469490955578,
      "learning_rate": 1.9219366778417378e-05,
      "loss": 0.8815,
      "step": 4990
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3254008729183663,
      "learning_rate": 1.921898253628331e-05,
      "loss": 0.1183,
      "step": 4991
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.42574827243096597,
      "learning_rate": 1.9218598203449857e-05,
      "loss": 0.3054,
      "step": 4992
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.30252431368627214,
      "learning_rate": 1.9218213779920803e-05,
      "loss": 0.2391,
      "step": 4993
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.117416585190844,
      "learning_rate": 1.9217829265699928e-05,
      "loss": 0.4119,
      "step": 4994
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8173077427537666,
      "learning_rate": 1.921744466079102e-05,
      "loss": 0.5023,
      "step": 4995
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9468714001896055,
      "learning_rate": 1.9217059965197852e-05,
      "loss": 0.3901,
      "step": 4996
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4340698861084751,
      "learning_rate": 1.9216675178924218e-05,
      "loss": 0.2728,
      "step": 4997
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5449090436958934,
      "learning_rate": 1.9216290301973904e-05,
      "loss": 0.2652,
      "step": 4998
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3794591213733395,
      "learning_rate": 1.9215905334350688e-05,
      "loss": 0.3402,
      "step": 4999
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.24380127706508237,
      "learning_rate": 1.9215520276058365e-05,
      "loss": 0.1182,
      "step": 5000
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2268989933007637,
      "learning_rate": 1.9215135127100723e-05,
      "loss": 0.4964,
      "step": 5001
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.45568475142495174,
      "learning_rate": 1.921474988748155e-05,
      "loss": 0.1968,
      "step": 5002
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8183944480428031,
      "learning_rate": 1.921436455720463e-05,
      "loss": 0.444,
      "step": 5003
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.540109582275481,
      "learning_rate": 1.9213979136273766e-05,
      "loss": 0.3148,
      "step": 5004
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.4707985547996187,
      "learning_rate": 1.921359362469274e-05,
      "loss": 0.751,
      "step": 5005
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4292242240772754,
      "learning_rate": 1.921320802246535e-05,
      "loss": 0.0815,
      "step": 5006
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4629670971823607,
      "learning_rate": 1.9212822329595387e-05,
      "loss": 0.2668,
      "step": 5007
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6603637157880013,
      "learning_rate": 1.921243654608665e-05,
      "loss": 0.9319,
      "step": 5008
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.25359633350678407,
      "learning_rate": 1.9212050671942924e-05,
      "loss": 0.0781,
      "step": 5009
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4054199725534929,
      "learning_rate": 1.9211664707168017e-05,
      "loss": 0.3377,
      "step": 5010
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.35140534862700107,
      "learning_rate": 1.921127865176572e-05,
      "loss": 0.2131,
      "step": 5011
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.726733461964332,
      "learning_rate": 1.9210892505739833e-05,
      "loss": 0.5721,
      "step": 5012
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7290465436074663,
      "learning_rate": 1.9210506269094153e-05,
      "loss": 0.424,
      "step": 5013
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.042322819688591,
      "learning_rate": 1.9210119941832485e-05,
      "loss": 0.6352,
      "step": 5014
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41365980323331786,
      "learning_rate": 1.9209733523958623e-05,
      "loss": 0.1768,
      "step": 5015
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.36525085417680214,
      "learning_rate": 1.9209347015476377e-05,
      "loss": 0.3002,
      "step": 5016
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.41459459573597446,
      "learning_rate": 1.920896041638954e-05,
      "loss": 0.2638,
      "step": 5017
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.765251292315028,
      "learning_rate": 1.9208573726701926e-05,
      "loss": 0.8304,
      "step": 5018
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5432974367465041,
      "learning_rate": 1.9208186946417327e-05,
      "loss": 0.2995,
      "step": 5019
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.33958102952856584,
      "learning_rate": 1.920780007553956e-05,
      "loss": 0.2124,
      "step": 5020
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0674473417933334,
      "learning_rate": 1.9207413114072425e-05,
      "loss": 0.5965,
      "step": 5021
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.29726881185723264,
      "learning_rate": 1.9207026062019727e-05,
      "loss": 0.2305,
      "step": 5022
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.6350378837397228,
      "learning_rate": 1.920663891938528e-05,
      "loss": 0.7895,
      "step": 5023
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3564178822423168,
      "learning_rate": 1.920625168617289e-05,
      "loss": 0.2026,
      "step": 5024
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5998995786750445,
      "learning_rate": 1.9205864362386363e-05,
      "loss": 0.3871,
      "step": 5025
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.9890885755026294,
      "learning_rate": 1.920547694802952e-05,
      "loss": 0.5297,
      "step": 5026
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1934334387797603,
      "learning_rate": 1.920508944310616e-05,
      "loss": 0.5901,
      "step": 5027
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3332664380345583,
      "learning_rate": 1.9204701847620102e-05,
      "loss": 0.2735,
      "step": 5028
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.1210719758981054,
      "learning_rate": 1.920431416157516e-05,
      "loss": 0.4361,
      "step": 5029
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3952272118778746,
      "learning_rate": 1.9203926384975145e-05,
      "loss": 0.2946,
      "step": 5030
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3326867518512778,
      "learning_rate": 1.9203538517823876e-05,
      "loss": 0.1773,
      "step": 5031
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.011809826029562,
      "learning_rate": 1.9203150560125164e-05,
      "loss": 0.474,
      "step": 5032
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4447635976554693,
      "learning_rate": 1.920276251188283e-05,
      "loss": 0.1639,
      "step": 5033
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3653358566581077,
      "learning_rate": 1.920237437310069e-05,
      "loss": 0.2867,
      "step": 5034
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3408438543389555,
      "learning_rate": 1.9201986143782564e-05,
      "loss": 0.2699,
      "step": 5035
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.046678499193023,
      "learning_rate": 1.920159782393227e-05,
      "loss": 0.6331,
      "step": 5036
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.6877659721234223,
      "learning_rate": 1.920120941355363e-05,
      "loss": 0.413,
      "step": 5037
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4761542538115695,
      "learning_rate": 1.920082091265046e-05,
      "loss": 0.3132,
      "step": 5038
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.26398454893280676,
      "learning_rate": 1.920043232122659e-05,
      "loss": 0.1451,
      "step": 5039
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.37171858802549107,
      "learning_rate": 1.920004363928584e-05,
      "loss": 0.3388,
      "step": 5040
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5124518461201881,
      "learning_rate": 1.919965486683203e-05,
      "loss": 0.375,
      "step": 5041
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.016076735838869,
      "learning_rate": 1.9199266003868994e-05,
      "loss": 0.5537,
      "step": 5042
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3308120962028379,
      "learning_rate": 1.9198877050400545e-05,
      "loss": 0.2341,
      "step": 5043
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.9303047642975912,
      "learning_rate": 1.919848800643052e-05,
      "loss": 0.8241,
      "step": 5044
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0804746439799704,
      "learning_rate": 1.9198098871962746e-05,
      "loss": 0.4305,
      "step": 5045
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.36937327735818654,
      "learning_rate": 1.9197709647001047e-05,
      "loss": 0.2967,
      "step": 5046
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4604795452357579,
      "learning_rate": 1.9197320331549253e-05,
      "loss": 0.297,
      "step": 5047
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.7176517735997698,
      "learning_rate": 1.9196930925611195e-05,
      "loss": 0.2582,
      "step": 5048
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.3253105774683465,
      "learning_rate": 1.9196541429190704e-05,
      "loss": 0.1406,
      "step": 5049
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.2853498046420941,
      "learning_rate": 1.9196151842291616e-05,
      "loss": 0.5649,
      "step": 5050
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4469175598930297,
      "learning_rate": 1.9195762164917758e-05,
      "loss": 0.3139,
      "step": 5051
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.4280590645165747,
      "learning_rate": 1.9195372397072965e-05,
      "loss": 0.1988,
      "step": 5052
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.40117761811773744,
      "learning_rate": 1.9194982538761072e-05,
      "loss": 0.3339,
      "step": 5053
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.8900383122943099,
      "learning_rate": 1.9194592589985918e-05,
      "loss": 0.4362,
      "step": 5054
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.0822992860928535,
      "learning_rate": 1.9194202550751334e-05,
      "loss": 0.5967,
      "step": 5055
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.2745472636019387,
      "learning_rate": 1.919381242106116e-05,
      "loss": 0.0737,
      "step": 5056
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.5174106296236356,
      "learning_rate": 1.919342220091924e-05,
      "loss": 0.3585,
      "step": 5057
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.24824972428965283,
      "learning_rate": 1.91930318903294e-05,
      "loss": 0.1863,
      "step": 5058
    },
    {
      "epoch": 0.15,
      "grad_norm": 2.058065008780143,
      "learning_rate": 1.9192641489295492e-05,
      "loss": 0.4892,
      "step": 5059
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.5682003681244303,
      "learning_rate": 1.9192250997821352e-05,
      "loss": 0.8777,
      "step": 5060
    },
    {
      "epoch": 0.15,
      "grad_norm": 0.34602498269689186,
      "learning_rate": 1.9191860415910818e-05,
      "loss": 0.2029,
      "step": 5061
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8337060757978552,
      "learning_rate": 1.919146974356774e-05,
      "loss": 0.5554,
      "step": 5062
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.470880199963917,
      "learning_rate": 1.9191078980795956e-05,
      "loss": 0.3312,
      "step": 5063
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5805105811414076,
      "learning_rate": 1.9190688127599316e-05,
      "loss": 0.4307,
      "step": 5064
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5765938705930801,
      "learning_rate": 1.919029718398166e-05,
      "loss": 0.2051,
      "step": 5065
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4113278548182686,
      "learning_rate": 1.9189906149946838e-05,
      "loss": 0.3174,
      "step": 5066
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23332431312976856,
      "learning_rate": 1.9189515025498694e-05,
      "loss": 0.0672,
      "step": 5067
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.810619444121057,
      "learning_rate": 1.9189123810641076e-05,
      "loss": 0.8171,
      "step": 5068
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3789746073260975,
      "learning_rate": 1.9188732505377836e-05,
      "loss": 0.2718,
      "step": 5069
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3725824765293319,
      "learning_rate": 1.918834110971282e-05,
      "loss": 0.2721,
      "step": 5070
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5095166478169919,
      "learning_rate": 1.9187949623649885e-05,
      "loss": 0.2905,
      "step": 5071
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6592522064187274,
      "learning_rate": 1.9187558047192875e-05,
      "loss": 0.4956,
      "step": 5072
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9655074665795381,
      "learning_rate": 1.918716638034565e-05,
      "loss": 0.4464,
      "step": 5073
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3358106745499261,
      "learning_rate": 1.9186774623112052e-05,
      "loss": 0.203,
      "step": 5074
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6563801611342388,
      "learning_rate": 1.9186382775495947e-05,
      "loss": 0.3652,
      "step": 5075
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.36019282535111924,
      "learning_rate": 1.9185990837501185e-05,
      "loss": 0.2557,
      "step": 5076
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.36252224090583884,
      "learning_rate": 1.9185598809131624e-05,
      "loss": 0.2011,
      "step": 5077
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7263183340097539,
      "learning_rate": 1.9185206690391118e-05,
      "loss": 0.3362,
      "step": 5078
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0309577199208657,
      "learning_rate": 1.9184814481283526e-05,
      "loss": 0.495,
      "step": 5079
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3924072013339525,
      "learning_rate": 1.9184422181812707e-05,
      "loss": 0.2888,
      "step": 5080
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4416417304245527,
      "learning_rate": 1.9184029791982518e-05,
      "loss": 0.371,
      "step": 5081
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4567249879436273,
      "learning_rate": 1.9183637311796824e-05,
      "loss": 0.2753,
      "step": 5082
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2039825679913303,
      "learning_rate": 1.9183244741259488e-05,
      "loss": 0.8936,
      "step": 5083
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2986909880741199,
      "learning_rate": 1.9182852080374364e-05,
      "loss": 0.1936,
      "step": 5084
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46566087137297973,
      "learning_rate": 1.9182459329145324e-05,
      "loss": 0.2762,
      "step": 5085
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3919996030289032,
      "learning_rate": 1.9182066487576224e-05,
      "loss": 0.2237,
      "step": 5086
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.37754828116456896,
      "learning_rate": 1.9181673555670933e-05,
      "loss": 0.2477,
      "step": 5087
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4767096533426312,
      "learning_rate": 1.9181280533433314e-05,
      "loss": 0.3163,
      "step": 5088
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6362600088365712,
      "learning_rate": 1.918088742086724e-05,
      "loss": 0.3375,
      "step": 5089
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.47380928376915,
      "learning_rate": 1.9180494217976574e-05,
      "loss": 0.6743,
      "step": 5090
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6077494326965597,
      "learning_rate": 1.9180100924765187e-05,
      "loss": 0.0357,
      "step": 5091
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7898876702885278,
      "learning_rate": 1.917970754123694e-05,
      "loss": 0.4718,
      "step": 5092
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3270287588076221,
      "learning_rate": 1.9179314067395717e-05,
      "loss": 0.2102,
      "step": 5093
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.39191537897623274,
      "learning_rate": 1.9178920503245376e-05,
      "loss": 0.3206,
      "step": 5094
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3793785683469603,
      "learning_rate": 1.91785268487898e-05,
      "loss": 0.1091,
      "step": 5095
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9501999017680935,
      "learning_rate": 1.9178133104032853e-05,
      "loss": 0.4745,
      "step": 5096
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4074081722412617,
      "learning_rate": 1.9177739268978414e-05,
      "loss": 0.2279,
      "step": 5097
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6662597587522854,
      "learning_rate": 1.9177345343630357e-05,
      "loss": 0.4065,
      "step": 5098
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4557909569700699,
      "learning_rate": 1.9176951327992555e-05,
      "loss": 0.2872,
      "step": 5099
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.30131026039613734,
      "learning_rate": 1.9176557222068885e-05,
      "loss": 0.1855,
      "step": 5100
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8042159064285723,
      "learning_rate": 1.917616302586323e-05,
      "loss": 0.3832,
      "step": 5101
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3992548243640558,
      "learning_rate": 1.917576873937946e-05,
      "loss": 0.2327,
      "step": 5102
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.152562226048098,
      "learning_rate": 1.917537436262146e-05,
      "loss": 0.7436,
      "step": 5103
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46250599740762016,
      "learning_rate": 1.9174979895593107e-05,
      "loss": 0.1955,
      "step": 5104
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.453253626598955,
      "learning_rate": 1.9174585338298284e-05,
      "loss": 0.3625,
      "step": 5105
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.33856679730744144,
      "learning_rate": 1.9174190690740867e-05,
      "loss": 0.0766,
      "step": 5106
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.440277028291311,
      "learning_rate": 1.917379595292475e-05,
      "loss": 0.3333,
      "step": 5107
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9724809532401635,
      "learning_rate": 1.9173401124853806e-05,
      "loss": 0.6405,
      "step": 5108
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1519764194588689,
      "learning_rate": 1.9173006206531922e-05,
      "loss": 0.6814,
      "step": 5109
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1243235226864103,
      "learning_rate": 1.9172611197962987e-05,
      "loss": 0.3104,
      "step": 5110
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4724520241126428,
      "learning_rate": 1.9172216099150885e-05,
      "loss": 0.2539,
      "step": 5111
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.313642195106894,
      "learning_rate": 1.9171820910099503e-05,
      "loss": 0.2578,
      "step": 5112
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8730802277180222,
      "learning_rate": 1.9171425630812726e-05,
      "loss": 0.412,
      "step": 5113
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1010076897858931,
      "learning_rate": 1.917103026129445e-05,
      "loss": 0.6539,
      "step": 5114
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3365878893471719,
      "learning_rate": 1.9170634801548557e-05,
      "loss": 0.1532,
      "step": 5115
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5873520395581123,
      "learning_rate": 1.9170239251578945e-05,
      "loss": 0.4002,
      "step": 5116
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.35866945517431464,
      "learning_rate": 1.9169843611389498e-05,
      "loss": 0.2711,
      "step": 5117
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5580712271805437,
      "learning_rate": 1.9169447880984117e-05,
      "loss": 0.2446,
      "step": 5118
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7136008609800203,
      "learning_rate": 1.9169052060366687e-05,
      "loss": 0.2105,
      "step": 5119
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41511087205543923,
      "learning_rate": 1.9168656149541103e-05,
      "loss": 0.2946,
      "step": 5120
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0242106487418836,
      "learning_rate": 1.916826014851127e-05,
      "loss": 0.4267,
      "step": 5121
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3440604601442974,
      "learning_rate": 1.9167864057281072e-05,
      "loss": 0.6833,
      "step": 5122
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3454032059061969,
      "learning_rate": 1.9167467875854414e-05,
      "loss": 0.2772,
      "step": 5123
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7899933853929412,
      "learning_rate": 1.9167071604235187e-05,
      "loss": 0.3393,
      "step": 5124
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4045506227724244,
      "learning_rate": 1.9166675242427296e-05,
      "loss": 0.2462,
      "step": 5125
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6379276472558537,
      "learning_rate": 1.9166278790434636e-05,
      "loss": 0.6171,
      "step": 5126
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41405617365996356,
      "learning_rate": 1.916588224826111e-05,
      "loss": 0.2061,
      "step": 5127
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46050134730065867,
      "learning_rate": 1.916548561591062e-05,
      "loss": 0.2451,
      "step": 5128
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.546318626363936,
      "learning_rate": 1.9165088893387064e-05,
      "loss": 0.3443,
      "step": 5129
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.34615163763609147,
      "learning_rate": 1.9164692080694347e-05,
      "loss": 0.2719,
      "step": 5130
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.774614859677587,
      "learning_rate": 1.9164295177836375e-05,
      "loss": 0.5273,
      "step": 5131
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9524372594716966,
      "learning_rate": 1.9163898184817052e-05,
      "loss": 0.4552,
      "step": 5132
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7075479492432637,
      "learning_rate": 1.9163501101640284e-05,
      "loss": 0.3034,
      "step": 5133
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4311004695976379,
      "learning_rate": 1.9163103928309978e-05,
      "loss": 0.2505,
      "step": 5134
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5437907035875464,
      "learning_rate": 1.9162706664830036e-05,
      "loss": 0.3952,
      "step": 5135
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.21374001164865342,
      "learning_rate": 1.9162309311204373e-05,
      "loss": 0.1321,
      "step": 5136
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7604186957255934,
      "learning_rate": 1.91619118674369e-05,
      "loss": 0.8177,
      "step": 5137
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.36946348273788193,
      "learning_rate": 1.916151433353152e-05,
      "loss": 0.1797,
      "step": 5138
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5759958313049285,
      "learning_rate": 1.9161116709492146e-05,
      "loss": 0.3813,
      "step": 5139
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5567864633077825,
      "learning_rate": 1.9160718995322694e-05,
      "loss": 0.4116,
      "step": 5140
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45248848993646573,
      "learning_rate": 1.9160321191027073e-05,
      "loss": 0.2751,
      "step": 5141
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8925179509772942,
      "learning_rate": 1.91599232966092e-05,
      "loss": 0.5218,
      "step": 5142
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.31793823190606924,
      "learning_rate": 1.9159525312072983e-05,
      "loss": 0.2275,
      "step": 5143
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3321021673418083,
      "learning_rate": 1.9159127237422347e-05,
      "loss": 0.7305,
      "step": 5144
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.18686719655836964,
      "learning_rate": 1.91587290726612e-05,
      "loss": 0.0723,
      "step": 5145
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.492498288173343,
      "learning_rate": 1.9158330817793466e-05,
      "loss": 0.358,
      "step": 5146
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3285204151776123,
      "learning_rate": 1.9157932472823057e-05,
      "loss": 0.1913,
      "step": 5147
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5715148898899,
      "learning_rate": 1.91575340377539e-05,
      "loss": 0.3898,
      "step": 5148
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2307540763939924,
      "learning_rate": 1.9157135512589905e-05,
      "loss": 0.3819,
      "step": 5149
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.9892658223103854,
      "learning_rate": 1.9156736897335e-05,
      "loss": 0.922,
      "step": 5150
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3834368305658514,
      "learning_rate": 1.91563381919931e-05,
      "loss": 0.1712,
      "step": 5151
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3866706638909795,
      "learning_rate": 1.9155939396568135e-05,
      "loss": 0.2576,
      "step": 5152
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5909132572930026,
      "learning_rate": 1.9155540511064026e-05,
      "loss": 0.3755,
      "step": 5153
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.23152927523839997,
      "learning_rate": 1.9155141535484696e-05,
      "loss": 0.1521,
      "step": 5154
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8758256104236549,
      "learning_rate": 1.9154742469834075e-05,
      "loss": 0.5591,
      "step": 5155
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3523454938608896,
      "learning_rate": 1.9154343314116078e-05,
      "loss": 0.1837,
      "step": 5156
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5910341998145671,
      "learning_rate": 1.9153944068334644e-05,
      "loss": 0.4006,
      "step": 5157
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7154163816196895,
      "learning_rate": 1.9153544732493695e-05,
      "loss": 0.3969,
      "step": 5158
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42955878011374116,
      "learning_rate": 1.9153145306597163e-05,
      "loss": 0.3021,
      "step": 5159
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.868591843756661,
      "learning_rate": 1.9152745790648973e-05,
      "loss": 0.5141,
      "step": 5160
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4483242633463516,
      "learning_rate": 1.9152346184653063e-05,
      "loss": 0.2659,
      "step": 5161
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5420145862553238,
      "learning_rate": 1.9151946488613356e-05,
      "loss": 0.2972,
      "step": 5162
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47247621853467114,
      "learning_rate": 1.9151546702533785e-05,
      "loss": 0.3188,
      "step": 5163
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4881033307220414,
      "learning_rate": 1.915114682641829e-05,
      "loss": 0.3069,
      "step": 5164
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3513392734362478,
      "learning_rate": 1.91507468602708e-05,
      "loss": 0.2207,
      "step": 5165
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4843407175124992,
      "learning_rate": 1.9150346804095256e-05,
      "loss": 0.3739,
      "step": 5166
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.28078305041406415,
      "learning_rate": 1.9149946657895583e-05,
      "loss": 0.1194,
      "step": 5167
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8931865652124984,
      "learning_rate": 1.9149546421675728e-05,
      "loss": 0.6663,
      "step": 5168
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5547963075258696,
      "learning_rate": 1.9149146095439622e-05,
      "loss": 0.2964,
      "step": 5169
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3933787662295733,
      "learning_rate": 1.914874567919121e-05,
      "loss": 0.2897,
      "step": 5170
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3817305515863149,
      "learning_rate": 1.9148345172934425e-05,
      "loss": 0.2952,
      "step": 5171
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.839094472153341,
      "learning_rate": 1.914794457667321e-05,
      "loss": 0.7294,
      "step": 5172
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7909712880439052,
      "learning_rate": 1.9147543890411507e-05,
      "loss": 0.4024,
      "step": 5173
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7297467286701864,
      "learning_rate": 1.9147143114153257e-05,
      "loss": 0.3524,
      "step": 5174
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24518986418947653,
      "learning_rate": 1.9146742247902402e-05,
      "loss": 0.1613,
      "step": 5175
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.559252380155616,
      "learning_rate": 1.914634129166289e-05,
      "loss": 0.8073,
      "step": 5176
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3468289454019369,
      "learning_rate": 1.9145940245438657e-05,
      "loss": 0.2573,
      "step": 5177
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43935444018249853,
      "learning_rate": 1.914553910923366e-05,
      "loss": 0.1792,
      "step": 5178
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4674886035417863,
      "learning_rate": 1.914513788305184e-05,
      "loss": 0.3363,
      "step": 5179
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0947888686669538,
      "learning_rate": 1.914473656689714e-05,
      "loss": 0.3783,
      "step": 5180
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0338393831604658,
      "learning_rate": 1.9144335160773516e-05,
      "loss": 0.5397,
      "step": 5181
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38625198846516356,
      "learning_rate": 1.9143933664684912e-05,
      "loss": 0.321,
      "step": 5182
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5032976027425117,
      "learning_rate": 1.9143532078635284e-05,
      "loss": 0.2985,
      "step": 5183
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5535859060631929,
      "learning_rate": 1.9143130402628575e-05,
      "loss": 0.237,
      "step": 5184
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3328720679318837,
      "learning_rate": 1.9142728636668742e-05,
      "loss": 0.148,
      "step": 5185
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0977999623789627,
      "learning_rate": 1.9142326780759733e-05,
      "loss": 0.4833,
      "step": 5186
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7533964376305087,
      "learning_rate": 1.914192483490551e-05,
      "loss": 0.8896,
      "step": 5187
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3505541354345481,
      "learning_rate": 1.914152279911002e-05,
      "loss": 0.182,
      "step": 5188
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.34555955713427583,
      "learning_rate": 1.914112067337722e-05,
      "loss": 0.2837,
      "step": 5189
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9305321322213994,
      "learning_rate": 1.914071845771107e-05,
      "loss": 0.437,
      "step": 5190
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9206162251576095,
      "learning_rate": 1.914031615211552e-05,
      "loss": 0.4203,
      "step": 5191
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1956824759757405,
      "learning_rate": 1.9139913756594533e-05,
      "loss": 0.4861,
      "step": 5192
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2960921691358299,
      "learning_rate": 1.913951127115207e-05,
      "loss": 0.1822,
      "step": 5193
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.34305395883424017,
      "learning_rate": 1.9139108695792084e-05,
      "loss": 0.2445,
      "step": 5194
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45361585214322175,
      "learning_rate": 1.9138706030518543e-05,
      "loss": 0.2277,
      "step": 5195
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8567596414169902,
      "learning_rate": 1.9138303275335403e-05,
      "loss": 0.9346,
      "step": 5196
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.320997686857979,
      "learning_rate": 1.9137900430246627e-05,
      "loss": 0.2105,
      "step": 5197
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9875154639178281,
      "learning_rate": 1.9137497495256183e-05,
      "loss": 0.5584,
      "step": 5198
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9174474830339706,
      "learning_rate": 1.913709447036803e-05,
      "loss": 0.4417,
      "step": 5199
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.043796656894427,
      "learning_rate": 1.9136691355586137e-05,
      "loss": 0.3763,
      "step": 5200
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.32567442425661114,
      "learning_rate": 1.9136288150914467e-05,
      "loss": 0.195,
      "step": 5201
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.46699608329403586,
      "learning_rate": 1.9135884856356988e-05,
      "loss": 0.3171,
      "step": 5202
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.21739926803931445,
      "learning_rate": 1.913548147191767e-05,
      "loss": 0.0774,
      "step": 5203
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.437819526268232,
      "learning_rate": 1.9135077997600473e-05,
      "loss": 0.4855,
      "step": 5204
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5668416473999784,
      "learning_rate": 1.913467443340938e-05,
      "loss": 0.4251,
      "step": 5205
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3082782720395241,
      "learning_rate": 1.913427077934835e-05,
      "loss": 0.2077,
      "step": 5206
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6091792920032668,
      "learning_rate": 1.913386703542136e-05,
      "loss": 0.3613,
      "step": 5207
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7915107748627416,
      "learning_rate": 1.9133463201632376e-05,
      "loss": 0.4665,
      "step": 5208
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.5136849213561976,
      "learning_rate": 1.913305927798538e-05,
      "loss": 0.9824,
      "step": 5209
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43640852324850665,
      "learning_rate": 1.9132655264484345e-05,
      "loss": 0.1396,
      "step": 5210
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4920763679129199,
      "learning_rate": 1.9132251161133238e-05,
      "loss": 0.3573,
      "step": 5211
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24669924649575262,
      "learning_rate": 1.913184696793604e-05,
      "loss": 0.1639,
      "step": 5212
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5251616402230118,
      "learning_rate": 1.9131442684896722e-05,
      "loss": 0.3766,
      "step": 5213
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.984709842906456,
      "learning_rate": 1.9131038312019275e-05,
      "loss": 0.4391,
      "step": 5214
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8168159782986476,
      "learning_rate": 1.9130633849307664e-05,
      "loss": 0.3556,
      "step": 5215
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3910552580542,
      "learning_rate": 1.9130229296765872e-05,
      "loss": 0.2971,
      "step": 5216
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.994126202154379,
      "learning_rate": 1.9129824654397882e-05,
      "loss": 0.4248,
      "step": 5217
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44905742513337626,
      "learning_rate": 1.9129419922207672e-05,
      "loss": 0.3086,
      "step": 5218
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7806808852096434,
      "learning_rate": 1.9129015100199228e-05,
      "loss": 0.188,
      "step": 5219
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38437402350588984,
      "learning_rate": 1.9128610188376525e-05,
      "loss": 0.2946,
      "step": 5220
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.25632439820307124,
      "learning_rate": 1.9128205186743553e-05,
      "loss": 0.1066,
      "step": 5221
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8826604365948816,
      "learning_rate": 1.912780009530429e-05,
      "loss": 0.8768,
      "step": 5222
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5317544915805742,
      "learning_rate": 1.9127394914062733e-05,
      "loss": 0.34,
      "step": 5223
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43620507736675346,
      "learning_rate": 1.912698964302286e-05,
      "loss": 0.2852,
      "step": 5224
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3958644716403739,
      "learning_rate": 1.912658428218866e-05,
      "loss": 0.2961,
      "step": 5225
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2742546566491026,
      "learning_rate": 1.9126178831564123e-05,
      "loss": 0.9652,
      "step": 5226
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1578192579187139,
      "learning_rate": 1.912577329115323e-05,
      "loss": 0.4328,
      "step": 5227
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0513015115337576,
      "learning_rate": 1.912536766095998e-05,
      "loss": 0.5024,
      "step": 5228
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.36058563986523773,
      "learning_rate": 1.9124961940988356e-05,
      "loss": 0.2249,
      "step": 5229
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2913773115548607,
      "learning_rate": 1.912455613124236e-05,
      "loss": 0.1743,
      "step": 5230
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.35068947204095235,
      "learning_rate": 1.9124150231725977e-05,
      "loss": 0.2649,
      "step": 5231
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6318355629797694,
      "learning_rate": 1.91237442424432e-05,
      "loss": 0.4222,
      "step": 5232
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4264593873758672,
      "learning_rate": 1.9123338163398023e-05,
      "loss": 0.2804,
      "step": 5233
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4770989337939122,
      "learning_rate": 1.912293199459445e-05,
      "loss": 0.2897,
      "step": 5234
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.6464199288391774,
      "learning_rate": 1.9122525736036466e-05,
      "loss": 0.9263,
      "step": 5235
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.35705176177356557,
      "learning_rate": 1.912211938772807e-05,
      "loss": 0.2592,
      "step": 5236
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9140935183071236,
      "learning_rate": 1.9121712949673266e-05,
      "loss": 0.5101,
      "step": 5237
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4381807326853674,
      "learning_rate": 1.9121306421876044e-05,
      "loss": 0.2006,
      "step": 5238
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5182365637855713,
      "learning_rate": 1.9120899804340412e-05,
      "loss": 0.3597,
      "step": 5239
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5350381773420068,
      "learning_rate": 1.9120493097070367e-05,
      "loss": 0.246,
      "step": 5240
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43851783188763427,
      "learning_rate": 1.9120086300069907e-05,
      "loss": 0.3908,
      "step": 5241
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41008453124202005,
      "learning_rate": 1.9119679413343037e-05,
      "loss": 0.0804,
      "step": 5242
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3044236554931601,
      "learning_rate": 1.911927243689376e-05,
      "loss": 0.2439,
      "step": 5243
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4624857774409827,
      "learning_rate": 1.9118865370726082e-05,
      "loss": 0.245,
      "step": 5244
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.006901783006388,
      "learning_rate": 1.9118458214844007e-05,
      "loss": 0.3023,
      "step": 5245
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3976934280922955,
      "learning_rate": 1.911805096925154e-05,
      "loss": 0.8317,
      "step": 5246
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3769116336877506,
      "learning_rate": 1.9117643633952684e-05,
      "loss": 0.2009,
      "step": 5247
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.37325184707096054,
      "learning_rate": 1.911723620895145e-05,
      "loss": 0.3108,
      "step": 5248
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.34323276085170706,
      "learning_rate": 1.911682869425185e-05,
      "loss": 0.1783,
      "step": 5249
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7616254900846279,
      "learning_rate": 1.9116421089857886e-05,
      "loss": 0.4987,
      "step": 5250
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5042780552978059,
      "learning_rate": 1.9116013395773573e-05,
      "loss": 0.1728,
      "step": 5251
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43506164954875837,
      "learning_rate": 1.911560561200292e-05,
      "loss": 0.337,
      "step": 5252
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.27296822243926994,
      "learning_rate": 1.9115197738549943e-05,
      "loss": 0.1068,
      "step": 5253
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41879428739255387,
      "learning_rate": 1.9114789775418648e-05,
      "loss": 0.3466,
      "step": 5254
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6252862433678517,
      "learning_rate": 1.911438172261305e-05,
      "loss": 0.1812,
      "step": 5255
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.405616435932017,
      "learning_rate": 1.9113973580137168e-05,
      "loss": 0.3115,
      "step": 5256
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0119505438088192,
      "learning_rate": 1.9113565347995016e-05,
      "loss": 0.3142,
      "step": 5257
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8392763326332591,
      "learning_rate": 1.911315702619061e-05,
      "loss": 0.4081,
      "step": 5258
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5286742887532512,
      "learning_rate": 1.911274861472796e-05,
      "loss": 0.4047,
      "step": 5259
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.31838202557151457,
      "learning_rate": 1.91123401136111e-05,
      "loss": 0.2133,
      "step": 5260
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6903515599502115,
      "learning_rate": 1.9111931522844033e-05,
      "loss": 0.4341,
      "step": 5261
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.31734706936757795,
      "learning_rate": 1.911152284243079e-05,
      "loss": 0.1707,
      "step": 5262
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.49337978695444584,
      "learning_rate": 1.911111407237538e-05,
      "loss": 0.2828,
      "step": 5263
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45661276919544586,
      "learning_rate": 1.911070521268184e-05,
      "loss": 0.0738,
      "step": 5264
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.103041544916632,
      "learning_rate": 1.911029626335418e-05,
      "loss": 0.5402,
      "step": 5265
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.28984523644667654,
      "learning_rate": 1.9109887224396425e-05,
      "loss": 0.2538,
      "step": 5266
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.944589845895908,
      "learning_rate": 1.9109478095812607e-05,
      "loss": 0.5297,
      "step": 5267
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0043535685817995,
      "learning_rate": 1.9109068877606746e-05,
      "loss": 0.4505,
      "step": 5268
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.079412950889729,
      "learning_rate": 1.9108659569782865e-05,
      "loss": 0.9237,
      "step": 5269
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.33593705750296016,
      "learning_rate": 1.9108250172345e-05,
      "loss": 0.2126,
      "step": 5270
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3240422036624776,
      "learning_rate": 1.9107840685297166e-05,
      "loss": 0.2169,
      "step": 5271
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3401583995791778,
      "learning_rate": 1.91074311086434e-05,
      "loss": 0.2474,
      "step": 5272
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.170670098315492,
      "learning_rate": 1.9107021442387733e-05,
      "loss": 0.3793,
      "step": 5273
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43373981858923444,
      "learning_rate": 1.9106611686534188e-05,
      "loss": 0.2359,
      "step": 5274
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8483748045380891,
      "learning_rate": 1.9106201841086803e-05,
      "loss": 0.3817,
      "step": 5275
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1083877256836312,
      "learning_rate": 1.910579190604961e-05,
      "loss": 0.5784,
      "step": 5276
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4255262294611194,
      "learning_rate": 1.9105381881426635e-05,
      "loss": 0.3041,
      "step": 5277
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4557383758150703,
      "learning_rate": 1.910497176722192e-05,
      "loss": 0.3223,
      "step": 5278
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.33311365874070137,
      "learning_rate": 1.91045615634395e-05,
      "loss": 0.2373,
      "step": 5279
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7769040797817566,
      "learning_rate": 1.9104151270083403e-05,
      "loss": 0.8539,
      "step": 5280
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2122862006168964,
      "learning_rate": 1.910374088715767e-05,
      "loss": 0.0758,
      "step": 5281
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0244972843052178,
      "learning_rate": 1.910333041466634e-05,
      "loss": 0.6041,
      "step": 5282
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.36639942113018825,
      "learning_rate": 1.910291985261345e-05,
      "loss": 0.2445,
      "step": 5283
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3674116206072699,
      "learning_rate": 1.910250920100304e-05,
      "loss": 0.2571,
      "step": 5284
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4820358681341577,
      "learning_rate": 1.910209845983915e-05,
      "loss": 0.3799,
      "step": 5285
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1646158987285353,
      "learning_rate": 1.910168762912582e-05,
      "loss": 0.5022,
      "step": 5286
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7041538094554413,
      "learning_rate": 1.9101276708867092e-05,
      "loss": 0.3524,
      "step": 5287
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4108745616362593,
      "learning_rate": 1.9100865699067008e-05,
      "loss": 0.2462,
      "step": 5288
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.237496308387654,
      "learning_rate": 1.9100454599729612e-05,
      "loss": 0.7001,
      "step": 5289
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.24453510616316912,
      "learning_rate": 1.910004341085895e-05,
      "loss": 0.1926,
      "step": 5290
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8848711266018324,
      "learning_rate": 1.9099632132459067e-05,
      "loss": 0.5805,
      "step": 5291
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.34502148182193815,
      "learning_rate": 1.9099220764534008e-05,
      "loss": 0.0765,
      "step": 5292
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4239139288889995,
      "learning_rate": 1.9098809307087823e-05,
      "loss": 0.3457,
      "step": 5293
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.73395285206083,
      "learning_rate": 1.9098397760124557e-05,
      "loss": 0.3962,
      "step": 5294
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38272442101872944,
      "learning_rate": 1.909798612364826e-05,
      "loss": 0.337,
      "step": 5295
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1911708327221526,
      "learning_rate": 1.9097574397662982e-05,
      "loss": 0.5151,
      "step": 5296
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3439695392525903,
      "learning_rate": 1.9097162582172772e-05,
      "loss": 0.2067,
      "step": 5297
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.887089198697536,
      "learning_rate": 1.9096750677181686e-05,
      "loss": 0.7906,
      "step": 5298
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.35577056495756154,
      "learning_rate": 1.9096338682693772e-05,
      "loss": 0.1431,
      "step": 5299
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9396725475315381,
      "learning_rate": 1.9095926598713085e-05,
      "loss": 0.4889,
      "step": 5300
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.40575653463210465,
      "learning_rate": 1.909551442524368e-05,
      "loss": 0.2465,
      "step": 5301
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.32413253831425276,
      "learning_rate": 1.909510216228961e-05,
      "loss": 0.2966,
      "step": 5302
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0420247402074165,
      "learning_rate": 1.9094689809854935e-05,
      "loss": 0.3684,
      "step": 5303
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8141925984196288,
      "learning_rate": 1.9094277367943704e-05,
      "loss": 0.8905,
      "step": 5304
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5944828997452262,
      "learning_rate": 1.9093864836559983e-05,
      "loss": 0.2876,
      "step": 5305
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.38273000222366044,
      "learning_rate": 1.9093452215707828e-05,
      "loss": 0.2843,
      "step": 5306
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47726450432521217,
      "learning_rate": 1.9093039505391297e-05,
      "loss": 0.3103,
      "step": 5307
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6184754195895193,
      "learning_rate": 1.9092626705614453e-05,
      "loss": 0.4274,
      "step": 5308
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6483268141257112,
      "learning_rate": 1.9092213816381355e-05,
      "loss": 0.2927,
      "step": 5309
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44770206752476877,
      "learning_rate": 1.909180083769606e-05,
      "loss": 0.2475,
      "step": 5310
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5972075821647992,
      "learning_rate": 1.9091387769562642e-05,
      "loss": 0.2605,
      "step": 5311
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5214114360293833,
      "learning_rate": 1.9090974611985163e-05,
      "loss": 0.1214,
      "step": 5312
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3879667573051892,
      "learning_rate": 1.909056136496768e-05,
      "loss": 0.3293,
      "step": 5313
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6239627869869236,
      "learning_rate": 1.9090148028514263e-05,
      "loss": 0.2971,
      "step": 5314
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0310644640968087,
      "learning_rate": 1.908973460262898e-05,
      "loss": 0.4326,
      "step": 5315
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.37515358504675406,
      "learning_rate": 1.9089321087315898e-05,
      "loss": 0.2395,
      "step": 5316
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7882987065938856,
      "learning_rate": 1.9088907482579084e-05,
      "loss": 0.5386,
      "step": 5317
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.402978289654539,
      "learning_rate": 1.9088493788422604e-05,
      "loss": 0.3347,
      "step": 5318
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1873293336849027,
      "learning_rate": 1.9088080004850534e-05,
      "loss": 0.5076,
      "step": 5319
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2742701570407147,
      "learning_rate": 1.9087666131866944e-05,
      "loss": 0.2289,
      "step": 5320
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3681272492826908,
      "learning_rate": 1.9087252169475902e-05,
      "loss": 0.2065,
      "step": 5321
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8223021408815623,
      "learning_rate": 1.9086838117681485e-05,
      "loss": 0.5372,
      "step": 5322
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1114350603774452,
      "learning_rate": 1.908642397648776e-05,
      "loss": 0.411,
      "step": 5323
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.47781119339959993,
      "learning_rate": 1.9086009745898814e-05,
      "loss": 0.2897,
      "step": 5324
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3861694022277106,
      "learning_rate": 1.9085595425918712e-05,
      "loss": 0.2867,
      "step": 5325
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8471638030867236,
      "learning_rate": 1.9085181016551528e-05,
      "loss": 0.4223,
      "step": 5326
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1989840107737038,
      "learning_rate": 1.908476651780135e-05,
      "loss": 0.3742,
      "step": 5327
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4715245698523628,
      "learning_rate": 1.9084351929672245e-05,
      "loss": 0.2471,
      "step": 5328
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.28618557338691464,
      "learning_rate": 1.9083937252168295e-05,
      "loss": 0.1769,
      "step": 5329
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8761775232752604,
      "learning_rate": 1.9083522485293583e-05,
      "loss": 0.8279,
      "step": 5330
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3607618879302265,
      "learning_rate": 1.9083107629052192e-05,
      "loss": 0.2574,
      "step": 5331
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8578976736292272,
      "learning_rate": 1.9082692683448196e-05,
      "loss": 0.9307,
      "step": 5332
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3732830561012342,
      "learning_rate": 1.9082277648485682e-05,
      "loss": 0.1879,
      "step": 5333
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6169451088515818,
      "learning_rate": 1.9081862524168732e-05,
      "loss": 0.4187,
      "step": 5334
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1696473703002594,
      "learning_rate": 1.908144731050143e-05,
      "loss": 0.4387,
      "step": 5335
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45282163616712845,
      "learning_rate": 1.908103200748786e-05,
      "loss": 0.3217,
      "step": 5336
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43166179474079647,
      "learning_rate": 1.908061661513211e-05,
      "loss": 0.3058,
      "step": 5337
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4028009528945251,
      "learning_rate": 1.9080201133438264e-05,
      "loss": 0.2463,
      "step": 5338
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3894261076658838,
      "learning_rate": 1.9079785562410414e-05,
      "loss": 0.2061,
      "step": 5339
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.5256152898972581,
      "learning_rate": 1.9079369902052644e-05,
      "loss": 0.622,
      "step": 5340
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0761852686257463,
      "learning_rate": 1.9078954152369046e-05,
      "loss": 0.5544,
      "step": 5341
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.31677659751393883,
      "learning_rate": 1.907853831336371e-05,
      "loss": 0.0793,
      "step": 5342
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.43342036677353013,
      "learning_rate": 1.9078122385040727e-05,
      "loss": 0.3602,
      "step": 5343
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3432048393583656,
      "learning_rate": 1.9077706367404188e-05,
      "loss": 0.3014,
      "step": 5344
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.4342732808753094,
      "learning_rate": 1.9077290260458186e-05,
      "loss": 0.6651,
      "step": 5345
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.29993808338043326,
      "learning_rate": 1.9076874064206818e-05,
      "loss": 0.0834,
      "step": 5346
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.44866419636357513,
      "learning_rate": 1.9076457778654175e-05,
      "loss": 0.28,
      "step": 5347
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.30328742423679633,
      "learning_rate": 1.9076041403804357e-05,
      "loss": 0.2048,
      "step": 5348
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.41356877500804023,
      "learning_rate": 1.9075624939661455e-05,
      "loss": 0.2835,
      "step": 5349
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.0655252888381734,
      "learning_rate": 1.9075208386229567e-05,
      "loss": 0.9297,
      "step": 5350
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3829182589659203,
      "learning_rate": 1.9074791743512793e-05,
      "loss": 0.2429,
      "step": 5351
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.6188387815009118,
      "learning_rate": 1.9074375011515233e-05,
      "loss": 0.3958,
      "step": 5352
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8481441188229332,
      "learning_rate": 1.9073958190240987e-05,
      "loss": 0.3868,
      "step": 5353
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.7427618447837716,
      "learning_rate": 1.9073541279694154e-05,
      "loss": 0.6843,
      "step": 5354
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.34316032149880366,
      "learning_rate": 1.9073124279878833e-05,
      "loss": 0.2372,
      "step": 5355
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.45516407108666246,
      "learning_rate": 1.9072707190799132e-05,
      "loss": 0.2795,
      "step": 5356
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.284632879005274,
      "learning_rate": 1.9072290012459153e-05,
      "loss": 0.1157,
      "step": 5357
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.2028212426248044,
      "learning_rate": 1.9071872744863e-05,
      "loss": 0.8127,
      "step": 5358
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.0935052244375028,
      "learning_rate": 1.9071455388014777e-05,
      "loss": 0.4757,
      "step": 5359
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.9143426960347206,
      "learning_rate": 1.907103794191859e-05,
      "loss": 0.388,
      "step": 5360
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.31130994342563206,
      "learning_rate": 1.9070620406578547e-05,
      "loss": 0.2607,
      "step": 5361
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8381966946041387,
      "learning_rate": 1.907020278199876e-05,
      "loss": 0.406,
      "step": 5362
    },
    {
      "epoch": 0.16,
      "grad_norm": 2.1812695731561678,
      "learning_rate": 1.9069785068183327e-05,
      "loss": 0.6643,
      "step": 5363
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.598996301267937,
      "learning_rate": 1.906936726513637e-05,
      "loss": 0.1882,
      "step": 5364
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5552629374034999,
      "learning_rate": 1.9068949372861993e-05,
      "loss": 0.2866,
      "step": 5365
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2620189433861452,
      "learning_rate": 1.9068531391364306e-05,
      "loss": 0.1341,
      "step": 5366
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.42235454023050206,
      "learning_rate": 1.9068113320647422e-05,
      "loss": 0.3482,
      "step": 5367
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.1340296061165653,
      "learning_rate": 1.906769516071546e-05,
      "loss": 0.4584,
      "step": 5368
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.812043773503385,
      "learning_rate": 1.906727691157252e-05,
      "loss": 0.4004,
      "step": 5369
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.39796498565223265,
      "learning_rate": 1.9066858573222734e-05,
      "loss": 0.2894,
      "step": 5370
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3299802770536662,
      "learning_rate": 1.906644014567021e-05,
      "loss": 0.7617,
      "step": 5371
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.344739497805747,
      "learning_rate": 1.9066021628919064e-05,
      "loss": 0.2723,
      "step": 5372
    },
    {
      "epoch": 0.16,
      "grad_norm": 3.040411600464091,
      "learning_rate": 1.906560302297341e-05,
      "loss": 0.9131,
      "step": 5373
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4047059690496943,
      "learning_rate": 1.9065184327837375e-05,
      "loss": 0.1936,
      "step": 5374
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.2908232177197997,
      "learning_rate": 1.906476554351507e-05,
      "loss": 0.1525,
      "step": 5375
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.7894044606258084,
      "learning_rate": 1.9064346670010626e-05,
      "loss": 0.529,
      "step": 5376
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8564345450979061,
      "learning_rate": 1.906392770732815e-05,
      "loss": 0.4174,
      "step": 5377
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.4706660569769257,
      "learning_rate": 1.906350865547177e-05,
      "loss": 0.3086,
      "step": 5378
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.32831062303289,
      "learning_rate": 1.9063089514445617e-05,
      "loss": 0.2388,
      "step": 5379
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.3838462465908115,
      "learning_rate": 1.9062670284253802e-05,
      "loss": 0.7392,
      "step": 5380
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.8702009076218948,
      "learning_rate": 1.9062250964900454e-05,
      "loss": 0.121,
      "step": 5381
    },
    {
      "epoch": 0.16,
      "grad_norm": 1.2590334659803701,
      "learning_rate": 1.90618315563897e-05,
      "loss": 0.7003,
      "step": 5382
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.31114237955461743,
      "learning_rate": 1.9061412058725665e-05,
      "loss": 0.1995,
      "step": 5383
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5409125762229307,
      "learning_rate": 1.9060992471912478e-05,
      "loss": 0.2445,
      "step": 5384
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.3635551498189094,
      "learning_rate": 1.9060572795954267e-05,
      "loss": 0.2698,
      "step": 5385
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.8480552089716116,
      "learning_rate": 1.9060153030855156e-05,
      "loss": 0.4854,
      "step": 5386
    },
    {
      "epoch": 0.16,
      "grad_norm": 0.5061594863423199,
      "learning_rate": 1.905973317661928e-05,
      "loss": 0.2492,
      "step": 5387
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3940686003582596,
      "learning_rate": 1.9059313233250767e-05,
      "loss": 0.244,
      "step": 5388
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.49983657008879995,
      "learning_rate": 1.905889320075375e-05,
      "loss": 0.2788,
      "step": 5389
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3952455585837849,
      "learning_rate": 1.9058473079132362e-05,
      "loss": 0.2743,
      "step": 5390
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.573422143878864,
      "learning_rate": 1.9058052868390736e-05,
      "loss": 0.6633,
      "step": 5391
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3447068233093513,
      "learning_rate": 1.9057632568533003e-05,
      "loss": 0.1711,
      "step": 5392
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.635641459162935,
      "learning_rate": 1.90572121795633e-05,
      "loss": 0.3891,
      "step": 5393
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4972176604352527,
      "learning_rate": 1.9056791701485764e-05,
      "loss": 0.2638,
      "step": 5394
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9953810078563337,
      "learning_rate": 1.905637113430453e-05,
      "loss": 0.6121,
      "step": 5395
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3439824668060247,
      "learning_rate": 1.905595047802374e-05,
      "loss": 0.1979,
      "step": 5396
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47211609669494076,
      "learning_rate": 1.905552973264753e-05,
      "loss": 0.3396,
      "step": 5397
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.30372980235445,
      "learning_rate": 1.9055108898180037e-05,
      "loss": 0.2099,
      "step": 5398
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1501220331857707,
      "learning_rate": 1.90546879746254e-05,
      "loss": 0.797,
      "step": 5399
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1421540209757874,
      "learning_rate": 1.9054266961987768e-05,
      "loss": 0.3946,
      "step": 5400
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.34262154619025215,
      "learning_rate": 1.9053845860271276e-05,
      "loss": 0.1814,
      "step": 5401
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48166476308046613,
      "learning_rate": 1.905342466948007e-05,
      "loss": 0.3751,
      "step": 5402
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.35019482025282506,
      "learning_rate": 1.9053003389618293e-05,
      "loss": 0.3021,
      "step": 5403
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5532433873330804,
      "learning_rate": 1.9052582020690093e-05,
      "loss": 0.6854,
      "step": 5404
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48338045277328573,
      "learning_rate": 1.9052160562699607e-05,
      "loss": 0.293,
      "step": 5405
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.391140366010394,
      "learning_rate": 1.905173901565099e-05,
      "loss": 0.2763,
      "step": 5406
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.27588202947949236,
      "learning_rate": 1.9051317379548387e-05,
      "loss": 0.109,
      "step": 5407
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4611534827171782,
      "learning_rate": 1.905089565439594e-05,
      "loss": 0.7073,
      "step": 5408
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36709912947249645,
      "learning_rate": 1.9050473840197808e-05,
      "loss": 0.25,
      "step": 5409
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.828436379053437,
      "learning_rate": 1.9050051936958137e-05,
      "loss": 0.3906,
      "step": 5410
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4211211769692202,
      "learning_rate": 1.9049629944681075e-05,
      "loss": 0.2912,
      "step": 5411
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0442658710692125,
      "learning_rate": 1.9049207863370776e-05,
      "loss": 0.6261,
      "step": 5412
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8576587586835045,
      "learning_rate": 1.9048785693031392e-05,
      "loss": 0.4584,
      "step": 5413
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.30857810893031407,
      "learning_rate": 1.904836343366708e-05,
      "loss": 0.2288,
      "step": 5414
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.49391365763834344,
      "learning_rate": 1.9047941085281987e-05,
      "loss": 0.3724,
      "step": 5415
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.231051737057049,
      "learning_rate": 1.9047518647880278e-05,
      "loss": 0.3278,
      "step": 5416
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.34499418582769953,
      "learning_rate": 1.90470961214661e-05,
      "loss": 0.1668,
      "step": 5417
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7900942513652877,
      "learning_rate": 1.904667350604361e-05,
      "loss": 0.4101,
      "step": 5418
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.46439724857959747,
      "learning_rate": 1.9046250801616976e-05,
      "loss": 0.2882,
      "step": 5419
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5450151340811695,
      "learning_rate": 1.9045828008190344e-05,
      "loss": 0.302,
      "step": 5420
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3655160802180227,
      "learning_rate": 1.9045405125767885e-05,
      "loss": 0.3266,
      "step": 5421
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0999310150209927,
      "learning_rate": 1.904498215435375e-05,
      "loss": 0.4868,
      "step": 5422
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9312013980979953,
      "learning_rate": 1.9044559093952103e-05,
      "loss": 0.5105,
      "step": 5423
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41766085122431373,
      "learning_rate": 1.9044135944567116e-05,
      "loss": 0.195,
      "step": 5424
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3756514041279413,
      "learning_rate": 1.9043712706202933e-05,
      "loss": 0.2444,
      "step": 5425
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.25411996876975423,
      "learning_rate": 1.9043289378863737e-05,
      "loss": 0.2077,
      "step": 5426
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9426572281655522,
      "learning_rate": 1.9042865962553677e-05,
      "loss": 0.4127,
      "step": 5427
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5683668728316541,
      "learning_rate": 1.9042442457276934e-05,
      "loss": 0.3661,
      "step": 5428
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3577415288180445,
      "learning_rate": 1.9042018863037663e-05,
      "loss": 0.3018,
      "step": 5429
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.490458001228684,
      "learning_rate": 1.9041595179840033e-05,
      "loss": 0.8243,
      "step": 5430
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.293333914721487,
      "learning_rate": 1.904117140768822e-05,
      "loss": 0.5391,
      "step": 5431
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36699179852797326,
      "learning_rate": 1.9040747546586384e-05,
      "loss": 0.3341,
      "step": 5432
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.33338213335232253,
      "learning_rate": 1.90403235965387e-05,
      "loss": 0.2064,
      "step": 5433
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.8513213949165943,
      "learning_rate": 1.9039899557549335e-05,
      "loss": 0.8728,
      "step": 5434
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.21932675434146157,
      "learning_rate": 1.9039475429622464e-05,
      "loss": 0.0729,
      "step": 5435
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8253046855574939,
      "learning_rate": 1.9039051212762264e-05,
      "loss": 0.5215,
      "step": 5436
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.53534092361582,
      "learning_rate": 1.90386269069729e-05,
      "loss": 0.2867,
      "step": 5437
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3399426917555765,
      "learning_rate": 1.9038202512258552e-05,
      "loss": 0.2855,
      "step": 5438
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4590999002527479,
      "learning_rate": 1.9037778028623394e-05,
      "loss": 0.2823,
      "step": 5439
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0581083525499706,
      "learning_rate": 1.9037353456071602e-05,
      "loss": 0.5126,
      "step": 5440
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3042269310252075,
      "learning_rate": 1.903692879460735e-05,
      "loss": 0.8293,
      "step": 5441
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3406915890480035,
      "learning_rate": 1.9036504044234823e-05,
      "loss": 0.2196,
      "step": 5442
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.275544944632749,
      "learning_rate": 1.9036079204958193e-05,
      "loss": 0.8481,
      "step": 5443
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.22029273852616837,
      "learning_rate": 1.9035654276781645e-05,
      "loss": 0.1685,
      "step": 5444
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9851751289949197,
      "learning_rate": 1.9035229259709354e-05,
      "loss": 0.5348,
      "step": 5445
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42838877523960545,
      "learning_rate": 1.9034804153745506e-05,
      "loss": 0.2487,
      "step": 5446
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.625187193826307,
      "learning_rate": 1.903437895889428e-05,
      "loss": 0.4048,
      "step": 5447
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4154626698460997,
      "learning_rate": 1.9033953675159865e-05,
      "loss": 0.3105,
      "step": 5448
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6311323554829293,
      "learning_rate": 1.9033528302546437e-05,
      "loss": 0.9359,
      "step": 5449
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.325008728608964,
      "learning_rate": 1.9033102841058188e-05,
      "loss": 0.2547,
      "step": 5450
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.12260199920691,
      "learning_rate": 1.90326772906993e-05,
      "loss": 0.6149,
      "step": 5451
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4114224921381841,
      "learning_rate": 1.903225165147396e-05,
      "loss": 0.2729,
      "step": 5452
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8313361225025206,
      "learning_rate": 1.903182592338636e-05,
      "loss": 0.5368,
      "step": 5453
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3883824480395531,
      "learning_rate": 1.9031400106440683e-05,
      "loss": 0.2545,
      "step": 5454
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2837531940141194,
      "learning_rate": 1.903097420064112e-05,
      "loss": 0.1826,
      "step": 5455
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4068621765227471,
      "learning_rate": 1.9030548205991858e-05,
      "loss": 0.3028,
      "step": 5456
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4162947989847886,
      "learning_rate": 1.9030122122497095e-05,
      "loss": 0.2649,
      "step": 5457
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.6519737737689115,
      "learning_rate": 1.9029695950161016e-05,
      "loss": 0.8213,
      "step": 5458
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9099826574819466,
      "learning_rate": 1.902926968898782e-05,
      "loss": 0.5287,
      "step": 5459
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4727336153251069,
      "learning_rate": 1.90288433389817e-05,
      "loss": 0.2802,
      "step": 5460
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5407108837916144,
      "learning_rate": 1.9028416900146842e-05,
      "loss": 0.2836,
      "step": 5461
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3758792352047178,
      "learning_rate": 1.902799037248745e-05,
      "loss": 0.3114,
      "step": 5462
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6886583465649719,
      "learning_rate": 1.902756375600772e-05,
      "loss": 0.4083,
      "step": 5463
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.37201902437912227,
      "learning_rate": 1.9027137050711846e-05,
      "loss": 0.1548,
      "step": 5464
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3658033948057156,
      "learning_rate": 1.9026710256604026e-05,
      "loss": 0.2344,
      "step": 5465
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.58385328089288,
      "learning_rate": 1.9026283373688464e-05,
      "loss": 0.7377,
      "step": 5466
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0272144538648993,
      "learning_rate": 1.9025856401969355e-05,
      "loss": 0.612,
      "step": 5467
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3461038973811328,
      "learning_rate": 1.90254293414509e-05,
      "loss": 0.263,
      "step": 5468
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43878249353212345,
      "learning_rate": 1.90250021921373e-05,
      "loss": 0.2552,
      "step": 5469
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5398338098007405,
      "learning_rate": 1.902457495403276e-05,
      "loss": 0.2377,
      "step": 5470
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6920254697291014,
      "learning_rate": 1.902414762714148e-05,
      "loss": 0.5007,
      "step": 5471
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.33593081872736513,
      "learning_rate": 1.9023720211467667e-05,
      "loss": 0.2236,
      "step": 5472
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.35117105737646204,
      "learning_rate": 1.902329270701553e-05,
      "loss": 0.3288,
      "step": 5473
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4173966359710735,
      "learning_rate": 1.9022865113789266e-05,
      "loss": 0.1965,
      "step": 5474
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.39693119283734346,
      "learning_rate": 1.9022437431793087e-05,
      "loss": 0.244,
      "step": 5475
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0071425282894655,
      "learning_rate": 1.9022009661031203e-05,
      "loss": 0.5434,
      "step": 5476
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0173011438738495,
      "learning_rate": 1.9021581801507815e-05,
      "loss": 0.6406,
      "step": 5477
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3904181598822159,
      "learning_rate": 1.9021153853227138e-05,
      "loss": 0.0759,
      "step": 5478
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4109299263415069,
      "learning_rate": 1.9020725816193383e-05,
      "loss": 0.3337,
      "step": 5479
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3886359566028988,
      "learning_rate": 1.9020297690410755e-05,
      "loss": 0.2964,
      "step": 5480
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8869047114971748,
      "learning_rate": 1.9019869475883475e-05,
      "loss": 0.0531,
      "step": 5481
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5908955456805736,
      "learning_rate": 1.9019441172615748e-05,
      "loss": 0.2941,
      "step": 5482
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3689063121844365,
      "learning_rate": 1.9019012780611793e-05,
      "loss": 0.2094,
      "step": 5483
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.47645586313992194,
      "learning_rate": 1.901858429987582e-05,
      "loss": 0.2481,
      "step": 5484
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43614690695057157,
      "learning_rate": 1.901815573041205e-05,
      "loss": 0.2997,
      "step": 5485
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5525921167101814,
      "learning_rate": 1.9017727072224693e-05,
      "loss": 0.3979,
      "step": 5486
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5639247965335983,
      "learning_rate": 1.901729832531797e-05,
      "loss": 0.2491,
      "step": 5487
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4669919881227505,
      "learning_rate": 1.9016869489696102e-05,
      "loss": 0.3439,
      "step": 5488
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8347813031013077,
      "learning_rate": 1.9016440565363307e-05,
      "loss": 0.3977,
      "step": 5489
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.669706714023701,
      "learning_rate": 1.9016011552323797e-05,
      "loss": 0.7495,
      "step": 5490
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3091894282536558,
      "learning_rate": 1.90155824505818e-05,
      "loss": 0.1816,
      "step": 5491
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38960365823855925,
      "learning_rate": 1.901515326014154e-05,
      "loss": 0.2877,
      "step": 5492
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2634942986677876,
      "learning_rate": 1.9014723981007233e-05,
      "loss": 0.1827,
      "step": 5493
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9593288231461926,
      "learning_rate": 1.9014294613183107e-05,
      "loss": 0.5919,
      "step": 5494
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8841970901213851,
      "learning_rate": 1.9013865156673382e-05,
      "loss": 0.5524,
      "step": 5495
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.383932727908597,
      "learning_rate": 1.9013435611482286e-05,
      "loss": 0.2408,
      "step": 5496
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6363675780107114,
      "learning_rate": 1.9013005977614046e-05,
      "loss": 0.3951,
      "step": 5497
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36563677948056944,
      "learning_rate": 1.901257625507289e-05,
      "loss": 0.3013,
      "step": 5498
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1634366249382995,
      "learning_rate": 1.901214644386304e-05,
      "loss": 0.6782,
      "step": 5499
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2759197696988375,
      "learning_rate": 1.9011716543988726e-05,
      "loss": 0.1064,
      "step": 5500
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1417037189383077,
      "learning_rate": 1.9011286555454185e-05,
      "loss": 0.5156,
      "step": 5501
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2936231533940475,
      "learning_rate": 1.9010856478263636e-05,
      "loss": 0.1939,
      "step": 5502
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5282313366639916,
      "learning_rate": 1.901042631242132e-05,
      "loss": 0.434,
      "step": 5503
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5786850354610441,
      "learning_rate": 1.9009996057931465e-05,
      "loss": 0.342,
      "step": 5504
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0707805032546545,
      "learning_rate": 1.9009565714798303e-05,
      "loss": 0.631,
      "step": 5505
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.34059804394772053,
      "learning_rate": 1.9009135283026067e-05,
      "loss": 0.2051,
      "step": 5506
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.05023975311101,
      "learning_rate": 1.9008704762618996e-05,
      "loss": 0.3827,
      "step": 5507
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1665537011758846,
      "learning_rate": 1.900827415358132e-05,
      "loss": 0.5764,
      "step": 5508
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3362577351053896,
      "learning_rate": 1.9007843455917282e-05,
      "loss": 0.2508,
      "step": 5509
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.49461717084873114,
      "learning_rate": 1.900741266963112e-05,
      "loss": 0.277,
      "step": 5510
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24196282027971042,
      "learning_rate": 1.900698179472706e-05,
      "loss": 0.1392,
      "step": 5511
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0870113823345546,
      "learning_rate": 1.9006550831209354e-05,
      "loss": 0.6122,
      "step": 5512
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9159848080050536,
      "learning_rate": 1.9006119779082236e-05,
      "loss": 0.4104,
      "step": 5513
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6604401877726351,
      "learning_rate": 1.9005688638349952e-05,
      "loss": 0.3892,
      "step": 5514
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2931900720719076,
      "learning_rate": 1.9005257409016735e-05,
      "loss": 0.2326,
      "step": 5515
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5043733363649748,
      "learning_rate": 1.9004826091086836e-05,
      "loss": 0.2875,
      "step": 5516
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.375390581773504,
      "learning_rate": 1.900439468456449e-05,
      "loss": 0.3438,
      "step": 5517
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2865931381185274,
      "learning_rate": 1.900396318945395e-05,
      "loss": 0.6994,
      "step": 5518
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.27403853213363955,
      "learning_rate": 1.9003531605759458e-05,
      "loss": 0.1449,
      "step": 5519
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4796536236773069,
      "learning_rate": 1.9003099933485258e-05,
      "loss": 0.2547,
      "step": 5520
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5110914068569795,
      "learning_rate": 1.90026681726356e-05,
      "loss": 0.4168,
      "step": 5521
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5736448318691668,
      "learning_rate": 1.9002236323214726e-05,
      "loss": 0.3808,
      "step": 5522
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6932596928211334,
      "learning_rate": 1.9001804385226894e-05,
      "loss": 0.3564,
      "step": 5523
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38995186930228637,
      "learning_rate": 1.9001372358676346e-05,
      "loss": 0.258,
      "step": 5524
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.484731861356747,
      "learning_rate": 1.9000940243567335e-05,
      "loss": 0.2951,
      "step": 5525
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.589797195366799,
      "learning_rate": 1.9000508039904116e-05,
      "loss": 0.0542,
      "step": 5526
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36298278336479417,
      "learning_rate": 1.900007574769093e-05,
      "loss": 0.3145,
      "step": 5527
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38980863612052924,
      "learning_rate": 1.8999643366932045e-05,
      "loss": 0.0753,
      "step": 5528
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38625253193843545,
      "learning_rate": 1.89992108976317e-05,
      "loss": 0.3243,
      "step": 5529
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6762905475039345,
      "learning_rate": 1.8998778339794163e-05,
      "loss": 0.4432,
      "step": 5530
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8288927298446818,
      "learning_rate": 1.8998345693423682e-05,
      "loss": 0.6362,
      "step": 5531
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.32112352868358224,
      "learning_rate": 1.8997912958524512e-05,
      "loss": 0.1099,
      "step": 5532
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3476845141274063,
      "learning_rate": 1.8997480135100916e-05,
      "loss": 0.212,
      "step": 5533
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3144405300440139,
      "learning_rate": 1.8997047223157152e-05,
      "loss": 0.264,
      "step": 5534
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.064119536294067,
      "learning_rate": 1.8996614222697473e-05,
      "loss": 0.464,
      "step": 5535
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4168240324648103,
      "learning_rate": 1.8996181133726147e-05,
      "loss": 0.7049,
      "step": 5536
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3415241549980413,
      "learning_rate": 1.899574795624743e-05,
      "loss": 0.1256,
      "step": 5537
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4807503178589569,
      "learning_rate": 1.8995314690265583e-05,
      "loss": 0.3407,
      "step": 5538
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.35605116910426016,
      "learning_rate": 1.8994881335784868e-05,
      "loss": 0.2927,
      "step": 5539
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2595716565392308,
      "learning_rate": 1.8994447892809556e-05,
      "loss": 0.6475,
      "step": 5540
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5931201020177476,
      "learning_rate": 1.8994014361343903e-05,
      "loss": 0.303,
      "step": 5541
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40539706865070846,
      "learning_rate": 1.899358074139218e-05,
      "loss": 0.2987,
      "step": 5542
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.29263505519104693,
      "learning_rate": 1.8993147032958646e-05,
      "loss": 0.0757,
      "step": 5543
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.172895853997447,
      "learning_rate": 1.8992713236047575e-05,
      "loss": 0.7611,
      "step": 5544
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.399453655338659,
      "learning_rate": 1.899227935066323e-05,
      "loss": 0.2558,
      "step": 5545
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36311641481390017,
      "learning_rate": 1.8991845376809884e-05,
      "loss": 0.2017,
      "step": 5546
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5441045856040495,
      "learning_rate": 1.8991411314491806e-05,
      "loss": 0.3992,
      "step": 5547
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4933332134011903,
      "learning_rate": 1.8990977163713263e-05,
      "loss": 0.4178,
      "step": 5548
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5438270191078296,
      "learning_rate": 1.8990542924478527e-05,
      "loss": 0.959,
      "step": 5549
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.38176167608534933,
      "learning_rate": 1.8990108596791874e-05,
      "loss": 0.261,
      "step": 5550
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5464752438254967,
      "learning_rate": 1.8989674180657573e-05,
      "loss": 0.3614,
      "step": 5551
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2598931232305183,
      "learning_rate": 1.89892396760799e-05,
      "loss": 0.1839,
      "step": 5552
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4863866766389686,
      "learning_rate": 1.8988805083063128e-05,
      "loss": 0.2505,
      "step": 5553
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.678503338128909,
      "learning_rate": 1.8988370401611535e-05,
      "loss": 0.4041,
      "step": 5554
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9711502250425674,
      "learning_rate": 1.8987935631729397e-05,
      "loss": 0.3961,
      "step": 5555
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.33265590704515247,
      "learning_rate": 1.898750077342099e-05,
      "loss": 0.2598,
      "step": 5556
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.37351706826881037,
      "learning_rate": 1.898706582669059e-05,
      "loss": 0.3348,
      "step": 5557
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9688418571760696,
      "learning_rate": 1.8986630791542485e-05,
      "loss": 0.5362,
      "step": 5558
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3220595865470985,
      "learning_rate": 1.8986195667980947e-05,
      "loss": 0.4024,
      "step": 5559
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4675236916019888,
      "learning_rate": 1.898576045601026e-05,
      "loss": 0.2919,
      "step": 5560
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.23062608353989572,
      "learning_rate": 1.8985325155634702e-05,
      "loss": 0.1287,
      "step": 5561
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5421987816331737,
      "learning_rate": 1.898488976685856e-05,
      "loss": 0.3856,
      "step": 5562
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4935599554220004,
      "learning_rate": 1.8984454289686117e-05,
      "loss": 0.2787,
      "step": 5563
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6801081943094697,
      "learning_rate": 1.8984018724121656e-05,
      "loss": 0.4372,
      "step": 5564
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3610642894658721,
      "learning_rate": 1.8983583070169462e-05,
      "loss": 0.2495,
      "step": 5565
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9801785915818805,
      "learning_rate": 1.8983147327833822e-05,
      "loss": 0.5477,
      "step": 5566
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.25436165344104,
      "learning_rate": 1.8982711497119023e-05,
      "loss": 0.5204,
      "step": 5567
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5576758279805386,
      "learning_rate": 1.8982275578029355e-05,
      "loss": 0.426,
      "step": 5568
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.30810136430469154,
      "learning_rate": 1.89818395705691e-05,
      "loss": 0.2033,
      "step": 5569
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.33654235700960894,
      "learning_rate": 1.8981403474742555e-05,
      "loss": 0.2337,
      "step": 5570
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.25296445038837434,
      "learning_rate": 1.8980967290554005e-05,
      "loss": 0.0713,
      "step": 5571
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9412332342863726,
      "learning_rate": 1.8980531018007745e-05,
      "loss": 0.414,
      "step": 5572
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7523772851756823,
      "learning_rate": 1.8980094657108067e-05,
      "loss": 0.4101,
      "step": 5573
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3743540851210223,
      "learning_rate": 1.8979658207859262e-05,
      "loss": 0.2532,
      "step": 5574
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5529949533576062,
      "learning_rate": 1.8979221670265625e-05,
      "loss": 0.3597,
      "step": 5575
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4629472574007185,
      "learning_rate": 1.8978785044331446e-05,
      "loss": 0.322,
      "step": 5576
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5653328447924566,
      "learning_rate": 1.897834833006103e-05,
      "loss": 0.7578,
      "step": 5577
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4626066601753923,
      "learning_rate": 1.897791152745867e-05,
      "loss": 0.0807,
      "step": 5578
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43630506017410686,
      "learning_rate": 1.897747463652866e-05,
      "loss": 0.306,
      "step": 5579
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2653035245307886,
      "learning_rate": 1.89770376572753e-05,
      "loss": 0.1808,
      "step": 5580
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5772011321714798,
      "learning_rate": 1.8976600589702892e-05,
      "loss": 0.4016,
      "step": 5581
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9691627897969038,
      "learning_rate": 1.8976163433815732e-05,
      "loss": 0.3918,
      "step": 5582
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4968664609915387,
      "learning_rate": 1.8975726189618123e-05,
      "loss": 0.2843,
      "step": 5583
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45511506231764354,
      "learning_rate": 1.8975288857114365e-05,
      "loss": 0.2941,
      "step": 5584
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9335894361229143,
      "learning_rate": 1.8974851436308764e-05,
      "loss": 0.5262,
      "step": 5585
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5597268228702367,
      "learning_rate": 1.897441392720562e-05,
      "loss": 0.3972,
      "step": 5586
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2890735313774368,
      "learning_rate": 1.897397632980924e-05,
      "loss": 0.2166,
      "step": 5587
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6085894066423174,
      "learning_rate": 1.897353864412393e-05,
      "loss": 0.3663,
      "step": 5588
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3710455743529032,
      "learning_rate": 1.8973100870153992e-05,
      "loss": 0.1874,
      "step": 5589
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0784460876461601,
      "learning_rate": 1.8972663007903733e-05,
      "loss": 0.6502,
      "step": 5590
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6401730231105943,
      "learning_rate": 1.8972225057377466e-05,
      "loss": 0.3063,
      "step": 5591
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4065516587476024,
      "learning_rate": 1.8971787018579496e-05,
      "loss": 0.2837,
      "step": 5592
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.37058046974915587,
      "learning_rate": 1.897134889151413e-05,
      "loss": 0.2697,
      "step": 5593
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.7406687998672592,
      "learning_rate": 1.897091067618569e-05,
      "loss": 0.7876,
      "step": 5594
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9997701190534029,
      "learning_rate": 1.897047237259847e-05,
      "loss": 0.4829,
      "step": 5595
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7973235503577215,
      "learning_rate": 1.8970033980756793e-05,
      "loss": 0.4034,
      "step": 5596
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.365197354040562,
      "learning_rate": 1.896959550066497e-05,
      "loss": 0.2715,
      "step": 5597
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40652584283091775,
      "learning_rate": 1.8969156932327316e-05,
      "loss": 0.2981,
      "step": 5598
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5048542750162215,
      "learning_rate": 1.8968718275748147e-05,
      "loss": 0.3905,
      "step": 5599
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.20953552677379936,
      "learning_rate": 1.8968279530931776e-05,
      "loss": 0.0747,
      "step": 5600
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45543362385871555,
      "learning_rate": 1.896784069788252e-05,
      "loss": 0.3142,
      "step": 5601
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9099677759652144,
      "learning_rate": 1.8967401776604696e-05,
      "loss": 0.3943,
      "step": 5602
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4572863189475225,
      "learning_rate": 1.896696276710262e-05,
      "loss": 0.7921,
      "step": 5603
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.34835526204617856,
      "learning_rate": 1.8966523669380616e-05,
      "loss": 0.2644,
      "step": 5604
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7863996544965038,
      "learning_rate": 1.8966084483443002e-05,
      "loss": 0.3937,
      "step": 5605
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4121828147028779,
      "learning_rate": 1.8965645209294096e-05,
      "loss": 0.2471,
      "step": 5606
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8017538861040253,
      "learning_rate": 1.8965205846938225e-05,
      "loss": 0.502,
      "step": 5607
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.24619115046869625,
      "learning_rate": 1.896476639637971e-05,
      "loss": 0.1393,
      "step": 5608
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.1790197238718876,
      "learning_rate": 1.8964326857622873e-05,
      "loss": 0.9124,
      "step": 5609
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.37477136772585823,
      "learning_rate": 1.8963887230672036e-05,
      "loss": 0.187,
      "step": 5610
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3226554982931164,
      "learning_rate": 1.8963447515531527e-05,
      "loss": 0.2697,
      "step": 5611
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5170149055214461,
      "learning_rate": 1.8963007712205675e-05,
      "loss": 0.8669,
      "step": 5612
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.233998628063405,
      "learning_rate": 1.89625678206988e-05,
      "loss": 0.313,
      "step": 5613
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7516745399048409,
      "learning_rate": 1.896212784101524e-05,
      "loss": 0.4791,
      "step": 5614
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3991596080606439,
      "learning_rate": 1.896168777315931e-05,
      "loss": 0.2516,
      "step": 5615
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.48157225392044756,
      "learning_rate": 1.8961247617135353e-05,
      "loss": 0.3959,
      "step": 5616
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4537452229924393,
      "learning_rate": 1.896080737294769e-05,
      "loss": 0.2675,
      "step": 5617
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3203389708836103,
      "learning_rate": 1.896036704060066e-05,
      "loss": 0.1708,
      "step": 5618
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3152356220282463,
      "learning_rate": 1.8959926620098583e-05,
      "loss": 0.2151,
      "step": 5619
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4522729343685514,
      "learning_rate": 1.8959486111445807e-05,
      "loss": 0.6896,
      "step": 5620
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0685780663071514,
      "learning_rate": 1.8959045514646654e-05,
      "loss": 0.5371,
      "step": 5621
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36523297472617783,
      "learning_rate": 1.895860482970546e-05,
      "loss": 0.3049,
      "step": 5622
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4442355734084517,
      "learning_rate": 1.8958164056626574e-05,
      "loss": 0.2255,
      "step": 5623
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41202417228277194,
      "learning_rate": 1.8957723195414315e-05,
      "loss": 0.3009,
      "step": 5624
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1108821856308424,
      "learning_rate": 1.895728224607303e-05,
      "loss": 0.5771,
      "step": 5625
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.216762305170793,
      "learning_rate": 1.8956841208607054e-05,
      "loss": 0.3248,
      "step": 5626
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2843270110483924,
      "learning_rate": 1.895640008302073e-05,
      "loss": 0.1977,
      "step": 5627
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.35569734671560665,
      "learning_rate": 1.895595886931839e-05,
      "loss": 0.2153,
      "step": 5628
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5707599896847174,
      "learning_rate": 1.8955517567504382e-05,
      "loss": 0.3981,
      "step": 5629
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2769277214853498,
      "learning_rate": 1.8955076177583044e-05,
      "loss": 0.5245,
      "step": 5630
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8876533135293503,
      "learning_rate": 1.8954634699558724e-05,
      "loss": 0.5114,
      "step": 5631
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6346569646996439,
      "learning_rate": 1.895419313343576e-05,
      "loss": 0.2224,
      "step": 5632
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44480717390199287,
      "learning_rate": 1.8953751479218496e-05,
      "loss": 0.328,
      "step": 5633
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4086829795869294,
      "learning_rate": 1.895330973691128e-05,
      "loss": 0.294,
      "step": 5634
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5171873162920565,
      "learning_rate": 1.8952867906518454e-05,
      "loss": 0.3208,
      "step": 5635
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2597273914095945,
      "learning_rate": 1.895242598804437e-05,
      "loss": 0.0991,
      "step": 5636
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7863491693730905,
      "learning_rate": 1.8951983981493374e-05,
      "loss": 0.3368,
      "step": 5637
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3872806726680926,
      "learning_rate": 1.895154188686981e-05,
      "loss": 0.2701,
      "step": 5638
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.43840664479421937,
      "learning_rate": 1.8951099704178038e-05,
      "loss": 0.297,
      "step": 5639
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5324567089239018,
      "learning_rate": 1.8950657433422396e-05,
      "loss": 0.4211,
      "step": 5640
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.743785091999847,
      "learning_rate": 1.8950215074607246e-05,
      "loss": 0.3397,
      "step": 5641
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5108645278333124,
      "learning_rate": 1.894977262773693e-05,
      "loss": 0.3075,
      "step": 5642
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9424006500594356,
      "learning_rate": 1.894933009281581e-05,
      "loss": 0.398,
      "step": 5643
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5165106018140688,
      "learning_rate": 1.8948887469848232e-05,
      "loss": 0.8351,
      "step": 5644
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2863461157586974,
      "learning_rate": 1.894844475883856e-05,
      "loss": 0.1777,
      "step": 5645
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.40801587443781834,
      "learning_rate": 1.894800195979114e-05,
      "loss": 0.2892,
      "step": 5646
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2937351337775535,
      "learning_rate": 1.8947559072710334e-05,
      "loss": 0.2014,
      "step": 5647
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9263105776630418,
      "learning_rate": 1.8947116097600497e-05,
      "loss": 0.804,
      "step": 5648
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9732683144169603,
      "learning_rate": 1.8946673034465986e-05,
      "loss": 0.4542,
      "step": 5649
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7322064073457067,
      "learning_rate": 1.8946229883311166e-05,
      "loss": 0.4407,
      "step": 5650
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3611893323050515,
      "learning_rate": 1.894578664414039e-05,
      "loss": 0.2369,
      "step": 5651
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.44375804856729106,
      "learning_rate": 1.8945343316958022e-05,
      "loss": 0.222,
      "step": 5652
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5087464901371356,
      "learning_rate": 1.8944899901768422e-05,
      "loss": 0.3583,
      "step": 5653
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2255807583710222,
      "learning_rate": 1.8944456398575955e-05,
      "loss": 0.1191,
      "step": 5654
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4352993811490495,
      "learning_rate": 1.894401280738498e-05,
      "loss": 0.2898,
      "step": 5655
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4714387846746363,
      "learning_rate": 1.894356912819987e-05,
      "loss": 0.2535,
      "step": 5656
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9119444254336293,
      "learning_rate": 1.8943125361024978e-05,
      "loss": 0.5363,
      "step": 5657
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3540232017606033,
      "learning_rate": 1.894268150586468e-05,
      "loss": 0.3015,
      "step": 5658
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.891563178678378,
      "learning_rate": 1.8942237562723334e-05,
      "loss": 0.4641,
      "step": 5659
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4155939517188905,
      "learning_rate": 1.8941793531605318e-05,
      "loss": 0.2083,
      "step": 5660
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6738586536516222,
      "learning_rate": 1.8941349412514995e-05,
      "loss": 0.3734,
      "step": 5661
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.2439600348617037,
      "learning_rate": 1.8940905205456732e-05,
      "loss": 0.3615,
      "step": 5662
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41408485282048024,
      "learning_rate": 1.89404609104349e-05,
      "loss": 0.3588,
      "step": 5663
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.29028114961776,
      "learning_rate": 1.8940016527453876e-05,
      "loss": 0.0745,
      "step": 5664
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.28923721837857636,
      "learning_rate": 1.8939572056518024e-05,
      "loss": 0.1864,
      "step": 5665
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.8959636794278569,
      "learning_rate": 1.893912749763172e-05,
      "loss": 0.5405,
      "step": 5666
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.796694728319158,
      "learning_rate": 1.893868285079934e-05,
      "loss": 0.407,
      "step": 5667
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5622270327259828,
      "learning_rate": 1.8938238116025257e-05,
      "loss": 0.2863,
      "step": 5668
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.351870986758467,
      "learning_rate": 1.893779329331385e-05,
      "loss": 0.211,
      "step": 5669
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.42586456863448974,
      "learning_rate": 1.8937348382669486e-05,
      "loss": 0.3025,
      "step": 5670
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.3637674630654417,
      "learning_rate": 1.893690338409655e-05,
      "loss": 0.323,
      "step": 5671
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4493070562547656,
      "learning_rate": 1.8936458297599414e-05,
      "loss": 0.8046,
      "step": 5672
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.31968277381090837,
      "learning_rate": 1.893601312318247e-05,
      "loss": 0.131,
      "step": 5673
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4056331453061536,
      "learning_rate": 1.8935567860850083e-05,
      "loss": 0.3237,
      "step": 5674
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9615216354132617,
      "learning_rate": 1.8935122510606637e-05,
      "loss": 0.4265,
      "step": 5675
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3714204888513133,
      "learning_rate": 1.8934677072456517e-05,
      "loss": 0.3298,
      "step": 5676
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3205061632520267,
      "learning_rate": 1.8934231546404106e-05,
      "loss": 0.1591,
      "step": 5677
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3314953187381917,
      "learning_rate": 1.8933785932453784e-05,
      "loss": 0.2101,
      "step": 5678
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7247560788070359,
      "learning_rate": 1.8933340230609938e-05,
      "loss": 0.2836,
      "step": 5679
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0902375310327423,
      "learning_rate": 1.893289444087695e-05,
      "loss": 0.4019,
      "step": 5680
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3855551373080008,
      "learning_rate": 1.8932448563259207e-05,
      "loss": 0.3094,
      "step": 5681
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.34103526650200383,
      "learning_rate": 1.8932002597761098e-05,
      "loss": 0.2022,
      "step": 5682
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5170822900930651,
      "learning_rate": 1.8931556544387004e-05,
      "loss": 0.3222,
      "step": 5683
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7576190173708203,
      "learning_rate": 1.8931110403141323e-05,
      "loss": 0.4443,
      "step": 5684
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.4323889842679438,
      "learning_rate": 1.8930664174028435e-05,
      "loss": 0.8582,
      "step": 5685
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2221756648007661,
      "learning_rate": 1.8930217857052734e-05,
      "loss": 0.0857,
      "step": 5686
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.002969638564514,
      "learning_rate": 1.8929771452218616e-05,
      "loss": 0.4263,
      "step": 5687
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2983913388618046,
      "learning_rate": 1.8929324959530463e-05,
      "loss": 0.252,
      "step": 5688
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0168377555276886,
      "learning_rate": 1.8928878378992675e-05,
      "loss": 0.6813,
      "step": 5689
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.9954980391369865,
      "learning_rate": 1.8928431710609646e-05,
      "loss": 0.4077,
      "step": 5690
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.7348143057718514,
      "learning_rate": 1.8927984954385766e-05,
      "loss": 0.2876,
      "step": 5691
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41722348782848473,
      "learning_rate": 1.892753811032543e-05,
      "loss": 0.2906,
      "step": 5692
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.406673325005671,
      "learning_rate": 1.892709117843304e-05,
      "loss": 0.328,
      "step": 5693
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5652743560441389,
      "learning_rate": 1.8926644158712988e-05,
      "loss": 0.4068,
      "step": 5694
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1328480888285364,
      "learning_rate": 1.8926197051169674e-05,
      "loss": 0.5257,
      "step": 5695
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.465478022881438,
      "learning_rate": 1.8925749855807495e-05,
      "loss": 0.2745,
      "step": 5696
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.2536014724112762,
      "learning_rate": 1.8925302572630856e-05,
      "loss": 0.144,
      "step": 5697
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.5803028855179564,
      "learning_rate": 1.8924855201644148e-05,
      "loss": 0.684,
      "step": 5698
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.544398661947307,
      "learning_rate": 1.892440774285178e-05,
      "loss": 0.2987,
      "step": 5699
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.4834746213398694,
      "learning_rate": 1.892396019625815e-05,
      "loss": 0.3419,
      "step": 5700
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.35024515475269224,
      "learning_rate": 1.8923512561867666e-05,
      "loss": 0.2536,
      "step": 5701
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1971665879360087,
      "learning_rate": 1.892306483968473e-05,
      "loss": 0.5569,
      "step": 5702
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.1104119922074736,
      "learning_rate": 1.8922617029713747e-05,
      "loss": 0.6021,
      "step": 5703
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.30395043799181026,
      "learning_rate": 1.892216913195912e-05,
      "loss": 0.2189,
      "step": 5704
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.45046244226481447,
      "learning_rate": 1.8921721146425254e-05,
      "loss": 0.2768,
      "step": 5705
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3834114362250208,
      "learning_rate": 1.8921273073116562e-05,
      "loss": 0.2492,
      "step": 5706
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.5233850650067362,
      "learning_rate": 1.8920824912037453e-05,
      "loss": 0.2484,
      "step": 5707
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.0839462643958018,
      "learning_rate": 1.892037666319233e-05,
      "loss": 0.4114,
      "step": 5708
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.6606346277476779,
      "learning_rate": 1.8919928326585606e-05,
      "loss": 0.3187,
      "step": 5709
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.36916786721123973,
      "learning_rate": 1.8919479902221694e-05,
      "loss": 0.2314,
      "step": 5710
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.9660506768042632,
      "learning_rate": 1.8919031390105003e-05,
      "loss": 0.7587,
      "step": 5711
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.3429794091712089,
      "learning_rate": 1.8918582790239947e-05,
      "loss": 0.2789,
      "step": 5712
    },
    {
      "epoch": 0.17,
      "grad_norm": 1.230500428605313,
      "learning_rate": 1.8918134102630937e-05,
      "loss": 0.7146,
      "step": 5713
    },
    {
      "epoch": 0.17,
      "grad_norm": 0.41830991723294275,
      "learning_rate": 1.891768532728239e-05,
      "loss": 0.077,
      "step": 5714
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3321926230356746,
      "learning_rate": 1.891723646419872e-05,
      "loss": 0.2678,
      "step": 5715
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.49329032503833603,
      "learning_rate": 1.8916787513384348e-05,
      "loss": 0.2089,
      "step": 5716
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38696718277622,
      "learning_rate": 1.8916338474843683e-05,
      "loss": 0.2931,
      "step": 5717
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9407667313211434,
      "learning_rate": 1.8915889348581145e-05,
      "loss": 0.5177,
      "step": 5718
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3134892340498379,
      "learning_rate": 1.8915440134601157e-05,
      "loss": 0.2125,
      "step": 5719
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.111399633834946,
      "learning_rate": 1.891499083290813e-05,
      "loss": 0.7371,
      "step": 5720
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7736980820109175,
      "learning_rate": 1.8914541443506494e-05,
      "loss": 0.5929,
      "step": 5721
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2495641230524706,
      "learning_rate": 1.8914091966400667e-05,
      "loss": 0.6638,
      "step": 5722
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.315263096050273,
      "learning_rate": 1.8913642401595064e-05,
      "loss": 0.221,
      "step": 5723
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5346304417715427,
      "learning_rate": 1.891319274909412e-05,
      "loss": 0.3475,
      "step": 5724
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3402528091382767,
      "learning_rate": 1.891274300890225e-05,
      "loss": 0.1453,
      "step": 5725
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8010069661929929,
      "learning_rate": 1.8912293181023883e-05,
      "loss": 0.5358,
      "step": 5726
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7227775742945464,
      "learning_rate": 1.8911843265463444e-05,
      "loss": 0.2512,
      "step": 5727
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3893144017177647,
      "learning_rate": 1.891139326222536e-05,
      "loss": 0.2918,
      "step": 5728
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5005128816743756,
      "learning_rate": 1.891094317131405e-05,
      "loss": 0.2681,
      "step": 5729
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5075397406858408,
      "learning_rate": 1.8910492992733957e-05,
      "loss": 0.3161,
      "step": 5730
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4429799906483074,
      "learning_rate": 1.8910042726489498e-05,
      "loss": 0.7135,
      "step": 5731
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3588362345727239,
      "learning_rate": 1.8909592372585105e-05,
      "loss": 0.2099,
      "step": 5732
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6427069427868369,
      "learning_rate": 1.8909141931025216e-05,
      "loss": 0.3346,
      "step": 5733
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5788893275859203,
      "learning_rate": 1.8908691401814252e-05,
      "loss": 0.2953,
      "step": 5734
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4378382474365456,
      "learning_rate": 1.8908240784956653e-05,
      "loss": 0.3284,
      "step": 5735
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.31257054929897926,
      "learning_rate": 1.890779008045685e-05,
      "loss": 0.112,
      "step": 5736
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.096003633540416,
      "learning_rate": 1.8907339288319276e-05,
      "loss": 0.4442,
      "step": 5737
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4246982797291724,
      "learning_rate": 1.890688840854837e-05,
      "loss": 0.259,
      "step": 5738
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.599054513590735,
      "learning_rate": 1.8906437441148563e-05,
      "loss": 0.8956,
      "step": 5739
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3571675073417206,
      "learning_rate": 1.8905986386124292e-05,
      "loss": 0.2554,
      "step": 5740
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1633699112230849,
      "learning_rate": 1.890553524348e-05,
      "loss": 0.5868,
      "step": 5741
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34886314151205267,
      "learning_rate": 1.890508401322012e-05,
      "loss": 0.219,
      "step": 5742
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6739412069190833,
      "learning_rate": 1.8904632695349092e-05,
      "loss": 0.407,
      "step": 5743
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.36918750686416985,
      "learning_rate": 1.890418128987136e-05,
      "loss": 0.2206,
      "step": 5744
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0063080944874516,
      "learning_rate": 1.890372979679136e-05,
      "loss": 0.3565,
      "step": 5745
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4729590724395832,
      "learning_rate": 1.890327821611354e-05,
      "loss": 0.2702,
      "step": 5746
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3220886597344595,
      "learning_rate": 1.8902826547842336e-05,
      "loss": 0.2601,
      "step": 5747
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6770144479100428,
      "learning_rate": 1.89023747919822e-05,
      "loss": 0.9341,
      "step": 5748
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9117527152080589,
      "learning_rate": 1.8901922948537567e-05,
      "loss": 0.3616,
      "step": 5749
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0087392777480506,
      "learning_rate": 1.8901471017512886e-05,
      "loss": 0.4456,
      "step": 5750
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3395233043390233,
      "learning_rate": 1.890101899891261e-05,
      "loss": 0.2583,
      "step": 5751
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8218365261173763,
      "learning_rate": 1.8900566892741173e-05,
      "loss": 0.505,
      "step": 5752
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.35341676627445606,
      "learning_rate": 1.8900114699003037e-05,
      "loss": 0.2827,
      "step": 5753
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3407065210151375,
      "learning_rate": 1.8899662417702637e-05,
      "loss": 0.1835,
      "step": 5754
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41661894259429716,
      "learning_rate": 1.8899210048844434e-05,
      "loss": 0.1908,
      "step": 5755
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5415116288138512,
      "learning_rate": 1.8898757592432873e-05,
      "loss": 0.2476,
      "step": 5756
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4961434714975723,
      "learning_rate": 1.889830504847241e-05,
      "loss": 0.8152,
      "step": 5757
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.47254687188644273,
      "learning_rate": 1.8897852416967492e-05,
      "loss": 0.2869,
      "step": 5758
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4052442799661843,
      "learning_rate": 1.8897399697922575e-05,
      "loss": 0.2615,
      "step": 5759
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4069403186643806,
      "learning_rate": 1.889694689134211e-05,
      "loss": 0.2362,
      "step": 5760
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0284580377755226,
      "learning_rate": 1.8896493997230553e-05,
      "loss": 0.5028,
      "step": 5761
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39966032011873553,
      "learning_rate": 1.8896041015592364e-05,
      "loss": 0.1895,
      "step": 5762
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5228651993236024,
      "learning_rate": 1.8895587946431996e-05,
      "loss": 0.2944,
      "step": 5763
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34440143450311633,
      "learning_rate": 1.8895134789753904e-05,
      "loss": 0.0807,
      "step": 5764
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3999031084081704,
      "learning_rate": 1.889468154556255e-05,
      "loss": 0.2907,
      "step": 5765
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4898219821472063,
      "learning_rate": 1.8894228213862395e-05,
      "loss": 0.3103,
      "step": 5766
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.046910693155963,
      "learning_rate": 1.8893774794657892e-05,
      "loss": 0.594,
      "step": 5767
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0438054136506032,
      "learning_rate": 1.889332128795351e-05,
      "loss": 0.412,
      "step": 5768
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33992856803003396,
      "learning_rate": 1.8892867693753706e-05,
      "loss": 0.2072,
      "step": 5769
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8892767957594302,
      "learning_rate": 1.8892414012062942e-05,
      "loss": 0.578,
      "step": 5770
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34510818082072486,
      "learning_rate": 1.8891960242885683e-05,
      "loss": 0.2652,
      "step": 5771
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5586286816131976,
      "learning_rate": 1.8891506386226394e-05,
      "loss": 0.2246,
      "step": 5772
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2995246650163875,
      "learning_rate": 1.889105244208954e-05,
      "loss": 0.1438,
      "step": 5773
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7139020028180106,
      "learning_rate": 1.8890598410479586e-05,
      "loss": 0.4273,
      "step": 5774
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0689472875400883,
      "learning_rate": 1.8890144291401e-05,
      "loss": 0.4693,
      "step": 5775
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2781442143455342,
      "learning_rate": 1.8889690084858247e-05,
      "loss": 0.6816,
      "step": 5776
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.30112422917876286,
      "learning_rate": 1.88892357908558e-05,
      "loss": 0.2615,
      "step": 5777
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.37339969575697357,
      "learning_rate": 1.8888781409398124e-05,
      "loss": 0.3006,
      "step": 5778
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.851548372201824,
      "learning_rate": 1.8888326940489695e-05,
      "loss": 0.3862,
      "step": 5779
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5056147071205799,
      "learning_rate": 1.888787238413498e-05,
      "loss": 0.7495,
      "step": 5780
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33185598401825933,
      "learning_rate": 1.8887417740338453e-05,
      "loss": 0.1084,
      "step": 5781
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.32069930759394877,
      "learning_rate": 1.888696300910458e-05,
      "loss": 0.2174,
      "step": 5782
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42391415673442556,
      "learning_rate": 1.8886508190437846e-05,
      "loss": 0.3283,
      "step": 5783
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1407847491459866,
      "learning_rate": 1.888605328434272e-05,
      "loss": 0.6016,
      "step": 5784
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8003468945203946,
      "learning_rate": 1.8885598290823676e-05,
      "loss": 0.5001,
      "step": 5785
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7097583907163315,
      "learning_rate": 1.8885143209885197e-05,
      "loss": 0.2969,
      "step": 5786
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5106814327270998,
      "learning_rate": 1.8884688041531753e-05,
      "loss": 0.2933,
      "step": 5787
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5001321279167029,
      "learning_rate": 1.8884232785767822e-05,
      "loss": 0.2548,
      "step": 5788
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41671267896970776,
      "learning_rate": 1.8883777442597887e-05,
      "loss": 0.3358,
      "step": 5789
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3516879567881835,
      "learning_rate": 1.8883322012026427e-05,
      "loss": 0.119,
      "step": 5790
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6500445350928021,
      "learning_rate": 1.888286649405792e-05,
      "loss": 0.2858,
      "step": 5791
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.402372071423342,
      "learning_rate": 1.8882410888696853e-05,
      "loss": 0.2462,
      "step": 5792
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.051671777597348,
      "learning_rate": 1.88819551959477e-05,
      "loss": 0.6421,
      "step": 5793
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3866035730166958,
      "learning_rate": 1.8881499415814955e-05,
      "loss": 0.3134,
      "step": 5794
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4117476024684587,
      "learning_rate": 1.8881043548303094e-05,
      "loss": 0.1879,
      "step": 5795
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.43548354022984437,
      "learning_rate": 1.8880587593416603e-05,
      "loss": 0.2936,
      "step": 5796
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4737233454685687,
      "learning_rate": 1.8880131551159972e-05,
      "loss": 0.3003,
      "step": 5797
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.211155720606596,
      "learning_rate": 1.8879675421537682e-05,
      "loss": 0.5947,
      "step": 5798
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2562162856694147,
      "learning_rate": 1.8879219204554224e-05,
      "loss": 0.1105,
      "step": 5799
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4074685309430849,
      "learning_rate": 1.8878762900214086e-05,
      "loss": 0.3039,
      "step": 5800
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.32368087796828804,
      "learning_rate": 1.887830650852176e-05,
      "loss": 0.2489,
      "step": 5801
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7856663671833799,
      "learning_rate": 1.887785002948173e-05,
      "loss": 0.5287,
      "step": 5802
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.322143399614894,
      "learning_rate": 1.887739346309849e-05,
      "loss": 0.4108,
      "step": 5803
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5087987983083969,
      "learning_rate": 1.8876936809376533e-05,
      "loss": 0.257,
      "step": 5804
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33439779717722995,
      "learning_rate": 1.887648006832035e-05,
      "loss": 0.216,
      "step": 5805
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5492242011525321,
      "learning_rate": 1.8876023239934435e-05,
      "loss": 0.3713,
      "step": 5806
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45862147918408586,
      "learning_rate": 1.8875566324223283e-05,
      "loss": 0.2466,
      "step": 5807
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0747086737221339,
      "learning_rate": 1.8875109321191393e-05,
      "loss": 0.4371,
      "step": 5808
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7655540438308541,
      "learning_rate": 1.8874652230843254e-05,
      "loss": 0.4478,
      "step": 5809
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.36904367024810875,
      "learning_rate": 1.8874195053183365e-05,
      "loss": 0.277,
      "step": 5810
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9097621985500347,
      "learning_rate": 1.8873737788216226e-05,
      "loss": 0.5273,
      "step": 5811
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34949199899661043,
      "learning_rate": 1.8873280435946336e-05,
      "loss": 0.2959,
      "step": 5812
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3042917217100446,
      "learning_rate": 1.887282299637819e-05,
      "loss": 0.1411,
      "step": 5813
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.31672329168407715,
      "learning_rate": 1.8872365469516295e-05,
      "loss": 0.1781,
      "step": 5814
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5869668682409448,
      "learning_rate": 1.887190785536515e-05,
      "loss": 0.3571,
      "step": 5815
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8852987611034561,
      "learning_rate": 1.8871450153929255e-05,
      "loss": 0.3832,
      "step": 5816
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2146928379277984,
      "learning_rate": 1.8870992365213113e-05,
      "loss": 0.7869,
      "step": 5817
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.318288175635469,
      "learning_rate": 1.887053448922123e-05,
      "loss": 0.2357,
      "step": 5818
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5388989920000691,
      "learning_rate": 1.887007652595811e-05,
      "loss": 0.3525,
      "step": 5819
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4239778941327852,
      "learning_rate": 1.8869618475428262e-05,
      "loss": 0.3358,
      "step": 5820
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9048154269387271,
      "learning_rate": 1.8869160337636185e-05,
      "loss": 0.6292,
      "step": 5821
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.24639280712903158,
      "learning_rate": 1.886870211258639e-05,
      "loss": 0.079,
      "step": 5822
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2565523967375899,
      "learning_rate": 1.8868243800283387e-05,
      "loss": 0.1412,
      "step": 5823
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4098204449687372,
      "learning_rate": 1.8867785400731682e-05,
      "loss": 0.3292,
      "step": 5824
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4553658107626168,
      "learning_rate": 1.8867326913935786e-05,
      "loss": 0.2468,
      "step": 5825
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4041343631712329,
      "learning_rate": 1.8866868339900213e-05,
      "loss": 0.7432,
      "step": 5826
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6491465455438663,
      "learning_rate": 1.8866409678629466e-05,
      "loss": 0.3388,
      "step": 5827
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.382603207125862,
      "learning_rate": 1.886595093012807e-05,
      "loss": 0.2777,
      "step": 5828
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7890695436202229,
      "learning_rate": 1.8865492094400527e-05,
      "loss": 0.4321,
      "step": 5829
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5775819508131632,
      "learning_rate": 1.8865033171451358e-05,
      "loss": 0.3364,
      "step": 5830
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2985903033086365,
      "learning_rate": 1.8864574161285073e-05,
      "loss": 0.2024,
      "step": 5831
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40092451815466496,
      "learning_rate": 1.8864115063906195e-05,
      "loss": 0.2282,
      "step": 5832
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3718300083647018,
      "learning_rate": 1.8863655879319235e-05,
      "loss": 0.2362,
      "step": 5833
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5263626847888492,
      "learning_rate": 1.8863196607528708e-05,
      "loss": 0.6541,
      "step": 5834
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8245996304254007,
      "learning_rate": 1.886273724853914e-05,
      "loss": 0.4169,
      "step": 5835
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.28690897827621864,
      "learning_rate": 1.8862277802355044e-05,
      "loss": 0.2304,
      "step": 5836
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41213479630656363,
      "learning_rate": 1.8861818268980946e-05,
      "loss": 0.3355,
      "step": 5837
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9750295672332377,
      "learning_rate": 1.886135864842136e-05,
      "loss": 0.3933,
      "step": 5838
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.545919143825364,
      "learning_rate": 1.8860898940680817e-05,
      "loss": 0.8467,
      "step": 5839
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2428753053380955,
      "learning_rate": 1.8860439145763833e-05,
      "loss": 0.0749,
      "step": 5840
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.46204344182107565,
      "learning_rate": 1.885997926367493e-05,
      "loss": 0.2801,
      "step": 5841
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5368987718743929,
      "learning_rate": 1.8859519294418637e-05,
      "loss": 0.2226,
      "step": 5842
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3712149672248326,
      "learning_rate": 1.885905923799948e-05,
      "loss": 0.3164,
      "step": 5843
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9275278688524679,
      "learning_rate": 1.885859909442198e-05,
      "loss": 0.4339,
      "step": 5844
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9837615184762896,
      "learning_rate": 1.8858138863690672e-05,
      "loss": 0.4821,
      "step": 5845
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34174576698425374,
      "learning_rate": 1.8857678545810078e-05,
      "loss": 0.1905,
      "step": 5846
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6080066347609459,
      "learning_rate": 1.8857218140784725e-05,
      "loss": 0.4323,
      "step": 5847
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3361151695621671,
      "learning_rate": 1.8856757648619146e-05,
      "loss": 0.268,
      "step": 5848
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5760474079931607,
      "learning_rate": 1.8856297069317872e-05,
      "loss": 0.025,
      "step": 5849
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3880439337609891,
      "learning_rate": 1.8855836402885433e-05,
      "loss": 0.2355,
      "step": 5850
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.27982840017724653,
      "learning_rate": 1.885537564932636e-05,
      "loss": 0.1953,
      "step": 5851
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8692222525902561,
      "learning_rate": 1.885491480864519e-05,
      "loss": 0.5999,
      "step": 5852
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0004297250752776,
      "learning_rate": 1.8854453880846454e-05,
      "loss": 0.4662,
      "step": 5853
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.36341300484352046,
      "learning_rate": 1.8853992865934688e-05,
      "loss": 0.3598,
      "step": 5854
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.29759888626072445,
      "learning_rate": 1.8853531763914424e-05,
      "loss": 0.2064,
      "step": 5855
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.626894746338574,
      "learning_rate": 1.8853070574790204e-05,
      "loss": 0.9246,
      "step": 5856
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0534819612224342,
      "learning_rate": 1.8852609298566565e-05,
      "loss": 0.5155,
      "step": 5857
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.8733804385706032,
      "learning_rate": 1.8852147935248038e-05,
      "loss": 0.7987,
      "step": 5858
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.24336486908653251,
      "learning_rate": 1.8851686484839168e-05,
      "loss": 0.0741,
      "step": 5859
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.35716620157257234,
      "learning_rate": 1.8851224947344497e-05,
      "loss": 0.3097,
      "step": 5860
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45394083726065343,
      "learning_rate": 1.8850763322768557e-05,
      "loss": 0.2737,
      "step": 5861
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8033608486474134,
      "learning_rate": 1.88503016111159e-05,
      "loss": 0.4629,
      "step": 5862
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2428266458315733,
      "learning_rate": 1.884983981239106e-05,
      "loss": 0.6701,
      "step": 5863
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3176096616933259,
      "learning_rate": 1.8849377926598584e-05,
      "loss": 0.2025,
      "step": 5864
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.039223119955562,
      "learning_rate": 1.8848915953743018e-05,
      "loss": 0.9046,
      "step": 5865
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34588604346663204,
      "learning_rate": 1.8848453893828904e-05,
      "loss": 0.2714,
      "step": 5866
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9881482875578063,
      "learning_rate": 1.8847991746860793e-05,
      "loss": 0.6502,
      "step": 5867
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3616289524648094,
      "learning_rate": 1.8847529512843225e-05,
      "loss": 0.2132,
      "step": 5868
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.630483887637716,
      "learning_rate": 1.884706719178075e-05,
      "loss": 0.3657,
      "step": 5869
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2541411922840081,
      "learning_rate": 1.8846604783677916e-05,
      "loss": 0.1857,
      "step": 5870
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.47142116208720425,
      "learning_rate": 1.8846142288539274e-05,
      "loss": 0.3854,
      "step": 5871
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4386768031659677,
      "learning_rate": 1.8845679706369375e-05,
      "loss": 0.2332,
      "step": 5872
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7300737134664521,
      "learning_rate": 1.8845217037172766e-05,
      "loss": 0.3103,
      "step": 5873
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44902193931011203,
      "learning_rate": 1.8844754280954002e-05,
      "loss": 0.2595,
      "step": 5874
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.282875225275305,
      "learning_rate": 1.8844291437717635e-05,
      "loss": 0.605,
      "step": 5875
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2104254938146275,
      "learning_rate": 1.8843828507468217e-05,
      "loss": 0.6135,
      "step": 5876
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2764740448735551,
      "learning_rate": 1.8843365490210307e-05,
      "loss": 0.164,
      "step": 5877
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4372016115485631,
      "learning_rate": 1.884290238594846e-05,
      "loss": 0.3142,
      "step": 5878
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.35905658330752965,
      "learning_rate": 1.8842439194687224e-05,
      "loss": 0.3543,
      "step": 5879
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5512427095312157,
      "learning_rate": 1.8841975916431165e-05,
      "loss": 0.3235,
      "step": 5880
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0125083517567102,
      "learning_rate": 1.8841512551184836e-05,
      "loss": 0.2947,
      "step": 5881
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.40909583763591134,
      "learning_rate": 1.8841049098952797e-05,
      "loss": 0.255,
      "step": 5882
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5513302489297852,
      "learning_rate": 1.884058555973961e-05,
      "loss": 0.3135,
      "step": 5883
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5563948299933407,
      "learning_rate": 1.884012193354983e-05,
      "loss": 0.3803,
      "step": 5884
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4795519274192508,
      "learning_rate": 1.8839658220388026e-05,
      "loss": 0.3084,
      "step": 5885
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7209564178493703,
      "learning_rate": 1.8839194420258752e-05,
      "loss": 0.4492,
      "step": 5886
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3626694107130301,
      "learning_rate": 1.8838730533166574e-05,
      "loss": 0.2528,
      "step": 5887
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.592238126465011,
      "learning_rate": 1.883826655911606e-05,
      "loss": 0.3956,
      "step": 5888
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3566065859042298,
      "learning_rate": 1.8837802498111772e-05,
      "loss": 0.2483,
      "step": 5889
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.29725985622414464,
      "learning_rate": 1.883733835015827e-05,
      "loss": 0.1807,
      "step": 5890
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4822770907866718,
      "learning_rate": 1.883687411526013e-05,
      "loss": 0.306,
      "step": 5891
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5290299652013325,
      "learning_rate": 1.8836409793421914e-05,
      "loss": 0.2678,
      "step": 5892
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4734236941169385,
      "learning_rate": 1.8835945384648192e-05,
      "loss": 0.8913,
      "step": 5893
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7556489751646189,
      "learning_rate": 1.883548088894353e-05,
      "loss": 0.3802,
      "step": 5894
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.36767768053808725,
      "learning_rate": 1.88350163063125e-05,
      "loss": 0.2868,
      "step": 5895
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3979876330552113,
      "learning_rate": 1.883455163675967e-05,
      "loss": 0.2046,
      "step": 5896
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5287266351713339,
      "learning_rate": 1.883408688028962e-05,
      "loss": 0.4067,
      "step": 5897
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3701906624593724,
      "learning_rate": 1.8833622036906912e-05,
      "loss": 0.1607,
      "step": 5898
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0984923787460499,
      "learning_rate": 1.8833157106616125e-05,
      "loss": 0.6684,
      "step": 5899
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4953053683809171,
      "learning_rate": 1.8832692089421832e-05,
      "loss": 0.0753,
      "step": 5900
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39034124028502537,
      "learning_rate": 1.8832226985328606e-05,
      "loss": 0.2575,
      "step": 5901
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4000971196160417,
      "learning_rate": 1.8831761794341025e-05,
      "loss": 0.3461,
      "step": 5902
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7820957610558078,
      "learning_rate": 1.8831296516463672e-05,
      "loss": 0.3691,
      "step": 5903
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0462504902447058,
      "learning_rate": 1.883083115170111e-05,
      "loss": 0.5023,
      "step": 5904
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3006625867632777,
      "learning_rate": 1.8830365700057928e-05,
      "loss": 0.1976,
      "step": 5905
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9721591964622093,
      "learning_rate": 1.8829900161538704e-05,
      "loss": 0.6077,
      "step": 5906
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3359947191523292,
      "learning_rate": 1.8829434536148016e-05,
      "loss": 0.2477,
      "step": 5907
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4464264036444935,
      "learning_rate": 1.8828968823890443e-05,
      "loss": 0.179,
      "step": 5908
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5983032519321073,
      "learning_rate": 1.8828503024770573e-05,
      "loss": 0.1829,
      "step": 5909
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4537197378269946,
      "learning_rate": 1.8828037138792984e-05,
      "loss": 0.3569,
      "step": 5910
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1975982676645918,
      "learning_rate": 1.882757116596226e-05,
      "loss": 0.5053,
      "step": 5911
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8347763302697498,
      "learning_rate": 1.882710510628299e-05,
      "loss": 0.4908,
      "step": 5912
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3428697484319597,
      "learning_rate": 1.8826638959759752e-05,
      "loss": 0.2758,
      "step": 5913
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.34855078079183754,
      "learning_rate": 1.8826172726397135e-05,
      "loss": 0.2501,
      "step": 5914
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.9348601690364158,
      "learning_rate": 1.882570640619973e-05,
      "loss": 0.84,
      "step": 5915
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9452130208060158,
      "learning_rate": 1.8825239999172117e-05,
      "loss": 0.378,
      "step": 5916
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.544724710125414,
      "learning_rate": 1.8824773505318893e-05,
      "loss": 0.1878,
      "step": 5917
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4306991435285207,
      "learning_rate": 1.882430692464464e-05,
      "loss": 0.1894,
      "step": 5918
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6218924359812918,
      "learning_rate": 1.8823840257153955e-05,
      "loss": 0.353,
      "step": 5919
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3765499472349416,
      "learning_rate": 1.8823373502851422e-05,
      "loss": 0.2814,
      "step": 5920
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9534309936797033,
      "learning_rate": 1.8822906661741643e-05,
      "loss": 0.5289,
      "step": 5921
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5460672923211362,
      "learning_rate": 1.88224397338292e-05,
      "loss": 0.3356,
      "step": 5922
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0611358069035342,
      "learning_rate": 1.8821972719118694e-05,
      "loss": 0.4925,
      "step": 5923
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4387286889609188,
      "learning_rate": 1.882150561761472e-05,
      "loss": 0.243,
      "step": 5924
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.49215736490279727,
      "learning_rate": 1.8821038429321867e-05,
      "loss": 0.3559,
      "step": 5925
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22585589473715292,
      "learning_rate": 1.882057115424474e-05,
      "loss": 0.1627,
      "step": 5926
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4466286683580651,
      "learning_rate": 1.8820103792387926e-05,
      "loss": 0.1621,
      "step": 5927
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4616952302106737,
      "learning_rate": 1.881963634375603e-05,
      "loss": 0.3275,
      "step": 5928
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0092424248649825,
      "learning_rate": 1.8819168808353652e-05,
      "loss": 0.4761,
      "step": 5929
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.571789402561749,
      "learning_rate": 1.881870118618539e-05,
      "loss": 0.4048,
      "step": 5930
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.48588461025156615,
      "learning_rate": 1.881823347725584e-05,
      "loss": 0.257,
      "step": 5931
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4364796783208994,
      "learning_rate": 1.881776568156961e-05,
      "loss": 0.2686,
      "step": 5932
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5076127332903971,
      "learning_rate": 1.88172977991313e-05,
      "loss": 0.245,
      "step": 5933
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.488072425867683,
      "learning_rate": 1.8816829829945513e-05,
      "loss": 0.8072,
      "step": 5934
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2766789867545962,
      "learning_rate": 1.8816361774016852e-05,
      "loss": 0.106,
      "step": 5935
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9557625918995404,
      "learning_rate": 1.8815893631349924e-05,
      "loss": 0.515,
      "step": 5936
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3245396225520332,
      "learning_rate": 1.8815425401949335e-05,
      "loss": 0.2294,
      "step": 5937
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5306179815876781,
      "learning_rate": 1.881495708581969e-05,
      "loss": 0.3917,
      "step": 5938
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9304053422535981,
      "learning_rate": 1.881448868296559e-05,
      "loss": 0.4448,
      "step": 5939
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3044949703674109,
      "learning_rate": 1.8814020193391657e-05,
      "loss": 0.1515,
      "step": 5940
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44265806040498834,
      "learning_rate": 1.881355161710249e-05,
      "loss": 0.2923,
      "step": 5941
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.49361077809719567,
      "learning_rate": 1.8813082954102705e-05,
      "loss": 0.255,
      "step": 5942
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5204660959897255,
      "learning_rate": 1.8812614204396906e-05,
      "loss": 0.3748,
      "step": 5943
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2912186674292848,
      "learning_rate": 1.8812145367989712e-05,
      "loss": 0.2059,
      "step": 5944
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7665974388542197,
      "learning_rate": 1.8811676444885735e-05,
      "loss": 0.3648,
      "step": 5945
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4062419881828839,
      "learning_rate": 1.881120743508958e-05,
      "loss": 0.2547,
      "step": 5946
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9229547024441899,
      "learning_rate": 1.8810738338605866e-05,
      "loss": 0.5211,
      "step": 5947
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4749241163729749,
      "learning_rate": 1.8810269155439216e-05,
      "loss": 0.3469,
      "step": 5948
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33091479489755565,
      "learning_rate": 1.8809799885594232e-05,
      "loss": 0.2357,
      "step": 5949
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3129586190414903,
      "learning_rate": 1.880933052907554e-05,
      "loss": 0.0779,
      "step": 5950
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4334581205095755,
      "learning_rate": 1.880886108588776e-05,
      "loss": 0.3377,
      "step": 5951
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9335240265094626,
      "learning_rate": 1.8808391556035503e-05,
      "loss": 0.3573,
      "step": 5952
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7371067127757694,
      "learning_rate": 1.8807921939523393e-05,
      "loss": 0.4332,
      "step": 5953
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5475721034112134,
      "learning_rate": 1.8807452236356044e-05,
      "loss": 0.37,
      "step": 5954
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.31348068302900806,
      "learning_rate": 1.8806982446538086e-05,
      "loss": 0.2021,
      "step": 5955
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5666051910261937,
      "learning_rate": 1.8806512570074136e-05,
      "loss": 0.3778,
      "step": 5956
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33050800044451095,
      "learning_rate": 1.8806042606968818e-05,
      "loss": 0.222,
      "step": 5957
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5014922507613916,
      "learning_rate": 1.8805572557226755e-05,
      "loss": 0.2392,
      "step": 5958
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.48719315607797836,
      "learning_rate": 1.880510242085257e-05,
      "loss": 0.1504,
      "step": 5959
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4293751974782547,
      "learning_rate": 1.8804632197850895e-05,
      "loss": 0.3184,
      "step": 5960
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3684700754566547,
      "learning_rate": 1.8804161888226347e-05,
      "loss": 0.2616,
      "step": 5961
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.4604336296693021,
      "learning_rate": 1.880369149198356e-05,
      "loss": 0.5634,
      "step": 5962
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5231835481118586,
      "learning_rate": 1.8803221009127157e-05,
      "loss": 0.3408,
      "step": 5963
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41142116525706957,
      "learning_rate": 1.8802750439661767e-05,
      "loss": 0.2765,
      "step": 5964
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7074043772698567,
      "learning_rate": 1.8802279783592026e-05,
      "loss": 0.401,
      "step": 5965
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.008280666091389,
      "learning_rate": 1.880180904092256e-05,
      "loss": 0.5123,
      "step": 5966
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3073582899291491,
      "learning_rate": 1.8801338211658e-05,
      "loss": 0.2655,
      "step": 5967
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3128664718828782,
      "learning_rate": 1.8800867295802975e-05,
      "loss": 0.1539,
      "step": 5968
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6039529915804976,
      "learning_rate": 1.8800396293362126e-05,
      "loss": 0.3573,
      "step": 5969
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.237512011716619,
      "learning_rate": 1.879992520434008e-05,
      "loss": 0.3119,
      "step": 5970
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8386586668986478,
      "learning_rate": 1.8799454028741474e-05,
      "loss": 0.546,
      "step": 5971
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3015733507366152,
      "learning_rate": 1.8798982766570945e-05,
      "loss": 0.2601,
      "step": 5972
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1067325440790559,
      "learning_rate": 1.879851141783313e-05,
      "loss": 0.4679,
      "step": 5973
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39392555942611385,
      "learning_rate": 1.8798039982532664e-05,
      "loss": 0.2893,
      "step": 5974
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.5349124273645636,
      "learning_rate": 1.8797568460674186e-05,
      "loss": 0.841,
      "step": 5975
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.27914163174765694,
      "learning_rate": 1.879709685226233e-05,
      "loss": 0.1248,
      "step": 5976
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.1431769457238272,
      "learning_rate": 1.8796625157301744e-05,
      "loss": 0.4625,
      "step": 5977
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3666044478317943,
      "learning_rate": 1.8796153375797066e-05,
      "loss": 0.2217,
      "step": 5978
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3331597333030613,
      "learning_rate": 1.8795681507752936e-05,
      "loss": 0.2519,
      "step": 5979
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7767920458101922,
      "learning_rate": 1.8795209553173996e-05,
      "loss": 0.4885,
      "step": 5980
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8009092903646166,
      "learning_rate": 1.879473751206489e-05,
      "loss": 0.3783,
      "step": 5981
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.476118704101056,
      "learning_rate": 1.8794265384430265e-05,
      "loss": 0.2665,
      "step": 5982
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.44129045537156286,
      "learning_rate": 1.8793793170274765e-05,
      "loss": 0.2938,
      "step": 5983
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5203540137063771,
      "learning_rate": 1.879332086960303e-05,
      "loss": 0.4058,
      "step": 5984
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.45370025455833984,
      "learning_rate": 1.8792848482419713e-05,
      "loss": 0.2496,
      "step": 5985
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2986430994955324,
      "learning_rate": 1.879237600872946e-05,
      "loss": 0.1531,
      "step": 5986
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.33490192506943267,
      "learning_rate": 1.8791903448536916e-05,
      "loss": 0.2238,
      "step": 5987
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.7972486705683943,
      "learning_rate": 1.8791430801846738e-05,
      "loss": 0.5752,
      "step": 5988
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.883565285615466,
      "learning_rate": 1.8790958068663566e-05,
      "loss": 0.4026,
      "step": 5989
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3519479232005216,
      "learning_rate": 1.8790485248992055e-05,
      "loss": 0.3267,
      "step": 5990
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3198817585407246,
      "learning_rate": 1.8790012342836864e-05,
      "loss": 0.208,
      "step": 5991
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9233708952949242,
      "learning_rate": 1.878953935020263e-05,
      "loss": 0.4984,
      "step": 5992
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.6327394171042597,
      "learning_rate": 1.8789066271094026e-05,
      "loss": 0.9572,
      "step": 5993
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.22322279798506905,
      "learning_rate": 1.878859310551569e-05,
      "loss": 0.1261,
      "step": 5994
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6971398143333231,
      "learning_rate": 1.8788119853472287e-05,
      "loss": 0.2911,
      "step": 5995
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4229088107990714,
      "learning_rate": 1.8787646514968465e-05,
      "loss": 0.2573,
      "step": 5996
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3912187828688404,
      "learning_rate": 1.8787173090008887e-05,
      "loss": 0.3279,
      "step": 5997
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9356913690538848,
      "learning_rate": 1.878669957859821e-05,
      "loss": 0.4597,
      "step": 5998
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.2794647600899154,
      "learning_rate": 1.8786225980741087e-05,
      "loss": 0.5828,
      "step": 5999
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.31705254826731155,
      "learning_rate": 1.8785752296442186e-05,
      "loss": 0.0742,
      "step": 6000
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4366349995044753,
      "learning_rate": 1.878527852570616e-05,
      "loss": 0.3361,
      "step": 6001
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4506610561124253,
      "learning_rate": 1.878480466853768e-05,
      "loss": 0.2935,
      "step": 6002
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5491901246978791,
      "learning_rate": 1.8784330724941392e-05,
      "loss": 0.4066,
      "step": 6003
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.18184213903236518,
      "learning_rate": 1.8783856694921972e-05,
      "loss": 0.0767,
      "step": 6004
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2831207885410229,
      "learning_rate": 1.878338257848408e-05,
      "loss": 0.2073,
      "step": 6005
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8845601386881328,
      "learning_rate": 1.8782908375632378e-05,
      "loss": 0.5517,
      "step": 6006
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.024905396568511,
      "learning_rate": 1.8782434086371538e-05,
      "loss": 0.4377,
      "step": 6007
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.5126735021994334,
      "learning_rate": 1.878195971070622e-05,
      "loss": 0.3502,
      "step": 6008
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.28318008227672514,
      "learning_rate": 1.8781485248641086e-05,
      "loss": 0.2308,
      "step": 6009
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.6325080700715932,
      "learning_rate": 1.878101070018082e-05,
      "loss": 0.4156,
      "step": 6010
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9014975300963832,
      "learning_rate": 1.8780536065330075e-05,
      "loss": 0.4697,
      "step": 6011
    },
    {
      "epoch": 0.18,
      "grad_norm": 2.1576475623516154,
      "learning_rate": 1.878006134409353e-05,
      "loss": 0.8794,
      "step": 6012
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2459124536709529,
      "learning_rate": 1.877958653647585e-05,
      "loss": 0.1323,
      "step": 6013
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.39852112048444577,
      "learning_rate": 1.8779111642481712e-05,
      "loss": 0.2964,
      "step": 6014
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.786028347273671,
      "learning_rate": 1.8778636662115787e-05,
      "loss": 0.2962,
      "step": 6015
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0095487092852982,
      "learning_rate": 1.877816159538274e-05,
      "loss": 0.5487,
      "step": 6016
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.0208898731673444,
      "learning_rate": 1.8777686442287252e-05,
      "loss": 0.3943,
      "step": 6017
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7779588970007901,
      "learning_rate": 1.8777211202834e-05,
      "loss": 0.4091,
      "step": 6018
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38614350367187095,
      "learning_rate": 1.8776735877027653e-05,
      "loss": 0.2613,
      "step": 6019
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.41281842918635114,
      "learning_rate": 1.877626046487289e-05,
      "loss": 0.252,
      "step": 6020
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.546287560011774,
      "learning_rate": 1.877578496637439e-05,
      "loss": 0.3612,
      "step": 6021
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.2157147151818842,
      "learning_rate": 1.877530938153683e-05,
      "loss": 0.1038,
      "step": 6022
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.42504811140844956,
      "learning_rate": 1.877483371036489e-05,
      "loss": 0.3222,
      "step": 6023
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.8669908508264091,
      "learning_rate": 1.877435795286325e-05,
      "loss": 0.4139,
      "step": 6024
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4660798693472364,
      "learning_rate": 1.8773882109036587e-05,
      "loss": 0.291,
      "step": 6025
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.35653237670325266,
      "learning_rate": 1.8773406178889583e-05,
      "loss": 0.2766,
      "step": 6026
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7139551592979582,
      "learning_rate": 1.8772930162426926e-05,
      "loss": 0.3138,
      "step": 6027
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3799929612663439,
      "learning_rate": 1.8772454059653293e-05,
      "loss": 0.2452,
      "step": 6028
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.3044663168082973,
      "learning_rate": 1.877197787057337e-05,
      "loss": 0.7796,
      "step": 6029
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9735006028284361,
      "learning_rate": 1.877150159519184e-05,
      "loss": 0.4128,
      "step": 6030
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.4921817235145729,
      "learning_rate": 1.8771025233513396e-05,
      "loss": 0.4004,
      "step": 6031
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3587737479796957,
      "learning_rate": 1.877054878554272e-05,
      "loss": 0.1737,
      "step": 6032
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.35966529325727803,
      "learning_rate": 1.8770072251284496e-05,
      "loss": 0.3016,
      "step": 6033
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.32325732048353667,
      "learning_rate": 1.8769595630743417e-05,
      "loss": 0.18,
      "step": 6034
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.134397348522157,
      "learning_rate": 1.876911892392417e-05,
      "loss": 0.3815,
      "step": 6035
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.9737820695480179,
      "learning_rate": 1.8768642130831445e-05,
      "loss": 0.5074,
      "step": 6036
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.3355708732365255,
      "learning_rate": 1.8768165251469935e-05,
      "loss": 0.2333,
      "step": 6037
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.38822756596478303,
      "learning_rate": 1.876768828584433e-05,
      "loss": 0.3221,
      "step": 6038
    },
    {
      "epoch": 0.18,
      "grad_norm": 1.041695932759111,
      "learning_rate": 1.876721123395932e-05,
      "loss": 0.0743,
      "step": 6039
    },
    {
      "epoch": 0.18,
      "grad_norm": 0.7306077186761195,
      "learning_rate": 1.8766734095819603e-05,
      "loss": 0.5188,
      "step": 6040
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2873659772566235,
      "learning_rate": 1.8766256871429874e-05,
      "loss": 0.185,
      "step": 6041
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7315472165905239,
      "learning_rate": 1.876577956079482e-05,
      "loss": 0.4892,
      "step": 6042
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.23183722931009768,
      "learning_rate": 1.8765302163919148e-05,
      "loss": 0.0759,
      "step": 6043
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38805491936530817,
      "learning_rate": 1.8764824680807545e-05,
      "loss": 0.3269,
      "step": 6044
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4321461336630743,
      "learning_rate": 1.8764347111464712e-05,
      "loss": 0.1739,
      "step": 6045
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.429168181891678,
      "learning_rate": 1.8763869455895355e-05,
      "loss": 0.2786,
      "step": 6046
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.0707566981946024,
      "learning_rate": 1.876339171410416e-05,
      "loss": 0.7703,
      "step": 6047
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1040601517572535,
      "learning_rate": 1.8762913886095835e-05,
      "loss": 0.4229,
      "step": 6048
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5549332601974327,
      "learning_rate": 1.876243597187508e-05,
      "loss": 0.4207,
      "step": 6049
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3916091007861669,
      "learning_rate": 1.8761957971446597e-05,
      "loss": 0.1934,
      "step": 6050
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4040549931857184,
      "learning_rate": 1.876147988481509e-05,
      "loss": 0.3442,
      "step": 6051
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34962841666094957,
      "learning_rate": 1.876100171198526e-05,
      "loss": 0.1199,
      "step": 6052
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5119262517413212,
      "learning_rate": 1.8760523452961813e-05,
      "loss": 0.2614,
      "step": 6053
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34392007404566394,
      "learning_rate": 1.876004510774945e-05,
      "loss": 0.1304,
      "step": 6054
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47740630717662874,
      "learning_rate": 1.8759566676352884e-05,
      "loss": 0.3167,
      "step": 6055
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3757429354062646,
      "learning_rate": 1.8759088158776823e-05,
      "loss": 0.2707,
      "step": 6056
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.221087700692395,
      "learning_rate": 1.8758609555025964e-05,
      "loss": 0.5676,
      "step": 6057
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2084954000148296,
      "learning_rate": 1.8758130865105023e-05,
      "loss": 0.43,
      "step": 6058
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3681083384220518,
      "learning_rate": 1.875765208901871e-05,
      "loss": 0.1767,
      "step": 6059
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5828921056622647,
      "learning_rate": 1.8757173226771733e-05,
      "loss": 0.3664,
      "step": 6060
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3095290368229466,
      "learning_rate": 1.875669427836881e-05,
      "loss": 0.1463,
      "step": 6061
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5725941582217289,
      "learning_rate": 1.875621524381464e-05,
      "loss": 0.4044,
      "step": 6062
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6650237020635021,
      "learning_rate": 1.8755736123113944e-05,
      "loss": 0.2939,
      "step": 6063
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4149593347361284,
      "learning_rate": 1.875525691627144e-05,
      "loss": 0.2953,
      "step": 6064
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2191784050515988,
      "learning_rate": 1.8754777623291833e-05,
      "loss": 0.4565,
      "step": 6065
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2909044358267017,
      "learning_rate": 1.8754298244179845e-05,
      "loss": 0.5707,
      "step": 6066
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5192524535675964,
      "learning_rate": 1.8753818778940188e-05,
      "loss": 0.2917,
      "step": 6067
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41490363907515526,
      "learning_rate": 1.8753339227577586e-05,
      "loss": 0.306,
      "step": 6068
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3776344549075244,
      "learning_rate": 1.8752859590096747e-05,
      "loss": 0.259,
      "step": 6069
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.9220278365107488,
      "learning_rate": 1.8752379866502402e-05,
      "loss": 0.8852,
      "step": 6070
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3203950899968169,
      "learning_rate": 1.875190005679926e-05,
      "loss": 0.1572,
      "step": 6071
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6298215729251727,
      "learning_rate": 1.8751420160992045e-05,
      "loss": 0.1706,
      "step": 6072
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34465368417951653,
      "learning_rate": 1.875094017908548e-05,
      "loss": 0.2684,
      "step": 6073
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.35614650832509664,
      "learning_rate": 1.875046011108428e-05,
      "loss": 0.2529,
      "step": 6074
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4128514512523715,
      "learning_rate": 1.8749979956993183e-05,
      "loss": 0.5302,
      "step": 6075
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.5241090127305874,
      "learning_rate": 1.87494997168169e-05,
      "loss": 0.4201,
      "step": 6076
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48384335862270644,
      "learning_rate": 1.874901939056016e-05,
      "loss": 0.2805,
      "step": 6077
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5015522512903364,
      "learning_rate": 1.874853897822769e-05,
      "loss": 0.252,
      "step": 6078
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.186819887870039,
      "learning_rate": 1.8748058479824215e-05,
      "loss": 0.6519,
      "step": 6079
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3341006625888737,
      "learning_rate": 1.874757789535446e-05,
      "loss": 0.2664,
      "step": 6080
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9892310259116655,
      "learning_rate": 1.8747097224823155e-05,
      "loss": 0.5714,
      "step": 6081
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4693998916209929,
      "learning_rate": 1.8746616468235034e-05,
      "loss": 0.1963,
      "step": 6082
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5012853886327795,
      "learning_rate": 1.8746135625594817e-05,
      "loss": 0.4058,
      "step": 6083
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3523004804390795,
      "learning_rate": 1.8745654696907243e-05,
      "loss": 0.1948,
      "step": 6084
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41343542373590875,
      "learning_rate": 1.8745173682177036e-05,
      "loss": 0.2485,
      "step": 6085
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40779507657079583,
      "learning_rate": 1.8744692581408936e-05,
      "loss": 0.2922,
      "step": 6086
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5152472107029407,
      "learning_rate": 1.8744211394607675e-05,
      "loss": 0.2469,
      "step": 6087
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.305153540744711,
      "learning_rate": 1.8743730121777985e-05,
      "loss": 0.6677,
      "step": 6088
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1407832152285913,
      "learning_rate": 1.8743248762924597e-05,
      "loss": 0.5257,
      "step": 6089
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0396954721733336,
      "learning_rate": 1.874276731805225e-05,
      "loss": 0.5508,
      "step": 6090
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3176192148583114,
      "learning_rate": 1.8742285787165686e-05,
      "loss": 0.1981,
      "step": 6091
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3673686350464564,
      "learning_rate": 1.8741804170269636e-05,
      "loss": 0.3418,
      "step": 6092
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5678808801710794,
      "learning_rate": 1.874132246736884e-05,
      "loss": 0.4161,
      "step": 6093
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44289610380173444,
      "learning_rate": 1.874084067846804e-05,
      "loss": 0.2082,
      "step": 6094
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5209593237507684,
      "learning_rate": 1.8740358803571967e-05,
      "loss": 0.1889,
      "step": 6095
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4360820352760978,
      "learning_rate": 1.8739876842685372e-05,
      "loss": 0.3208,
      "step": 6096
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1745411809165287,
      "learning_rate": 1.8739394795812993e-05,
      "loss": 0.0515,
      "step": 6097
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3772649058708786,
      "learning_rate": 1.8738912662959573e-05,
      "loss": 0.2952,
      "step": 6098
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7837787461212853,
      "learning_rate": 1.8738430444129853e-05,
      "loss": 0.5666,
      "step": 6099
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3393902963393227,
      "learning_rate": 1.873794813932858e-05,
      "loss": 0.2007,
      "step": 6100
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9475144175348658,
      "learning_rate": 1.8737465748560497e-05,
      "loss": 0.5778,
      "step": 6101
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40269866302341806,
      "learning_rate": 1.8736983271830353e-05,
      "loss": 0.228,
      "step": 6102
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3366489099395249,
      "learning_rate": 1.873650070914289e-05,
      "loss": 0.2711,
      "step": 6103
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45944171880596235,
      "learning_rate": 1.873601806050286e-05,
      "loss": 0.1848,
      "step": 6104
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6728723098876679,
      "learning_rate": 1.873553532591501e-05,
      "loss": 0.4402,
      "step": 6105
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8461034805747535,
      "learning_rate": 1.8735052505384087e-05,
      "loss": 0.3536,
      "step": 6106
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9535778242865793,
      "learning_rate": 1.8734569598914846e-05,
      "loss": 0.6127,
      "step": 6107
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7300841369315415,
      "learning_rate": 1.8734086606512035e-05,
      "loss": 0.3474,
      "step": 6108
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1368551467626014,
      "learning_rate": 1.8733603528180403e-05,
      "loss": 0.4938,
      "step": 6109
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.29611661584502774,
      "learning_rate": 1.8733120363924707e-05,
      "loss": 0.2515,
      "step": 6110
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.28030979017012136,
      "learning_rate": 1.8732637113749703e-05,
      "loss": 0.1257,
      "step": 6111
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4876035050370896,
      "learning_rate": 1.8732153777660137e-05,
      "loss": 0.2491,
      "step": 6112
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5712001922421451,
      "learning_rate": 1.8731670355660773e-05,
      "loss": 0.2313,
      "step": 6113
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41771675473190983,
      "learning_rate": 1.873118684775636e-05,
      "loss": 0.2918,
      "step": 6114
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.36963106606032387,
      "learning_rate": 1.873070325395166e-05,
      "loss": 0.2688,
      "step": 6115
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4076029785448856,
      "learning_rate": 1.8730219574251426e-05,
      "loss": 0.6007,
      "step": 6116
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7549281512947983,
      "learning_rate": 1.872973580866042e-05,
      "loss": 0.3981,
      "step": 6117
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4995420537660822,
      "learning_rate": 1.87292519571834e-05,
      "loss": 0.2858,
      "step": 6118
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4496154152545755,
      "learning_rate": 1.8728768019825126e-05,
      "loss": 0.293,
      "step": 6119
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.1014617098160406,
      "learning_rate": 1.8728283996590365e-05,
      "loss": 0.9689,
      "step": 6120
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2804967329249628,
      "learning_rate": 1.872779988748387e-05,
      "loss": 0.2089,
      "step": 6121
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.49347258983881337,
      "learning_rate": 1.872731569251041e-05,
      "loss": 0.1839,
      "step": 6122
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38509869162826965,
      "learning_rate": 1.8726831411674746e-05,
      "loss": 0.2213,
      "step": 6123
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.20972213279478,
      "learning_rate": 1.872634704498164e-05,
      "loss": 0.3823,
      "step": 6124
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1031847781882096,
      "learning_rate": 1.8725862592435862e-05,
      "loss": 0.5741,
      "step": 6125
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5412700885404875,
      "learning_rate": 1.872537805404218e-05,
      "loss": 0.343,
      "step": 6126
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4673410897060837,
      "learning_rate": 1.8724893429805356e-05,
      "loss": 0.3005,
      "step": 6127
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.35981346731688296,
      "learning_rate": 1.8724408719730158e-05,
      "loss": 0.241,
      "step": 6128
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8891822819309148,
      "learning_rate": 1.8723923923821358e-05,
      "loss": 0.9098,
      "step": 6129
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2798001769646577,
      "learning_rate": 1.8723439042083722e-05,
      "loss": 0.1218,
      "step": 6130
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7269554370755121,
      "learning_rate": 1.872295407452202e-05,
      "loss": 0.327,
      "step": 6131
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38739881515794405,
      "learning_rate": 1.8722469021141032e-05,
      "loss": 0.2489,
      "step": 6132
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0200080767776203,
      "learning_rate": 1.872198388194552e-05,
      "loss": 0.6307,
      "step": 6133
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.42002698196957766,
      "learning_rate": 1.8721498656940263e-05,
      "loss": 0.3287,
      "step": 6134
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9677120269139616,
      "learning_rate": 1.872101334613003e-05,
      "loss": 0.6274,
      "step": 6135
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2625311086966073,
      "learning_rate": 1.87205279495196e-05,
      "loss": 0.0776,
      "step": 6136
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41617952851835033,
      "learning_rate": 1.8720042467113742e-05,
      "loss": 0.2547,
      "step": 6137
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6354262356336156,
      "learning_rate": 1.871955689891724e-05,
      "loss": 0.7611,
      "step": 6138
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.29574017792884727,
      "learning_rate": 1.871907124493487e-05,
      "loss": 0.2179,
      "step": 6139
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45623972598248086,
      "learning_rate": 1.8718585505171408e-05,
      "loss": 0.2969,
      "step": 6140
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2940816653945335,
      "learning_rate": 1.871809967963163e-05,
      "loss": 0.2126,
      "step": 6141
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6911184172408515,
      "learning_rate": 1.871761376832032e-05,
      "loss": 0.5379,
      "step": 6142
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7934057290729712,
      "learning_rate": 1.871712777124226e-05,
      "loss": 0.4396,
      "step": 6143
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.686769664951621,
      "learning_rate": 1.8716641688402225e-05,
      "loss": 0.8551,
      "step": 6144
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.33474357045464187,
      "learning_rate": 1.8716155519805003e-05,
      "loss": 0.2326,
      "step": 6145
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.45708943974945354,
      "learning_rate": 1.8715669265455375e-05,
      "loss": 0.3372,
      "step": 6146
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8034617265470813,
      "learning_rate": 1.8715182925358122e-05,
      "loss": 0.4942,
      "step": 6147
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4480432237056735,
      "learning_rate": 1.8714696499518034e-05,
      "loss": 0.7363,
      "step": 6148
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2785537280827098,
      "learning_rate": 1.8714209987939896e-05,
      "loss": 0.0742,
      "step": 6149
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.30813291747198857,
      "learning_rate": 1.871372339062849e-05,
      "loss": 0.2643,
      "step": 6150
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9852612547024999,
      "learning_rate": 1.8713236707588608e-05,
      "loss": 0.587,
      "step": 6151
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3212218614297947,
      "learning_rate": 1.8712749938825036e-05,
      "loss": 0.3128,
      "step": 6152
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3392252103946685,
      "learning_rate": 1.8712263084342563e-05,
      "loss": 0.7066,
      "step": 6153
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3344488026832723,
      "learning_rate": 1.8711776144145977e-05,
      "loss": 0.1899,
      "step": 6154
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6417308632521942,
      "learning_rate": 1.871128911824007e-05,
      "loss": 0.4069,
      "step": 6155
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2774359300373879,
      "learning_rate": 1.8710802006629638e-05,
      "loss": 0.491,
      "step": 6156
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3782774889875095,
      "learning_rate": 1.8710314809319467e-05,
      "loss": 0.3155,
      "step": 6157
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.26231376030034426,
      "learning_rate": 1.8709827526314356e-05,
      "loss": 0.1292,
      "step": 6158
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0814070941182057,
      "learning_rate": 1.8709340157619093e-05,
      "loss": 0.4536,
      "step": 6159
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37532987395574396,
      "learning_rate": 1.8708852703238478e-05,
      "loss": 0.2939,
      "step": 6160
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6272118155362558,
      "learning_rate": 1.8708365163177302e-05,
      "loss": 0.5422,
      "step": 6161
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.33922429529138975,
      "learning_rate": 1.8707877537440367e-05,
      "loss": 0.282,
      "step": 6162
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.33573112793152926,
      "learning_rate": 1.8707389826032463e-05,
      "loss": 0.08,
      "step": 6163
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4663292287954858,
      "learning_rate": 1.8706902028958397e-05,
      "loss": 0.3147,
      "step": 6164
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9575677107906226,
      "learning_rate": 1.870641414622296e-05,
      "loss": 0.4925,
      "step": 6165
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1411712437162456,
      "learning_rate": 1.8705926177830963e-05,
      "loss": 0.6306,
      "step": 6166
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7178290321780224,
      "learning_rate": 1.8705438123787193e-05,
      "loss": 0.413,
      "step": 6167
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4330144945860777,
      "learning_rate": 1.870494998409646e-05,
      "loss": 0.2695,
      "step": 6168
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.28623105941464955,
      "learning_rate": 1.8704461758763566e-05,
      "loss": 0.2721,
      "step": 6169
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3982223431093788,
      "learning_rate": 1.870397344779331e-05,
      "loss": 0.1842,
      "step": 6170
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0104446969398695,
      "learning_rate": 1.87034850511905e-05,
      "loss": 0.3165,
      "step": 6171
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9555693706646422,
      "learning_rate": 1.8702996568959945e-05,
      "loss": 0.5437,
      "step": 6172
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3468891293660396,
      "learning_rate": 1.870250800110644e-05,
      "loss": 0.2388,
      "step": 6173
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3885223104513158,
      "learning_rate": 1.8702019347634804e-05,
      "loss": 0.7611,
      "step": 6174
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3468230267999982,
      "learning_rate": 1.870153060854983e-05,
      "loss": 0.2694,
      "step": 6175
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.661488872405015,
      "learning_rate": 1.8701041783856342e-05,
      "loss": 0.4034,
      "step": 6176
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43758140072720964,
      "learning_rate": 1.870055287355914e-05,
      "loss": 0.2854,
      "step": 6177
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4556089037794458,
      "learning_rate": 1.8700063877663034e-05,
      "loss": 0.294,
      "step": 6178
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4394627739467201,
      "learning_rate": 1.869957479617284e-05,
      "loss": 0.1964,
      "step": 6179
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34670056331023624,
      "learning_rate": 1.8699085629093363e-05,
      "loss": 0.2815,
      "step": 6180
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7730389220256029,
      "learning_rate": 1.869859637642942e-05,
      "loss": 0.4404,
      "step": 6181
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34422393979969695,
      "learning_rate": 1.8698107038185826e-05,
      "loss": 0.2374,
      "step": 6182
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2235984796989312,
      "learning_rate": 1.869761761436739e-05,
      "loss": 0.7498,
      "step": 6183
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7582301401093604,
      "learning_rate": 1.8697128104978933e-05,
      "loss": 0.4068,
      "step": 6184
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.85613828865739,
      "learning_rate": 1.8696638510025267e-05,
      "loss": 0.5105,
      "step": 6185
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.44615581179948177,
      "learning_rate": 1.869614882951121e-05,
      "loss": 0.0772,
      "step": 6186
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41208218166105,
      "learning_rate": 1.8695659063441578e-05,
      "loss": 0.2974,
      "step": 6187
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.24220183305199816,
      "learning_rate": 1.869516921182119e-05,
      "loss": 0.1669,
      "step": 6188
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3499576464592244,
      "learning_rate": 1.869467927465487e-05,
      "loss": 0.6172,
      "step": 6189
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7784579484307603,
      "learning_rate": 1.869418925194743e-05,
      "loss": 0.7714,
      "step": 6190
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.312437052156377,
      "learning_rate": 1.8693699143703697e-05,
      "loss": 0.2087,
      "step": 6191
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9057244988315589,
      "learning_rate": 1.8693208949928495e-05,
      "loss": 0.5825,
      "step": 6192
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3577098028655242,
      "learning_rate": 1.869271867062664e-05,
      "loss": 0.3314,
      "step": 6193
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9689173096699598,
      "learning_rate": 1.8692228305802957e-05,
      "loss": 0.6216,
      "step": 6194
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37681064672092046,
      "learning_rate": 1.8691737855462274e-05,
      "loss": 0.2134,
      "step": 6195
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5356420944179163,
      "learning_rate": 1.869124731960941e-05,
      "loss": 0.3981,
      "step": 6196
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.26187493345526275,
      "learning_rate": 1.8690756698249198e-05,
      "loss": 0.1153,
      "step": 6197
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5683972447303651,
      "learning_rate": 1.869026599138646e-05,
      "loss": 0.3772,
      "step": 6198
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.382976813405383,
      "learning_rate": 1.868977519902603e-05,
      "loss": 0.2565,
      "step": 6199
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3796053986422622,
      "learning_rate": 1.8689284321172732e-05,
      "loss": 0.2849,
      "step": 6200
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3114116250240477,
      "learning_rate": 1.8688793357831393e-05,
      "loss": 0.4105,
      "step": 6201
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8521887662422315,
      "learning_rate": 1.8688302309006846e-05,
      "loss": 0.5484,
      "step": 6202
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6860838807645604,
      "learning_rate": 1.868781117470392e-05,
      "loss": 0.4248,
      "step": 6203
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.25583312772757116,
      "learning_rate": 1.8687319954927454e-05,
      "loss": 0.2069,
      "step": 6204
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40592371594184096,
      "learning_rate": 1.8686828649682272e-05,
      "loss": 0.3274,
      "step": 6205
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2908244769470341,
      "learning_rate": 1.8686337258973214e-05,
      "loss": 0.0759,
      "step": 6206
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5302677714347163,
      "learning_rate": 1.868584578280511e-05,
      "loss": 0.3317,
      "step": 6207
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5253394041834963,
      "learning_rate": 1.86853542211828e-05,
      "loss": 0.2938,
      "step": 6208
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5212700408106631,
      "learning_rate": 1.8684862574111113e-05,
      "loss": 0.3328,
      "step": 6209
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.482868296116695,
      "learning_rate": 1.8684370841594896e-05,
      "loss": 0.3304,
      "step": 6210
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4020365639798683,
      "learning_rate": 1.8683879023638973e-05,
      "loss": 0.3554,
      "step": 6211
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7602246773202742,
      "learning_rate": 1.8683387120248198e-05,
      "loss": 0.4142,
      "step": 6212
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0106226781459138,
      "learning_rate": 1.8682895131427397e-05,
      "loss": 0.4576,
      "step": 6213
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3120614954357067,
      "learning_rate": 1.8682403057181423e-05,
      "loss": 0.2213,
      "step": 6214
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4952544135442319,
      "learning_rate": 1.8681910897515107e-05,
      "loss": 0.216,
      "step": 6215
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3542173544307042,
      "learning_rate": 1.8681418652433297e-05,
      "loss": 0.2868,
      "step": 6216
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0570069455689095,
      "learning_rate": 1.8680926321940832e-05,
      "loss": 0.5402,
      "step": 6217
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43851726495327115,
      "learning_rate": 1.868043390604256e-05,
      "loss": 0.2764,
      "step": 6218
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4912564397190085,
      "learning_rate": 1.867994140474332e-05,
      "loss": 0.3604,
      "step": 6219
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4618645848772921,
      "learning_rate": 1.867944881804796e-05,
      "loss": 0.3047,
      "step": 6220
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1787037084733587,
      "learning_rate": 1.867895614596133e-05,
      "loss": 0.3218,
      "step": 6221
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37730677833875786,
      "learning_rate": 1.867846338848827e-05,
      "loss": 0.2994,
      "step": 6222
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3881246513598848,
      "learning_rate": 1.8677970545633635e-05,
      "loss": 0.2239,
      "step": 6223
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5018237267579067,
      "learning_rate": 1.867747761740227e-05,
      "loss": 0.8094,
      "step": 6224
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1397662001239335,
      "learning_rate": 1.867698460379902e-05,
      "loss": 0.551,
      "step": 6225
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8672392798225634,
      "learning_rate": 1.8676491504828745e-05,
      "loss": 0.6522,
      "step": 6226
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3176443643569876,
      "learning_rate": 1.867599832049629e-05,
      "loss": 0.2107,
      "step": 6227
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47521494251432767,
      "learning_rate": 1.867550505080651e-05,
      "loss": 0.415,
      "step": 6228
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3976291463537992,
      "learning_rate": 1.8675011695764257e-05,
      "loss": 0.2772,
      "step": 6229
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.25122320870938575,
      "learning_rate": 1.8674518255374384e-05,
      "loss": 0.0973,
      "step": 6230
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6977236459540774,
      "learning_rate": 1.8674024729641743e-05,
      "loss": 0.3643,
      "step": 6231
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3432680138239853,
      "learning_rate": 1.8673531118571197e-05,
      "loss": 0.2264,
      "step": 6232
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7737372815904806,
      "learning_rate": 1.8673037422167598e-05,
      "loss": 0.6933,
      "step": 6233
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.35278432010585464,
      "learning_rate": 1.86725436404358e-05,
      "loss": 0.2833,
      "step": 6234
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7227818941410866,
      "learning_rate": 1.867204977338067e-05,
      "loss": 0.5468,
      "step": 6235
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.31759235198283675,
      "learning_rate": 1.867155582100705e-05,
      "loss": 0.1747,
      "step": 6236
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6124008912902631,
      "learning_rate": 1.8671061783319818e-05,
      "loss": 0.426,
      "step": 6237
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.26029090944557776,
      "learning_rate": 1.8670567660323826e-05,
      "loss": 0.158,
      "step": 6238
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5090628767397022,
      "learning_rate": 1.8670073452023935e-05,
      "loss": 0.3539,
      "step": 6239
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4081503769781361,
      "learning_rate": 1.866957915842501e-05,
      "loss": 0.2604,
      "step": 6240
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.43742447338563517,
      "learning_rate": 1.8669084779531913e-05,
      "loss": 0.2869,
      "step": 6241
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.49397601548101133,
      "learning_rate": 1.8668590315349508e-05,
      "loss": 0.2698,
      "step": 6242
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6455864637212193,
      "learning_rate": 1.8668095765882656e-05,
      "loss": 0.4157,
      "step": 6243
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.076269198254353,
      "learning_rate": 1.8667601131136228e-05,
      "loss": 0.5331,
      "step": 6244
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3345554003584397,
      "learning_rate": 1.8667106411115087e-05,
      "loss": 0.2208,
      "step": 6245
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3769954293751695,
      "learning_rate": 1.8666611605824102e-05,
      "loss": 0.3075,
      "step": 6246
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2771512135439875,
      "learning_rate": 1.8666116715268137e-05,
      "loss": 0.1992,
      "step": 6247
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5250291837470824,
      "learning_rate": 1.866562173945207e-05,
      "loss": 0.2787,
      "step": 6248
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6869214159253968,
      "learning_rate": 1.8665126678380762e-05,
      "loss": 0.1727,
      "step": 6249
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38052092229763373,
      "learning_rate": 1.8664631532059086e-05,
      "loss": 0.2866,
      "step": 6250
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.1507900717600124,
      "learning_rate": 1.8664136300491916e-05,
      "loss": 0.5303,
      "step": 6251
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6321370248915759,
      "learning_rate": 1.866364098368412e-05,
      "loss": 0.3908,
      "step": 6252
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5566718446047482,
      "learning_rate": 1.8663145581640575e-05,
      "loss": 0.3408,
      "step": 6253
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6501979748944025,
      "learning_rate": 1.866265009436615e-05,
      "loss": 0.3584,
      "step": 6254
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3856157084631518,
      "learning_rate": 1.8662154521865726e-05,
      "loss": 0.2662,
      "step": 6255
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2691852972518657,
      "learning_rate": 1.8661658864144178e-05,
      "loss": 0.1229,
      "step": 6256
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4561887309988745,
      "learning_rate": 1.8661163121206375e-05,
      "loss": 0.2348,
      "step": 6257
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.2972450609819134,
      "learning_rate": 1.86606672930572e-05,
      "loss": 0.2355,
      "step": 6258
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.49590516656547834,
      "learning_rate": 1.866017137970153e-05,
      "loss": 0.3273,
      "step": 6259
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.7528845845990884,
      "learning_rate": 1.865967538114425e-05,
      "loss": 0.3285,
      "step": 6260
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9190418524350153,
      "learning_rate": 1.8659179297390227e-05,
      "loss": 0.5723,
      "step": 6261
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7460844430562309,
      "learning_rate": 1.8658683128444352e-05,
      "loss": 0.4362,
      "step": 6262
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7631311949639207,
      "learning_rate": 1.8658186874311506e-05,
      "loss": 0.3378,
      "step": 6263
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.32121930383787467,
      "learning_rate": 1.8657690534996565e-05,
      "loss": 0.2531,
      "step": 6264
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5404862552444508,
      "learning_rate": 1.8657194110504418e-05,
      "loss": 0.4158,
      "step": 6265
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6797884527725317,
      "learning_rate": 1.8656697600839947e-05,
      "loss": 0.1152,
      "step": 6266
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4835664467963472,
      "learning_rate": 1.8656201006008034e-05,
      "loss": 0.6937,
      "step": 6267
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.35460358449890433,
      "learning_rate": 1.8655704326013565e-05,
      "loss": 0.1793,
      "step": 6268
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6182578060931375,
      "learning_rate": 1.8655207560861436e-05,
      "loss": 0.3765,
      "step": 6269
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3677467036366645,
      "learning_rate": 1.8654710710556522e-05,
      "loss": 0.3477,
      "step": 6270
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8744082938094931,
      "learning_rate": 1.8654213775103716e-05,
      "loss": 0.4527,
      "step": 6271
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.0173025932756063,
      "learning_rate": 1.865371675450791e-05,
      "loss": 0.4545,
      "step": 6272
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.36894366482702634,
      "learning_rate": 1.8653219648773988e-05,
      "loss": 0.2584,
      "step": 6273
    },
    {
      "epoch": 0.19,
      "grad_norm": 7.381016409044067,
      "learning_rate": 1.865272245790685e-05,
      "loss": 0.7513,
      "step": 6274
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3296025024976637,
      "learning_rate": 1.8652225181911373e-05,
      "loss": 0.0761,
      "step": 6275
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.40519837471982906,
      "learning_rate": 1.8651727820792463e-05,
      "loss": 0.3391,
      "step": 6276
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37161094069368533,
      "learning_rate": 1.865123037455501e-05,
      "loss": 0.2113,
      "step": 6277
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8100229970087589,
      "learning_rate": 1.86507328432039e-05,
      "loss": 0.5564,
      "step": 6278
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7106037496013654,
      "learning_rate": 1.8650235226744034e-05,
      "loss": 0.4101,
      "step": 6279
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8511305573425472,
      "learning_rate": 1.8649737525180312e-05,
      "loss": 0.6286,
      "step": 6280
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6506966411797713,
      "learning_rate": 1.8649239738517623e-05,
      "loss": 0.2094,
      "step": 6281
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.31708420939263654,
      "learning_rate": 1.864874186676087e-05,
      "loss": 0.2672,
      "step": 6282
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.6663188548660177,
      "learning_rate": 1.8648243909914945e-05,
      "loss": 0.7308,
      "step": 6283
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3943484425854638,
      "learning_rate": 1.8647745867984755e-05,
      "loss": 0.0784,
      "step": 6284
    },
    {
      "epoch": 0.19,
      "grad_norm": 13.733582529629462,
      "learning_rate": 1.864724774097519e-05,
      "loss": 0.5781,
      "step": 6285
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5349540186397717,
      "learning_rate": 1.8646749528891163e-05,
      "loss": 0.2167,
      "step": 6286
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8295385960522963,
      "learning_rate": 1.8646251231737563e-05,
      "loss": 0.6025,
      "step": 6287
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.38761329396709737,
      "learning_rate": 1.8645752849519303e-05,
      "loss": 0.3067,
      "step": 6288
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2334030241404077,
      "learning_rate": 1.8645254382241283e-05,
      "loss": 0.7106,
      "step": 6289
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.427696243320715,
      "learning_rate": 1.86447558299084e-05,
      "loss": 0.1592,
      "step": 6290
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46472531765400044,
      "learning_rate": 1.8644257192525567e-05,
      "loss": 0.3368,
      "step": 6291
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8229433499516053,
      "learning_rate": 1.864375847009769e-05,
      "loss": 0.4245,
      "step": 6292
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5776505964954551,
      "learning_rate": 1.864325966262967e-05,
      "loss": 0.257,
      "step": 6293
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46950291177592257,
      "learning_rate": 1.8642760770126416e-05,
      "loss": 0.2264,
      "step": 6294
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5027703834011237,
      "learning_rate": 1.8642261792592838e-05,
      "loss": 0.1914,
      "step": 6295
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6453134290179617,
      "learning_rate": 1.864176273003385e-05,
      "loss": 0.3794,
      "step": 6296
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.073773500683213,
      "learning_rate": 1.8641263582454355e-05,
      "loss": 0.4472,
      "step": 6297
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4081081505046995,
      "learning_rate": 1.8640764349859262e-05,
      "loss": 0.8913,
      "step": 6298
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4192174475206819,
      "learning_rate": 1.864026503225349e-05,
      "loss": 0.2243,
      "step": 6299
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4295455519532564,
      "learning_rate": 1.8639765629641943e-05,
      "loss": 0.2926,
      "step": 6300
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4106810279046402,
      "learning_rate": 1.8639266142029543e-05,
      "loss": 0.2162,
      "step": 6301
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3848079940826288,
      "learning_rate": 1.8638766569421198e-05,
      "loss": 0.7179,
      "step": 6302
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4032011130742023,
      "learning_rate": 1.8638266911821824e-05,
      "loss": 0.2008,
      "step": 6303
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4406386541336442,
      "learning_rate": 1.8637767169236342e-05,
      "loss": 0.2954,
      "step": 6304
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.34893398079203714,
      "learning_rate": 1.8637267341669664e-05,
      "loss": 0.3075,
      "step": 6305
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5469677038976131,
      "learning_rate": 1.8636767429126702e-05,
      "loss": 0.3602,
      "step": 6306
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8422985222928283,
      "learning_rate": 1.8636267431612384e-05,
      "loss": 0.4721,
      "step": 6307
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.433315434034574,
      "learning_rate": 1.8635767349131628e-05,
      "loss": 0.172,
      "step": 6308
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.41549541388559375,
      "learning_rate": 1.8635267181689344e-05,
      "loss": 0.3067,
      "step": 6309
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.683263011902728,
      "learning_rate": 1.8634766929290465e-05,
      "loss": 0.0504,
      "step": 6310
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.48412936915060784,
      "learning_rate": 1.8634266591939908e-05,
      "loss": 0.3627,
      "step": 6311
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4215505710218744,
      "learning_rate": 1.8633766169642595e-05,
      "loss": 0.3349,
      "step": 6312
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6879614851163006,
      "learning_rate": 1.863326566240345e-05,
      "loss": 0.3026,
      "step": 6313
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.470071742654345,
      "learning_rate": 1.8632765070227393e-05,
      "loss": 0.2901,
      "step": 6314
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6160839047928862,
      "learning_rate": 1.8632264393119354e-05,
      "loss": 0.17,
      "step": 6315
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.958276376182741,
      "learning_rate": 1.8631763631084258e-05,
      "loss": 0.6542,
      "step": 6316
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.368501833360978,
      "learning_rate": 1.8631262784127034e-05,
      "loss": 0.3013,
      "step": 6317
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.39824751885094595,
      "learning_rate": 1.8630761852252606e-05,
      "loss": 0.2002,
      "step": 6318
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5694563358419615,
      "learning_rate": 1.86302608354659e-05,
      "loss": 0.361,
      "step": 6319
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.729091491170244,
      "learning_rate": 1.862975973377185e-05,
      "loss": 0.3943,
      "step": 6320
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8642056878180497,
      "learning_rate": 1.8629258547175383e-05,
      "loss": 0.3811,
      "step": 6321
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5650174801224557,
      "learning_rate": 1.8628757275681434e-05,
      "loss": 0.3348,
      "step": 6322
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3077773241652298,
      "learning_rate": 1.862825591929493e-05,
      "loss": 0.2578,
      "step": 6323
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3024164043403427,
      "learning_rate": 1.8627754478020803e-05,
      "loss": 0.196,
      "step": 6324
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2002646848904153,
      "learning_rate": 1.862725295186399e-05,
      "loss": 0.6085,
      "step": 6325
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0745280353029187,
      "learning_rate": 1.8626751340829424e-05,
      "loss": 0.7952,
      "step": 6326
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.862679687850583,
      "learning_rate": 1.8626249644922038e-05,
      "loss": 0.2117,
      "step": 6327
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.4079680719703374,
      "learning_rate": 1.8625747864146776e-05,
      "loss": 0.7288,
      "step": 6328
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.49551622750715385,
      "learning_rate": 1.862524599850856e-05,
      "loss": 0.2986,
      "step": 6329
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5418686227674712,
      "learning_rate": 1.8624744048012343e-05,
      "loss": 0.3853,
      "step": 6330
    },
    {
      "epoch": 0.19,
      "grad_norm": 5.947276522230348,
      "learning_rate": 1.862424201266305e-05,
      "loss": 0.2319,
      "step": 6331
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4467009222007985,
      "learning_rate": 1.862373989246563e-05,
      "loss": 0.3183,
      "step": 6332
    },
    {
      "epoch": 0.19,
      "grad_norm": 6.3182331472752145,
      "learning_rate": 1.8623237687425017e-05,
      "loss": 0.1116,
      "step": 6333
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.3480787490592696,
      "learning_rate": 1.862273539754616e-05,
      "loss": 0.6649,
      "step": 6334
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.745972047721449,
      "learning_rate": 1.8622233022833985e-05,
      "loss": 0.3022,
      "step": 6335
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5887483227621811,
      "learning_rate": 1.862173056329345e-05,
      "loss": 0.2209,
      "step": 6336
    },
    {
      "epoch": 0.19,
      "grad_norm": 2.0936623629830775,
      "learning_rate": 1.8621228018929493e-05,
      "loss": 0.8389,
      "step": 6337
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8679344705859244,
      "learning_rate": 1.862072538974706e-05,
      "loss": 0.4206,
      "step": 6338
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7891182231230115,
      "learning_rate": 1.862022267575109e-05,
      "loss": 0.5374,
      "step": 6339
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.9163695526214712,
      "learning_rate": 1.8619719876946535e-05,
      "loss": 0.2182,
      "step": 6340
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6514522295759408,
      "learning_rate": 1.861921699333834e-05,
      "loss": 0.3741,
      "step": 6341
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7752651365102224,
      "learning_rate": 1.861871402493145e-05,
      "loss": 0.2553,
      "step": 6342
    },
    {
      "epoch": 0.19,
      "grad_norm": 3.4602113570503867,
      "learning_rate": 1.861821097173082e-05,
      "loss": 0.8228,
      "step": 6343
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7421252885015422,
      "learning_rate": 1.8617707833741386e-05,
      "loss": 0.1292,
      "step": 6344
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.046213434970419,
      "learning_rate": 1.8617204610968114e-05,
      "loss": 0.4834,
      "step": 6345
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.47363315568846387,
      "learning_rate": 1.8616701303415947e-05,
      "loss": 0.323,
      "step": 6346
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4059509672792334,
      "learning_rate": 1.8616197911089836e-05,
      "loss": 0.3111,
      "step": 6347
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.8883938073782316,
      "learning_rate": 1.8615694433994734e-05,
      "loss": 0.8014,
      "step": 6348
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.6008842073257329,
      "learning_rate": 1.8615190872135596e-05,
      "loss": 0.0847,
      "step": 6349
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4398062600923459,
      "learning_rate": 1.8614687225517377e-05,
      "loss": 0.3058,
      "step": 6350
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5123201618985711,
      "learning_rate": 1.8614183494145027e-05,
      "loss": 0.1742,
      "step": 6351
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.735585043224765,
      "learning_rate": 1.8613679678023508e-05,
      "loss": 0.8198,
      "step": 6352
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3238427205792943,
      "learning_rate": 1.8613175777157772e-05,
      "loss": 0.219,
      "step": 6353
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.5212872812756258,
      "learning_rate": 1.861267179155278e-05,
      "loss": 0.2675,
      "step": 6354
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4737171264753768,
      "learning_rate": 1.8612167721213488e-05,
      "loss": 0.3261,
      "step": 6355
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7020096228281101,
      "learning_rate": 1.861166356614486e-05,
      "loss": 0.5125,
      "step": 6356
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.8931423189452027,
      "learning_rate": 1.8611159326351847e-05,
      "loss": 0.3457,
      "step": 6357
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.2213366791017295,
      "learning_rate": 1.8610655001839418e-05,
      "loss": 0.589,
      "step": 6358
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.3054168808515658,
      "learning_rate": 1.861015059261253e-05,
      "loss": 0.2351,
      "step": 6359
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4551331925659815,
      "learning_rate": 1.8609646098676148e-05,
      "loss": 0.2544,
      "step": 6360
    },
    {
      "epoch": 0.19,
      "grad_norm": 1.5665593520748058,
      "learning_rate": 1.8609141520035238e-05,
      "loss": 0.9741,
      "step": 6361
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4424392140799255,
      "learning_rate": 1.8608636856694756e-05,
      "loss": 0.2523,
      "step": 6362
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4937050383344157,
      "learning_rate": 1.8608132108659677e-05,
      "loss": 0.3561,
      "step": 6363
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.46608425480588395,
      "learning_rate": 1.8607627275934957e-05,
      "loss": 0.2858,
      "step": 6364
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.4223030124188503,
      "learning_rate": 1.860712235852557e-05,
      "loss": 0.324,
      "step": 6365
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.37978127819732393,
      "learning_rate": 1.8606617356436484e-05,
      "loss": 0.126,
      "step": 6366
    },
    {
      "epoch": 0.19,
      "grad_norm": 0.7939766355798772,
      "learning_rate": 1.8606112269672662e-05,
      "loss": 0.3262,
      "step": 6367
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39998620212978414,
      "learning_rate": 1.8605607098239074e-05,
      "loss": 0.2262,
      "step": 6368
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.6443504510621998,
      "learning_rate": 1.860510184214069e-05,
      "loss": 0.7809,
      "step": 6369
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5213671014821792,
      "learning_rate": 1.860459650138249e-05,
      "loss": 0.329,
      "step": 6370
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5094888338316382,
      "learning_rate": 1.8604091075969432e-05,
      "loss": 0.4092,
      "step": 6371
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.30311541999215436,
      "learning_rate": 1.8603585565906495e-05,
      "loss": 0.0738,
      "step": 6372
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4457463352868783,
      "learning_rate": 1.860307997119865e-05,
      "loss": 0.2881,
      "step": 6373
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3965196288885414,
      "learning_rate": 1.860257429185088e-05,
      "loss": 0.1821,
      "step": 6374
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.990993129118004,
      "learning_rate": 1.8602068527868153e-05,
      "loss": 0.3636,
      "step": 6375
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5955333129627772,
      "learning_rate": 1.860156267925544e-05,
      "loss": 0.4164,
      "step": 6376
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3315558818569055,
      "learning_rate": 1.8601056746017726e-05,
      "loss": 0.2011,
      "step": 6377
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5748579297954352,
      "learning_rate": 1.8600550728159987e-05,
      "loss": 0.4456,
      "step": 6378
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1307521760667976,
      "learning_rate": 1.86000446256872e-05,
      "loss": 0.5935,
      "step": 6379
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7054998362486838,
      "learning_rate": 1.8599538438604344e-05,
      "loss": 0.4803,
      "step": 6380
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48229114891436614,
      "learning_rate": 1.8599032166916397e-05,
      "loss": 0.2655,
      "step": 6381
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4354600598989372,
      "learning_rate": 1.8598525810628345e-05,
      "loss": 0.3143,
      "step": 6382
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.29438432874359505,
      "learning_rate": 1.8598019369745164e-05,
      "loss": 0.2218,
      "step": 6383
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5264874128192708,
      "learning_rate": 1.859751284427184e-05,
      "loss": 0.2453,
      "step": 6384
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7798694272081758,
      "learning_rate": 1.8597006234213356e-05,
      "loss": 0.2496,
      "step": 6385
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3763334282299684,
      "learning_rate": 1.8596499539574696e-05,
      "loss": 0.3072,
      "step": 6386
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0838206587637027,
      "learning_rate": 1.8595992760360847e-05,
      "loss": 0.4225,
      "step": 6387
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5208504665418049,
      "learning_rate": 1.859548589657679e-05,
      "loss": 0.3346,
      "step": 6388
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5783018988156928,
      "learning_rate": 1.8594978948227513e-05,
      "loss": 0.3783,
      "step": 6389
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35256572524993446,
      "learning_rate": 1.859447191531801e-05,
      "loss": 0.2291,
      "step": 6390
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6041641506142165,
      "learning_rate": 1.8593964797853262e-05,
      "loss": 0.3788,
      "step": 6391
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39668664869347775,
      "learning_rate": 1.859345759583826e-05,
      "loss": 0.1849,
      "step": 6392
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49580933097175045,
      "learning_rate": 1.8592950309277993e-05,
      "loss": 0.2373,
      "step": 6393
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.34415327373294835,
      "learning_rate": 1.8592442938177453e-05,
      "loss": 0.2562,
      "step": 6394
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0918910847527006,
      "learning_rate": 1.8591935482541636e-05,
      "loss": 0.5309,
      "step": 6395
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39733714088047023,
      "learning_rate": 1.8591427942375528e-05,
      "loss": 0.2462,
      "step": 6396
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9313573903902855,
      "learning_rate": 1.8590920317684125e-05,
      "loss": 0.5659,
      "step": 6397
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8162508959128169,
      "learning_rate": 1.859041260847242e-05,
      "loss": 0.4336,
      "step": 6398
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1011315547456633,
      "learning_rate": 1.858990481474541e-05,
      "loss": 0.5005,
      "step": 6399
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3412518548072308,
      "learning_rate": 1.858939693650809e-05,
      "loss": 0.2211,
      "step": 6400
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2759028824904146,
      "learning_rate": 1.8588888973765463e-05,
      "loss": 0.2111,
      "step": 6401
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5723947104361927,
      "learning_rate": 1.8588380926522512e-05,
      "loss": 0.2806,
      "step": 6402
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4849222088697294,
      "learning_rate": 1.8587872794784247e-05,
      "loss": 0.0296,
      "step": 6403
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4600940020368573,
      "learning_rate": 1.8587364578555666e-05,
      "loss": 0.2881,
      "step": 6404
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4971932139277737,
      "learning_rate": 1.858685627784176e-05,
      "loss": 0.3446,
      "step": 6405
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4635741556903944,
      "learning_rate": 1.858634789264754e-05,
      "loss": 0.3811,
      "step": 6406
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0174356673540943,
      "learning_rate": 1.8585839422978005e-05,
      "loss": 0.4694,
      "step": 6407
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0700057366214355,
      "learning_rate": 1.8585330868838156e-05,
      "loss": 0.6039,
      "step": 6408
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32436521977151656,
      "learning_rate": 1.8584822230232997e-05,
      "loss": 0.2368,
      "step": 6409
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5442023583734545,
      "learning_rate": 1.858431350716753e-05,
      "loss": 0.3303,
      "step": 6410
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.27132737001542256,
      "learning_rate": 1.8583804699646766e-05,
      "loss": 0.1172,
      "step": 6411
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3548065168128845,
      "learning_rate": 1.8583295807675705e-05,
      "loss": 0.2967,
      "step": 6412
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.34784260615532203,
      "learning_rate": 1.8582786831259353e-05,
      "loss": 0.1743,
      "step": 6413
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.42508715229596733,
      "learning_rate": 1.8582277770402727e-05,
      "loss": 0.2976,
      "step": 6414
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8137490113465483,
      "learning_rate": 1.858176862511082e-05,
      "loss": 0.5699,
      "step": 6415
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.673017239958787,
      "learning_rate": 1.8581259395388653e-05,
      "loss": 0.5029,
      "step": 6416
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7584979624766299,
      "learning_rate": 1.8580750081241234e-05,
      "loss": 0.4132,
      "step": 6417
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35491059221065685,
      "learning_rate": 1.858024068267357e-05,
      "loss": 0.2382,
      "step": 6418
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3710790590161798,
      "learning_rate": 1.8579731199690676e-05,
      "loss": 0.3127,
      "step": 6419
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.25002286701460663,
      "learning_rate": 1.8579221632297563e-05,
      "loss": 0.1186,
      "step": 6420
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7130106158208392,
      "learning_rate": 1.8578711980499243e-05,
      "loss": 0.527,
      "step": 6421
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36224554219953764,
      "learning_rate": 1.8578202244300734e-05,
      "loss": 0.0782,
      "step": 6422
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4007680142680958,
      "learning_rate": 1.857769242370705e-05,
      "loss": 0.3369,
      "step": 6423
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44397663737566917,
      "learning_rate": 1.8577182518723204e-05,
      "loss": 0.322,
      "step": 6424
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5433242005363422,
      "learning_rate": 1.8576672529354214e-05,
      "loss": 0.3782,
      "step": 6425
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2494652774682968,
      "learning_rate": 1.85761624556051e-05,
      "loss": 0.5165,
      "step": 6426
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.31596338056059453,
      "learning_rate": 1.8575652297480876e-05,
      "loss": 0.2056,
      "step": 6427
    },
    {
      "epoch": 0.2,
      "grad_norm": 3.281828058277467,
      "learning_rate": 1.8575142054986563e-05,
      "loss": 0.833,
      "step": 6428
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2173650860459251,
      "learning_rate": 1.857463172812718e-05,
      "loss": 0.1493,
      "step": 6429
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5757486066258702,
      "learning_rate": 1.8574121316907744e-05,
      "loss": 0.3827,
      "step": 6430
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.34735816411354403,
      "learning_rate": 1.857361082133329e-05,
      "loss": 0.1686,
      "step": 6431
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5192640841732221,
      "learning_rate": 1.857310024140883e-05,
      "loss": 0.4036,
      "step": 6432
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7372932708873422,
      "learning_rate": 1.8572589577139382e-05,
      "loss": 0.4138,
      "step": 6433
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.808184922704792,
      "learning_rate": 1.857207882852998e-05,
      "loss": 0.7813,
      "step": 6434
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7694295122276811,
      "learning_rate": 1.857156799558565e-05,
      "loss": 0.2893,
      "step": 6435
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43817258583031266,
      "learning_rate": 1.8571057078311413e-05,
      "loss": 0.3077,
      "step": 6436
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.34963039256433204,
      "learning_rate": 1.857054607671229e-05,
      "loss": 0.2771,
      "step": 6437
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9633234805104198,
      "learning_rate": 1.8570034990793324e-05,
      "loss": 0.957,
      "step": 6438
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.391576984769437,
      "learning_rate": 1.8569523820559524e-05,
      "loss": 0.1755,
      "step": 6439
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5086328652882129,
      "learning_rate": 1.8569012566015937e-05,
      "loss": 0.235,
      "step": 6440
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5812241151675546,
      "learning_rate": 1.8568501227167582e-05,
      "loss": 0.3879,
      "step": 6441
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.44705006351170623,
      "learning_rate": 1.8567989804019492e-05,
      "loss": 0.2702,
      "step": 6442
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5244753313317813,
      "learning_rate": 1.85674782965767e-05,
      "loss": 0.3773,
      "step": 6443
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36881608159040025,
      "learning_rate": 1.8566966704844238e-05,
      "loss": 0.173,
      "step": 6444
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46163940583127494,
      "learning_rate": 1.8566455028827133e-05,
      "loss": 0.2947,
      "step": 6445
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9412680642551198,
      "learning_rate": 1.856594326853043e-05,
      "loss": 0.3612,
      "step": 6446
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5211271962327746,
      "learning_rate": 1.856543142395916e-05,
      "loss": 0.2529,
      "step": 6447
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3965544395694289,
      "learning_rate": 1.856491949511835e-05,
      "loss": 0.2988,
      "step": 6448
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8904064933321072,
      "learning_rate": 1.856440748201305e-05,
      "loss": 0.4041,
      "step": 6449
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4453178230001695,
      "learning_rate": 1.856389538464829e-05,
      "loss": 0.2661,
      "step": 6450
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5869703443893961,
      "learning_rate": 1.8563383203029108e-05,
      "loss": 0.267,
      "step": 6451
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.507329142663605,
      "learning_rate": 1.8562870937160545e-05,
      "loss": 0.6665,
      "step": 6452
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4646867037675367,
      "learning_rate": 1.856235858704764e-05,
      "loss": 0.2448,
      "step": 6453
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4717214044686057,
      "learning_rate": 1.8561846152695437e-05,
      "loss": 0.2491,
      "step": 6454
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3952031929044106,
      "learning_rate": 1.8561333634108968e-05,
      "loss": 0.2765,
      "step": 6455
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0938923009603012,
      "learning_rate": 1.8560821031293288e-05,
      "loss": 0.5313,
      "step": 6456
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9358526074335896,
      "learning_rate": 1.8560308344253432e-05,
      "loss": 0.369,
      "step": 6457
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43535627885720707,
      "learning_rate": 1.8559795572994442e-05,
      "loss": 0.2887,
      "step": 6458
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.28887313499828127,
      "learning_rate": 1.855928271752137e-05,
      "loss": 0.1684,
      "step": 6459
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.45055221575858084,
      "learning_rate": 1.8558769777839257e-05,
      "loss": 0.314,
      "step": 6460
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4124025933226747,
      "learning_rate": 1.8558256753953152e-05,
      "loss": 0.526,
      "step": 6461
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8733701942766943,
      "learning_rate": 1.8557743645868102e-05,
      "loss": 0.7604,
      "step": 6462
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37260758196912275,
      "learning_rate": 1.8557230453589153e-05,
      "loss": 0.2152,
      "step": 6463
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.469324071768783,
      "learning_rate": 1.8556717177121353e-05,
      "loss": 0.7944,
      "step": 6464
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8794883811268193,
      "learning_rate": 1.8556203816469754e-05,
      "loss": 0.3878,
      "step": 6465
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38479981043609135,
      "learning_rate": 1.855569037163941e-05,
      "loss": 0.3003,
      "step": 6466
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8014666594904422,
      "learning_rate": 1.8555176842635368e-05,
      "loss": 0.362,
      "step": 6467
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41158272809139584,
      "learning_rate": 1.8554663229462678e-05,
      "loss": 0.2558,
      "step": 6468
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3464614873130001,
      "learning_rate": 1.85541495321264e-05,
      "loss": 0.1859,
      "step": 6469
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.504789476911091,
      "learning_rate": 1.8553635750631582e-05,
      "loss": 0.6885,
      "step": 6470
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4772156362653101,
      "learning_rate": 1.8553121884983284e-05,
      "loss": 0.3328,
      "step": 6471
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3608810562769184,
      "learning_rate": 1.8552607935186554e-05,
      "loss": 0.1815,
      "step": 6472
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5299964192579208,
      "learning_rate": 1.8552093901246457e-05,
      "loss": 0.352,
      "step": 6473
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0050428116949348,
      "learning_rate": 1.8551579783168045e-05,
      "loss": 0.4314,
      "step": 6474
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9771446699798215,
      "learning_rate": 1.8551065580956375e-05,
      "loss": 0.5811,
      "step": 6475
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3681157632614913,
      "learning_rate": 1.855055129461651e-05,
      "loss": 0.0766,
      "step": 6476
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.434650848437527,
      "learning_rate": 1.8550036924153507e-05,
      "loss": 0.3268,
      "step": 6477
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.22148214770546912,
      "learning_rate": 1.8549522469572426e-05,
      "loss": 0.1766,
      "step": 6478
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.300240898162794,
      "learning_rate": 1.8549007930878337e-05,
      "loss": 0.7132,
      "step": 6479
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4588688071321219,
      "learning_rate": 1.854849330807629e-05,
      "loss": 0.8502,
      "step": 6480
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.342157861669953,
      "learning_rate": 1.854797860117135e-05,
      "loss": 0.1817,
      "step": 6481
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7132616707000795,
      "learning_rate": 1.8547463810168586e-05,
      "loss": 0.4248,
      "step": 6482
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8823900906162114,
      "learning_rate": 1.8546948935073062e-05,
      "loss": 0.3891,
      "step": 6483
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5287760981494851,
      "learning_rate": 1.854643397588984e-05,
      "loss": 0.3771,
      "step": 6484
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4987396751097748,
      "learning_rate": 1.854591893262399e-05,
      "loss": 0.1852,
      "step": 6485
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36256029568681863,
      "learning_rate": 1.8545403805280578e-05,
      "loss": 0.2762,
      "step": 6486
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.25086970901900807,
      "learning_rate": 1.854488859386467e-05,
      "loss": 0.1329,
      "step": 6487
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.828696838981188,
      "learning_rate": 1.8544373298381336e-05,
      "loss": 0.9407,
      "step": 6488
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2761628011343886,
      "learning_rate": 1.854385791883565e-05,
      "loss": 0.2351,
      "step": 6489
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8840869930703383,
      "learning_rate": 1.8543342455232674e-05,
      "loss": 0.3233,
      "step": 6490
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.45841914396571787,
      "learning_rate": 1.8542826907577486e-05,
      "loss": 0.297,
      "step": 6491
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0312905399360999,
      "learning_rate": 1.8542311275875155e-05,
      "loss": 0.4135,
      "step": 6492
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.2071737283644333,
      "learning_rate": 1.8541795560130754e-05,
      "loss": 0.6179,
      "step": 6493
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35627140792252104,
      "learning_rate": 1.854127976034936e-05,
      "loss": 0.0815,
      "step": 6494
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3350130251940854,
      "learning_rate": 1.8540763876536046e-05,
      "loss": 0.2711,
      "step": 6495
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38089533546105614,
      "learning_rate": 1.8540247908695883e-05,
      "loss": 0.2683,
      "step": 6496
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3679956824454245,
      "learning_rate": 1.8539731856833955e-05,
      "loss": 0.8704,
      "step": 6497
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5120172444920296,
      "learning_rate": 1.853921572095533e-05,
      "loss": 0.2612,
      "step": 6498
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4519691396982143,
      "learning_rate": 1.8538699501065094e-05,
      "loss": 0.2958,
      "step": 6499
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4961066037776845,
      "learning_rate": 1.8538183197168323e-05,
      "loss": 0.3015,
      "step": 6500
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0175714101450004,
      "learning_rate": 1.8537666809270095e-05,
      "loss": 0.6434,
      "step": 6501
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3726380538265535,
      "learning_rate": 1.8537150337375492e-05,
      "loss": 0.2869,
      "step": 6502
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.4176119221857526,
      "learning_rate": 1.8536633781489594e-05,
      "loss": 0.7216,
      "step": 6503
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3670485499338874,
      "learning_rate": 1.853611714161749e-05,
      "loss": 0.1822,
      "step": 6504
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.47915937276754345,
      "learning_rate": 1.8535600417764252e-05,
      "loss": 0.2889,
      "step": 6505
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.49214900707088804,
      "learning_rate": 1.8535083609934966e-05,
      "loss": 0.3082,
      "step": 6506
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5510575097987412,
      "learning_rate": 1.853456671813472e-05,
      "loss": 0.3274,
      "step": 6507
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4379713058376259,
      "learning_rate": 1.85340497423686e-05,
      "loss": 0.2752,
      "step": 6508
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4280221541449544,
      "learning_rate": 1.8533532682641693e-05,
      "loss": 0.2952,
      "step": 6509
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5137378825575154,
      "learning_rate": 1.853301553895908e-05,
      "loss": 0.315,
      "step": 6510
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0800361884329555,
      "learning_rate": 1.8532498311325856e-05,
      "loss": 0.3109,
      "step": 6511
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.2589831976508847,
      "learning_rate": 1.8531980999747103e-05,
      "loss": 0.7208,
      "step": 6512
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3229299414009196,
      "learning_rate": 1.8531463604227913e-05,
      "loss": 0.2022,
      "step": 6513
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3685021600635243,
      "learning_rate": 1.853094612477338e-05,
      "loss": 0.3105,
      "step": 6514
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.902871302601657,
      "learning_rate": 1.853042856138859e-05,
      "loss": 0.5162,
      "step": 6515
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7528186886659651,
      "learning_rate": 1.8529910914078635e-05,
      "loss": 0.5352,
      "step": 6516
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6718440165250757,
      "learning_rate": 1.8529393182848613e-05,
      "loss": 0.2839,
      "step": 6517
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.326436460083556,
      "learning_rate": 1.8528875367703612e-05,
      "loss": 0.1888,
      "step": 6518
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7094184638808443,
      "learning_rate": 1.852835746864873e-05,
      "loss": 0.2827,
      "step": 6519
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3650046783393798,
      "learning_rate": 1.852783948568906e-05,
      "loss": 0.2654,
      "step": 6520
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3755760621672453,
      "learning_rate": 1.85273214188297e-05,
      "loss": 0.7463,
      "step": 6521
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3195537584065778,
      "learning_rate": 1.8526803268075745e-05,
      "loss": 0.2081,
      "step": 6522
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.8542512394756367,
      "learning_rate": 1.8526285033432297e-05,
      "loss": 0.8257,
      "step": 6523
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0156766502298915,
      "learning_rate": 1.852576671490445e-05,
      "loss": 0.4403,
      "step": 6524
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39891470459255557,
      "learning_rate": 1.85252483124973e-05,
      "loss": 0.3506,
      "step": 6525
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3651497653866017,
      "learning_rate": 1.8524729826215956e-05,
      "loss": 0.2362,
      "step": 6526
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6430304275418787,
      "learning_rate": 1.8524211256065515e-05,
      "loss": 0.3862,
      "step": 6527
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2872857769750348,
      "learning_rate": 1.8523692602051078e-05,
      "loss": 0.1089,
      "step": 6528
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5182747606157665,
      "learning_rate": 1.852317386417775e-05,
      "loss": 0.707,
      "step": 6529
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4456801692912804,
      "learning_rate": 1.8522655042450633e-05,
      "loss": 0.2718,
      "step": 6530
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.34244821171639317,
      "learning_rate": 1.8522136136874827e-05,
      "loss": 0.1869,
      "step": 6531
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3857066177725719,
      "learning_rate": 1.8521617147455445e-05,
      "loss": 0.326,
      "step": 6532
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7971237802830043,
      "learning_rate": 1.8521098074197588e-05,
      "loss": 0.43,
      "step": 6533
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9357284447483202,
      "learning_rate": 1.852057891710637e-05,
      "loss": 0.5083,
      "step": 6534
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4321118489408309,
      "learning_rate": 1.8520059676186887e-05,
      "loss": 0.1614,
      "step": 6535
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40825862504552884,
      "learning_rate": 1.8519540351444253e-05,
      "loss": 0.2993,
      "step": 6536
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.27594323029683715,
      "learning_rate": 1.851902094288358e-05,
      "loss": 0.2026,
      "step": 6537
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48893037456134625,
      "learning_rate": 1.851850145050998e-05,
      "loss": 0.2298,
      "step": 6538
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9377476740793298,
      "learning_rate": 1.8517981874328553e-05,
      "loss": 0.3905,
      "step": 6539
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.48525952694444574,
      "learning_rate": 1.851746221434442e-05,
      "loss": 0.2694,
      "step": 6540
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5929263974600292,
      "learning_rate": 1.851694247056269e-05,
      "loss": 0.3313,
      "step": 6541
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1267977311161397,
      "learning_rate": 1.8516422642988478e-05,
      "loss": 0.5505,
      "step": 6542
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3225643106514539,
      "learning_rate": 1.85159027316269e-05,
      "loss": 0.3059,
      "step": 6543
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4256267089041922,
      "learning_rate": 1.8515382736483065e-05,
      "loss": 0.1677,
      "step": 6544
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4073003225711898,
      "learning_rate": 1.8514862657562096e-05,
      "loss": 0.2943,
      "step": 6545
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2697903623771002,
      "learning_rate": 1.8514342494869108e-05,
      "loss": 0.0754,
      "step": 6546
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5016394415144747,
      "learning_rate": 1.8513822248409213e-05,
      "loss": 0.241,
      "step": 6547
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36913579092104704,
      "learning_rate": 1.8513301918187534e-05,
      "loss": 0.2596,
      "step": 6548
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46581158843219767,
      "learning_rate": 1.8512781504209187e-05,
      "loss": 0.3122,
      "step": 6549
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.47926767456236513,
      "learning_rate": 1.8512261006479296e-05,
      "loss": 0.258,
      "step": 6550
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9863552267833552,
      "learning_rate": 1.8511740425002983e-05,
      "loss": 0.513,
      "step": 6551
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.01132696952183,
      "learning_rate": 1.8511219759785367e-05,
      "loss": 0.4898,
      "step": 6552
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9867825080845308,
      "learning_rate": 1.851069901083157e-05,
      "loss": 0.4923,
      "step": 6553
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4080284004703133,
      "learning_rate": 1.8510178178146713e-05,
      "loss": 0.203,
      "step": 6554
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3877577393498521,
      "learning_rate": 1.8509657261735927e-05,
      "loss": 0.3302,
      "step": 6555
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.21582692191436023,
      "learning_rate": 1.850913626160433e-05,
      "loss": 0.1521,
      "step": 6556
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9813267046646084,
      "learning_rate": 1.8508615177757052e-05,
      "loss": 0.3426,
      "step": 6557
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.761402443240451,
      "learning_rate": 1.850809401019922e-05,
      "loss": 0.403,
      "step": 6558
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6133158575018133,
      "learning_rate": 1.8507572758935953e-05,
      "loss": 0.33,
      "step": 6559
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0404463543107212,
      "learning_rate": 1.850705142397239e-05,
      "loss": 0.6304,
      "step": 6560
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.34120604982944036,
      "learning_rate": 1.8506530005313656e-05,
      "loss": 0.2634,
      "step": 6561
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3899482545464696,
      "learning_rate": 1.850600850296488e-05,
      "loss": 0.709,
      "step": 6562
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.30870384286874036,
      "learning_rate": 1.8505486916931193e-05,
      "loss": 0.2072,
      "step": 6563
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4977509643487186,
      "learning_rate": 1.850496524721773e-05,
      "loss": 0.2859,
      "step": 6564
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3086483645561713,
      "learning_rate": 1.8504443493829617e-05,
      "loss": 0.1168,
      "step": 6565
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.818600439713565,
      "learning_rate": 1.8503921656771995e-05,
      "loss": 0.546,
      "step": 6566
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2696817044256038,
      "learning_rate": 1.8503399736049988e-05,
      "loss": 0.2261,
      "step": 6567
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3978251675010188,
      "learning_rate": 1.850287773166874e-05,
      "loss": 0.3517,
      "step": 6568
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6513915669155785,
      "learning_rate": 1.8502355643633382e-05,
      "loss": 0.4141,
      "step": 6569
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5950285030307365,
      "learning_rate": 1.8501833471949052e-05,
      "loss": 0.924,
      "step": 6570
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4865204290184775,
      "learning_rate": 1.850131121662089e-05,
      "loss": 0.1609,
      "step": 6571
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.31647443435311984,
      "learning_rate": 1.8500788877654023e-05,
      "loss": 0.2244,
      "step": 6572
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5198597333012926,
      "learning_rate": 1.85002664550536e-05,
      "loss": 0.3539,
      "step": 6573
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2513479246038506,
      "learning_rate": 1.8499743948824765e-05,
      "loss": 0.1554,
      "step": 6574
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7598062610083933,
      "learning_rate": 1.8499221358972646e-05,
      "loss": 0.5162,
      "step": 6575
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3427409430727179,
      "learning_rate": 1.8498698685502393e-05,
      "loss": 0.2313,
      "step": 6576
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6047752295450283,
      "learning_rate": 1.8498175928419146e-05,
      "loss": 0.3859,
      "step": 6577
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6403434920192972,
      "learning_rate": 1.8497653087728048e-05,
      "loss": 0.4275,
      "step": 6578
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3699117216208612,
      "learning_rate": 1.8497130163434243e-05,
      "loss": 0.31,
      "step": 6579
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5628855498224148,
      "learning_rate": 1.8496607155542877e-05,
      "loss": 0.1734,
      "step": 6580
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9867487996249977,
      "learning_rate": 1.849608406405909e-05,
      "loss": 0.4469,
      "step": 6581
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3913128423855416,
      "learning_rate": 1.8495560888988034e-05,
      "loss": 0.2262,
      "step": 6582
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32970211791343246,
      "learning_rate": 1.8495037630334854e-05,
      "loss": 0.2032,
      "step": 6583
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43539309969594386,
      "learning_rate": 1.84945142881047e-05,
      "loss": 0.31,
      "step": 6584
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32937457925534863,
      "learning_rate": 1.849399086230272e-05,
      "loss": 0.0778,
      "step": 6585
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4509757737428039,
      "learning_rate": 1.849346735293406e-05,
      "loss": 0.3506,
      "step": 6586
    },
    {
      "epoch": 0.2,
      "grad_norm": 2.273064768990808,
      "learning_rate": 1.8492943760003878e-05,
      "loss": 0.4285,
      "step": 6587
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.7313863792685893,
      "learning_rate": 1.8492420083517317e-05,
      "loss": 0.9189,
      "step": 6588
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1164931967979361,
      "learning_rate": 1.8491896323479535e-05,
      "loss": 0.3037,
      "step": 6589
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5034131853585584,
      "learning_rate": 1.849137247989568e-05,
      "loss": 0.3122,
      "step": 6590
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32438993542163613,
      "learning_rate": 1.849084855277091e-05,
      "loss": 0.2674,
      "step": 6591
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5362170720324773,
      "learning_rate": 1.849032454211038e-05,
      "loss": 0.2512,
      "step": 6592
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8978214825842724,
      "learning_rate": 1.8489800447919243e-05,
      "loss": 0.409,
      "step": 6593
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4447072454420104,
      "learning_rate": 1.8489276270202653e-05,
      "loss": 0.2426,
      "step": 6594
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3348129297920704,
      "learning_rate": 1.8488752008965773e-05,
      "loss": 0.2159,
      "step": 6595
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5571911230804745,
      "learning_rate": 1.8488227664213755e-05,
      "loss": 0.8486,
      "step": 6596
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32298288681996457,
      "learning_rate": 1.848770323595176e-05,
      "loss": 0.2591,
      "step": 6597
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0455169609866466,
      "learning_rate": 1.8487178724184948e-05,
      "loss": 0.392,
      "step": 6598
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46157100000706536,
      "learning_rate": 1.848665412891848e-05,
      "loss": 0.3169,
      "step": 6599
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4748322797702185,
      "learning_rate": 1.8486129450157518e-05,
      "loss": 0.2535,
      "step": 6600
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8183586053361425,
      "learning_rate": 1.848560468790722e-05,
      "loss": 0.5007,
      "step": 6601
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.39530906476580296,
      "learning_rate": 1.848507984217275e-05,
      "loss": 0.2953,
      "step": 6602
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6880825587961397,
      "learning_rate": 1.8484554912959273e-05,
      "loss": 0.3747,
      "step": 6603
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36695684810932266,
      "learning_rate": 1.8484029900271953e-05,
      "loss": 0.2347,
      "step": 6604
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40110339363558806,
      "learning_rate": 1.8483504804115957e-05,
      "loss": 0.1955,
      "step": 6605
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.1594188807849268,
      "learning_rate": 1.8482979624496448e-05,
      "loss": 0.5883,
      "step": 6606
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.583852831196037,
      "learning_rate": 1.8482454361418596e-05,
      "loss": 0.6458,
      "step": 6607
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3170423570981277,
      "learning_rate": 1.8481929014887562e-05,
      "loss": 0.0781,
      "step": 6608
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36760335306007214,
      "learning_rate": 1.8481403584908526e-05,
      "loss": 0.291,
      "step": 6609
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4769336797795783,
      "learning_rate": 1.8480878071486644e-05,
      "loss": 0.3226,
      "step": 6610
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4399223873067674,
      "learning_rate": 1.8480352474627098e-05,
      "loss": 0.2294,
      "step": 6611
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.575775159968767,
      "learning_rate": 1.8479826794335053e-05,
      "loss": 0.2601,
      "step": 6612
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3441546486116451,
      "learning_rate": 1.847930103061568e-05,
      "loss": 0.216,
      "step": 6613
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5564059976414913,
      "learning_rate": 1.8478775183474155e-05,
      "loss": 0.4095,
      "step": 6614
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.46686813731510457,
      "learning_rate": 1.847824925291565e-05,
      "loss": 0.2934,
      "step": 6615
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4517424217391501,
      "learning_rate": 1.847772323894534e-05,
      "loss": 0.6617,
      "step": 6616
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.32981541875961057,
      "learning_rate": 1.8477197141568396e-05,
      "loss": 0.197,
      "step": 6617
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5785584524647782,
      "learning_rate": 1.847667096079e-05,
      "loss": 0.4095,
      "step": 6618
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.686722650934753,
      "learning_rate": 1.8476144696615327e-05,
      "loss": 0.3953,
      "step": 6619
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5719684483833702,
      "learning_rate": 1.8475618349049547e-05,
      "loss": 0.3589,
      "step": 6620
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.381098939462964,
      "learning_rate": 1.847509191809785e-05,
      "loss": 0.2431,
      "step": 6621
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3963679907955551,
      "learning_rate": 1.847456540376541e-05,
      "loss": 0.2952,
      "step": 6622
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4276507455037887,
      "learning_rate": 1.8474038806057407e-05,
      "loss": 0.1384,
      "step": 6623
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3268475000259117,
      "learning_rate": 1.847351212497902e-05,
      "loss": 0.144,
      "step": 6624
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8930792422875478,
      "learning_rate": 1.8472985360535433e-05,
      "loss": 0.5263,
      "step": 6625
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2901512415495129,
      "learning_rate": 1.847245851273183e-05,
      "loss": 0.2165,
      "step": 6626
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3900641512386994,
      "learning_rate": 1.847193158157339e-05,
      "loss": 0.3172,
      "step": 6627
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6667253696073105,
      "learning_rate": 1.84714045670653e-05,
      "loss": 0.3901,
      "step": 6628
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.710322551160504,
      "learning_rate": 1.8470877469212746e-05,
      "loss": 0.7786,
      "step": 6629
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5300893139333127,
      "learning_rate": 1.847035028802091e-05,
      "loss": 0.24,
      "step": 6630
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0361385434734605,
      "learning_rate": 1.8469823023494983e-05,
      "loss": 0.4638,
      "step": 6631
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.318968038086149,
      "learning_rate": 1.846929567564015e-05,
      "loss": 0.2598,
      "step": 6632
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5467352281026217,
      "learning_rate": 1.84687682444616e-05,
      "loss": 0.4158,
      "step": 6633
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.38673917391725876,
      "learning_rate": 1.8468240729964517e-05,
      "loss": 0.144,
      "step": 6634
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0370619807355894,
      "learning_rate": 1.8467713132154096e-05,
      "loss": 0.4725,
      "step": 6635
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35564621925237005,
      "learning_rate": 1.846718545103553e-05,
      "loss": 0.2569,
      "step": 6636
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7712313515648086,
      "learning_rate": 1.8466657686614007e-05,
      "loss": 0.3995,
      "step": 6637
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4255610721393431,
      "learning_rate": 1.846612983889472e-05,
      "loss": 0.3278,
      "step": 6638
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9701099485789304,
      "learning_rate": 1.8465601907882863e-05,
      "loss": 0.2996,
      "step": 6639
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4074350128011739,
      "learning_rate": 1.8465073893583625e-05,
      "loss": 0.2841,
      "step": 6640
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3405694897593524,
      "learning_rate": 1.846454579600221e-05,
      "loss": 0.2048,
      "step": 6641
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5073186306329553,
      "learning_rate": 1.8464017615143805e-05,
      "loss": 0.3115,
      "step": 6642
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.9864026474137517,
      "learning_rate": 1.8463489351013608e-05,
      "loss": 0.4017,
      "step": 6643
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35497115181560923,
      "learning_rate": 1.846296100361682e-05,
      "loss": 0.2778,
      "step": 6644
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.33076891620422,
      "learning_rate": 1.8462432572958634e-05,
      "loss": 0.2464,
      "step": 6645
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0580793690324426,
      "learning_rate": 1.8461904059044256e-05,
      "loss": 0.5387,
      "step": 6646
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.3946684689043858,
      "learning_rate": 1.846137546187888e-05,
      "loss": 0.3369,
      "step": 6647
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0637042458919441,
      "learning_rate": 1.846084678146771e-05,
      "loss": 0.6546,
      "step": 6648
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3436796210315844,
      "learning_rate": 1.846031801781594e-05,
      "loss": 0.2035,
      "step": 6649
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.40042601122033183,
      "learning_rate": 1.8459789170928784e-05,
      "loss": 0.272,
      "step": 6650
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5612184632463594,
      "learning_rate": 1.8459260240811435e-05,
      "loss": 0.3949,
      "step": 6651
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.854106024134122,
      "learning_rate": 1.8458731227469097e-05,
      "loss": 0.4161,
      "step": 6652
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5722155173295308,
      "learning_rate": 1.8458202130906985e-05,
      "loss": 0.3419,
      "step": 6653
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3257479703997544,
      "learning_rate": 1.8457672951130292e-05,
      "loss": 0.1944,
      "step": 6654
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7016205729213898,
      "learning_rate": 1.845714368814423e-05,
      "loss": 0.3363,
      "step": 6655
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.35727456668805324,
      "learning_rate": 1.8456614341954008e-05,
      "loss": 0.2632,
      "step": 6656
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.467511199484814,
      "learning_rate": 1.845608491256483e-05,
      "loss": 0.693,
      "step": 6657
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.2508625629789519,
      "learning_rate": 1.8455555399981907e-05,
      "loss": 0.0777,
      "step": 6658
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4429945132726838,
      "learning_rate": 1.8455025804210448e-05,
      "loss": 0.352,
      "step": 6659
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7867007442030941,
      "learning_rate": 1.8454496125255665e-05,
      "loss": 0.4796,
      "step": 6660
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4414284451161902,
      "learning_rate": 1.8453966363122764e-05,
      "loss": 0.4179,
      "step": 6661
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.41388926318999153,
      "learning_rate": 1.845343651781696e-05,
      "loss": 0.3047,
      "step": 6662
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3060651550411444,
      "learning_rate": 1.8452906589343465e-05,
      "loss": 0.1993,
      "step": 6663
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.33842782267470595,
      "learning_rate": 1.8452376577707496e-05,
      "loss": 0.1691,
      "step": 6664
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.8531542891759752,
      "learning_rate": 1.8451846482914267e-05,
      "loss": 0.3642,
      "step": 6665
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.581455294731298,
      "learning_rate": 1.8451316304968986e-05,
      "loss": 0.7661,
      "step": 6666
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3545169456728426,
      "learning_rate": 1.845078604387688e-05,
      "loss": 0.1998,
      "step": 6667
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4242816330791993,
      "learning_rate": 1.8450255699643155e-05,
      "loss": 0.3408,
      "step": 6668
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5543873583798383,
      "learning_rate": 1.8449725272273038e-05,
      "loss": 0.3407,
      "step": 6669
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0208332657518362,
      "learning_rate": 1.8449194761771742e-05,
      "loss": 0.5825,
      "step": 6670
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5831718378604815,
      "learning_rate": 1.8448664168144487e-05,
      "loss": 0.309,
      "step": 6671
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.36466122039865184,
      "learning_rate": 1.8448133491396495e-05,
      "loss": 0.2937,
      "step": 6672
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.24061581302643797,
      "learning_rate": 1.8447602731532986e-05,
      "loss": 0.1576,
      "step": 6673
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5401460250843432,
      "learning_rate": 1.8447071888559184e-05,
      "loss": 0.3805,
      "step": 6674
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.0211687436758443,
      "learning_rate": 1.8446540962480305e-05,
      "loss": 0.3012,
      "step": 6675
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3418958026682784,
      "learning_rate": 1.844600995330158e-05,
      "loss": 0.1842,
      "step": 6676
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.6925940033947444,
      "learning_rate": 1.8445478861028227e-05,
      "loss": 0.4421,
      "step": 6677
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.9246107161349888,
      "learning_rate": 1.8444947685665472e-05,
      "loss": 0.4206,
      "step": 6678
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5478834337818054,
      "learning_rate": 1.844441642721855e-05,
      "loss": 0.4031,
      "step": 6679
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3108226992688729,
      "learning_rate": 1.8443885085692674e-05,
      "loss": 0.2091,
      "step": 6680
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4466913953595052,
      "learning_rate": 1.844335366109308e-05,
      "loss": 0.3391,
      "step": 6681
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.418168425705559,
      "learning_rate": 1.8442822153424998e-05,
      "loss": 0.1376,
      "step": 6682
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.45666739190701583,
      "learning_rate": 1.8442290562693647e-05,
      "loss": 0.2748,
      "step": 6683
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.010633411317867,
      "learning_rate": 1.8441758888904268e-05,
      "loss": 0.4915,
      "step": 6684
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.7221391883582755,
      "learning_rate": 1.844122713206209e-05,
      "loss": 0.3333,
      "step": 6685
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.3009297488228649,
      "learning_rate": 1.8440695292172337e-05,
      "loss": 0.2648,
      "step": 6686
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4508270759364024,
      "learning_rate": 1.8440163369240248e-05,
      "loss": 0.3941,
      "step": 6687
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.5859238564161522,
      "learning_rate": 1.8439631363271052e-05,
      "loss": 0.4448,
      "step": 6688
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.4841636537272177,
      "learning_rate": 1.843909927426999e-05,
      "loss": 0.0268,
      "step": 6689
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.43271574601059415,
      "learning_rate": 1.843856710224229e-05,
      "loss": 0.271,
      "step": 6690
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.29168198431469783,
      "learning_rate": 1.8438034847193194e-05,
      "loss": 0.1355,
      "step": 6691
    },
    {
      "epoch": 0.2,
      "grad_norm": 0.37299618128841405,
      "learning_rate": 1.843750250912793e-05,
      "loss": 0.3126,
      "step": 6692
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.5695764167353894,
      "learning_rate": 1.8436970088051742e-05,
      "loss": 0.1319,
      "step": 6693
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6948603516717183,
      "learning_rate": 1.8436437583969867e-05,
      "loss": 0.4525,
      "step": 6694
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2968829062555281,
      "learning_rate": 1.843590499688754e-05,
      "loss": 0.2567,
      "step": 6695
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6649990121000865,
      "learning_rate": 1.8435372326810007e-05,
      "loss": 0.5288,
      "step": 6696
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46141794056464575,
      "learning_rate": 1.8434839573742505e-05,
      "loss": 0.2914,
      "step": 6697
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6014230409742665,
      "learning_rate": 1.8434306737690274e-05,
      "loss": 0.4005,
      "step": 6698
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2956799595067866,
      "learning_rate": 1.843377381865856e-05,
      "loss": 0.2097,
      "step": 6699
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.49957932246015835,
      "learning_rate": 1.8433240816652604e-05,
      "loss": 0.2959,
      "step": 6700
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21580426165229685,
      "learning_rate": 1.8432707731677647e-05,
      "loss": 0.0729,
      "step": 6701
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6069189347391447,
      "learning_rate": 1.843217456373894e-05,
      "loss": 0.3966,
      "step": 6702
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38656716940122526,
      "learning_rate": 1.8431641312841723e-05,
      "loss": 0.29,
      "step": 6703
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.29968780699822417,
      "learning_rate": 1.8431107978991244e-05,
      "loss": 0.2738,
      "step": 6704
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9955377576537559,
      "learning_rate": 1.8430574562192753e-05,
      "loss": 0.5756,
      "step": 6705
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8806811157279998,
      "learning_rate": 1.8430041062451495e-05,
      "loss": 0.6628,
      "step": 6706
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1472251665661888,
      "learning_rate": 1.842950747977272e-05,
      "loss": 0.693,
      "step": 6707
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3493662619492655,
      "learning_rate": 1.8428973814161674e-05,
      "loss": 0.19,
      "step": 6708
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3351828241350882,
      "learning_rate": 1.842844006562361e-05,
      "loss": 0.2246,
      "step": 6709
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2659649139814447,
      "learning_rate": 1.8427906234163783e-05,
      "loss": 0.2201,
      "step": 6710
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0863775074083133,
      "learning_rate": 1.842737231978744e-05,
      "loss": 0.6022,
      "step": 6711
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5073877943420685,
      "learning_rate": 1.8426838322499834e-05,
      "loss": 0.2061,
      "step": 6712
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40619978996191847,
      "learning_rate": 1.8426304242306217e-05,
      "loss": 0.3233,
      "step": 6713
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6713708353277495,
      "learning_rate": 1.842577007921185e-05,
      "loss": 0.4147,
      "step": 6714
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.32069694566935036,
      "learning_rate": 1.8425235833221984e-05,
      "loss": 0.2675,
      "step": 6715
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4176727061972876,
      "learning_rate": 1.8424701504341875e-05,
      "loss": 0.6805,
      "step": 6716
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3156910078013256,
      "learning_rate": 1.842416709257678e-05,
      "loss": 0.1754,
      "step": 6717
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5523588338357531,
      "learning_rate": 1.8423632597931957e-05,
      "loss": 0.4289,
      "step": 6718
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2823136502120306,
      "learning_rate": 1.8423098020412665e-05,
      "loss": 0.1593,
      "step": 6719
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.048791362208935,
      "learning_rate": 1.8422563360024166e-05,
      "loss": 0.5913,
      "step": 6720
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3063824080332653,
      "learning_rate": 1.8422028616771714e-05,
      "loss": 0.2618,
      "step": 6721
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3958245158236742,
      "learning_rate": 1.8421493790660578e-05,
      "loss": 0.3086,
      "step": 6722
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7830096771853788,
      "learning_rate": 1.8420958881696006e-05,
      "loss": 0.4322,
      "step": 6723
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.1951741584095252,
      "learning_rate": 1.8420423889883277e-05,
      "loss": 0.8088,
      "step": 6724
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7973844870361748,
      "learning_rate": 1.8419888815227646e-05,
      "loss": 0.3902,
      "step": 6725
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4770342266712422,
      "learning_rate": 1.841935365773438e-05,
      "loss": 0.2992,
      "step": 6726
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37203891105694126,
      "learning_rate": 1.8418818417408736e-05,
      "loss": 0.2447,
      "step": 6727
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3013012100820928,
      "learning_rate": 1.841828309425599e-05,
      "loss": 0.181,
      "step": 6728
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8141523768746858,
      "learning_rate": 1.8417747688281406e-05,
      "loss": 0.4954,
      "step": 6729
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.27217015930708127,
      "learning_rate": 1.8417212199490252e-05,
      "loss": 0.1581,
      "step": 6730
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41860639808913264,
      "learning_rate": 1.841667662788779e-05,
      "loss": 0.3171,
      "step": 6731
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1169787648943394,
      "learning_rate": 1.8416140973479297e-05,
      "loss": 0.5254,
      "step": 6732
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5584826095982756,
      "learning_rate": 1.841560523627004e-05,
      "loss": 0.4474,
      "step": 6733
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41058554261156455,
      "learning_rate": 1.8415069416265287e-05,
      "loss": 0.2954,
      "step": 6734
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42744126411326083,
      "learning_rate": 1.8414533513470314e-05,
      "loss": 0.2673,
      "step": 6735
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.49005903444201726,
      "learning_rate": 1.841399752789039e-05,
      "loss": 0.2301,
      "step": 6736
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5081134331138715,
      "learning_rate": 1.8413461459530792e-05,
      "loss": 0.2962,
      "step": 6737
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5335398916711916,
      "learning_rate": 1.841292530839679e-05,
      "loss": 0.3432,
      "step": 6738
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33669709978301754,
      "learning_rate": 1.841238907449366e-05,
      "loss": 0.2539,
      "step": 6739
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4140735784359715,
      "learning_rate": 1.8411852757826684e-05,
      "loss": 0.197,
      "step": 6740
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.478726196272729,
      "learning_rate": 1.8411316358401126e-05,
      "loss": 0.3004,
      "step": 6741
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.761812594848473,
      "learning_rate": 1.841077987622227e-05,
      "loss": 0.8857,
      "step": 6742
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.776689483059052,
      "learning_rate": 1.84102433112954e-05,
      "loss": 0.1603,
      "step": 6743
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8011428255003545,
      "learning_rate": 1.8409706663625785e-05,
      "loss": 0.3921,
      "step": 6744
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.29910029900308005,
      "learning_rate": 1.8409169933218707e-05,
      "loss": 0.244,
      "step": 6745
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5308740809974135,
      "learning_rate": 1.8408633120079452e-05,
      "loss": 0.4132,
      "step": 6746
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9137194436142269,
      "learning_rate": 1.8408096224213294e-05,
      "loss": 0.4082,
      "step": 6747
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36187517190253526,
      "learning_rate": 1.8407559245625522e-05,
      "loss": 0.182,
      "step": 6748
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2964649429974557,
      "learning_rate": 1.8407022184321412e-05,
      "loss": 0.2043,
      "step": 6749
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.598966015323481,
      "learning_rate": 1.8406485040306256e-05,
      "loss": 0.9314,
      "step": 6750
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4341825861973618,
      "learning_rate": 1.840594781358533e-05,
      "loss": 0.26,
      "step": 6751
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5105635480899733,
      "learning_rate": 1.8405410504163926e-05,
      "loss": 0.3544,
      "step": 6752
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.511121852872921,
      "learning_rate": 1.8404873112047328e-05,
      "loss": 0.1958,
      "step": 6753
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5300543836174266,
      "learning_rate": 1.840433563724082e-05,
      "loss": 0.3047,
      "step": 6754
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.246292962189518,
      "learning_rate": 1.8403798079749693e-05,
      "loss": 0.6169,
      "step": 6755
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.42768479338689885,
      "learning_rate": 1.8403260439579238e-05,
      "loss": 0.3032,
      "step": 6756
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.32988719910047143,
      "learning_rate": 1.840272271673474e-05,
      "loss": 0.2551,
      "step": 6757
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.24427068874293142,
      "learning_rate": 1.840218491122149e-05,
      "loss": 0.1746,
      "step": 6758
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.9788211887951332,
      "learning_rate": 1.8401647023044784e-05,
      "loss": 0.9168,
      "step": 6759
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.207228302968287,
      "learning_rate": 1.8401109052209905e-05,
      "loss": 0.6357,
      "step": 6760
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.6384776035146644,
      "learning_rate": 1.8400570998722155e-05,
      "loss": 0.5929,
      "step": 6761
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38854279122443947,
      "learning_rate": 1.840003286258682e-05,
      "loss": 0.24,
      "step": 6762
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5984023679873963,
      "learning_rate": 1.83994946438092e-05,
      "loss": 0.4034,
      "step": 6763
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34714901053988917,
      "learning_rate": 1.839895634239459e-05,
      "loss": 0.3174,
      "step": 6764
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8397629754126564,
      "learning_rate": 1.839841795834828e-05,
      "loss": 0.5634,
      "step": 6765
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.27548783794827564,
      "learning_rate": 1.8397879491675573e-05,
      "loss": 0.14,
      "step": 6766
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.010001128932886,
      "learning_rate": 1.8397340942381763e-05,
      "loss": 0.4154,
      "step": 6767
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4354674600161744,
      "learning_rate": 1.839680231047215e-05,
      "loss": 0.2686,
      "step": 6768
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3157691017632941,
      "learning_rate": 1.8396263595952033e-05,
      "loss": 0.2368,
      "step": 6769
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3241969375721918,
      "learning_rate": 1.839572479882671e-05,
      "loss": 0.6905,
      "step": 6770
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5495583177876635,
      "learning_rate": 1.8395185919101487e-05,
      "loss": 0.3299,
      "step": 6771
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3845456578322045,
      "learning_rate": 1.8394646956781662e-05,
      "loss": 0.2988,
      "step": 6772
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7095086574520418,
      "learning_rate": 1.8394107911872536e-05,
      "loss": 0.3997,
      "step": 6773
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.142264547489486,
      "learning_rate": 1.8393568784379416e-05,
      "loss": 0.9013,
      "step": 6774
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3585487257491621,
      "learning_rate": 1.8393029574307607e-05,
      "loss": 0.2531,
      "step": 6775
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.459261110211856,
      "learning_rate": 1.839249028166241e-05,
      "loss": 0.2739,
      "step": 6776
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47726043488365655,
      "learning_rate": 1.8391950906449126e-05,
      "loss": 0.3032,
      "step": 6777
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5229526050901638,
      "learning_rate": 1.8391411448673073e-05,
      "loss": 0.2445,
      "step": 6778
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5766094841325303,
      "learning_rate": 1.839087190833955e-05,
      "loss": 0.2588,
      "step": 6779
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47958332021853645,
      "learning_rate": 1.8390332285453874e-05,
      "loss": 0.3437,
      "step": 6780
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.31993931637738976,
      "learning_rate": 1.8389792580021345e-05,
      "loss": 0.2338,
      "step": 6781
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7997204492099818,
      "learning_rate": 1.8389252792047277e-05,
      "loss": 0.3783,
      "step": 6782
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2154143806972042,
      "learning_rate": 1.8388712921536974e-05,
      "loss": 0.639,
      "step": 6783
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9320480582785123,
      "learning_rate": 1.838817296849576e-05,
      "loss": 0.4666,
      "step": 6784
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5065594087481825,
      "learning_rate": 1.838763293292894e-05,
      "loss": 0.2502,
      "step": 6785
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48321948084247884,
      "learning_rate": 1.838709281484182e-05,
      "loss": 0.2505,
      "step": 6786
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.39429858906137405,
      "learning_rate": 1.8386552614239728e-05,
      "loss": 0.3037,
      "step": 6787
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40047973726870345,
      "learning_rate": 1.838601233112797e-05,
      "loss": 0.1837,
      "step": 6788
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6575213826877677,
      "learning_rate": 1.8385471965511863e-05,
      "loss": 0.4091,
      "step": 6789
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.35617275520330755,
      "learning_rate": 1.8384931517396725e-05,
      "loss": 0.2386,
      "step": 6790
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0005074122291902,
      "learning_rate": 1.838439098678787e-05,
      "loss": 0.5996,
      "step": 6791
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4946734400010549,
      "learning_rate": 1.838385037369062e-05,
      "loss": 0.2643,
      "step": 6792
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5132853642893301,
      "learning_rate": 1.8383309678110287e-05,
      "loss": 0.4124,
      "step": 6793
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3364138988994637,
      "learning_rate": 1.83827689000522e-05,
      "loss": 0.0789,
      "step": 6794
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3915601557779156,
      "learning_rate": 1.838222803952167e-05,
      "loss": 0.2652,
      "step": 6795
    },
    {
      "epoch": 0.21,
      "grad_norm": 4.140068066284717,
      "learning_rate": 1.8381687096524024e-05,
      "loss": 0.5518,
      "step": 6796
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6526634877805367,
      "learning_rate": 1.8381146071064585e-05,
      "loss": 0.4068,
      "step": 6797
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41365987186709835,
      "learning_rate": 1.8380604963148668e-05,
      "loss": 0.3264,
      "step": 6798
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.318627520548894,
      "learning_rate": 1.8380063772781605e-05,
      "loss": 0.2022,
      "step": 6799
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3308446550875103,
      "learning_rate": 1.8379522499968718e-05,
      "loss": 0.168,
      "step": 6800
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6827054330903589,
      "learning_rate": 1.8378981144715328e-05,
      "loss": 0.3945,
      "step": 6801
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3306012885259415,
      "learning_rate": 1.837843970702677e-05,
      "loss": 0.6745,
      "step": 6802
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4716554482973314,
      "learning_rate": 1.837789818690836e-05,
      "loss": 0.169,
      "step": 6803
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.47555068767898995,
      "learning_rate": 1.8377356584365433e-05,
      "loss": 0.3659,
      "step": 6804
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4151562001726728,
      "learning_rate": 1.8376814899403315e-05,
      "loss": 0.3033,
      "step": 6805
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8998441379050057,
      "learning_rate": 1.8376273132027336e-05,
      "loss": 0.4714,
      "step": 6806
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3602331819491998,
      "learning_rate": 1.837573128224283e-05,
      "loss": 0.1887,
      "step": 6807
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.24338994118277724,
      "learning_rate": 1.8375189350055115e-05,
      "loss": 0.168,
      "step": 6808
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0064420916841286,
      "learning_rate": 1.8374647335469537e-05,
      "loss": 0.9195,
      "step": 6809
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41961688722457624,
      "learning_rate": 1.837410523849142e-05,
      "loss": 0.2427,
      "step": 6810
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46173636261965456,
      "learning_rate": 1.8373563059126103e-05,
      "loss": 0.3691,
      "step": 6811
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3533410620307274,
      "learning_rate": 1.8373020797378917e-05,
      "loss": 0.1987,
      "step": 6812
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.657679001337824,
      "learning_rate": 1.8372478453255194e-05,
      "loss": 0.368,
      "step": 6813
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0039103089506876,
      "learning_rate": 1.8371936026760276e-05,
      "loss": 0.459,
      "step": 6814
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7969153699318503,
      "learning_rate": 1.8371393517899497e-05,
      "loss": 0.5195,
      "step": 6815
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.314024715722439,
      "learning_rate": 1.8370850926678195e-05,
      "loss": 0.2394,
      "step": 6816
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0510168447837611,
      "learning_rate": 1.8370308253101706e-05,
      "loss": 0.4961,
      "step": 6817
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2376164803209813,
      "learning_rate": 1.836976549717537e-05,
      "loss": 0.1601,
      "step": 6818
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4080558217325414,
      "learning_rate": 1.8369222658904528e-05,
      "loss": 0.7817,
      "step": 6819
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8818731102386888,
      "learning_rate": 1.836867973829452e-05,
      "loss": 0.5261,
      "step": 6820
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4011040842788304,
      "learning_rate": 1.836813673535068e-05,
      "loss": 0.0776,
      "step": 6821
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41136044263966426,
      "learning_rate": 1.8367593650078364e-05,
      "loss": 0.3513,
      "step": 6822
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.40374741340466297,
      "learning_rate": 1.8367050482482905e-05,
      "loss": 0.3188,
      "step": 6823
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8927104905046411,
      "learning_rate": 1.8366507232569654e-05,
      "loss": 0.572,
      "step": 6824
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2402185209797454,
      "learning_rate": 1.8365963900343953e-05,
      "loss": 0.1244,
      "step": 6825
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44675618342359913,
      "learning_rate": 1.8365420485811143e-05,
      "loss": 0.2856,
      "step": 6826
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.22671053834750177,
      "learning_rate": 1.8364876988976572e-05,
      "loss": 0.1484,
      "step": 6827
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36781310031421105,
      "learning_rate": 1.8364333409845592e-05,
      "loss": 0.3165,
      "step": 6828
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.225769227882264,
      "learning_rate": 1.836378974842355e-05,
      "loss": 0.355,
      "step": 6829
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7531787827743569,
      "learning_rate": 1.8363246004715788e-05,
      "loss": 0.4838,
      "step": 6830
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3506353289049739,
      "learning_rate": 1.836270217872766e-05,
      "loss": 0.268,
      "step": 6831
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6950660644309689,
      "learning_rate": 1.8362158270464515e-05,
      "loss": 0.5161,
      "step": 6832
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9497582358601975,
      "learning_rate": 1.836161427993171e-05,
      "loss": 0.3735,
      "step": 6833
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34014521101468226,
      "learning_rate": 1.836107020713459e-05,
      "loss": 0.2825,
      "step": 6834
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.45340136528032765,
      "learning_rate": 1.8360526052078512e-05,
      "loss": 0.3049,
      "step": 6835
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.28885264675855554,
      "learning_rate": 1.8359981814768827e-05,
      "loss": 0.1648,
      "step": 6836
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5347498643400372,
      "learning_rate": 1.8359437495210888e-05,
      "loss": 0.2812,
      "step": 6837
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8485858797676444,
      "learning_rate": 1.8358893093410054e-05,
      "loss": 0.4543,
      "step": 6838
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5984389755850791,
      "learning_rate": 1.8358348609371675e-05,
      "loss": 0.415,
      "step": 6839
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37365013974143757,
      "learning_rate": 1.835780404310112e-05,
      "loss": 0.237,
      "step": 6840
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3737943058866219,
      "learning_rate": 1.835725939460373e-05,
      "loss": 0.3459,
      "step": 6841
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6968523547053255,
      "learning_rate": 1.835671466388488e-05,
      "loss": 0.4448,
      "step": 6842
    },
    {
      "epoch": 0.21,
      "grad_norm": 2.0177590797504514,
      "learning_rate": 1.835616985094992e-05,
      "loss": 0.8465,
      "step": 6843
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2949260092529241,
      "learning_rate": 1.835562495580421e-05,
      "loss": 0.075,
      "step": 6844
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2145448650646786,
      "learning_rate": 1.835507997845311e-05,
      "loss": 0.2345,
      "step": 6845
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2862609759172337,
      "learning_rate": 1.8354534918901987e-05,
      "loss": 0.2327,
      "step": 6846
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9343463940346154,
      "learning_rate": 1.83539897771562e-05,
      "loss": 0.4242,
      "step": 6847
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1006224100362871,
      "learning_rate": 1.835344455322111e-05,
      "loss": 0.5514,
      "step": 6848
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3253543459542789,
      "learning_rate": 1.8352899247102088e-05,
      "loss": 0.2149,
      "step": 6849
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9720181287292425,
      "learning_rate": 1.835235385880449e-05,
      "loss": 0.6201,
      "step": 6850
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.39014578775559844,
      "learning_rate": 1.835180838833369e-05,
      "loss": 0.2668,
      "step": 6851
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48691810270360136,
      "learning_rate": 1.8351262835695053e-05,
      "loss": 0.3596,
      "step": 6852
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.35589251454724485,
      "learning_rate": 1.8350717200893938e-05,
      "loss": 0.2076,
      "step": 6853
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3378125456729122,
      "learning_rate": 1.8350171483935723e-05,
      "loss": 0.2523,
      "step": 6854
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5500712216672783,
      "learning_rate": 1.8349625684825775e-05,
      "loss": 0.2005,
      "step": 6855
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9903354787408802,
      "learning_rate": 1.834907980356946e-05,
      "loss": 0.5351,
      "step": 6856
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6124358425456495,
      "learning_rate": 1.8348533840172152e-05,
      "loss": 0.3504,
      "step": 6857
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.43077528389880926,
      "learning_rate": 1.834798779463922e-05,
      "loss": 0.2944,
      "step": 6858
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3405951706818827,
      "learning_rate": 1.8347441666976035e-05,
      "loss": 0.254,
      "step": 6859
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2505974858023363,
      "learning_rate": 1.8346895457187976e-05,
      "loss": 0.5104,
      "step": 6860
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3649502265584892,
      "learning_rate": 1.8346349165280412e-05,
      "loss": 0.6129,
      "step": 6861
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34000867332332385,
      "learning_rate": 1.8345802791258716e-05,
      "loss": 0.194,
      "step": 6862
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3627987395467736,
      "learning_rate": 1.834525633512827e-05,
      "loss": 0.2313,
      "step": 6863
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2548131842502117,
      "learning_rate": 1.8344709796894443e-05,
      "loss": 0.1955,
      "step": 6864
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8297769863865622,
      "learning_rate": 1.834416317656262e-05,
      "loss": 0.5966,
      "step": 6865
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.510428035224335,
      "learning_rate": 1.834361647413817e-05,
      "loss": 0.3014,
      "step": 6866
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4361025306896745,
      "learning_rate": 1.834306968962648e-05,
      "loss": 0.3537,
      "step": 6867
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1251331919971979,
      "learning_rate": 1.834252282303292e-05,
      "loss": 0.6111,
      "step": 6868
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.8040241914215929,
      "learning_rate": 1.8341975874362875e-05,
      "loss": 0.8234,
      "step": 6869
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33684903464251104,
      "learning_rate": 1.834142884362173e-05,
      "loss": 0.2783,
      "step": 6870
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.027746531505169,
      "learning_rate": 1.834088173081486e-05,
      "loss": 0.4634,
      "step": 6871
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36862867017244544,
      "learning_rate": 1.8340334535947654e-05,
      "loss": 0.2304,
      "step": 6872
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34397936147928754,
      "learning_rate": 1.833978725902549e-05,
      "loss": 0.2367,
      "step": 6873
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4135809638782946,
      "learning_rate": 1.8339239900053756e-05,
      "loss": 0.3399,
      "step": 6874
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4277298155587927,
      "learning_rate": 1.8338692459037836e-05,
      "loss": 0.2706,
      "step": 6875
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.435909528455114,
      "learning_rate": 1.8338144935983116e-05,
      "loss": 0.3142,
      "step": 6876
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3564414488372408,
      "learning_rate": 1.8337597330894983e-05,
      "loss": 0.2706,
      "step": 6877
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3534757021774555,
      "learning_rate": 1.8337049643778824e-05,
      "loss": 0.713,
      "step": 6878
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9124201762613323,
      "learning_rate": 1.8336501874640027e-05,
      "loss": 0.0336,
      "step": 6879
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8896262859609718,
      "learning_rate": 1.833595402348398e-05,
      "loss": 0.5096,
      "step": 6880
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.30887488851891703,
      "learning_rate": 1.8335406090316078e-05,
      "loss": 0.2487,
      "step": 6881
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.35735848736441966,
      "learning_rate": 1.8334858075141704e-05,
      "loss": 0.3281,
      "step": 6882
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5830630357527671,
      "learning_rate": 1.833430997796626e-05,
      "loss": 0.3902,
      "step": 6883
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4091996949219916,
      "learning_rate": 1.8333761798795123e-05,
      "loss": 0.1827,
      "step": 6884
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33202651357909113,
      "learning_rate": 1.83332135376337e-05,
      "loss": 0.2031,
      "step": 6885
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0060966026547795,
      "learning_rate": 1.833266519448738e-05,
      "loss": 0.5986,
      "step": 6886
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1683103665664327,
      "learning_rate": 1.8332116769361558e-05,
      "loss": 0.6466,
      "step": 6887
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.32969449480238033,
      "learning_rate": 1.833156826226163e-05,
      "loss": 0.275,
      "step": 6888
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7421997357628274,
      "learning_rate": 1.833101967319299e-05,
      "loss": 0.3708,
      "step": 6889
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46842909065351535,
      "learning_rate": 1.8330471002161043e-05,
      "loss": 0.2395,
      "step": 6890
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.830986810537084,
      "learning_rate": 1.8329922249171176e-05,
      "loss": 0.5759,
      "step": 6891
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4486539899462104,
      "learning_rate": 1.8329373414228792e-05,
      "loss": 0.2547,
      "step": 6892
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2812307779154014,
      "learning_rate": 1.8328824497339293e-05,
      "loss": 0.2385,
      "step": 6893
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36009266872149437,
      "learning_rate": 1.8328275498508082e-05,
      "loss": 0.1848,
      "step": 6894
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4790904056269485,
      "learning_rate": 1.832772641774055e-05,
      "loss": 0.3847,
      "step": 6895
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9421340696689717,
      "learning_rate": 1.8327177255042112e-05,
      "loss": 0.5335,
      "step": 6896
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2972147534210048,
      "learning_rate": 1.8326628010418158e-05,
      "loss": 0.5996,
      "step": 6897
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2075417704375973,
      "learning_rate": 1.8326078683874104e-05,
      "loss": 0.26,
      "step": 6898
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3510468883756344,
      "learning_rate": 1.832552927541534e-05,
      "loss": 0.2573,
      "step": 6899
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3898794334813042,
      "learning_rate": 1.8324979785047287e-05,
      "loss": 0.3634,
      "step": 6900
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37590865842134386,
      "learning_rate": 1.8324430212775343e-05,
      "loss": 0.1828,
      "step": 6901
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.48225148759176095,
      "learning_rate": 1.832388055860491e-05,
      "loss": 0.2735,
      "step": 6902
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.31383219978786714,
      "learning_rate": 1.8323330822541406e-05,
      "loss": 0.1933,
      "step": 6903
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5532898945534506,
      "learning_rate": 1.8322781004590232e-05,
      "loss": 0.4064,
      "step": 6904
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3208965159583171,
      "learning_rate": 1.8322231104756803e-05,
      "loss": 0.2716,
      "step": 6905
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4918298083584156,
      "learning_rate": 1.832168112304652e-05,
      "loss": 0.7766,
      "step": 6906
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6602159947068706,
      "learning_rate": 1.8321131059464808e-05,
      "loss": 0.3217,
      "step": 6907
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.35886717834172754,
      "learning_rate": 1.8320580914017062e-05,
      "loss": 0.274,
      "step": 6908
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5435810242721902,
      "learning_rate": 1.832003068670871e-05,
      "loss": 0.3754,
      "step": 6909
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3803874753248842,
      "learning_rate": 1.8319480377545155e-05,
      "loss": 0.6902,
      "step": 6910
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2788193291842219,
      "learning_rate": 1.8318929986531817e-05,
      "loss": 0.2281,
      "step": 6911
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36134893624150793,
      "learning_rate": 1.8318379513674108e-05,
      "loss": 0.1948,
      "step": 6912
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5635728782182233,
      "learning_rate": 1.831782895897744e-05,
      "loss": 0.4047,
      "step": 6913
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.27980847498460404,
      "learning_rate": 1.831727832244724e-05,
      "loss": 0.1012,
      "step": 6914
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1066899499851228,
      "learning_rate": 1.8316727604088917e-05,
      "loss": 0.5836,
      "step": 6915
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.634871094607335,
      "learning_rate": 1.8316176803907888e-05,
      "loss": 0.2712,
      "step": 6916
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4081394079953595,
      "learning_rate": 1.8315625921909582e-05,
      "loss": 0.3384,
      "step": 6917
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38991098805841246,
      "learning_rate": 1.8315074958099406e-05,
      "loss": 0.258,
      "step": 6918
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4283456384595254,
      "learning_rate": 1.8314523912482785e-05,
      "loss": 0.7687,
      "step": 6919
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9250041447985847,
      "learning_rate": 1.831397278506515e-05,
      "loss": 0.4668,
      "step": 6920
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4522951847139405,
      "learning_rate": 1.8313421575851908e-05,
      "loss": 0.2966,
      "step": 6921
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.474140427551373,
      "learning_rate": 1.831287028484849e-05,
      "loss": 0.2554,
      "step": 6922
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3226527556076716,
      "learning_rate": 1.831231891206032e-05,
      "loss": 0.2521,
      "step": 6923
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37075776831312657,
      "learning_rate": 1.831176745749282e-05,
      "loss": 0.2517,
      "step": 6924
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9502774751984038,
      "learning_rate": 1.831121592115142e-05,
      "loss": 0.4114,
      "step": 6925
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4205199848880448,
      "learning_rate": 1.831066430304154e-05,
      "loss": 0.3119,
      "step": 6926
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.49394739937121607,
      "learning_rate": 1.8310112603168613e-05,
      "loss": 0.2487,
      "step": 6927
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1699987060479093,
      "learning_rate": 1.830956082153806e-05,
      "loss": 0.6846,
      "step": 6928
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3518492440973091,
      "learning_rate": 1.8309008958155317e-05,
      "loss": 0.2446,
      "step": 6929
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7765931814509918,
      "learning_rate": 1.830845701302581e-05,
      "loss": 0.3937,
      "step": 6930
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33646254866633507,
      "learning_rate": 1.8307904986154967e-05,
      "loss": 0.2607,
      "step": 6931
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4373437497017467,
      "learning_rate": 1.8307352877548223e-05,
      "loss": 0.3199,
      "step": 6932
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7241667661375901,
      "learning_rate": 1.8306800687211005e-05,
      "loss": 0.4054,
      "step": 6933
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4020568997903735,
      "learning_rate": 1.830624841514875e-05,
      "loss": 0.3068,
      "step": 6934
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2910563129607602,
      "learning_rate": 1.8305696061366892e-05,
      "loss": 0.2114,
      "step": 6935
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3685102983976428,
      "learning_rate": 1.8305143625870862e-05,
      "loss": 0.337,
      "step": 6936
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2230591643659543,
      "learning_rate": 1.8304591108666094e-05,
      "loss": 0.3195,
      "step": 6937
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8977945129057233,
      "learning_rate": 1.8304038509758028e-05,
      "loss": 0.4943,
      "step": 6938
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.630095209055685,
      "learning_rate": 1.83034858291521e-05,
      "loss": 0.3483,
      "step": 6939
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.35436859459866155,
      "learning_rate": 1.8302933066853743e-05,
      "loss": 0.2504,
      "step": 6940
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9301724673129069,
      "learning_rate": 1.83023802228684e-05,
      "loss": 0.573,
      "step": 6941
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3307184060042875,
      "learning_rate": 1.830182729720151e-05,
      "loss": 0.3054,
      "step": 6942
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2876100212744487,
      "learning_rate": 1.830127428985851e-05,
      "loss": 0.1993,
      "step": 6943
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.32143661207090585,
      "learning_rate": 1.830072120084484e-05,
      "loss": 0.2204,
      "step": 6944
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4061317772213713,
      "learning_rate": 1.8300168030165947e-05,
      "loss": 0.8128,
      "step": 6945
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8755509190728785,
      "learning_rate": 1.8299614777827267e-05,
      "loss": 0.3475,
      "step": 6946
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37698218843305137,
      "learning_rate": 1.829906144383425e-05,
      "loss": 0.3318,
      "step": 6947
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.36678739162961654,
      "learning_rate": 1.829850802819233e-05,
      "loss": 0.1261,
      "step": 6948
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4716613073306391,
      "learning_rate": 1.829795453090696e-05,
      "loss": 0.3353,
      "step": 6949
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6814557124344893,
      "learning_rate": 1.8297400951983587e-05,
      "loss": 0.4157,
      "step": 6950
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8322991094864335,
      "learning_rate": 1.8296847291427648e-05,
      "loss": 0.536,
      "step": 6951
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.21873099245283217,
      "learning_rate": 1.8296293549244597e-05,
      "loss": 0.1432,
      "step": 6952
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33820562480332794,
      "learning_rate": 1.8295739725439882e-05,
      "loss": 0.1783,
      "step": 6953
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3888171053497484,
      "learning_rate": 1.8295185820018945e-05,
      "loss": 0.3149,
      "step": 6954
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4065884641611393,
      "learning_rate": 1.8294631832987244e-05,
      "loss": 0.3452,
      "step": 6955
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5471293383364773,
      "learning_rate": 1.8294077764350228e-05,
      "loss": 0.9208,
      "step": 6956
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6649830404028284,
      "learning_rate": 1.829352361411334e-05,
      "loss": 0.3018,
      "step": 6957
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3805537178127791,
      "learning_rate": 1.8292969382282043e-05,
      "loss": 0.2979,
      "step": 6958
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.34495188701302454,
      "learning_rate": 1.8292415068861782e-05,
      "loss": 0.3061,
      "step": 6959
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9737221737154045,
      "learning_rate": 1.8291860673858014e-05,
      "loss": 0.591,
      "step": 6960
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.27732377938263253,
      "learning_rate": 1.829130619727619e-05,
      "loss": 0.1349,
      "step": 6961
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.46881909472555844,
      "learning_rate": 1.829075163912177e-05,
      "loss": 0.3116,
      "step": 6962
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33130199814432054,
      "learning_rate": 1.8290196999400204e-05,
      "loss": 0.1879,
      "step": 6963
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.4425861026825484,
      "learning_rate": 1.8289642278116956e-05,
      "loss": 0.7061,
      "step": 6964
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.32590675273928815,
      "learning_rate": 1.828908747527748e-05,
      "loss": 0.2796,
      "step": 6965
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4050830198466434,
      "learning_rate": 1.828853259088723e-05,
      "loss": 0.1729,
      "step": 6966
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3898582632129041,
      "learning_rate": 1.8287977624951675e-05,
      "loss": 0.3452,
      "step": 6967
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7087403448743472,
      "learning_rate": 1.8287422577476263e-05,
      "loss": 0.4056,
      "step": 6968
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8302737863963465,
      "learning_rate": 1.8286867448466462e-05,
      "loss": 0.5409,
      "step": 6969
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3556081207577703,
      "learning_rate": 1.8286312237927734e-05,
      "loss": 0.2561,
      "step": 6970
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.463724071711072,
      "learning_rate": 1.828575694586554e-05,
      "loss": 0.2851,
      "step": 6971
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3673162628384787,
      "learning_rate": 1.828520157228534e-05,
      "loss": 0.1609,
      "step": 6972
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5557613935722561,
      "learning_rate": 1.8284646117192602e-05,
      "loss": 0.2405,
      "step": 6973
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7807428588992216,
      "learning_rate": 1.828409058059279e-05,
      "loss": 0.4483,
      "step": 6974
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6584910411135735,
      "learning_rate": 1.828353496249137e-05,
      "loss": 0.4019,
      "step": 6975
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37834908774347303,
      "learning_rate": 1.8282979262893806e-05,
      "loss": 0.1912,
      "step": 6976
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38586937148609757,
      "learning_rate": 1.8282423481805567e-05,
      "loss": 0.3201,
      "step": 6977
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.4554669713923302,
      "learning_rate": 1.8281867619232118e-05,
      "loss": 0.3001,
      "step": 6978
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.30726042156281963,
      "learning_rate": 1.8281311675178935e-05,
      "loss": 0.1448,
      "step": 6979
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.7843180866862033,
      "learning_rate": 1.8280755649651482e-05,
      "loss": 0.3923,
      "step": 6980
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.285484203297156,
      "learning_rate": 1.8280199542655228e-05,
      "loss": 0.1943,
      "step": 6981
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2633305893368267,
      "learning_rate": 1.8279643354195647e-05,
      "loss": 0.6867,
      "step": 6982
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.37561852289601694,
      "learning_rate": 1.827908708427821e-05,
      "loss": 0.2744,
      "step": 6983
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.2383371121920659,
      "learning_rate": 1.827853073290839e-05,
      "loss": 0.5266,
      "step": 6984
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3336652317935265,
      "learning_rate": 1.827797430009166e-05,
      "loss": 0.208,
      "step": 6985
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9087116441799824,
      "learning_rate": 1.8277417785833494e-05,
      "loss": 0.5947,
      "step": 6986
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.000023942993965,
      "learning_rate": 1.8276861190139375e-05,
      "loss": 0.3438,
      "step": 6987
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5350884663941307,
      "learning_rate": 1.8276304513014765e-05,
      "loss": 0.3821,
      "step": 6988
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.33016613332073963,
      "learning_rate": 1.8275747754465152e-05,
      "loss": 0.1944,
      "step": 6989
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.30913963441695397,
      "learning_rate": 1.8275190914496005e-05,
      "loss": 0.2365,
      "step": 6990
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2653528692934671,
      "learning_rate": 1.8274633993112815e-05,
      "loss": 0.0772,
      "step": 6991
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.9462043136751579,
      "learning_rate": 1.8274076990321045e-05,
      "loss": 0.427,
      "step": 6992
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.702266391413519,
      "learning_rate": 1.8273519906126186e-05,
      "loss": 0.4696,
      "step": 6993
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3390775980070791,
      "learning_rate": 1.8272962740533718e-05,
      "loss": 0.2314,
      "step": 6994
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.5549023768836546,
      "learning_rate": 1.8272405493549115e-05,
      "loss": 0.3825,
      "step": 6995
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.38441647848538285,
      "learning_rate": 1.8271848165177867e-05,
      "loss": 0.2669,
      "step": 6996
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.5620573326062537,
      "learning_rate": 1.8271290755425458e-05,
      "loss": 0.7531,
      "step": 6997
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.26889263655375834,
      "learning_rate": 1.8270733264297368e-05,
      "loss": 0.1363,
      "step": 6998
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.6109834512053272,
      "learning_rate": 1.827017569179908e-05,
      "loss": 0.3831,
      "step": 6999
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.31959267090669435,
      "learning_rate": 1.8269618037936086e-05,
      "loss": 0.2013,
      "step": 7000
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3820330068558199,
      "learning_rate": 1.8269060302713866e-05,
      "loss": 0.3489,
      "step": 7001
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.0739672258559254,
      "learning_rate": 1.826850248613791e-05,
      "loss": 0.3439,
      "step": 7002
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.050669737769629,
      "learning_rate": 1.8267944588213704e-05,
      "loss": 0.5401,
      "step": 7003
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3801769796848347,
      "learning_rate": 1.8267386608946743e-05,
      "loss": 0.2632,
      "step": 7004
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.16130361867476,
      "learning_rate": 1.826682854834251e-05,
      "loss": 0.3585,
      "step": 7005
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3839784946829149,
      "learning_rate": 1.8266270406406496e-05,
      "loss": 0.3205,
      "step": 7006
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.32617617086948164,
      "learning_rate": 1.8265712183144196e-05,
      "loss": 0.0774,
      "step": 7007
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.41383255189090434,
      "learning_rate": 1.82651538785611e-05,
      "loss": 0.3026,
      "step": 7008
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3051102254016016,
      "learning_rate": 1.8264595492662698e-05,
      "loss": 0.1793,
      "step": 7009
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8082714297543142,
      "learning_rate": 1.826403702545449e-05,
      "loss": 0.5007,
      "step": 7010
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.1068281321385873,
      "learning_rate": 1.8263478476941964e-05,
      "loss": 0.4947,
      "step": 7011
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.44192985533772833,
      "learning_rate": 1.8262919847130616e-05,
      "loss": 0.2812,
      "step": 7012
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.29366043534698094,
      "learning_rate": 1.826236113602595e-05,
      "loss": 0.2549,
      "step": 7013
    },
    {
      "epoch": 0.21,
      "grad_norm": 1.3578255750398935,
      "learning_rate": 1.8261802343633448e-05,
      "loss": 0.7587,
      "step": 7014
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8602790875957066,
      "learning_rate": 1.826124346995862e-05,
      "loss": 0.3798,
      "step": 7015
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.8253712851928446,
      "learning_rate": 1.8260684515006964e-05,
      "loss": 0.4336,
      "step": 7016
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.2455399747327774,
      "learning_rate": 1.826012547878397e-05,
      "loss": 0.1746,
      "step": 7017
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.820023782773309,
      "learning_rate": 1.8259566361295145e-05,
      "loss": 0.3982,
      "step": 7018
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.364432722785566,
      "learning_rate": 1.8259007162545994e-05,
      "loss": 0.3146,
      "step": 7019
    },
    {
      "epoch": 0.21,
      "grad_norm": 0.3487556303717827,
      "learning_rate": 1.8258447882542005e-05,
      "loss": 0.2052,
      "step": 7020
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.44709314082971846,
      "learning_rate": 1.825788852128869e-05,
      "loss": 0.2762,
      "step": 7021
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5256823858485737,
      "learning_rate": 1.8257329078791555e-05,
      "loss": 0.2825,
      "step": 7022
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8522490048999662,
      "learning_rate": 1.8256769555056096e-05,
      "loss": 0.7574,
      "step": 7023
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33795234594710577,
      "learning_rate": 1.825620995008782e-05,
      "loss": 0.2466,
      "step": 7024
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6950519411062251,
      "learning_rate": 1.8255650263892236e-05,
      "loss": 0.3473,
      "step": 7025
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4197068713686661,
      "learning_rate": 1.8255090496474847e-05,
      "loss": 0.2436,
      "step": 7026
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9156440028349082,
      "learning_rate": 1.8254530647841164e-05,
      "loss": 0.5477,
      "step": 7027
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.306926499138619,
      "learning_rate": 1.8253970717996693e-05,
      "loss": 0.1837,
      "step": 7028
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5119387435126571,
      "learning_rate": 1.8253410706946936e-05,
      "loss": 0.3731,
      "step": 7029
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33541283284666445,
      "learning_rate": 1.8252850614697416e-05,
      "loss": 0.0793,
      "step": 7030
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35200339374091644,
      "learning_rate": 1.8252290441253632e-05,
      "loss": 0.2591,
      "step": 7031
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1142373022495256,
      "learning_rate": 1.82517301866211e-05,
      "loss": 0.7117,
      "step": 7032
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.026190082754865,
      "learning_rate": 1.825116985080533e-05,
      "loss": 0.3966,
      "step": 7033
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0613644355608922,
      "learning_rate": 1.8250609433811836e-05,
      "loss": 0.5011,
      "step": 7034
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.29964243670532276,
      "learning_rate": 1.8250048935646134e-05,
      "loss": 0.1995,
      "step": 7035
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4582047986158241,
      "learning_rate": 1.8249488356313736e-05,
      "loss": 0.3709,
      "step": 7036
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4227276368385357,
      "learning_rate": 1.824892769582016e-05,
      "loss": 0.2771,
      "step": 7037
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5201604292188932,
      "learning_rate": 1.8248366954170912e-05,
      "loss": 0.3017,
      "step": 7038
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3193963330623748,
      "learning_rate": 1.824780613137152e-05,
      "loss": 0.1977,
      "step": 7039
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5948319767822299,
      "learning_rate": 1.8247245227427496e-05,
      "loss": 0.4464,
      "step": 7040
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3170778247665007,
      "learning_rate": 1.8246684242344358e-05,
      "loss": 0.1287,
      "step": 7041
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5389200998457165,
      "learning_rate": 1.824612317612763e-05,
      "loss": 0.4031,
      "step": 7042
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4452325963877996,
      "learning_rate": 1.8245562028782826e-05,
      "loss": 0.2879,
      "step": 7043
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3734484755004167,
      "learning_rate": 1.8245000800315474e-05,
      "loss": 0.2545,
      "step": 7044
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9334136429548302,
      "learning_rate": 1.8244439490731087e-05,
      "loss": 0.5633,
      "step": 7045
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2830208880782122,
      "learning_rate": 1.8243878100035194e-05,
      "loss": 0.1131,
      "step": 7046
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5086894927162895,
      "learning_rate": 1.8243316628233314e-05,
      "loss": 0.4292,
      "step": 7047
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2755962314260759,
      "learning_rate": 1.8242755075330975e-05,
      "loss": 0.2077,
      "step": 7048
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.645108208042563,
      "learning_rate": 1.8242193441333697e-05,
      "loss": 0.3822,
      "step": 7049
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33413007517655774,
      "learning_rate": 1.8241631726247008e-05,
      "loss": 0.1871,
      "step": 7050
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9055801553649723,
      "learning_rate": 1.8241069930076433e-05,
      "loss": 0.5585,
      "step": 7051
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6405104635934176,
      "learning_rate": 1.8240508052827505e-05,
      "loss": 0.3676,
      "step": 7052
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0810199535476248,
      "learning_rate": 1.8239946094505743e-05,
      "loss": 0.4509,
      "step": 7053
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.30951780778629834,
      "learning_rate": 1.823938405511668e-05,
      "loss": 0.2627,
      "step": 7054
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5549102267716054,
      "learning_rate": 1.8238821934665847e-05,
      "loss": 0.3978,
      "step": 7055
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3421848733293806,
      "learning_rate": 1.8238259733158775e-05,
      "loss": 0.5827,
      "step": 7056
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3320620328732575,
      "learning_rate": 1.823769745060099e-05,
      "loss": 0.0754,
      "step": 7057
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38606761697211694,
      "learning_rate": 1.8237135086998027e-05,
      "loss": 0.3252,
      "step": 7058
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2480342562136844,
      "learning_rate": 1.8236572642355422e-05,
      "loss": 0.1412,
      "step": 7059
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3511665816223418,
      "learning_rate": 1.82360101166787e-05,
      "loss": 0.3403,
      "step": 7060
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7313612150121954,
      "learning_rate": 1.8235447509973405e-05,
      "loss": 0.3984,
      "step": 7061
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42584273299565756,
      "learning_rate": 1.8234884822245063e-05,
      "loss": 0.2694,
      "step": 7062
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4311377326611056,
      "learning_rate": 1.823432205349922e-05,
      "loss": 0.2884,
      "step": 7063
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0841915265443693,
      "learning_rate": 1.8233759203741405e-05,
      "loss": 0.7571,
      "step": 7064
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7144501533868965,
      "learning_rate": 1.8233196272977153e-05,
      "loss": 0.5404,
      "step": 7065
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3368007778413266,
      "learning_rate": 1.8232633261212014e-05,
      "loss": 0.2922,
      "step": 7066
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.32927719895235436,
      "learning_rate": 1.8232070168451518e-05,
      "loss": 0.2321,
      "step": 7067
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3450220803268068,
      "learning_rate": 1.8231506994701205e-05,
      "loss": 0.2166,
      "step": 7068
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7351345260587651,
      "learning_rate": 1.8230943739966617e-05,
      "loss": 0.4039,
      "step": 7069
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6478306346902262,
      "learning_rate": 1.8230380404253297e-05,
      "loss": 0.325,
      "step": 7070
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4377220360007126,
      "learning_rate": 1.822981698756679e-05,
      "loss": 0.3181,
      "step": 7071
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3651685872232428,
      "learning_rate": 1.822925348991263e-05,
      "loss": 0.2603,
      "step": 7072
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.163936256614741,
      "learning_rate": 1.8228689911296372e-05,
      "loss": 0.965,
      "step": 7073
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5206560022614712,
      "learning_rate": 1.8228126251723552e-05,
      "loss": 0.6356,
      "step": 7074
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6544140471166645,
      "learning_rate": 1.822756251119972e-05,
      "loss": 0.3708,
      "step": 7075
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3726894638689028,
      "learning_rate": 1.8226998689730416e-05,
      "loss": 0.2239,
      "step": 7076
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4989943070336479,
      "learning_rate": 1.8226434787321196e-05,
      "loss": 0.3216,
      "step": 7077
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.389355039640661,
      "learning_rate": 1.8225870803977604e-05,
      "loss": 0.313,
      "step": 7078
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43099881352331676,
      "learning_rate": 1.8225306739705185e-05,
      "loss": 0.326,
      "step": 7079
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.26035153628965974,
      "learning_rate": 1.8224742594509498e-05,
      "loss": 0.0751,
      "step": 7080
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4113094100799634,
      "learning_rate": 1.8224178368396083e-05,
      "loss": 0.3177,
      "step": 7081
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8738375778087814,
      "learning_rate": 1.8223614061370493e-05,
      "loss": 0.4147,
      "step": 7082
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3107447206451694,
      "learning_rate": 1.8223049673438287e-05,
      "loss": 0.275,
      "step": 7083
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8133614893206658,
      "learning_rate": 1.822248520460501e-05,
      "loss": 0.9304,
      "step": 7084
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2978375254098335,
      "learning_rate": 1.822192065487622e-05,
      "loss": 0.1996,
      "step": 7085
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7301861428030035,
      "learning_rate": 1.8221356024257463e-05,
      "loss": 0.5222,
      "step": 7086
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.32456613208373736,
      "learning_rate": 1.8220791312754307e-05,
      "loss": 0.2365,
      "step": 7087
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5031792479216264,
      "learning_rate": 1.8220226520372296e-05,
      "loss": 0.2305,
      "step": 7088
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3502210467938086,
      "learning_rate": 1.8219661647116994e-05,
      "loss": 0.1757,
      "step": 7089
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37595171592745474,
      "learning_rate": 1.8219096692993957e-05,
      "loss": 0.2984,
      "step": 7090
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0031322434464383,
      "learning_rate": 1.8218531658008743e-05,
      "loss": 0.3293,
      "step": 7091
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6307549078016776,
      "learning_rate": 1.821796654216691e-05,
      "loss": 0.8756,
      "step": 7092
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6462841458935925,
      "learning_rate": 1.8217401345474014e-05,
      "loss": 0.2949,
      "step": 7093
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37447002165937526,
      "learning_rate": 1.8216836067935624e-05,
      "loss": 0.2867,
      "step": 7094
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5328261056617489,
      "learning_rate": 1.8216270709557296e-05,
      "loss": 0.3359,
      "step": 7095
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4793567400840274,
      "learning_rate": 1.821570527034459e-05,
      "loss": 0.3083,
      "step": 7096
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3405410719253005,
      "learning_rate": 1.8215139750303077e-05,
      "loss": 0.1903,
      "step": 7097
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35545784611737,
      "learning_rate": 1.8214574149438315e-05,
      "loss": 0.2009,
      "step": 7098
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5457294711612296,
      "learning_rate": 1.8214008467755867e-05,
      "loss": 0.3559,
      "step": 7099
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8105210569028345,
      "learning_rate": 1.8213442705261305e-05,
      "loss": 0.411,
      "step": 7100
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.557642903530384,
      "learning_rate": 1.8212876861960186e-05,
      "loss": 0.4463,
      "step": 7101
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35005948044910357,
      "learning_rate": 1.8212310937858087e-05,
      "loss": 0.2161,
      "step": 7102
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4374948998822711,
      "learning_rate": 1.8211744932960568e-05,
      "loss": 0.3361,
      "step": 7103
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6377123623409675,
      "learning_rate": 1.8211178847273202e-05,
      "loss": 0.4022,
      "step": 7104
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0266964080354926,
      "learning_rate": 1.8210612680801553e-05,
      "loss": 0.5882,
      "step": 7105
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2097873312483628,
      "learning_rate": 1.8210046433551198e-05,
      "loss": 0.077,
      "step": 7106
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6798180335241759,
      "learning_rate": 1.8209480105527702e-05,
      "loss": 0.3131,
      "step": 7107
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3416019792040446,
      "learning_rate": 1.8208913696736644e-05,
      "loss": 0.2614,
      "step": 7108
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4251649309718241,
      "learning_rate": 1.820834720718359e-05,
      "loss": 0.2473,
      "step": 7109
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.004017402471403,
      "learning_rate": 1.8207780636874116e-05,
      "loss": 0.6136,
      "step": 7110
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9150277446240239,
      "learning_rate": 1.8207213985813793e-05,
      "loss": 0.3986,
      "step": 7111
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4420956499419072,
      "learning_rate": 1.8206647254008198e-05,
      "loss": 0.2824,
      "step": 7112
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4044473407688743,
      "learning_rate": 1.820608044146291e-05,
      "loss": 0.2892,
      "step": 7113
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5340103392309383,
      "learning_rate": 1.82055135481835e-05,
      "loss": 0.4259,
      "step": 7114
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2733194247571547,
      "learning_rate": 1.820494657417555e-05,
      "loss": 0.0748,
      "step": 7115
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4409610735393057,
      "learning_rate": 1.8204379519444636e-05,
      "loss": 0.2135,
      "step": 7116
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3587336478901706,
      "learning_rate": 1.820381238399633e-05,
      "loss": 0.2267,
      "step": 7117
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2880689032434227,
      "learning_rate": 1.8203245167836226e-05,
      "loss": 0.6777,
      "step": 7118
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4347556521147177,
      "learning_rate": 1.8202677870969893e-05,
      "loss": 0.2772,
      "step": 7119
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6544025580971625,
      "learning_rate": 1.820211049340292e-05,
      "loss": 0.392,
      "step": 7120
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2718074172745072,
      "learning_rate": 1.820154303514088e-05,
      "loss": 0.1883,
      "step": 7121
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.05783316372632,
      "learning_rate": 1.8200975496189365e-05,
      "loss": 0.4326,
      "step": 7122
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.3146292394731955,
      "learning_rate": 1.8200407876553955e-05,
      "loss": 0.9214,
      "step": 7123
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2585683767750293,
      "learning_rate": 1.8199840176240232e-05,
      "loss": 0.1214,
      "step": 7124
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3814830965807831,
      "learning_rate": 1.8199272395253783e-05,
      "loss": 0.2857,
      "step": 7125
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3797687276502303,
      "learning_rate": 1.8198704533600192e-05,
      "loss": 0.2583,
      "step": 7126
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9733870677587364,
      "learning_rate": 1.819813659128505e-05,
      "loss": 0.7618,
      "step": 7127
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.109158914093853,
      "learning_rate": 1.8197568568313943e-05,
      "loss": 0.4684,
      "step": 7128
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9528716204460695,
      "learning_rate": 1.8197000464692462e-05,
      "loss": 0.5786,
      "step": 7129
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35160766473073907,
      "learning_rate": 1.819643228042619e-05,
      "loss": 0.1778,
      "step": 7130
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5923414188726118,
      "learning_rate": 1.8195864015520718e-05,
      "loss": 0.3913,
      "step": 7131
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3131682411262597,
      "learning_rate": 1.8195295669981643e-05,
      "loss": 0.2608,
      "step": 7132
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9157019762137161,
      "learning_rate": 1.8194727243814552e-05,
      "loss": 0.5984,
      "step": 7133
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2097464525522414,
      "learning_rate": 1.8194158737025036e-05,
      "loss": 0.0707,
      "step": 7134
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4510922608900804,
      "learning_rate": 1.819359014961869e-05,
      "loss": 0.3466,
      "step": 7135
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3607223967137859,
      "learning_rate": 1.8193021481601112e-05,
      "loss": 0.2156,
      "step": 7136
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4093135775431525,
      "learning_rate": 1.819245273297789e-05,
      "loss": 0.3065,
      "step": 7137
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0241468711788047,
      "learning_rate": 1.8191883903754625e-05,
      "loss": 0.7067,
      "step": 7138
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3528063284446589,
      "learning_rate": 1.8191314993936905e-05,
      "loss": 0.197,
      "step": 7139
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5671724590299805,
      "learning_rate": 1.8190746003530336e-05,
      "loss": 0.4216,
      "step": 7140
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1254721886052592,
      "learning_rate": 1.8190176932540515e-05,
      "loss": 0.0776,
      "step": 7141
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4103680442697661,
      "learning_rate": 1.8189607780973036e-05,
      "loss": 0.7074,
      "step": 7142
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23675698313868362,
      "learning_rate": 1.8189038548833503e-05,
      "loss": 0.1955,
      "step": 7143
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3768299664134257,
      "learning_rate": 1.818846923612751e-05,
      "loss": 0.2942,
      "step": 7144
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.429283219816672,
      "learning_rate": 1.8187899842860666e-05,
      "loss": 0.2201,
      "step": 7145
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9467024212907965,
      "learning_rate": 1.8187330369038567e-05,
      "loss": 0.5667,
      "step": 7146
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7000764841489158,
      "learning_rate": 1.818676081466682e-05,
      "loss": 0.4183,
      "step": 7147
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.433749419534851,
      "learning_rate": 1.8186191179751024e-05,
      "loss": 0.3075,
      "step": 7148
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33734919330641316,
      "learning_rate": 1.818562146429679e-05,
      "loss": 0.2704,
      "step": 7149
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4458777496243099,
      "learning_rate": 1.818505166830971e-05,
      "loss": 0.275,
      "step": 7150
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3938472652194032,
      "learning_rate": 1.8184481791795404e-05,
      "loss": 0.7201,
      "step": 7151
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.23978451429488243,
      "learning_rate": 1.8183911834759474e-05,
      "loss": 0.0769,
      "step": 7152
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2929143851995744,
      "learning_rate": 1.8183341797207526e-05,
      "loss": 0.2345,
      "step": 7153
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8537189898313561,
      "learning_rate": 1.8182771679145168e-05,
      "loss": 0.421,
      "step": 7154
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3569057918837523,
      "learning_rate": 1.818220148057801e-05,
      "loss": 0.3487,
      "step": 7155
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8105285367700497,
      "learning_rate": 1.818163120151166e-05,
      "loss": 0.4474,
      "step": 7156
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43690117675376405,
      "learning_rate": 1.818106084195173e-05,
      "loss": 0.3147,
      "step": 7157
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42932643494303363,
      "learning_rate": 1.8180490401903834e-05,
      "loss": 0.227,
      "step": 7158
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4484373493359717,
      "learning_rate": 1.817991988137358e-05,
      "loss": 0.7385,
      "step": 7159
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4251839338870545,
      "learning_rate": 1.8179349280366585e-05,
      "loss": 0.2714,
      "step": 7160
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33964296180091763,
      "learning_rate": 1.8178778598888458e-05,
      "loss": 0.245,
      "step": 7161
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41593998718765357,
      "learning_rate": 1.8178207836944816e-05,
      "loss": 0.2032,
      "step": 7162
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5739370803258913,
      "learning_rate": 1.8177636994541274e-05,
      "loss": 0.3616,
      "step": 7163
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42142697559260306,
      "learning_rate": 1.817706607168345e-05,
      "loss": 0.2804,
      "step": 7164
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8008240042118243,
      "learning_rate": 1.817649506837696e-05,
      "loss": 0.5121,
      "step": 7165
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6696621104003937,
      "learning_rate": 1.8175923984627418e-05,
      "loss": 0.421,
      "step": 7166
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.30241207151703553,
      "learning_rate": 1.8175352820440447e-05,
      "loss": 0.2519,
      "step": 7167
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.48005314986007436,
      "learning_rate": 1.8174781575821667e-05,
      "loss": 0.3994,
      "step": 7168
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6406129281117512,
      "learning_rate": 1.8174210250776695e-05,
      "loss": 0.1438,
      "step": 7169
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.798393569265637,
      "learning_rate": 1.8173638845311152e-05,
      "loss": 0.547,
      "step": 7170
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2992979536709758,
      "learning_rate": 1.817306735943066e-05,
      "loss": 0.1979,
      "step": 7171
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6358640952381045,
      "learning_rate": 1.8172495793140844e-05,
      "loss": 0.4766,
      "step": 7172
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.34415643355581893,
      "learning_rate": 1.8171924146447325e-05,
      "loss": 0.2513,
      "step": 7173
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.31732981310116654,
      "learning_rate": 1.8171352419355725e-05,
      "loss": 0.1981,
      "step": 7174
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.385582398909287,
      "learning_rate": 1.817078061187167e-05,
      "loss": 0.1662,
      "step": 7175
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3529077972141788,
      "learning_rate": 1.8170208724000795e-05,
      "loss": 0.2433,
      "step": 7176
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.7056392368850024,
      "learning_rate": 1.8169636755748706e-05,
      "loss": 0.7268,
      "step": 7177
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4222065514291206,
      "learning_rate": 1.816906470712105e-05,
      "loss": 0.255,
      "step": 7178
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.43367795812578136,
      "learning_rate": 1.8168492578123448e-05,
      "loss": 0.3595,
      "step": 7179
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3414499474725369,
      "learning_rate": 1.8167920368761527e-05,
      "loss": 0.2557,
      "step": 7180
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1242496745180652,
      "learning_rate": 1.8167348079040916e-05,
      "loss": 0.5422,
      "step": 7181
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38762994687021257,
      "learning_rate": 1.816677570896725e-05,
      "loss": 0.2286,
      "step": 7182
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.6030369543834213,
      "learning_rate": 1.816620325854616e-05,
      "loss": 0.9648,
      "step": 7183
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3628474108959583,
      "learning_rate": 1.8165630727783268e-05,
      "loss": 0.2041,
      "step": 7184
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5707321245238722,
      "learning_rate": 1.8165058116684218e-05,
      "loss": 0.3962,
      "step": 7185
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2523428039477341,
      "learning_rate": 1.8164485425254637e-05,
      "loss": 0.2098,
      "step": 7186
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9826526449834778,
      "learning_rate": 1.8163912653500163e-05,
      "loss": 0.6379,
      "step": 7187
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8105837316003809,
      "learning_rate": 1.8163339801426433e-05,
      "loss": 0.43,
      "step": 7188
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3211523909113257,
      "learning_rate": 1.816276686903908e-05,
      "loss": 0.1837,
      "step": 7189
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5599104586942517,
      "learning_rate": 1.8162193856343738e-05,
      "loss": 0.4111,
      "step": 7190
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2524892907459518,
      "learning_rate": 1.8161620763346045e-05,
      "loss": 0.2285,
      "step": 7191
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.093301443149453,
      "learning_rate": 1.8161047590051645e-05,
      "loss": 0.8964,
      "step": 7192
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4350078840565971,
      "learning_rate": 1.816047433646617e-05,
      "loss": 0.2325,
      "step": 7193
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3725907242973409,
      "learning_rate": 1.815990100259527e-05,
      "loss": 0.2907,
      "step": 7194
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.287284488665144,
      "learning_rate": 1.8159327588444574e-05,
      "loss": 0.1544,
      "step": 7195
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46558002119758884,
      "learning_rate": 1.815875409401973e-05,
      "loss": 0.3824,
      "step": 7196
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.47185154044795274,
      "learning_rate": 1.8158180519326376e-05,
      "loss": 0.3631,
      "step": 7197
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.48036371282092594,
      "learning_rate": 1.815760686437016e-05,
      "loss": 0.3342,
      "step": 7198
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4415360425722207,
      "learning_rate": 1.815703312915672e-05,
      "loss": 0.308,
      "step": 7199
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2367196325317378,
      "learning_rate": 1.815645931369171e-05,
      "loss": 0.702,
      "step": 7200
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0044325190597592,
      "learning_rate": 1.8155885417980767e-05,
      "loss": 0.6126,
      "step": 7201
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35578202893017513,
      "learning_rate": 1.8155311442029536e-05,
      "loss": 0.218,
      "step": 7202
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.31561700969190365,
      "learning_rate": 1.8154737385843672e-05,
      "loss": 0.2878,
      "step": 7203
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2808395774026221,
      "learning_rate": 1.815416324942882e-05,
      "loss": 0.0744,
      "step": 7204
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7569300340325625,
      "learning_rate": 1.8153589032790622e-05,
      "loss": 0.5169,
      "step": 7205
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.715613230863787,
      "learning_rate": 1.8153014735934732e-05,
      "loss": 0.4067,
      "step": 7206
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46845197648225023,
      "learning_rate": 1.8152440358866805e-05,
      "loss": 0.3363,
      "step": 7207
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5035767562260541,
      "learning_rate": 1.8151865901592483e-05,
      "loss": 0.2569,
      "step": 7208
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38258878508495014,
      "learning_rate": 1.8151291364117423e-05,
      "loss": 0.3256,
      "step": 7209
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9989700057256212,
      "learning_rate": 1.815071674644728e-05,
      "loss": 0.5782,
      "step": 7210
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7919085788241639,
      "learning_rate": 1.8150142048587696e-05,
      "loss": 0.3935,
      "step": 7211
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3964244698530908,
      "learning_rate": 1.8149567270544342e-05,
      "loss": 0.2421,
      "step": 7212
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.316247717122348,
      "learning_rate": 1.8148992412322855e-05,
      "loss": 0.2213,
      "step": 7213
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41201901498335164,
      "learning_rate": 1.8148417473928907e-05,
      "loss": 0.3028,
      "step": 7214
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9533994052023698,
      "learning_rate": 1.8147842455368142e-05,
      "loss": 0.3541,
      "step": 7215
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7882503112203338,
      "learning_rate": 1.8147267356646222e-05,
      "loss": 0.3845,
      "step": 7216
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40051627007886015,
      "learning_rate": 1.814669217776881e-05,
      "loss": 0.2626,
      "step": 7217
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.8491684923313172,
      "learning_rate": 1.8146116918741553e-05,
      "loss": 0.8361,
      "step": 7218
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0858022053762972,
      "learning_rate": 1.8145541579570117e-05,
      "loss": 0.6213,
      "step": 7219
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37454985838363913,
      "learning_rate": 1.814496616026017e-05,
      "loss": 0.3156,
      "step": 7220
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33864951832997225,
      "learning_rate": 1.814439066081736e-05,
      "loss": 0.1946,
      "step": 7221
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5125622418737551,
      "learning_rate": 1.8143815081247358e-05,
      "loss": 0.3026,
      "step": 7222
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8000892261370026,
      "learning_rate": 1.8143239421555822e-05,
      "loss": 0.4115,
      "step": 7223
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5763068628124838,
      "learning_rate": 1.8142663681748417e-05,
      "loss": 0.2611,
      "step": 7224
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3932586599992182,
      "learning_rate": 1.8142087861830807e-05,
      "loss": 0.1869,
      "step": 7225
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40650168349005916,
      "learning_rate": 1.814151196180866e-05,
      "loss": 0.3237,
      "step": 7226
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.34177801844164574,
      "learning_rate": 1.8140935981687635e-05,
      "loss": 0.2549,
      "step": 7227
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9929901515683505,
      "learning_rate": 1.8140359921473406e-05,
      "loss": 0.539,
      "step": 7228
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9655020149385061,
      "learning_rate": 1.8139783781171635e-05,
      "loss": 0.4987,
      "step": 7229
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3113096533013287,
      "learning_rate": 1.8139207560788e-05,
      "loss": 0.2534,
      "step": 7230
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0637658897238111,
      "learning_rate": 1.8138631260328154e-05,
      "loss": 0.5966,
      "step": 7231
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.41192683004998143,
      "learning_rate": 1.813805487979778e-05,
      "loss": 0.3382,
      "step": 7232
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.29436120975911556,
      "learning_rate": 1.8137478419202542e-05,
      "loss": 0.204,
      "step": 7233
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3556543085682115,
      "learning_rate": 1.8136901878548118e-05,
      "loss": 0.1709,
      "step": 7234
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5856990693155825,
      "learning_rate": 1.813632525784017e-05,
      "loss": 0.3411,
      "step": 7235
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.500913318871112,
      "learning_rate": 1.8135748557084382e-05,
      "loss": 0.3995,
      "step": 7236
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4145171651338944,
      "learning_rate": 1.813517177628642e-05,
      "loss": 0.8501,
      "step": 7237
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3339967702513058,
      "learning_rate": 1.8134594915451964e-05,
      "loss": 0.2633,
      "step": 7238
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0041772705299077,
      "learning_rate": 1.8134017974586683e-05,
      "loss": 0.4358,
      "step": 7239
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4254170181458644,
      "learning_rate": 1.813344095369626e-05,
      "loss": 0.2929,
      "step": 7240
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3800350505809601,
      "learning_rate": 1.813286385278637e-05,
      "loss": 0.2246,
      "step": 7241
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.530067861299744,
      "learning_rate": 1.8132286671862686e-05,
      "loss": 0.3241,
      "step": 7242
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.24364828336602745,
      "learning_rate": 1.8131709410930888e-05,
      "loss": 0.0769,
      "step": 7243
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4361200482179376,
      "learning_rate": 1.813113206999666e-05,
      "loss": 0.3142,
      "step": 7244
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37237245740618163,
      "learning_rate": 1.813055464906568e-05,
      "loss": 0.278,
      "step": 7245
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.380926602216434,
      "learning_rate": 1.8129977148143628e-05,
      "loss": 0.8771,
      "step": 7246
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7112279039416315,
      "learning_rate": 1.8129399567236183e-05,
      "loss": 0.3678,
      "step": 7247
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.42009770050364786,
      "learning_rate": 1.812882190634903e-05,
      "loss": 0.2722,
      "step": 7248
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3974066935317239,
      "learning_rate": 1.8128244165487853e-05,
      "loss": 0.3194,
      "step": 7249
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33324616620002984,
      "learning_rate": 1.8127666344658336e-05,
      "loss": 0.3,
      "step": 7250
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40243215986494246,
      "learning_rate": 1.8127088443866164e-05,
      "loss": 0.1162,
      "step": 7251
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4421168051853839,
      "learning_rate": 1.8126510463117023e-05,
      "loss": 0.2538,
      "step": 7252
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3100683303997154,
      "learning_rate": 1.8125932402416598e-05,
      "loss": 0.2207,
      "step": 7253
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.0621864805030352,
      "learning_rate": 1.8125354261770574e-05,
      "loss": 0.3937,
      "step": 7254
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1876969409245537,
      "learning_rate": 1.812477604118464e-05,
      "loss": 0.6064,
      "step": 7255
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.34220201820050655,
      "learning_rate": 1.8124197740664488e-05,
      "loss": 0.2845,
      "step": 7256
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4237510650223017,
      "learning_rate": 1.8123619360215807e-05,
      "loss": 0.2935,
      "step": 7257
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4198150632109733,
      "learning_rate": 1.812304089984428e-05,
      "loss": 0.2688,
      "step": 7258
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3854496201213444,
      "learning_rate": 1.812246235955561e-05,
      "loss": 0.746,
      "step": 7259
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2681129399423957,
      "learning_rate": 1.812188373935548e-05,
      "loss": 0.097,
      "step": 7260
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9323966028840888,
      "learning_rate": 1.8121305039249585e-05,
      "loss": 0.3662,
      "step": 7261
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3416684155097528,
      "learning_rate": 1.812072625924362e-05,
      "loss": 0.221,
      "step": 7262
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3659840868103293,
      "learning_rate": 1.8120147399343276e-05,
      "loss": 0.3189,
      "step": 7263
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.842222347422627,
      "learning_rate": 1.8119568459554254e-05,
      "loss": 0.423,
      "step": 7264
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7197118397439249,
      "learning_rate": 1.8118989439882238e-05,
      "loss": 0.5252,
      "step": 7265
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5089710380248523,
      "learning_rate": 1.811841034033294e-05,
      "loss": 0.0744,
      "step": 7266
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3417129729619734,
      "learning_rate": 1.8117831160912048e-05,
      "loss": 0.2487,
      "step": 7267
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3841232747694358,
      "learning_rate": 1.811725190162526e-05,
      "loss": 0.3307,
      "step": 7268
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.22208530632740656,
      "learning_rate": 1.811667256247828e-05,
      "loss": 0.074,
      "step": 7269
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9897433348551798,
      "learning_rate": 1.8116093143476802e-05,
      "loss": 0.901,
      "step": 7270
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2847857689952007,
      "learning_rate": 1.811551364462653e-05,
      "loss": 0.1987,
      "step": 7271
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.2618375741670962,
      "learning_rate": 1.8114934065933165e-05,
      "loss": 0.8182,
      "step": 7272
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6388118473717801,
      "learning_rate": 1.8114354407402405e-05,
      "loss": 0.3656,
      "step": 7273
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5055541940442217,
      "learning_rate": 1.811377466903996e-05,
      "loss": 0.3884,
      "step": 7274
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.36277069982235005,
      "learning_rate": 1.811319485085153e-05,
      "loss": 0.2443,
      "step": 7275
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5355947400675386,
      "learning_rate": 1.8112614952842817e-05,
      "loss": 0.3933,
      "step": 7276
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.982239807258231,
      "learning_rate": 1.811203497501953e-05,
      "loss": 0.3809,
      "step": 7277
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.3119610291794812,
      "learning_rate": 1.8111454917387374e-05,
      "loss": 0.701,
      "step": 7278
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.20771990153243414,
      "learning_rate": 1.8110874779952053e-05,
      "loss": 0.135,
      "step": 7279
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.27684086226424526,
      "learning_rate": 1.8110294562719283e-05,
      "loss": 0.2379,
      "step": 7280
    },
    {
      "epoch": 0.22,
      "grad_norm": 2.2719827654787816,
      "learning_rate": 1.8109714265694763e-05,
      "loss": 0.8318,
      "step": 7281
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5857596705076955,
      "learning_rate": 1.8109133888884204e-05,
      "loss": 0.3958,
      "step": 7282
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8879049356031276,
      "learning_rate": 1.810855343229332e-05,
      "loss": 0.6522,
      "step": 7283
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3446846691000784,
      "learning_rate": 1.8107972895927816e-05,
      "loss": 0.201,
      "step": 7284
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6184976886619031,
      "learning_rate": 1.810739227979341e-05,
      "loss": 0.3667,
      "step": 7285
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.31330029709320695,
      "learning_rate": 1.810681158389581e-05,
      "loss": 0.233,
      "step": 7286
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.68217698589973,
      "learning_rate": 1.810623080824073e-05,
      "loss": 0.7624,
      "step": 7287
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.17951274502542003,
      "learning_rate": 1.8105649952833883e-05,
      "loss": 0.0955,
      "step": 7288
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5207139181941662,
      "learning_rate": 1.810506901768099e-05,
      "loss": 0.3193,
      "step": 7289
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5166472431657608,
      "learning_rate": 1.8104488002787755e-05,
      "loss": 0.3095,
      "step": 7290
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9151104058541424,
      "learning_rate": 1.8103906908159904e-05,
      "loss": 0.5298,
      "step": 7291
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3641332065297594,
      "learning_rate": 1.8103325733803147e-05,
      "loss": 0.2916,
      "step": 7292
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.27166190231281545,
      "learning_rate": 1.8102744479723207e-05,
      "loss": 0.0747,
      "step": 7293
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3846214729330775,
      "learning_rate": 1.81021631459258e-05,
      "loss": 0.3052,
      "step": 7294
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.1221016780738093,
      "learning_rate": 1.8101581732416648e-05,
      "loss": 0.3572,
      "step": 7295
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.516211797471458,
      "learning_rate": 1.8101000239201466e-05,
      "loss": 0.3475,
      "step": 7296
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3551486400588688,
      "learning_rate": 1.810041866628598e-05,
      "loss": 0.2473,
      "step": 7297
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.46082338112963855,
      "learning_rate": 1.8099837013675906e-05,
      "loss": 0.3043,
      "step": 7298
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35903022625391967,
      "learning_rate": 1.8099255281376974e-05,
      "loss": 0.2513,
      "step": 7299
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8745762020291382,
      "learning_rate": 1.8098673469394905e-05,
      "loss": 0.5513,
      "step": 7300
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8931991710412907,
      "learning_rate": 1.809809157773542e-05,
      "loss": 0.5005,
      "step": 7301
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9884365556771328,
      "learning_rate": 1.8097509606404245e-05,
      "loss": 0.6045,
      "step": 7302
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3939345512754664,
      "learning_rate": 1.8096927555407106e-05,
      "loss": 0.2249,
      "step": 7303
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4038998249466474,
      "learning_rate": 1.809634542474973e-05,
      "loss": 0.3384,
      "step": 7304
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9093526991055209,
      "learning_rate": 1.8095763214437845e-05,
      "loss": 0.0325,
      "step": 7305
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.355599584924677,
      "learning_rate": 1.8095180924477174e-05,
      "loss": 0.1889,
      "step": 7306
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.38811517726305217,
      "learning_rate": 1.809459855487345e-05,
      "loss": 0.2894,
      "step": 7307
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8415778852448912,
      "learning_rate": 1.8094016105632405e-05,
      "loss": 0.4068,
      "step": 7308
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5168426178996128,
      "learning_rate": 1.8093433576759766e-05,
      "loss": 0.2767,
      "step": 7309
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3570369781294178,
      "learning_rate": 1.8092850968261262e-05,
      "loss": 0.2885,
      "step": 7310
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.747451450182691,
      "learning_rate": 1.8092268280142627e-05,
      "loss": 0.3715,
      "step": 7311
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37986421228992157,
      "learning_rate": 1.8091685512409597e-05,
      "loss": 0.255,
      "step": 7312
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.178681098458306,
      "learning_rate": 1.80911026650679e-05,
      "loss": 0.6229,
      "step": 7313
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.755200468803727,
      "learning_rate": 1.8090519738123275e-05,
      "loss": 0.3846,
      "step": 7314
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3708205753042441,
      "learning_rate": 1.808993673158145e-05,
      "loss": 0.3538,
      "step": 7315
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4563526436028016,
      "learning_rate": 1.808935364544817e-05,
      "loss": 0.0718,
      "step": 7316
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.36515002676017155,
      "learning_rate": 1.8088770479729165e-05,
      "loss": 0.3427,
      "step": 7317
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.40899441710392936,
      "learning_rate": 1.8088187234430173e-05,
      "loss": 0.1921,
      "step": 7318
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.4844828682730156,
      "learning_rate": 1.8087603909556936e-05,
      "loss": 0.7987,
      "step": 7319
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.2821609951309701,
      "learning_rate": 1.8087020505115192e-05,
      "loss": 0.0702,
      "step": 7320
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33523897450390044,
      "learning_rate": 1.8086437021110678e-05,
      "loss": 0.2169,
      "step": 7321
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37774135116594,
      "learning_rate": 1.8085853457549133e-05,
      "loss": 0.3084,
      "step": 7322
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7716645238742471,
      "learning_rate": 1.8085269814436307e-05,
      "loss": 0.4053,
      "step": 7323
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9864194676806092,
      "learning_rate": 1.808468609177793e-05,
      "loss": 0.5832,
      "step": 7324
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3322469064013921,
      "learning_rate": 1.8084102289579752e-05,
      "loss": 0.2261,
      "step": 7325
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4612250705012509,
      "learning_rate": 1.808351840784752e-05,
      "loss": 0.3978,
      "step": 7326
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3274645051176754,
      "learning_rate": 1.808293444658697e-05,
      "loss": 0.2532,
      "step": 7327
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.9725879726116347,
      "learning_rate": 1.8082350405803854e-05,
      "loss": 0.8168,
      "step": 7328
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4516636809644997,
      "learning_rate": 1.808176628550391e-05,
      "loss": 0.1207,
      "step": 7329
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.3960544758548881,
      "learning_rate": 1.8081182085692897e-05,
      "loss": 0.3054,
      "step": 7330
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37577843390591165,
      "learning_rate": 1.808059780637655e-05,
      "loss": 0.147,
      "step": 7331
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.9994999668740769,
      "learning_rate": 1.8080013447560627e-05,
      "loss": 0.6834,
      "step": 7332
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.37405887537860993,
      "learning_rate": 1.807942900925087e-05,
      "loss": 0.2998,
      "step": 7333
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.33068836812835756,
      "learning_rate": 1.8078844491453033e-05,
      "loss": 0.2398,
      "step": 7334
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.532625755528432,
      "learning_rate": 1.8078259894172866e-05,
      "loss": 0.3332,
      "step": 7335
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.8867962420941908,
      "learning_rate": 1.807767521741612e-05,
      "loss": 0.3444,
      "step": 7336
    },
    {
      "epoch": 0.22,
      "grad_norm": 1.5365286636836506,
      "learning_rate": 1.8077090461188548e-05,
      "loss": 0.8411,
      "step": 7337
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.36745194870959863,
      "learning_rate": 1.8076505625495904e-05,
      "loss": 0.0765,
      "step": 7338
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.35059360315498667,
      "learning_rate": 1.8075920710343936e-05,
      "loss": 0.3075,
      "step": 7339
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4732650158179052,
      "learning_rate": 1.8075335715738404e-05,
      "loss": 0.2749,
      "step": 7340
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7733208544460091,
      "learning_rate": 1.8074750641685064e-05,
      "loss": 0.5056,
      "step": 7341
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7286061344825047,
      "learning_rate": 1.8074165488189667e-05,
      "loss": 0.398,
      "step": 7342
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.7932210912460775,
      "learning_rate": 1.8073580255257974e-05,
      "loss": 0.3244,
      "step": 7343
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.4230382048076764,
      "learning_rate": 1.8072994942895747e-05,
      "loss": 0.2628,
      "step": 7344
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.6701487632231901,
      "learning_rate": 1.8072409551108734e-05,
      "loss": 0.3949,
      "step": 7345
    },
    {
      "epoch": 0.22,
      "grad_norm": 0.5028935422148343,
      "learning_rate": 1.80718240799027e-05,
      "loss": 0.2831,
      "step": 7346
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2834798467081654,
      "learning_rate": 1.8071238529283406e-05,
      "loss": 0.0661,
      "step": 7347
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5647039868656598,
      "learning_rate": 1.807065289925661e-05,
      "loss": 0.295,
      "step": 7348
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3258219979676322,
      "learning_rate": 1.8070067189828074e-05,
      "loss": 0.2098,
      "step": 7349
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9367453113964593,
      "learning_rate": 1.8069481401003566e-05,
      "loss": 0.5023,
      "step": 7350
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30650619189891054,
      "learning_rate": 1.8068895532788842e-05,
      "loss": 0.244,
      "step": 7351
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1229477750257397,
      "learning_rate": 1.806830958518967e-05,
      "loss": 0.3118,
      "step": 7352
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39770326734286804,
      "learning_rate": 1.8067723558211815e-05,
      "loss": 0.2296,
      "step": 7353
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.223052749205996,
      "learning_rate": 1.806713745186104e-05,
      "loss": 0.747,
      "step": 7354
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0029956074101853,
      "learning_rate": 1.806655126614311e-05,
      "loss": 0.5225,
      "step": 7355
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.33996721005099917,
      "learning_rate": 1.8065965001063796e-05,
      "loss": 0.2664,
      "step": 7356
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30346017585504825,
      "learning_rate": 1.8065378656628863e-05,
      "loss": 0.2053,
      "step": 7357
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3410135556919138,
      "learning_rate": 1.806479223284408e-05,
      "loss": 0.2754,
      "step": 7358
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7014593272055517,
      "learning_rate": 1.806420572971522e-05,
      "loss": 0.3936,
      "step": 7359
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5739696362610961,
      "learning_rate": 1.806361914724805e-05,
      "loss": 0.3903,
      "step": 7360
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6925016969351299,
      "learning_rate": 1.806303248544834e-05,
      "loss": 0.435,
      "step": 7361
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38944080635451606,
      "learning_rate": 1.8062445744321866e-05,
      "loss": 0.2436,
      "step": 7362
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5412411628427815,
      "learning_rate": 1.8061858923874396e-05,
      "loss": 0.3993,
      "step": 7363
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4385249546550189,
      "learning_rate": 1.8061272024111707e-05,
      "loss": 0.3042,
      "step": 7364
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0312284508080365,
      "learning_rate": 1.8060685045039568e-05,
      "loss": 0.6023,
      "step": 7365
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3236748408396939,
      "learning_rate": 1.806009798666376e-05,
      "loss": 0.1837,
      "step": 7366
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4925979651528479,
      "learning_rate": 1.8059510848990057e-05,
      "loss": 0.3887,
      "step": 7367
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2803273213540187,
      "learning_rate": 1.805892363202423e-05,
      "loss": 0.1783,
      "step": 7368
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4062458134730876,
      "learning_rate": 1.805833633577206e-05,
      "loss": 0.3353,
      "step": 7369
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30589355844474303,
      "learning_rate": 1.805774896023933e-05,
      "loss": 0.0776,
      "step": 7370
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39931138862104204,
      "learning_rate": 1.8057161505431814e-05,
      "loss": 0.3094,
      "step": 7371
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8766680945080957,
      "learning_rate": 1.8056573971355292e-05,
      "loss": 0.4558,
      "step": 7372
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9472266919061837,
      "learning_rate": 1.8055986358015544e-05,
      "loss": 0.4691,
      "step": 7373
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4744045575297844,
      "learning_rate": 1.8055398665418348e-05,
      "loss": 0.3974,
      "step": 7374
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.36114283494880123,
      "learning_rate": 1.8054810893569496e-05,
      "loss": 0.1801,
      "step": 7375
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.36679683628952703,
      "learning_rate": 1.805422304247476e-05,
      "loss": 0.3358,
      "step": 7376
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4337227049680882,
      "learning_rate": 1.8053635112139924e-05,
      "loss": 0.2382,
      "step": 7377
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4748699867667745,
      "learning_rate": 1.805304710257078e-05,
      "loss": 0.275,
      "step": 7378
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6770617287203958,
      "learning_rate": 1.805245901377311e-05,
      "loss": 0.2792,
      "step": 7379
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.390345527872432,
      "learning_rate": 1.8051870845752695e-05,
      "loss": 0.2842,
      "step": 7380
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.33546580306042717,
      "learning_rate": 1.8051282598515326e-05,
      "loss": 0.2743,
      "step": 7381
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.6198555093593574,
      "learning_rate": 1.805069427206679e-05,
      "loss": 0.8215,
      "step": 7382
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7841667382482823,
      "learning_rate": 1.8050105866412877e-05,
      "loss": 0.3903,
      "step": 7383
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4219821330838212,
      "learning_rate": 1.804951738155937e-05,
      "loss": 0.2787,
      "step": 7384
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41032334440791407,
      "learning_rate": 1.8048928817512066e-05,
      "loss": 0.3252,
      "step": 7385
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2981772146196244,
      "learning_rate": 1.8048340174276752e-05,
      "loss": 0.1142,
      "step": 7386
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2997010981533213,
      "learning_rate": 1.8047751451859215e-05,
      "loss": 0.2622,
      "step": 7387
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3863696998481998,
      "learning_rate": 1.8047162650265254e-05,
      "loss": 0.077,
      "step": 7388
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4291282332775214,
      "learning_rate": 1.8046573769500658e-05,
      "loss": 0.3311,
      "step": 7389
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9979091914870092,
      "learning_rate": 1.804598480957122e-05,
      "loss": 0.5191,
      "step": 7390
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9808421177175063,
      "learning_rate": 1.804539577048274e-05,
      "loss": 0.6539,
      "step": 7391
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3588950033396483,
      "learning_rate": 1.8044806652241008e-05,
      "loss": 0.2883,
      "step": 7392
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4141421171003674,
      "learning_rate": 1.804421745485182e-05,
      "loss": 0.2782,
      "step": 7393
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5058776696136362,
      "learning_rate": 1.8043628178320972e-05,
      "loss": 0.2833,
      "step": 7394
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1568464928868116,
      "learning_rate": 1.8043038822654265e-05,
      "loss": 0.7517,
      "step": 7395
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30356961963694945,
      "learning_rate": 1.8042449387857497e-05,
      "loss": 0.1222,
      "step": 7396
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9543061880015561,
      "learning_rate": 1.8041859873936464e-05,
      "loss": 0.404,
      "step": 7397
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4339639062593038,
      "learning_rate": 1.8041270280896967e-05,
      "loss": 0.2017,
      "step": 7398
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2758181212700948,
      "learning_rate": 1.804068060874481e-05,
      "loss": 0.2544,
      "step": 7399
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0087899788938828,
      "learning_rate": 1.8040090857485787e-05,
      "loss": 0.7072,
      "step": 7400
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7088825469134569,
      "learning_rate": 1.8039501027125707e-05,
      "loss": 0.3834,
      "step": 7401
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8211866990243285,
      "learning_rate": 1.803891111767037e-05,
      "loss": 0.4535,
      "step": 7402
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3878237448196551,
      "learning_rate": 1.803832112912558e-05,
      "loss": 0.2355,
      "step": 7403
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.501413451998563,
      "learning_rate": 1.8037731061497147e-05,
      "loss": 0.3418,
      "step": 7404
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.26805102661511554,
      "learning_rate": 1.8037140914790863e-05,
      "loss": 0.1748,
      "step": 7405
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5441901184536199,
      "learning_rate": 1.8036550689012547e-05,
      "loss": 0.3379,
      "step": 7406
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41490192581158664,
      "learning_rate": 1.8035960384167997e-05,
      "loss": 0.2104,
      "step": 7407
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2505798502093288,
      "learning_rate": 1.803537000026303e-05,
      "loss": 0.6748,
      "step": 7408
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2052469674909752,
      "learning_rate": 1.8034779537303446e-05,
      "loss": 0.4307,
      "step": 7409
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4846153556835692,
      "learning_rate": 1.8034188995295055e-05,
      "loss": 0.4346,
      "step": 7410
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.334970967009428,
      "learning_rate": 1.8033598374243673e-05,
      "loss": 0.215,
      "step": 7411
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4392158253825152,
      "learning_rate": 1.8033007674155106e-05,
      "loss": 0.2733,
      "step": 7412
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.699154427004081,
      "learning_rate": 1.8032416895035162e-05,
      "loss": 0.8817,
      "step": 7413
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23245683176919657,
      "learning_rate": 1.803182603688966e-05,
      "loss": 0.1039,
      "step": 7414
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9671497464008875,
      "learning_rate": 1.8031235099724415e-05,
      "loss": 0.4822,
      "step": 7415
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3693838575210054,
      "learning_rate": 1.803064408354523e-05,
      "loss": 0.2379,
      "step": 7416
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5405708477436255,
      "learning_rate": 1.8030052988357932e-05,
      "loss": 0.3816,
      "step": 7417
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7183528133997863,
      "learning_rate": 1.8029461814168324e-05,
      "loss": 0.3566,
      "step": 7418
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0417326824496742,
      "learning_rate": 1.8028870560982234e-05,
      "loss": 0.6013,
      "step": 7419
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3786905648007343,
      "learning_rate": 1.8028279228805474e-05,
      "loss": 0.2024,
      "step": 7420
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.895858808321278,
      "learning_rate": 1.8027687817643856e-05,
      "loss": 0.375,
      "step": 7421
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.36721058556822284,
      "learning_rate": 1.802709632750321e-05,
      "loss": 0.2584,
      "step": 7422
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5069249669112483,
      "learning_rate": 1.8026504758389345e-05,
      "loss": 0.352,
      "step": 7423
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22859151318770018,
      "learning_rate": 1.8025913110308083e-05,
      "loss": 0.0988,
      "step": 7424
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4125063655792575,
      "learning_rate": 1.8025321383265254e-05,
      "loss": 0.2063,
      "step": 7425
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5972375210828726,
      "learning_rate": 1.8024729577266665e-05,
      "loss": 0.4037,
      "step": 7426
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7964679334438967,
      "learning_rate": 1.802413769231815e-05,
      "loss": 0.4296,
      "step": 7427
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40804660053087205,
      "learning_rate": 1.802354572842553e-05,
      "loss": 0.3293,
      "step": 7428
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5345516422895822,
      "learning_rate": 1.8022953685594622e-05,
      "loss": 0.1636,
      "step": 7429
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3853861603897169,
      "learning_rate": 1.8022361563831258e-05,
      "loss": 0.2801,
      "step": 7430
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7123716891574751,
      "learning_rate": 1.8021769363141262e-05,
      "loss": 0.3778,
      "step": 7431
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5425732324777843,
      "learning_rate": 1.8021177083530462e-05,
      "loss": 0.214,
      "step": 7432
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3255784558724589,
      "learning_rate": 1.8020584725004676e-05,
      "loss": 0.1824,
      "step": 7433
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.422263117107498,
      "learning_rate": 1.8019992287569744e-05,
      "loss": 0.3003,
      "step": 7434
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.31089271091318615,
      "learning_rate": 1.8019399771231487e-05,
      "loss": 0.2855,
      "step": 7435
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7509625386466335,
      "learning_rate": 1.8018807175995733e-05,
      "loss": 0.5321,
      "step": 7436
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0172672491440533,
      "learning_rate": 1.801821450186832e-05,
      "loss": 0.5805,
      "step": 7437
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5310258986373052,
      "learning_rate": 1.8017621748855075e-05,
      "loss": 0.1281,
      "step": 7438
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.403723015698685,
      "learning_rate": 1.8017028916961826e-05,
      "loss": 0.2966,
      "step": 7439
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38272692748949,
      "learning_rate": 1.8016436006194407e-05,
      "loss": 0.2421,
      "step": 7440
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4975038367723492,
      "learning_rate": 1.801584301655866e-05,
      "loss": 0.3741,
      "step": 7441
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38349505214578367,
      "learning_rate": 1.8015249948060404e-05,
      "loss": 0.2308,
      "step": 7442
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4643100894916054,
      "learning_rate": 1.8014656800705485e-05,
      "loss": 0.29,
      "step": 7443
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4457085441586125,
      "learning_rate": 1.8014063574499737e-05,
      "loss": 0.2853,
      "step": 7444
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48500329509484885,
      "learning_rate": 1.801347026944899e-05,
      "loss": 0.2864,
      "step": 7445
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3667169183958443,
      "learning_rate": 1.801287688555909e-05,
      "loss": 0.2873,
      "step": 7446
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6980120720796701,
      "learning_rate": 1.801228342283587e-05,
      "loss": 0.376,
      "step": 7447
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3705005191372932,
      "learning_rate": 1.801168988128517e-05,
      "loss": 0.2437,
      "step": 7448
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3501309723132635,
      "learning_rate": 1.8011096260912823e-05,
      "loss": 0.7575,
      "step": 7449
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2725424744683896,
      "learning_rate": 1.8010502561724683e-05,
      "loss": 0.1103,
      "step": 7450
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37891872725576514,
      "learning_rate": 1.8009908783726576e-05,
      "loss": 0.2968,
      "step": 7451
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9047612442989468,
      "learning_rate": 1.8009314926924355e-05,
      "loss": 0.4327,
      "step": 7452
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3253903631367732,
      "learning_rate": 1.8008720991323855e-05,
      "loss": 0.2594,
      "step": 7453
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39100425404816874,
      "learning_rate": 1.8008126976930927e-05,
      "loss": 0.2881,
      "step": 7454
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9010964958107779,
      "learning_rate": 1.800753288375141e-05,
      "loss": 0.5974,
      "step": 7455
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.417451686048712,
      "learning_rate": 1.8006938711791146e-05,
      "loss": 0.6495,
      "step": 7456
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.29177819247061576,
      "learning_rate": 1.800634446105599e-05,
      "loss": 0.1959,
      "step": 7457
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5299960488006681,
      "learning_rate": 1.800575013155178e-05,
      "loss": 0.4121,
      "step": 7458
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4098511410639224,
      "learning_rate": 1.8005155723284363e-05,
      "loss": 0.2597,
      "step": 7459
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6651454687539519,
      "learning_rate": 1.8004561236259594e-05,
      "loss": 0.4911,
      "step": 7460
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3862611980187123,
      "learning_rate": 1.8003966670483315e-05,
      "loss": 0.2311,
      "step": 7461
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4452566412986854,
      "learning_rate": 1.8003372025961378e-05,
      "loss": 0.3798,
      "step": 7462
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.33881213260772436,
      "learning_rate": 1.8002777302699638e-05,
      "loss": 0.1361,
      "step": 7463
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3518702473729437,
      "learning_rate": 1.8002182500703936e-05,
      "loss": 0.2768,
      "step": 7464
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49321812971490037,
      "learning_rate": 1.8001587619980133e-05,
      "loss": 0.3124,
      "step": 7465
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3492546836084863,
      "learning_rate": 1.8000992660534077e-05,
      "loss": 0.2184,
      "step": 7466
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.086448714345618,
      "learning_rate": 1.800039762237162e-05,
      "loss": 0.8731,
      "step": 7467
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5510737223808273,
      "learning_rate": 1.7999802505498626e-05,
      "loss": 0.3914,
      "step": 7468
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40949872454462743,
      "learning_rate": 1.7999207309920934e-05,
      "loss": 0.3843,
      "step": 7469
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2757979983035616,
      "learning_rate": 1.7998612035644414e-05,
      "loss": 0.2071,
      "step": 7470
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5652153445846331,
      "learning_rate": 1.7998016682674915e-05,
      "loss": 0.3314,
      "step": 7471
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8772383719478977,
      "learning_rate": 1.7997421251018298e-05,
      "loss": 0.3379,
      "step": 7472
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.412253938659921,
      "learning_rate": 1.7996825740680414e-05,
      "loss": 0.773,
      "step": 7473
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24298446046376632,
      "learning_rate": 1.7996230151667132e-05,
      "loss": 0.0726,
      "step": 7474
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0557545518248344,
      "learning_rate": 1.7995634483984306e-05,
      "loss": 0.4842,
      "step": 7475
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.31227877398842335,
      "learning_rate": 1.79950387376378e-05,
      "loss": 0.2598,
      "step": 7476
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46036839367027355,
      "learning_rate": 1.7994442912633467e-05,
      "loss": 0.3305,
      "step": 7477
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7364392979785224,
      "learning_rate": 1.7993847008977176e-05,
      "loss": 0.5217,
      "step": 7478
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.32299980082500296,
      "learning_rate": 1.7993251026674787e-05,
      "loss": 0.078,
      "step": 7479
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38966839398326947,
      "learning_rate": 1.7992654965732162e-05,
      "loss": 0.2973,
      "step": 7480
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4538827690548013,
      "learning_rate": 1.7992058826155173e-05,
      "loss": 0.2956,
      "step": 7481
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4678303985489787,
      "learning_rate": 1.7991462607949676e-05,
      "loss": 0.3624,
      "step": 7482
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.17525150527740635,
      "learning_rate": 1.7990866311121542e-05,
      "loss": 0.0723,
      "step": 7483
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45269833696911915,
      "learning_rate": 1.7990269935676634e-05,
      "loss": 0.3035,
      "step": 7484
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49398319111379985,
      "learning_rate": 1.798967348162082e-05,
      "loss": 0.2894,
      "step": 7485
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8100335494192706,
      "learning_rate": 1.7989076948959977e-05,
      "loss": 0.4802,
      "step": 7486
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.404899220045761,
      "learning_rate": 1.798848033769996e-05,
      "loss": 0.2524,
      "step": 7487
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43741571122332595,
      "learning_rate": 1.7987883647846646e-05,
      "loss": 0.3509,
      "step": 7488
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2988309213411934,
      "learning_rate": 1.79872868794059e-05,
      "loss": 0.2273,
      "step": 7489
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9241957475802943,
      "learning_rate": 1.7986690032383602e-05,
      "loss": 0.5229,
      "step": 7490
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.543612374936658,
      "learning_rate": 1.798609310678562e-05,
      "loss": 0.8781,
      "step": 7491
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23261576155454333,
      "learning_rate": 1.7985496102617826e-05,
      "loss": 0.0735,
      "step": 7492
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37092534308281333,
      "learning_rate": 1.7984899019886092e-05,
      "loss": 0.2824,
      "step": 7493
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5220945623638372,
      "learning_rate": 1.7984301858596295e-05,
      "loss": 0.3114,
      "step": 7494
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5589132204777455,
      "learning_rate": 1.798370461875431e-05,
      "loss": 0.4045,
      "step": 7495
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7406847742898331,
      "learning_rate": 1.798310730036601e-05,
      "loss": 0.3778,
      "step": 7496
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6658892214698202,
      "learning_rate": 1.7982509903437275e-05,
      "loss": 0.3297,
      "step": 7497
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4001127412601414,
      "learning_rate": 1.7981912427973985e-05,
      "loss": 0.2389,
      "step": 7498
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7761192088497124,
      "learning_rate": 1.7981314873982007e-05,
      "loss": 0.8983,
      "step": 7499
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.35783286161068484,
      "learning_rate": 1.7980717241467228e-05,
      "loss": 0.2929,
      "step": 7500
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30259768772418494,
      "learning_rate": 1.798011953043553e-05,
      "loss": 0.1772,
      "step": 7501
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.28414023216707934,
      "learning_rate": 1.797952174089279e-05,
      "loss": 0.0797,
      "step": 7502
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4225344271825197,
      "learning_rate": 1.797892387284489e-05,
      "loss": 0.3082,
      "step": 7503
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8509700682753378,
      "learning_rate": 1.797832592629771e-05,
      "loss": 0.524,
      "step": 7504
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30924984743519934,
      "learning_rate": 1.7977727901257133e-05,
      "loss": 0.2567,
      "step": 7505
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9644741630674591,
      "learning_rate": 1.7977129797729044e-05,
      "loss": 1.0473,
      "step": 7506
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2768779279502241,
      "learning_rate": 1.7976531615719333e-05,
      "loss": 0.2029,
      "step": 7507
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.217419312368172,
      "learning_rate": 1.7975933355233874e-05,
      "loss": 0.7401,
      "step": 7508
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6844118731515594,
      "learning_rate": 1.797533501627856e-05,
      "loss": 0.4568,
      "step": 7509
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9107118569314553,
      "learning_rate": 1.797473659885927e-05,
      "loss": 0.5891,
      "step": 7510
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3010479321046864,
      "learning_rate": 1.7974138102981905e-05,
      "loss": 0.2313,
      "step": 7511
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.32413819480307526,
      "learning_rate": 1.797353952865234e-05,
      "loss": 0.3027,
      "step": 7512
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.27630342380201367,
      "learning_rate": 1.7972940875876474e-05,
      "loss": 0.191,
      "step": 7513
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4710249117278429,
      "learning_rate": 1.7972342144660192e-05,
      "loss": 0.754,
      "step": 7514
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6083416117268501,
      "learning_rate": 1.797174333500938e-05,
      "loss": 0.1937,
      "step": 7515
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41271655452601147,
      "learning_rate": 1.797114444692994e-05,
      "loss": 0.3165,
      "step": 7516
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.058115986316846,
      "learning_rate": 1.7970545480427753e-05,
      "loss": 0.5986,
      "step": 7517
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30435803817149204,
      "learning_rate": 1.796994643550872e-05,
      "loss": 0.2527,
      "step": 7518
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.948955916933511,
      "learning_rate": 1.7969347312178732e-05,
      "loss": 0.4882,
      "step": 7519
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3034834611376469,
      "learning_rate": 1.796874811044368e-05,
      "loss": 0.2202,
      "step": 7520
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5136907438177121,
      "learning_rate": 1.7968148830309467e-05,
      "loss": 0.4116,
      "step": 7521
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2726768128673828,
      "learning_rate": 1.7967549471781977e-05,
      "loss": 0.1117,
      "step": 7522
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2914583608368482,
      "learning_rate": 1.796695003486712e-05,
      "loss": 0.2624,
      "step": 7523
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3311803908280811,
      "learning_rate": 1.7966350519570784e-05,
      "loss": 0.0742,
      "step": 7524
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4169315230177576,
      "learning_rate": 1.796575092589887e-05,
      "loss": 0.3101,
      "step": 7525
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.132276698370354,
      "learning_rate": 1.796515125385728e-05,
      "loss": 0.4819,
      "step": 7526
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2322540236930701,
      "learning_rate": 1.7964551503451905e-05,
      "loss": 0.702,
      "step": 7527
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4746321434002081,
      "learning_rate": 1.7963951674688657e-05,
      "loss": 0.3335,
      "step": 7528
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6788053606614727,
      "learning_rate": 1.796335176757343e-05,
      "loss": 0.3081,
      "step": 7529
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.30210687589598184,
      "learning_rate": 1.796275178211213e-05,
      "loss": 0.236,
      "step": 7530
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3350447281505577,
      "learning_rate": 1.7962151718310656e-05,
      "loss": 0.171,
      "step": 7531
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5084531852545662,
      "learning_rate": 1.7961551576174915e-05,
      "loss": 0.2655,
      "step": 7532
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9069052940518136,
      "learning_rate": 1.796095135571081e-05,
      "loss": 0.4089,
      "step": 7533
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43382744174505583,
      "learning_rate": 1.7960351056924244e-05,
      "loss": 0.3259,
      "step": 7534
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.35469036093220635,
      "learning_rate": 1.7959750679821125e-05,
      "loss": 0.2704,
      "step": 7535
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5136982190833034,
      "learning_rate": 1.795915022440736e-05,
      "loss": 0.4073,
      "step": 7536
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7875647817999214,
      "learning_rate": 1.7958549690688854e-05,
      "loss": 0.3713,
      "step": 7537
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7457539351394168,
      "learning_rate": 1.7957949078671523e-05,
      "loss": 0.4649,
      "step": 7538
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3080256148924592,
      "learning_rate": 1.795734838836127e-05,
      "loss": 0.227,
      "step": 7539
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5246178967751972,
      "learning_rate": 1.7956747619764e-05,
      "loss": 0.287,
      "step": 7540
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24933359648196,
      "learning_rate": 1.7956146772885633e-05,
      "loss": 0.1688,
      "step": 7541
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.501618397334483,
      "learning_rate": 1.7955545847732076e-05,
      "loss": 0.374,
      "step": 7542
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.292989784895839,
      "learning_rate": 1.795494484430924e-05,
      "loss": 0.2066,
      "step": 7543
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9709236900708333,
      "learning_rate": 1.7954343762623037e-05,
      "loss": 0.5968,
      "step": 7544
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1237063679240296,
      "learning_rate": 1.7953742602679387e-05,
      "loss": 0.504,
      "step": 7545
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48039294215324274,
      "learning_rate": 1.7953141364484198e-05,
      "loss": 0.3271,
      "step": 7546
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4050524653638162,
      "learning_rate": 1.7952540048043388e-05,
      "loss": 0.3141,
      "step": 7547
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40305761008894714,
      "learning_rate": 1.795193865336287e-05,
      "loss": 0.2341,
      "step": 7548
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.3295343522637375,
      "learning_rate": 1.7951337180448565e-05,
      "loss": 0.6969,
      "step": 7549
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.27722484399689074,
      "learning_rate": 1.7950735629306387e-05,
      "loss": 0.0968,
      "step": 7550
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2091035731453958,
      "learning_rate": 1.795013399994226e-05,
      "loss": 0.8107,
      "step": 7551
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.33523288607468993,
      "learning_rate": 1.7949532292362094e-05,
      "loss": 0.1758,
      "step": 7552
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5836503798465188,
      "learning_rate": 1.7948930506571814e-05,
      "loss": 0.3713,
      "step": 7553
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3677993105259925,
      "learning_rate": 1.7948328642577343e-05,
      "loss": 0.286,
      "step": 7554
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8409017718638773,
      "learning_rate": 1.7947726700384597e-05,
      "loss": 0.5849,
      "step": 7555
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.28848910414705453,
      "learning_rate": 1.7947124679999502e-05,
      "loss": 0.067,
      "step": 7556
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.337073493760141,
      "learning_rate": 1.7946522581427977e-05,
      "loss": 0.2171,
      "step": 7557
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0492491168263156,
      "learning_rate": 1.7945920404675952e-05,
      "loss": 0.6343,
      "step": 7558
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.20476036839561346,
      "learning_rate": 1.7945318149749345e-05,
      "loss": 0.1781,
      "step": 7559
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.9285124843622399,
      "learning_rate": 1.7944715816654086e-05,
      "loss": 0.9433,
      "step": 7560
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3001822322890924,
      "learning_rate": 1.7944113405396096e-05,
      "loss": 0.1934,
      "step": 7561
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45250322258829767,
      "learning_rate": 1.7943510915981303e-05,
      "loss": 0.3798,
      "step": 7562
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5448032904141483,
      "learning_rate": 1.7942908348415638e-05,
      "loss": 0.427,
      "step": 7563
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.8047146203971092,
      "learning_rate": 1.7942305702705028e-05,
      "loss": 1.0481,
      "step": 7564
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3233985340738077,
      "learning_rate": 1.79417029788554e-05,
      "loss": 0.232,
      "step": 7565
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3938831480170537,
      "learning_rate": 1.7941100176872687e-05,
      "loss": 0.2951,
      "step": 7566
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.045886060417602,
      "learning_rate": 1.7940497296762815e-05,
      "loss": 0.2951,
      "step": 7567
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4763641002030674,
      "learning_rate": 1.7939894338531718e-05,
      "loss": 0.6544,
      "step": 7568
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.22707337011709403,
      "learning_rate": 1.7939291302185328e-05,
      "loss": 0.1566,
      "step": 7569
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.34234476831694577,
      "learning_rate": 1.7938688187729578e-05,
      "loss": 0.2326,
      "step": 7570
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.33525963137208314,
      "learning_rate": 1.79380849951704e-05,
      "loss": 0.2929,
      "step": 7571
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5886597026208907,
      "learning_rate": 1.793748172451373e-05,
      "loss": 0.3722,
      "step": 7572
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5531042257565533,
      "learning_rate": 1.7936878375765507e-05,
      "loss": 0.9491,
      "step": 7573
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4922971375361269,
      "learning_rate": 1.793627494893166e-05,
      "loss": 0.192,
      "step": 7574
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41491823946438705,
      "learning_rate": 1.7935671444018133e-05,
      "loss": 0.3237,
      "step": 7575
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5768867475327607,
      "learning_rate": 1.793506786103085e-05,
      "loss": 0.0419,
      "step": 7576
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3866986043187647,
      "learning_rate": 1.793446419997577e-05,
      "loss": 0.3325,
      "step": 7577
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.27151582352723963,
      "learning_rate": 1.793386046085881e-05,
      "loss": 0.1577,
      "step": 7578
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4351424535936898,
      "learning_rate": 1.7933256643685927e-05,
      "loss": 0.2687,
      "step": 7579
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43463243297203596,
      "learning_rate": 1.793265274846305e-05,
      "loss": 0.2841,
      "step": 7580
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6896275541544867,
      "learning_rate": 1.7932048775196125e-05,
      "loss": 0.4919,
      "step": 7581
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.39145724426385015,
      "learning_rate": 1.7931444723891097e-05,
      "loss": 0.292,
      "step": 7582
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4041494064464306,
      "learning_rate": 1.793084059455391e-05,
      "loss": 0.2472,
      "step": 7583
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.46594416047321185,
      "learning_rate": 1.7930236387190495e-05,
      "loss": 0.3322,
      "step": 7584
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.48332727066278675,
      "learning_rate": 1.7929632101806807e-05,
      "loss": 0.2751,
      "step": 7585
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.41935917157991226,
      "learning_rate": 1.7929027738408793e-05,
      "loss": 0.2464,
      "step": 7586
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8061261698285903,
      "learning_rate": 1.7928423297002395e-05,
      "loss": 0.403,
      "step": 7587
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5785762637739028,
      "learning_rate": 1.7927818777593556e-05,
      "loss": 0.3844,
      "step": 7588
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.27392239227990656,
      "learning_rate": 1.792721418018823e-05,
      "loss": 0.2578,
      "step": 7589
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8793789579302188,
      "learning_rate": 1.792660950479236e-05,
      "loss": 0.5595,
      "step": 7590
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9064562975284768,
      "learning_rate": 1.7926004751411904e-05,
      "loss": 0.4904,
      "step": 7591
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.45773424610548685,
      "learning_rate": 1.7925399920052798e-05,
      "loss": 0.2868,
      "step": 7592
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3301691076506678,
      "learning_rate": 1.7924795010721002e-05,
      "loss": 0.2078,
      "step": 7593
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.365230661742902,
      "learning_rate": 1.7924190023422466e-05,
      "loss": 0.737,
      "step": 7594
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.37339468164239004,
      "learning_rate": 1.792358495816314e-05,
      "loss": 0.2531,
      "step": 7595
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7282562916497491,
      "learning_rate": 1.792297981494898e-05,
      "loss": 0.401,
      "step": 7596
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7483088722872114,
      "learning_rate": 1.7922374593785937e-05,
      "loss": 0.3886,
      "step": 7597
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2653385648844634,
      "learning_rate": 1.7921769294679964e-05,
      "loss": 0.1945,
      "step": 7598
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.660058542993765,
      "learning_rate": 1.792116391763702e-05,
      "loss": 0.9248,
      "step": 7599
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3088725665297211,
      "learning_rate": 1.7920558462663054e-05,
      "loss": 0.2119,
      "step": 7600
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.43110029551953843,
      "learning_rate": 1.7919952929764034e-05,
      "loss": 0.3367,
      "step": 7601
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.34419996708837364,
      "learning_rate": 1.791934731894591e-05,
      "loss": 0.219,
      "step": 7602
    },
    {
      "epoch": 0.23,
      "grad_norm": 2.027591604287134,
      "learning_rate": 1.791874163021464e-05,
      "loss": 0.9581,
      "step": 7603
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9331225436413212,
      "learning_rate": 1.7918135863576185e-05,
      "loss": 0.391,
      "step": 7604
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7061167527057922,
      "learning_rate": 1.79175300190365e-05,
      "loss": 0.4746,
      "step": 7605
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3196000637735577,
      "learning_rate": 1.7916924096601557e-05,
      "loss": 0.1952,
      "step": 7606
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4113704588046879,
      "learning_rate": 1.7916318096277304e-05,
      "loss": 0.3118,
      "step": 7607
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38890366764859713,
      "learning_rate": 1.791571201806971e-05,
      "loss": 0.2667,
      "step": 7608
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.18835937753513554,
      "learning_rate": 1.7915105861984737e-05,
      "loss": 0.1047,
      "step": 7609
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.4547420903155155,
      "learning_rate": 1.7914499628028347e-05,
      "loss": 0.7102,
      "step": 7610
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3599810601606856,
      "learning_rate": 1.791389331620651e-05,
      "loss": 0.177,
      "step": 7611
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5822043359295357,
      "learning_rate": 1.7913286926525182e-05,
      "loss": 0.4071,
      "step": 7612
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.36249938146201577,
      "learning_rate": 1.7912680458990333e-05,
      "loss": 0.314,
      "step": 7613
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8882807309913825,
      "learning_rate": 1.791207391360793e-05,
      "loss": 0.5544,
      "step": 7614
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5060304348576099,
      "learning_rate": 1.791146729038394e-05,
      "loss": 0.191,
      "step": 7615
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3715551655901115,
      "learning_rate": 1.7910860589324337e-05,
      "loss": 0.29,
      "step": 7616
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8458037134014456,
      "learning_rate": 1.7910253810435078e-05,
      "loss": 0.6115,
      "step": 7617
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5055834336895471,
      "learning_rate": 1.7909646953722146e-05,
      "loss": 0.3899,
      "step": 7618
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.23210803795507975,
      "learning_rate": 1.79090400191915e-05,
      "loss": 0.149,
      "step": 7619
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.413534279628904,
      "learning_rate": 1.7908433006849115e-05,
      "loss": 0.2846,
      "step": 7620
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.455587488096378,
      "learning_rate": 1.7907825916700967e-05,
      "loss": 0.3175,
      "step": 7621
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7642987729875766,
      "learning_rate": 1.7907218748753025e-05,
      "loss": 0.4069,
      "step": 7622
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8402502208446296,
      "learning_rate": 1.790661150301126e-05,
      "loss": 0.5528,
      "step": 7623
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3027984843555016,
      "learning_rate": 1.7906004179481652e-05,
      "loss": 0.1922,
      "step": 7624
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.32541046002485546,
      "learning_rate": 1.7905396778170174e-05,
      "loss": 0.2996,
      "step": 7625
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1098566049477603,
      "learning_rate": 1.79047892990828e-05,
      "loss": 0.5066,
      "step": 7626
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5717704770281857,
      "learning_rate": 1.7904181742225507e-05,
      "loss": 0.8336,
      "step": 7627
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.19026167495007495,
      "learning_rate": 1.7903574107604275e-05,
      "loss": 0.0712,
      "step": 7628
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4382860501516933,
      "learning_rate": 1.790296639522508e-05,
      "loss": 0.2714,
      "step": 7629
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.49008401381659383,
      "learning_rate": 1.79023586050939e-05,
      "loss": 0.3019,
      "step": 7630
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.34742157369763954,
      "learning_rate": 1.7901750737216717e-05,
      "loss": 0.3431,
      "step": 7631
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7332568770426389,
      "learning_rate": 1.790114279159951e-05,
      "loss": 0.4318,
      "step": 7632
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6542206459840227,
      "learning_rate": 1.790053476824826e-05,
      "loss": 0.3562,
      "step": 7633
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.38057550194014794,
      "learning_rate": 1.789992666716895e-05,
      "loss": 0.244,
      "step": 7634
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.1487197593081633,
      "learning_rate": 1.789931848836756e-05,
      "loss": 0.6817,
      "step": 7635
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3510061488401593,
      "learning_rate": 1.789871023185008e-05,
      "loss": 0.3096,
      "step": 7636
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2706953553653957,
      "learning_rate": 1.7898101897622485e-05,
      "loss": 0.0743,
      "step": 7637
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6797395842636526,
      "learning_rate": 1.7897493485690766e-05,
      "loss": 0.3038,
      "step": 7638
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3775164732894287,
      "learning_rate": 1.7896884996060908e-05,
      "loss": 0.3017,
      "step": 7639
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.6610455179241519,
      "learning_rate": 1.78962764287389e-05,
      "loss": 0.5038,
      "step": 7640
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7398875711322429,
      "learning_rate": 1.789566778373072e-05,
      "loss": 0.37,
      "step": 7641
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.350846491275607,
      "learning_rate": 1.7895059061042366e-05,
      "loss": 0.3018,
      "step": 7642
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3080548731158835,
      "learning_rate": 1.7894450260679827e-05,
      "loss": 0.2068,
      "step": 7643
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.5316250017039883,
      "learning_rate": 1.7893841382649082e-05,
      "loss": 0.831,
      "step": 7644
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.039140138438703,
      "learning_rate": 1.789323242695613e-05,
      "loss": 0.3943,
      "step": 7645
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4827414326623909,
      "learning_rate": 1.789262339360696e-05,
      "loss": 0.3306,
      "step": 7646
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.8579100219714099,
      "learning_rate": 1.7892014282607565e-05,
      "loss": 0.3077,
      "step": 7647
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2744368206098246,
      "learning_rate": 1.789140509396394e-05,
      "loss": 0.2486,
      "step": 7648
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.9580056811052934,
      "learning_rate": 1.7890795827682067e-05,
      "loss": 0.5923,
      "step": 7649
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.24474487424154848,
      "learning_rate": 1.7890186483767955e-05,
      "loss": 0.067,
      "step": 7650
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0656002669194176,
      "learning_rate": 1.788957706222759e-05,
      "loss": 0.479,
      "step": 7651
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.31537433756888456,
      "learning_rate": 1.788896756306697e-05,
      "loss": 0.23,
      "step": 7652
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.7761540454539018,
      "learning_rate": 1.788835798629209e-05,
      "loss": 0.8088,
      "step": 7653
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.34841344240165073,
      "learning_rate": 1.788774833190895e-05,
      "loss": 0.2715,
      "step": 7654
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.892454462938831,
      "learning_rate": 1.7887138599923547e-05,
      "loss": 0.6054,
      "step": 7655
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.35865833820316567,
      "learning_rate": 1.7886528790341878e-05,
      "loss": 0.2229,
      "step": 7656
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7001610877308245,
      "learning_rate": 1.7885918903169944e-05,
      "loss": 0.4044,
      "step": 7657
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.31838040274339596,
      "learning_rate": 1.7885308938413746e-05,
      "loss": 0.1746,
      "step": 7658
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.0502478353462354,
      "learning_rate": 1.7884698896079282e-05,
      "loss": 0.5548,
      "step": 7659
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.31729118801262374,
      "learning_rate": 1.7884088776172557e-05,
      "loss": 0.2392,
      "step": 7660
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3760752536834935,
      "learning_rate": 1.7883478578699573e-05,
      "loss": 0.1981,
      "step": 7661
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5573252028636971,
      "learning_rate": 1.7882868303666333e-05,
      "loss": 0.4393,
      "step": 7662
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.7434757146993091,
      "learning_rate": 1.7882257951078838e-05,
      "loss": 0.4388,
      "step": 7663
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.5766100209911614,
      "learning_rate": 1.7881647520943098e-05,
      "loss": 0.4897,
      "step": 7664
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2950013059630338,
      "learning_rate": 1.7881037013265117e-05,
      "loss": 0.0758,
      "step": 7665
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.40071250371784967,
      "learning_rate": 1.78804264280509e-05,
      "loss": 0.3043,
      "step": 7666
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.2243738743766273,
      "learning_rate": 1.7879815765306454e-05,
      "loss": 0.1911,
      "step": 7667
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2929650038121667,
      "learning_rate": 1.787920502503779e-05,
      "loss": 0.5684,
      "step": 7668
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.2315934604204737,
      "learning_rate": 1.787859420725091e-05,
      "loss": 0.4061,
      "step": 7669
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4570128340453459,
      "learning_rate": 1.787798331195183e-05,
      "loss": 0.2985,
      "step": 7670
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.4476250576787664,
      "learning_rate": 1.7877372339146557e-05,
      "loss": 0.3071,
      "step": 7671
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.3561718431793509,
      "learning_rate": 1.7876761288841106e-05,
      "loss": 0.3282,
      "step": 7672
    },
    {
      "epoch": 0.23,
      "grad_norm": 0.739060351177689,
      "learning_rate": 1.7876150161041482e-05,
      "loss": 0.3918,
      "step": 7673
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41972604074241393,
      "learning_rate": 1.7875538955753704e-05,
      "loss": 0.2225,
      "step": 7674
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44367169109114934,
      "learning_rate": 1.787492767298378e-05,
      "loss": 0.3321,
      "step": 7675
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.336485113262334,
      "learning_rate": 1.787431631273773e-05,
      "loss": 0.129,
      "step": 7676
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4297315805252307,
      "learning_rate": 1.787370487502156e-05,
      "loss": 0.2777,
      "step": 7677
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3190814036966383,
      "learning_rate": 1.7873093359841294e-05,
      "loss": 0.2582,
      "step": 7678
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4113128077097584,
      "learning_rate": 1.7872481767202945e-05,
      "loss": 0.293,
      "step": 7679
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4337718424245792,
      "learning_rate": 1.787187009711253e-05,
      "loss": 0.2573,
      "step": 7680
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9781843019637796,
      "learning_rate": 1.7871258349576065e-05,
      "loss": 0.6253,
      "step": 7681
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7691003328537223,
      "learning_rate": 1.7870646524599573e-05,
      "loss": 0.3994,
      "step": 7682
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6917829197109248,
      "learning_rate": 1.7870034622189068e-05,
      "loss": 0.3411,
      "step": 7683
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3537610536784701,
      "learning_rate": 1.7869422642350575e-05,
      "loss": 0.2357,
      "step": 7684
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3499902899484958,
      "learning_rate": 1.7868810585090114e-05,
      "loss": 0.3305,
      "step": 7685
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22065846103697986,
      "learning_rate": 1.78681984504137e-05,
      "loss": 0.1007,
      "step": 7686
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5499603849556594,
      "learning_rate": 1.7867586238327366e-05,
      "loss": 0.2871,
      "step": 7687
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.104574786675452,
      "learning_rate": 1.7866973948837125e-05,
      "loss": 0.5617,
      "step": 7688
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42973023331987464,
      "learning_rate": 1.786636158194901e-05,
      "loss": 0.2855,
      "step": 7689
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3848883909469563,
      "learning_rate": 1.786574913766904e-05,
      "loss": 0.3417,
      "step": 7690
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9303794630714178,
      "learning_rate": 1.786513661600324e-05,
      "loss": 0.4434,
      "step": 7691
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5514938426882128,
      "learning_rate": 1.7864524016957637e-05,
      "loss": 0.6821,
      "step": 7692
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3451323864618234,
      "learning_rate": 1.7863911340538263e-05,
      "loss": 0.2054,
      "step": 7693
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5056469484181884,
      "learning_rate": 1.786329858675114e-05,
      "loss": 0.2474,
      "step": 7694
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.27192403795592524,
      "learning_rate": 1.7862685755602297e-05,
      "loss": 0.1961,
      "step": 7695
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5107631019454263,
      "learning_rate": 1.7862072847097765e-05,
      "loss": 0.3562,
      "step": 7696
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.355990367340125,
      "learning_rate": 1.7861459861243574e-05,
      "loss": 0.1984,
      "step": 7697
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5337525838981916,
      "learning_rate": 1.7860846798045758e-05,
      "loss": 0.4313,
      "step": 7698
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7761142780698899,
      "learning_rate": 1.7860233657510337e-05,
      "loss": 0.4217,
      "step": 7699
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6286289452799831,
      "learning_rate": 1.7859620439643356e-05,
      "loss": 0.9444,
      "step": 7700
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.35259160727331007,
      "learning_rate": 1.785900714445084e-05,
      "loss": 0.2348,
      "step": 7701
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3045252315457975,
      "learning_rate": 1.7858393771938827e-05,
      "loss": 0.2433,
      "step": 7702
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3761408928800705,
      "learning_rate": 1.785778032211335e-05,
      "loss": 0.8298,
      "step": 7703
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2925884391400186,
      "learning_rate": 1.7857166794980448e-05,
      "loss": 0.0977,
      "step": 7704
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1506709246329765,
      "learning_rate": 1.7856553190546153e-05,
      "loss": 0.6938,
      "step": 7705
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3275322277448384,
      "learning_rate": 1.78559395088165e-05,
      "loss": 0.2388,
      "step": 7706
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48288899962828197,
      "learning_rate": 1.785532574979753e-05,
      "loss": 0.3537,
      "step": 7707
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3317706953666825,
      "learning_rate": 1.7854711913495276e-05,
      "loss": 0.2642,
      "step": 7708
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9353090600820928,
      "learning_rate": 1.7854097999915788e-05,
      "loss": 0.6229,
      "step": 7709
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48799026040569177,
      "learning_rate": 1.7853484009065096e-05,
      "loss": 0.1727,
      "step": 7710
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0335717437598373,
      "learning_rate": 1.7852869940949245e-05,
      "loss": 0.5147,
      "step": 7711
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3652766559819465,
      "learning_rate": 1.7852255795574275e-05,
      "loss": 0.2338,
      "step": 7712
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.27341719646687124,
      "learning_rate": 1.785164157294623e-05,
      "loss": 0.2069,
      "step": 7713
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48293517163696315,
      "learning_rate": 1.7851027273071153e-05,
      "loss": 0.326,
      "step": 7714
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.32320333091819065,
      "learning_rate": 1.7850412895955084e-05,
      "loss": 0.077,
      "step": 7715
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42174287860602483,
      "learning_rate": 1.784979844160407e-05,
      "loss": 0.3536,
      "step": 7716
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9431141717433882,
      "learning_rate": 1.7849183910024157e-05,
      "loss": 0.4234,
      "step": 7717
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0667694337064286,
      "learning_rate": 1.784856930122139e-05,
      "loss": 0.6362,
      "step": 7718
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3215343518340723,
      "learning_rate": 1.7847954615201815e-05,
      "loss": 0.2493,
      "step": 7719
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42822848636601285,
      "learning_rate": 1.7847339851971482e-05,
      "loss": 0.2744,
      "step": 7720
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.950810079964002,
      "learning_rate": 1.7846725011536438e-05,
      "loss": 0.2672,
      "step": 7721
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.379897173382812,
      "learning_rate": 1.7846110093902726e-05,
      "loss": 0.2085,
      "step": 7722
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0171672704678298,
      "learning_rate": 1.7845495099076407e-05,
      "loss": 0.386,
      "step": 7723
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.762142508809841,
      "learning_rate": 1.7844880027063524e-05,
      "loss": 0.4364,
      "step": 7724
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.32899649020767563,
      "learning_rate": 1.7844264877870126e-05,
      "loss": 0.2172,
      "step": 7725
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36531053862506574,
      "learning_rate": 1.7843649651502276e-05,
      "loss": 0.3117,
      "step": 7726
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1211533512232346,
      "learning_rate": 1.7843034347966015e-05,
      "loss": 0.5328,
      "step": 7727
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.636353765818781,
      "learning_rate": 1.7842418967267404e-05,
      "loss": 0.2345,
      "step": 7728
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3646854981327333,
      "learning_rate": 1.7841803509412493e-05,
      "loss": 0.2771,
      "step": 7729
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9940150633925722,
      "learning_rate": 1.784118797440734e-05,
      "loss": 0.0657,
      "step": 7730
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4664657045337739,
      "learning_rate": 1.7840572362257994e-05,
      "loss": 0.3525,
      "step": 7731
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48729137158679364,
      "learning_rate": 1.7839956672970524e-05,
      "loss": 0.2726,
      "step": 7732
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7090699833457346,
      "learning_rate": 1.7839340906550978e-05,
      "loss": 0.354,
      "step": 7733
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3461931493028392,
      "learning_rate": 1.783872506300542e-05,
      "loss": 0.2129,
      "step": 7734
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9766571397627982,
      "learning_rate": 1.7838109142339898e-05,
      "loss": 0.6568,
      "step": 7735
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2651265695429993,
      "learning_rate": 1.7837493144560485e-05,
      "loss": 0.1324,
      "step": 7736
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3691069628739938,
      "learning_rate": 1.7836877069673236e-05,
      "loss": 0.3327,
      "step": 7737
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2975037689107071,
      "learning_rate": 1.783626091768421e-05,
      "loss": 0.0734,
      "step": 7738
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4080879800423615,
      "learning_rate": 1.7835644688599466e-05,
      "loss": 0.3364,
      "step": 7739
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8130887280121629,
      "learning_rate": 1.783502838242508e-05,
      "loss": 0.398,
      "step": 7740
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.35528752736235264,
      "learning_rate": 1.7834411999167102e-05,
      "loss": 0.2691,
      "step": 7741
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.074644789566046,
      "learning_rate": 1.7833795538831598e-05,
      "loss": 0.5294,
      "step": 7742
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.32718644374863404,
      "learning_rate": 1.783317900142464e-05,
      "loss": 0.2247,
      "step": 7743
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4303116333108268,
      "learning_rate": 1.7832562386952293e-05,
      "loss": 0.343,
      "step": 7744
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24585152016654838,
      "learning_rate": 1.7831945695420613e-05,
      "loss": 0.1255,
      "step": 7745
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9927849103296311,
      "learning_rate": 1.783132892683568e-05,
      "loss": 0.9337,
      "step": 7746
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.32223803588301186,
      "learning_rate": 1.7830712081203553e-05,
      "loss": 0.1842,
      "step": 7747
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5334148132568625,
      "learning_rate": 1.7830095158530304e-05,
      "loss": 0.4113,
      "step": 7748
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3243733587392236,
      "learning_rate": 1.7829478158822004e-05,
      "loss": 0.2929,
      "step": 7749
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9666497441459736,
      "learning_rate": 1.7828861082084717e-05,
      "loss": 0.5742,
      "step": 7750
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6235818799323996,
      "learning_rate": 1.7828243928324523e-05,
      "loss": 0.2788,
      "step": 7751
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3550737545409415,
      "learning_rate": 1.7827626697547488e-05,
      "loss": 0.2866,
      "step": 7752
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.873012799971897,
      "learning_rate": 1.7827009389759685e-05,
      "loss": 0.6578,
      "step": 7753
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3124113518354228,
      "learning_rate": 1.782639200496719e-05,
      "loss": 0.0743,
      "step": 7754
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39966196901753626,
      "learning_rate": 1.7825774543176076e-05,
      "loss": 0.347,
      "step": 7755
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3339842870781399,
      "learning_rate": 1.7825157004392414e-05,
      "loss": 0.1835,
      "step": 7756
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5454994605730561,
      "learning_rate": 1.7824539388622282e-05,
      "loss": 0.4233,
      "step": 7757
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6827395449860999,
      "learning_rate": 1.782392169587176e-05,
      "loss": 0.404,
      "step": 7758
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7051147310713094,
      "learning_rate": 1.7823303926146923e-05,
      "loss": 0.5002,
      "step": 7759
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39949433426061615,
      "learning_rate": 1.7822686079453845e-05,
      "loss": 0.124,
      "step": 7760
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.40516829953964406,
      "learning_rate": 1.7822068155798608e-05,
      "loss": 0.3209,
      "step": 7761
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36760991186978836,
      "learning_rate": 1.7821450155187293e-05,
      "loss": 0.2885,
      "step": 7762
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5382739309569616,
      "learning_rate": 1.7820832077625977e-05,
      "loss": 0.2963,
      "step": 7763
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3621023245731064,
      "learning_rate": 1.7820213923120736e-05,
      "loss": 0.1548,
      "step": 7764
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.208987588349441,
      "learning_rate": 1.7819595691677663e-05,
      "loss": 0.353,
      "step": 7765
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3990013024091082,
      "learning_rate": 1.7818977383302833e-05,
      "loss": 0.2832,
      "step": 7766
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3327770615195697,
      "learning_rate": 1.781835899800233e-05,
      "loss": 0.2949,
      "step": 7767
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8255960263814224,
      "learning_rate": 1.781774053578224e-05,
      "loss": 0.5205,
      "step": 7768
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0473295938883018,
      "learning_rate": 1.7817121996648646e-05,
      "loss": 0.4132,
      "step": 7769
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.45919634143166405,
      "learning_rate": 1.7816503380607634e-05,
      "loss": 0.3092,
      "step": 7770
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.418406691160763,
      "learning_rate": 1.7815884687665287e-05,
      "loss": 0.2789,
      "step": 7771
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.33807861010032614,
      "learning_rate": 1.7815265917827696e-05,
      "loss": 0.2514,
      "step": 7772
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.27266217849626156,
      "learning_rate": 1.7814647071100948e-05,
      "loss": 0.17,
      "step": 7773
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6395294883389183,
      "learning_rate": 1.7814028147491132e-05,
      "loss": 0.3482,
      "step": 7774
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.398737751995873,
      "learning_rate": 1.7813409147004336e-05,
      "loss": 0.2932,
      "step": 7775
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.812434475139184,
      "learning_rate": 1.7812790069646646e-05,
      "loss": 0.4931,
      "step": 7776
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1351945708299849,
      "learning_rate": 1.781217091542416e-05,
      "loss": 0.0935,
      "step": 7777
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5136726943238443,
      "learning_rate": 1.7811551684342963e-05,
      "loss": 0.3869,
      "step": 7778
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3141820335946851,
      "learning_rate": 1.7810932376409153e-05,
      "loss": 0.196,
      "step": 7779
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.440541119964686,
      "learning_rate": 1.781031299162882e-05,
      "loss": 0.3195,
      "step": 7780
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8292113277162645,
      "learning_rate": 1.780969353000806e-05,
      "loss": 0.9045,
      "step": 7781
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.26754394156228406,
      "learning_rate": 1.780907399155296e-05,
      "loss": 0.1573,
      "step": 7782
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5908626886560329,
      "learning_rate": 1.7808454376269624e-05,
      "loss": 0.3796,
      "step": 7783
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3385305490913639,
      "learning_rate": 1.7807834684164143e-05,
      "loss": 0.2606,
      "step": 7784
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5088844793641093,
      "learning_rate": 1.7807214915242616e-05,
      "loss": 0.3901,
      "step": 7785
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46087295279308904,
      "learning_rate": 1.7806595069511142e-05,
      "loss": 0.2978,
      "step": 7786
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2030494597025652,
      "learning_rate": 1.7805975146975812e-05,
      "loss": 0.6913,
      "step": 7787
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36218872675134706,
      "learning_rate": 1.7805355147642736e-05,
      "loss": 0.1795,
      "step": 7788
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5785013764035385,
      "learning_rate": 1.7804735071518002e-05,
      "loss": 0.3868,
      "step": 7789
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2809494636752412,
      "learning_rate": 1.780411491860772e-05,
      "loss": 0.1359,
      "step": 7790
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.34305131173048514,
      "learning_rate": 1.7803494688917985e-05,
      "loss": 0.3332,
      "step": 7791
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38016331389427155,
      "learning_rate": 1.7802874382454902e-05,
      "loss": 0.2118,
      "step": 7792
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3398586532669434,
      "learning_rate": 1.7802253999224577e-05,
      "loss": 0.2659,
      "step": 7793
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4901560322976382,
      "learning_rate": 1.7801633539233107e-05,
      "loss": 0.3159,
      "step": 7794
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3571294232862854,
      "learning_rate": 1.7801013002486597e-05,
      "loss": 0.4187,
      "step": 7795
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3570114367482675,
      "learning_rate": 1.7800392388991158e-05,
      "loss": 0.3124,
      "step": 7796
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.34750975716105414,
      "learning_rate": 1.779977169875289e-05,
      "loss": 0.1844,
      "step": 7797
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5217798850728339,
      "learning_rate": 1.7799150931777903e-05,
      "loss": 0.3617,
      "step": 7798
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.724223316844156,
      "learning_rate": 1.7798530088072298e-05,
      "loss": 0.4053,
      "step": 7799
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9384416521052018,
      "learning_rate": 1.7797909167642192e-05,
      "loss": 0.5748,
      "step": 7800
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5463191290293087,
      "learning_rate": 1.7797288170493692e-05,
      "loss": 0.3036,
      "step": 7801
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36249571714783174,
      "learning_rate": 1.77966670966329e-05,
      "loss": 0.2665,
      "step": 7802
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.21731027935289549,
      "learning_rate": 1.779604594606594e-05,
      "loss": 0.1704,
      "step": 7803
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.9377480518127865,
      "learning_rate": 1.7795424718798906e-05,
      "loss": 0.7213,
      "step": 7804
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.879339545271794,
      "learning_rate": 1.7794803414837922e-05,
      "loss": 0.4559,
      "step": 7805
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.35697427692904177,
      "learning_rate": 1.7794182034189096e-05,
      "loss": 0.2112,
      "step": 7806
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4374062337561542,
      "learning_rate": 1.7793560576858544e-05,
      "loss": 0.7906,
      "step": 7807
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0208607062510555,
      "learning_rate": 1.779293904285238e-05,
      "loss": 0.4085,
      "step": 7808
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.34503734694312066,
      "learning_rate": 1.7792317432176714e-05,
      "loss": 0.3514,
      "step": 7809
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42454000426602484,
      "learning_rate": 1.7791695744837667e-05,
      "loss": 0.2575,
      "step": 7810
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43688052093969637,
      "learning_rate": 1.7791073980841353e-05,
      "loss": 0.3526,
      "step": 7811
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.19063651112055913,
      "learning_rate": 1.779045214019389e-05,
      "loss": 0.07,
      "step": 7812
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.6189384279941708,
      "learning_rate": 1.7789830222901395e-05,
      "loss": 0.7243,
      "step": 7813
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3257808420696845,
      "learning_rate": 1.778920822896999e-05,
      "loss": 0.2593,
      "step": 7814
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4669286691139953,
      "learning_rate": 1.7788586158405786e-05,
      "loss": 0.2821,
      "step": 7815
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.47946050654350303,
      "learning_rate": 1.778796401121491e-05,
      "loss": 0.2667,
      "step": 7816
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9543707278321083,
      "learning_rate": 1.7787341787403485e-05,
      "loss": 0.5923,
      "step": 7817
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7199019483080242,
      "learning_rate": 1.7786719486977628e-05,
      "loss": 0.3992,
      "step": 7818
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5566354629914029,
      "learning_rate": 1.778609710994346e-05,
      "loss": 0.1799,
      "step": 7819
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5117792470790257,
      "learning_rate": 1.7785474656307108e-05,
      "loss": 0.3561,
      "step": 7820
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24371143264538275,
      "learning_rate": 1.7784852126074693e-05,
      "loss": 0.2224,
      "step": 7821
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.49407415385977466,
      "learning_rate": 1.7784229519252346e-05,
      "loss": 0.311,
      "step": 7822
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9770188889432093,
      "learning_rate": 1.7783606835846183e-05,
      "loss": 0.5134,
      "step": 7823
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8773599905212429,
      "learning_rate": 1.7782984075862336e-05,
      "loss": 0.3955,
      "step": 7824
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3530877807384853,
      "learning_rate": 1.778236123930693e-05,
      "loss": 0.2884,
      "step": 7825
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5173635147921599,
      "learning_rate": 1.7781738326186096e-05,
      "loss": 0.4123,
      "step": 7826
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43015905870012305,
      "learning_rate": 1.778111533650596e-05,
      "loss": 0.3093,
      "step": 7827
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.2693707909648224,
      "learning_rate": 1.778049227027265e-05,
      "loss": 0.9057,
      "step": 7828
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2823109367809505,
      "learning_rate": 1.7779869127492297e-05,
      "loss": 0.2012,
      "step": 7829
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4632681122908306,
      "learning_rate": 1.7779245908171027e-05,
      "loss": 0.2241,
      "step": 7830
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.22107823140248717,
      "learning_rate": 1.777862261231498e-05,
      "loss": 0.1123,
      "step": 7831
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.30803421739780223,
      "learning_rate": 1.7777999239930284e-05,
      "loss": 0.2824,
      "step": 7832
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7187065173075978,
      "learning_rate": 1.777737579102307e-05,
      "loss": 0.3518,
      "step": 7833
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.33327995521395726,
      "learning_rate": 1.777675226559948e-05,
      "loss": 0.2717,
      "step": 7834
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9153200875004078,
      "learning_rate": 1.777612866366564e-05,
      "loss": 0.5799,
      "step": 7835
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.537485603781387,
      "learning_rate": 1.7775504985227684e-05,
      "loss": 0.405,
      "step": 7836
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0320423729661514,
      "learning_rate": 1.7774881230291754e-05,
      "loss": 0.4556,
      "step": 7837
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.25891711218146823,
      "learning_rate": 1.7774257398863982e-05,
      "loss": 0.2221,
      "step": 7838
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5887322990365143,
      "learning_rate": 1.777363349095051e-05,
      "loss": 0.4199,
      "step": 7839
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.219095852678232,
      "learning_rate": 1.7773009506557475e-05,
      "loss": 0.1001,
      "step": 7840
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3869403212715918,
      "learning_rate": 1.7772385445691012e-05,
      "loss": 0.951,
      "step": 7841
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36729763160305134,
      "learning_rate": 1.7771761308357264e-05,
      "loss": 0.2389,
      "step": 7842
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5995019057229138,
      "learning_rate": 1.7771137094562373e-05,
      "loss": 0.3517,
      "step": 7843
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4822067232606013,
      "learning_rate": 1.7770512804312473e-05,
      "loss": 0.3193,
      "step": 7844
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42189332202762964,
      "learning_rate": 1.776988843761372e-05,
      "loss": 0.2801,
      "step": 7845
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.856108848669235,
      "learning_rate": 1.776926399447224e-05,
      "loss": 0.4406,
      "step": 7846
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.38487737093638286,
      "learning_rate": 1.7768639474894187e-05,
      "loss": 0.2065,
      "step": 7847
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5691215403107467,
      "learning_rate": 1.7768014878885704e-05,
      "loss": 0.38,
      "step": 7848
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2108170468096789,
      "learning_rate": 1.7767390206452932e-05,
      "loss": 0.1066,
      "step": 7849
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.37281505410193444,
      "learning_rate": 1.7766765457602023e-05,
      "loss": 0.3045,
      "step": 7850
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6404755742894682,
      "learning_rate": 1.7766140632339116e-05,
      "loss": 0.3221,
      "step": 7851
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3828129891082265,
      "learning_rate": 1.7765515730670365e-05,
      "loss": 0.3214,
      "step": 7852
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5949982625925143,
      "learning_rate": 1.7764890752601912e-05,
      "loss": 0.3978,
      "step": 7853
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.3760040762489854,
      "learning_rate": 1.776426569813991e-05,
      "loss": 0.774,
      "step": 7854
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4229924544279558,
      "learning_rate": 1.776364056729051e-05,
      "loss": 0.2446,
      "step": 7855
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4586908605829354,
      "learning_rate": 1.7763015360059855e-05,
      "loss": 0.2862,
      "step": 7856
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.34828483133301424,
      "learning_rate": 1.7762390076454102e-05,
      "loss": 0.242,
      "step": 7857
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.17816792945223642,
      "learning_rate": 1.7761764716479405e-05,
      "loss": 0.0723,
      "step": 7858
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0333007055711085,
      "learning_rate": 1.776113928014191e-05,
      "loss": 0.6713,
      "step": 7859
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5288526593271009,
      "learning_rate": 1.776051376744777e-05,
      "loss": 0.2749,
      "step": 7860
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3762243336239337,
      "learning_rate": 1.7759888178403146e-05,
      "loss": 0.3324,
      "step": 7861
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42739532935451435,
      "learning_rate": 1.775926251301419e-05,
      "loss": 0.2615,
      "step": 7862
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5001788263999295,
      "learning_rate": 1.775863677128705e-05,
      "loss": 0.3678,
      "step": 7863
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9966225879230817,
      "learning_rate": 1.7758010953227896e-05,
      "loss": 0.3451,
      "step": 7864
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4569041234444045,
      "learning_rate": 1.7757385058842873e-05,
      "loss": 0.3345,
      "step": 7865
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.45813766444915394,
      "learning_rate": 1.7756759088138142e-05,
      "loss": 0.2569,
      "step": 7866
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4441868319493745,
      "learning_rate": 1.7756133041119866e-05,
      "loss": 0.281,
      "step": 7867
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.407047384255247,
      "learning_rate": 1.77555069177942e-05,
      "loss": 0.302,
      "step": 7868
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6016751256804059,
      "learning_rate": 1.7754880718167307e-05,
      "loss": 0.3851,
      "step": 7869
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3371160934375201,
      "learning_rate": 1.7754254442245346e-05,
      "loss": 0.2436,
      "step": 7870
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7780815395187289,
      "learning_rate": 1.7753628090034475e-05,
      "loss": 0.464,
      "step": 7871
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.44583881235831685,
      "learning_rate": 1.7753001661540857e-05,
      "loss": 0.2411,
      "step": 7872
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4307962000736147,
      "learning_rate": 1.775237515677066e-05,
      "loss": 0.2482,
      "step": 7873
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6592562637137845,
      "learning_rate": 1.775174857573005e-05,
      "loss": 0.4097,
      "step": 7874
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.30305281419144203,
      "learning_rate": 1.7751121918425185e-05,
      "loss": 0.2607,
      "step": 7875
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8956860081534941,
      "learning_rate": 1.7750495184862232e-05,
      "loss": 0.5645,
      "step": 7876
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8676637254259691,
      "learning_rate": 1.774986837504735e-05,
      "loss": 0.4019,
      "step": 7877
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.507477469915349,
      "learning_rate": 1.774924148898672e-05,
      "loss": 0.8698,
      "step": 7878
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.28553028655351514,
      "learning_rate": 1.7748614526686503e-05,
      "loss": 0.2021,
      "step": 7879
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3651619277061621,
      "learning_rate": 1.7747987488152865e-05,
      "loss": 0.3213,
      "step": 7880
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2280405315172813,
      "learning_rate": 1.7747360373391974e-05,
      "loss": 0.0867,
      "step": 7881
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4225480747061532,
      "learning_rate": 1.7746733182410006e-05,
      "loss": 0.5973,
      "step": 7882
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.32853937124600524,
      "learning_rate": 1.7746105915213126e-05,
      "loss": 0.2079,
      "step": 7883
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5866448465365479,
      "learning_rate": 1.7745478571807512e-05,
      "loss": 0.4084,
      "step": 7884
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.765779223394691,
      "learning_rate": 1.774485115219932e-05,
      "loss": 0.3909,
      "step": 7885
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4172353135893598,
      "learning_rate": 1.7744223656394747e-05,
      "loss": 0.3149,
      "step": 7886
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7020338210263903,
      "learning_rate": 1.7743596084399945e-05,
      "loss": 0.4434,
      "step": 7887
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3162575079430621,
      "learning_rate": 1.7742968436221102e-05,
      "loss": 0.2323,
      "step": 7888
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.124781076472008,
      "learning_rate": 1.7742340711864386e-05,
      "loss": 0.9122,
      "step": 7889
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2450837342342176,
      "learning_rate": 1.774171291133597e-05,
      "loss": 0.0911,
      "step": 7890
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48120029549861476,
      "learning_rate": 1.774108503464204e-05,
      "loss": 0.3794,
      "step": 7891
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.27206904915112673,
      "learning_rate": 1.7740457081788766e-05,
      "loss": 0.2201,
      "step": 7892
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6819342868155309,
      "learning_rate": 1.773982905278233e-05,
      "loss": 0.3938,
      "step": 7893
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8248634115023927,
      "learning_rate": 1.773920094762891e-05,
      "loss": 0.4016,
      "step": 7894
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8240448681021538,
      "learning_rate": 1.7738572766334682e-05,
      "loss": 0.4868,
      "step": 7895
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.47021491708838337,
      "learning_rate": 1.773794450890583e-05,
      "loss": 0.2918,
      "step": 7896
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0939568393029793,
      "learning_rate": 1.7737316175348533e-05,
      "loss": 0.5631,
      "step": 7897
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.29265280895136453,
      "learning_rate": 1.7736687765668975e-05,
      "loss": 0.249,
      "step": 7898
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2510510387128907,
      "learning_rate": 1.7736059279873336e-05,
      "loss": 0.0867,
      "step": 7899
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9563356897114728,
      "learning_rate": 1.7735430717967798e-05,
      "loss": 0.5536,
      "step": 7900
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.24962958163135107,
      "learning_rate": 1.7734802079958552e-05,
      "loss": 0.0744,
      "step": 7901
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.39159746446437266,
      "learning_rate": 1.7734173365851775e-05,
      "loss": 0.3425,
      "step": 7902
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.47639886633289796,
      "learning_rate": 1.7733544575653653e-05,
      "loss": 0.3593,
      "step": 7903
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4649325129400992,
      "learning_rate": 1.773291570937038e-05,
      "loss": 0.3827,
      "step": 7904
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1206091030164436,
      "learning_rate": 1.7732286767008135e-05,
      "loss": 0.5257,
      "step": 7905
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42143464936021163,
      "learning_rate": 1.773165774857311e-05,
      "loss": 0.2899,
      "step": 7906
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4303511844011698,
      "learning_rate": 1.773102865407149e-05,
      "loss": 0.2778,
      "step": 7907
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3496333776416877,
      "learning_rate": 1.773039948350947e-05,
      "loss": 0.1882,
      "step": 7908
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42480971934228123,
      "learning_rate": 1.772977023689323e-05,
      "loss": 0.2596,
      "step": 7909
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41988021920753726,
      "learning_rate": 1.7729140914228974e-05,
      "loss": 0.3306,
      "step": 7910
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3162281489318738,
      "learning_rate": 1.7728511515522885e-05,
      "loss": 0.2544,
      "step": 7911
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6554782600587483,
      "learning_rate": 1.772788204078115e-05,
      "loss": 0.3932,
      "step": 7912
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7804592703841231,
      "learning_rate": 1.7727252490009977e-05,
      "loss": 0.5271,
      "step": 7913
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5181108851844618,
      "learning_rate": 1.7726622863215546e-05,
      "loss": 0.286,
      "step": 7914
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3744692677734104,
      "learning_rate": 1.772599316040406e-05,
      "loss": 0.3015,
      "step": 7915
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.48527329761335986,
      "learning_rate": 1.772536338158171e-05,
      "loss": 0.2837,
      "step": 7916
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4949753668244649,
      "learning_rate": 1.7724733526754693e-05,
      "loss": 0.3847,
      "step": 7917
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.272653806332698,
      "learning_rate": 1.7724103595929203e-05,
      "loss": 0.1693,
      "step": 7918
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4945014083563522,
      "learning_rate": 1.7723473589111445e-05,
      "loss": 0.3516,
      "step": 7919
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.33984801391524666,
      "learning_rate": 1.772284350630761e-05,
      "loss": 0.2282,
      "step": 7920
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7252131235471113,
      "learning_rate": 1.77222133475239e-05,
      "loss": 0.5163,
      "step": 7921
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3264286401597622,
      "learning_rate": 1.7721583112766513e-05,
      "loss": 0.2669,
      "step": 7922
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.218254808537742,
      "learning_rate": 1.7720952802041654e-05,
      "loss": 0.717,
      "step": 7923
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.43661345285636577,
      "learning_rate": 1.772032241535552e-05,
      "loss": 0.0754,
      "step": 7924
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3944560635129934,
      "learning_rate": 1.7719691952714312e-05,
      "loss": 0.2586,
      "step": 7925
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.45316570741878665,
      "learning_rate": 1.771906141412423e-05,
      "loss": 0.2087,
      "step": 7926
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.31428848575873725,
      "learning_rate": 1.771843079959149e-05,
      "loss": 0.287,
      "step": 7927
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9578601095542091,
      "learning_rate": 1.7717800109122284e-05,
      "loss": 0.525,
      "step": 7928
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3148909321352345,
      "learning_rate": 1.7717169342722824e-05,
      "loss": 0.197,
      "step": 7929
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3903045222568506,
      "learning_rate": 1.771653850039931e-05,
      "loss": 0.293,
      "step": 7930
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9594591299183927,
      "learning_rate": 1.771590758215795e-05,
      "loss": 0.4039,
      "step": 7931
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.1168380019686515,
      "learning_rate": 1.7715276588004953e-05,
      "loss": 0.8224,
      "step": 7932
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3812898144974382,
      "learning_rate": 1.771464551794653e-05,
      "loss": 0.1917,
      "step": 7933
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.28097639852245193,
      "learning_rate": 1.7714014371988883e-05,
      "loss": 0.2741,
      "step": 7934
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3188594031046075,
      "learning_rate": 1.7713383150138222e-05,
      "loss": 0.1422,
      "step": 7935
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9288199633088736,
      "learning_rate": 1.7712751852400765e-05,
      "loss": 0.5768,
      "step": 7936
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4652355562632208,
      "learning_rate": 1.7712120478782715e-05,
      "loss": 0.3399,
      "step": 7937
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.23221834078428155,
      "learning_rate": 1.7711489029290285e-05,
      "loss": 0.1712,
      "step": 7938
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5092740560120216,
      "learning_rate": 1.771085750392969e-05,
      "loss": 0.3896,
      "step": 7939
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4143633329988027,
      "learning_rate": 1.7710225902707143e-05,
      "loss": 0.2598,
      "step": 7940
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.5658955909213905,
      "learning_rate": 1.7709594225628853e-05,
      "loss": 0.7237,
      "step": 7941
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.350010869620559,
      "learning_rate": 1.7708962472701036e-05,
      "loss": 0.2022,
      "step": 7942
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.592795252719187,
      "learning_rate": 1.7708330643929917e-05,
      "loss": 0.3599,
      "step": 7943
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.215177119991427,
      "learning_rate": 1.7707698739321698e-05,
      "loss": 0.3336,
      "step": 7944
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3804684282870898,
      "learning_rate": 1.7707066758882607e-05,
      "loss": 0.3285,
      "step": 7945
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4641561919698084,
      "learning_rate": 1.7706434702618858e-05,
      "loss": 0.2783,
      "step": 7946
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.41533676763202126,
      "learning_rate": 1.7705802570536665e-05,
      "loss": 0.3442,
      "step": 7947
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2219849619286681,
      "learning_rate": 1.770517036264225e-05,
      "loss": 0.0712,
      "step": 7948
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.1201768216142116,
      "learning_rate": 1.7704538078941838e-05,
      "loss": 0.6111,
      "step": 7949
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4226124250289989,
      "learning_rate": 1.7703905719441642e-05,
      "loss": 0.3075,
      "step": 7950
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.29380392494890895,
      "learning_rate": 1.7703273284147888e-05,
      "loss": 0.0724,
      "step": 7951
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3323754451455694,
      "learning_rate": 1.7702640773066796e-05,
      "loss": 0.3286,
      "step": 7952
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8410755047385975,
      "learning_rate": 1.770200818620459e-05,
      "loss": 0.424,
      "step": 7953
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7748535987337104,
      "learning_rate": 1.7701375523567495e-05,
      "loss": 0.5255,
      "step": 7954
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2523822391662739,
      "learning_rate": 1.770074278516173e-05,
      "loss": 0.1157,
      "step": 7955
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.42363459236306067,
      "learning_rate": 1.7700109970993527e-05,
      "loss": 0.2843,
      "step": 7956
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.26121429899264614,
      "learning_rate": 1.7699477081069106e-05,
      "loss": 0.2,
      "step": 7957
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5367885922044201,
      "learning_rate": 1.7698844115394698e-05,
      "loss": 0.3984,
      "step": 7958
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0322859815491754,
      "learning_rate": 1.7698211073976527e-05,
      "loss": 0.5116,
      "step": 7959
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9379773658778736,
      "learning_rate": 1.7697577956820823e-05,
      "loss": 0.5028,
      "step": 7960
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.31158239944451827,
      "learning_rate": 1.769694476393381e-05,
      "loss": 0.2169,
      "step": 7961
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.828339888898166,
      "learning_rate": 1.769631149532173e-05,
      "loss": 0.5358,
      "step": 7962
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.33939277815125146,
      "learning_rate": 1.76956781509908e-05,
      "loss": 0.2655,
      "step": 7963
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6031401454856555,
      "learning_rate": 1.7695044730947258e-05,
      "loss": 0.2836,
      "step": 7964
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3518634206254818,
      "learning_rate": 1.769441123519733e-05,
      "loss": 0.283,
      "step": 7965
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2752589535918921,
      "learning_rate": 1.7693777663747256e-05,
      "loss": 0.0988,
      "step": 7966
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.29724884301312,
      "learning_rate": 1.7693144016603264e-05,
      "loss": 0.6788,
      "step": 7967
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.46908252807992507,
      "learning_rate": 1.7692510293771593e-05,
      "loss": 0.3384,
      "step": 7968
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3962650940633841,
      "learning_rate": 1.769187649525847e-05,
      "loss": 0.3393,
      "step": 7969
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3588745893969064,
      "learning_rate": 1.7691242621070136e-05,
      "loss": 0.2491,
      "step": 7970
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7748717257884884,
      "learning_rate": 1.769060867121283e-05,
      "loss": 0.5195,
      "step": 7971
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5790113412480384,
      "learning_rate": 1.7689974645692786e-05,
      "loss": 0.4034,
      "step": 7972
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.4190369460038934,
      "learning_rate": 1.7689340544516234e-05,
      "loss": 0.6534,
      "step": 7973
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.5091634644453686,
      "learning_rate": 1.7688706367689425e-05,
      "loss": 0.0728,
      "step": 7974
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.33557153695055164,
      "learning_rate": 1.768807211521859e-05,
      "loss": 0.2663,
      "step": 7975
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.2795281041954112,
      "learning_rate": 1.7687437787109975e-05,
      "loss": 0.2224,
      "step": 7976
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.8451818611374041,
      "learning_rate": 1.7686803383369815e-05,
      "loss": 0.4934,
      "step": 7977
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0831399777443793,
      "learning_rate": 1.7686168904004355e-05,
      "loss": 0.4508,
      "step": 7978
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36014025394956883,
      "learning_rate": 1.7685534349019836e-05,
      "loss": 0.2579,
      "step": 7979
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.7455939003385911,
      "learning_rate": 1.76848997184225e-05,
      "loss": 0.5366,
      "step": 7980
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.35051232225883605,
      "learning_rate": 1.7684265012218596e-05,
      "loss": 0.2685,
      "step": 7981
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8587829030282124,
      "learning_rate": 1.768363023041436e-05,
      "loss": 0.8309,
      "step": 7982
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3942528928140743,
      "learning_rate": 1.7682995373016044e-05,
      "loss": 0.1992,
      "step": 7983
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36469964518184755,
      "learning_rate": 1.7682360440029892e-05,
      "loss": 0.2506,
      "step": 7984
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3183513814705782,
      "learning_rate": 1.768172543146215e-05,
      "loss": 0.1566,
      "step": 7985
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.8499630838563814,
      "learning_rate": 1.7681090347319062e-05,
      "loss": 0.8028,
      "step": 7986
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3537966398047537,
      "learning_rate": 1.7680455187606883e-05,
      "loss": 0.2647,
      "step": 7987
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.3883200231599898,
      "learning_rate": 1.7679819952331857e-05,
      "loss": 0.2913,
      "step": 7988
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9441609816631633,
      "learning_rate": 1.7679184641500238e-05,
      "loss": 0.4256,
      "step": 7989
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.2038850383718676,
      "learning_rate": 1.767854925511827e-05,
      "loss": 0.2941,
      "step": 7990
    },
    {
      "epoch": 0.24,
      "grad_norm": 2.260600532555801,
      "learning_rate": 1.767791379319221e-05,
      "loss": 0.8459,
      "step": 7991
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.33191311719464656,
      "learning_rate": 1.767727825572831e-05,
      "loss": 0.1721,
      "step": 7992
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.31839835927944593,
      "learning_rate": 1.7676642642732816e-05,
      "loss": 0.2959,
      "step": 7993
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.34412463113421715,
      "learning_rate": 1.7676006954211988e-05,
      "loss": 0.0823,
      "step": 7994
    },
    {
      "epoch": 0.24,
      "grad_norm": 1.0726007357853784,
      "learning_rate": 1.7675371190172077e-05,
      "loss": 0.6184,
      "step": 7995
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.6362426222757341,
      "learning_rate": 1.767473535061934e-05,
      "loss": 0.289,
      "step": 7996
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.4416249240950768,
      "learning_rate": 1.767409943556003e-05,
      "loss": 0.3584,
      "step": 7997
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.9695852731765638,
      "learning_rate": 1.7673463445000405e-05,
      "loss": 0.4953,
      "step": 7998
    },
    {
      "epoch": 0.24,
      "grad_norm": 0.36594158122956916,
      "learning_rate": 1.7672827378946723e-05,
      "loss": 0.3045,
      "step": 7999
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.147292179532662,
      "learning_rate": 1.7672191237405238e-05,
      "loss": 0.4021,
      "step": 8000
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6851542367751866,
      "learning_rate": 1.767155502038221e-05,
      "loss": 0.3875,
      "step": 8001
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3577854587039427,
      "learning_rate": 1.7670918727883902e-05,
      "loss": 0.237,
      "step": 8002
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2451872626732952,
      "learning_rate": 1.7670282359916574e-05,
      "loss": 0.1532,
      "step": 8003
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.37383995919751595,
      "learning_rate": 1.7669645916486482e-05,
      "loss": 0.3428,
      "step": 8004
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8753152762862098,
      "learning_rate": 1.766900939759989e-05,
      "loss": 0.4,
      "step": 8005
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41831317056540007,
      "learning_rate": 1.7668372803263057e-05,
      "loss": 0.2893,
      "step": 8006
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5131553903768329,
      "learning_rate": 1.7667736133482254e-05,
      "loss": 0.3123,
      "step": 8007
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3455615864251398,
      "learning_rate": 1.7667099388263738e-05,
      "loss": 0.7124,
      "step": 8008
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1066840732298384,
      "learning_rate": 1.7666462567613773e-05,
      "loss": 0.339,
      "step": 8009
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7309713997623777,
      "learning_rate": 1.7665825671538628e-05,
      "loss": 0.3896,
      "step": 8010
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26781285871658583,
      "learning_rate": 1.766518870004457e-05,
      "loss": 0.2561,
      "step": 8011
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43688224816839494,
      "learning_rate": 1.766455165313786e-05,
      "loss": 0.212,
      "step": 8012
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.71929231136543,
      "learning_rate": 1.7663914530824774e-05,
      "loss": 0.3851,
      "step": 8013
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6944417559791437,
      "learning_rate": 1.7663277333111568e-05,
      "loss": 0.5209,
      "step": 8014
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3235003470343044,
      "learning_rate": 1.7662640060004524e-05,
      "loss": 0.2123,
      "step": 8015
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.29446622262839456,
      "learning_rate": 1.7662002711509902e-05,
      "loss": 0.2016,
      "step": 8016
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.495703929968842,
      "learning_rate": 1.766136528763398e-05,
      "loss": 0.3803,
      "step": 8017
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8631367371458124,
      "learning_rate": 1.7660727788383023e-05,
      "loss": 0.3583,
      "step": 8018
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6192688303912733,
      "learning_rate": 1.7660090213763303e-05,
      "loss": 0.3781,
      "step": 8019
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3606016266541925,
      "learning_rate": 1.76594525637811e-05,
      "loss": 0.2528,
      "step": 8020
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0739318514645635,
      "learning_rate": 1.7658814838442677e-05,
      "loss": 0.5567,
      "step": 8021
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.33222482907591727,
      "learning_rate": 1.7658177037754315e-05,
      "loss": 0.2945,
      "step": 8022
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.781394845286113,
      "learning_rate": 1.765753916172229e-05,
      "loss": 0.501,
      "step": 8023
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3189768459677551,
      "learning_rate": 1.7656901210352873e-05,
      "loss": 0.2138,
      "step": 8024
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6973505866296386,
      "learning_rate": 1.7656263183652343e-05,
      "loss": 0.8942,
      "step": 8025
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.19753862613129167,
      "learning_rate": 1.765562508162698e-05,
      "loss": 0.0918,
      "step": 8026
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4711336570762514,
      "learning_rate": 1.7654986904283055e-05,
      "loss": 0.8189,
      "step": 8027
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3396461942178484,
      "learning_rate": 1.765434865162685e-05,
      "loss": 0.2009,
      "step": 8028
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2960167989550478,
      "learning_rate": 1.7653710323664643e-05,
      "loss": 0.2378,
      "step": 8029
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9126286818291116,
      "learning_rate": 1.7653071920402717e-05,
      "loss": 0.4761,
      "step": 8030
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1675560740319193,
      "learning_rate": 1.765243344184735e-05,
      "loss": 0.3947,
      "step": 8031
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4764602367788258,
      "learning_rate": 1.765179488800483e-05,
      "loss": 1.0446,
      "step": 8032
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3640676650262728,
      "learning_rate": 1.765115625888143e-05,
      "loss": 0.1868,
      "step": 8033
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40552704641090515,
      "learning_rate": 1.7650517554483437e-05,
      "loss": 0.2867,
      "step": 8034
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26361079465594567,
      "learning_rate": 1.7649878774817132e-05,
      "loss": 0.2237,
      "step": 8035
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0500161609763978,
      "learning_rate": 1.7649239919888806e-05,
      "loss": 0.6113,
      "step": 8036
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.37264455166714444,
      "learning_rate": 1.7648600989704743e-05,
      "loss": 0.0712,
      "step": 8037
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3575338679099074,
      "learning_rate": 1.7647961984271225e-05,
      "loss": 0.2857,
      "step": 8038
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0251297505726245,
      "learning_rate": 1.7647322903594538e-05,
      "loss": 0.448,
      "step": 8039
    },
    {
      "epoch": 0.25,
      "grad_norm": 7.1646562535099605,
      "learning_rate": 1.7646683747680975e-05,
      "loss": 0.8434,
      "step": 8040
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3081201284794167,
      "learning_rate": 1.764604451653682e-05,
      "loss": 0.2791,
      "step": 8041
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2940348869133615,
      "learning_rate": 1.7645405210168362e-05,
      "loss": 0.2056,
      "step": 8042
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.436646691421849,
      "learning_rate": 1.764476582858189e-05,
      "loss": 0.7518,
      "step": 8043
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3571234050968418,
      "learning_rate": 1.7644126371783703e-05,
      "loss": 0.0766,
      "step": 8044
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.158428869094983,
      "learning_rate": 1.764348683978008e-05,
      "loss": 0.658,
      "step": 8045
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2884230120431923,
      "learning_rate": 1.764284723257732e-05,
      "loss": 0.2283,
      "step": 8046
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3864735057302668,
      "learning_rate": 1.7642207550181717e-05,
      "loss": 0.3386,
      "step": 8047
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7152204897972578,
      "learning_rate": 1.7641567792599558e-05,
      "loss": 0.4142,
      "step": 8048
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7402766827592488,
      "learning_rate": 1.764092795983714e-05,
      "loss": 0.5484,
      "step": 8049
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9330657523695047,
      "learning_rate": 1.764028805190076e-05,
      "loss": 0.6557,
      "step": 8050
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4079208198961184,
      "learning_rate": 1.763964806879671e-05,
      "loss": 0.2762,
      "step": 8051
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4732050973297604,
      "learning_rate": 1.7639008010531293e-05,
      "loss": 0.2341,
      "step": 8052
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2771190736106658,
      "learning_rate": 1.76383678771108e-05,
      "loss": 0.2407,
      "step": 8053
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4375828159633248,
      "learning_rate": 1.763772766854153e-05,
      "loss": 0.2121,
      "step": 8054
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6792890052867022,
      "learning_rate": 1.7637087384829777e-05,
      "loss": 0.2296,
      "step": 8055
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.46280972981306345,
      "learning_rate": 1.7636447025981853e-05,
      "loss": 0.3544,
      "step": 8056
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7134320796932792,
      "learning_rate": 1.7635806592004047e-05,
      "loss": 0.377,
      "step": 8057
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.36455571780560775,
      "learning_rate": 1.7635166082902662e-05,
      "loss": 0.3101,
      "step": 8058
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9188763507142721,
      "learning_rate": 1.7634525498684004e-05,
      "loss": 0.5912,
      "step": 8059
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9436679300237959,
      "learning_rate": 1.763388483935437e-05,
      "loss": 0.4017,
      "step": 8060
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.378238655149799,
      "learning_rate": 1.7633244104920067e-05,
      "loss": 0.2603,
      "step": 8061
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.32313444631443144,
      "learning_rate": 1.7632603295387395e-05,
      "loss": 0.1756,
      "step": 8062
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.207560383514924,
      "learning_rate": 1.763196241076266e-05,
      "loss": 0.3999,
      "step": 8063
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41170310540706917,
      "learning_rate": 1.763132145105217e-05,
      "loss": 0.3345,
      "step": 8064
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28716203177476796,
      "learning_rate": 1.763068041626223e-05,
      "loss": 0.1984,
      "step": 8065
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7113183742872906,
      "learning_rate": 1.763003930639914e-05,
      "loss": 0.5381,
      "step": 8066
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8078059218036322,
      "learning_rate": 1.7629398121469216e-05,
      "loss": 0.052,
      "step": 8067
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8971178049683474,
      "learning_rate": 1.7628756861478758e-05,
      "loss": 0.6523,
      "step": 8068
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6382155339215502,
      "learning_rate": 1.7628115526434086e-05,
      "loss": 0.3644,
      "step": 8069
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2797544256288475,
      "learning_rate": 1.7627474116341502e-05,
      "loss": 0.2468,
      "step": 8070
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4285913424346095,
      "learning_rate": 1.7626832631207315e-05,
      "loss": 0.6701,
      "step": 8071
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5516902181703767,
      "learning_rate": 1.762619107103784e-05,
      "loss": 0.2546,
      "step": 8072
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.715106703020615,
      "learning_rate": 1.762554943583939e-05,
      "loss": 0.4139,
      "step": 8073
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.280000114583518,
      "learning_rate": 1.7624907725618276e-05,
      "loss": 0.1778,
      "step": 8074
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9237650874862307,
      "learning_rate": 1.7624265940380807e-05,
      "loss": 0.9039,
      "step": 8075
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3327335657994812,
      "learning_rate": 1.7623624080133302e-05,
      "loss": 0.2656,
      "step": 8076
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5072009894084566,
      "learning_rate": 1.7622982144882077e-05,
      "loss": 0.9975,
      "step": 8077
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.35300205098652065,
      "learning_rate": 1.7622340134633446e-05,
      "loss": 0.2181,
      "step": 8078
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5525468954323415,
      "learning_rate": 1.7621698049393722e-05,
      "loss": 0.3992,
      "step": 8079
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7311491714285645,
      "learning_rate": 1.7621055889169225e-05,
      "loss": 0.405,
      "step": 8080
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6536983322754419,
      "learning_rate": 1.762041365396627e-05,
      "loss": 0.5016,
      "step": 8081
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23647112338697387,
      "learning_rate": 1.7619771343791186e-05,
      "loss": 0.193,
      "step": 8082
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39876098139685634,
      "learning_rate": 1.761912895865028e-05,
      "loss": 0.2118,
      "step": 8083
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.36849084958850953,
      "learning_rate": 1.7618486498549873e-05,
      "loss": 0.2637,
      "step": 8084
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9766997869665802,
      "learning_rate": 1.761784396349629e-05,
      "loss": 0.4681,
      "step": 8085
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5503600851735182,
      "learning_rate": 1.7617201353495855e-05,
      "loss": 0.946,
      "step": 8086
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28069692712536665,
      "learning_rate": 1.7616558668554887e-05,
      "loss": 0.0706,
      "step": 8087
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.32247681586719495,
      "learning_rate": 1.7615915908679707e-05,
      "loss": 0.2973,
      "step": 8088
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3934372135951736,
      "learning_rate": 1.7615273073876638e-05,
      "loss": 0.2902,
      "step": 8089
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6993270141193145,
      "learning_rate": 1.7614630164152008e-05,
      "loss": 0.4847,
      "step": 8090
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3462674854180128,
      "learning_rate": 1.7613987179512145e-05,
      "loss": 0.1747,
      "step": 8091
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41028376443063214,
      "learning_rate": 1.7613344119963365e-05,
      "loss": 0.29,
      "step": 8092
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3285136324356252,
      "learning_rate": 1.7612700985512004e-05,
      "loss": 0.1948,
      "step": 8093
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38972603472249484,
      "learning_rate": 1.7612057776164383e-05,
      "loss": 0.3155,
      "step": 8094
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3679202387813645,
      "learning_rate": 1.7611414491926836e-05,
      "loss": 0.6087,
      "step": 8095
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3552821639277278,
      "learning_rate": 1.7610771132805687e-05,
      "loss": 0.1311,
      "step": 8096
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40623295615147587,
      "learning_rate": 1.761012769880727e-05,
      "loss": 0.3137,
      "step": 8097
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9218685056901115,
      "learning_rate": 1.760948418993791e-05,
      "loss": 0.4264,
      "step": 8098
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7794631576976523,
      "learning_rate": 1.760884060620394e-05,
      "loss": 0.576,
      "step": 8099
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.24044777261291037,
      "learning_rate": 1.7608196947611694e-05,
      "loss": 0.2085,
      "step": 8100
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4482850212433984,
      "learning_rate": 1.7607553214167502e-05,
      "loss": 0.2889,
      "step": 8101
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.25530723981611997,
      "learning_rate": 1.7606909405877697e-05,
      "loss": 0.1413,
      "step": 8102
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0137033936694246,
      "learning_rate": 1.7606265522748615e-05,
      "loss": 0.6491,
      "step": 8103
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.995919461976135,
      "learning_rate": 1.7605621564786594e-05,
      "loss": 0.5153,
      "step": 8104
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39798412186263404,
      "learning_rate": 1.760497753199796e-05,
      "loss": 0.3154,
      "step": 8105
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3789306476478145,
      "learning_rate": 1.7604333424389057e-05,
      "loss": 0.2206,
      "step": 8106
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.521940089059492,
      "learning_rate": 1.760368924196622e-05,
      "loss": 0.4046,
      "step": 8107
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9892781139301852,
      "learning_rate": 1.7603044984735786e-05,
      "loss": 0.4361,
      "step": 8108
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2708657852529877,
      "learning_rate": 1.7602400652704093e-05,
      "loss": 0.1102,
      "step": 8109
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9656505162463083,
      "learning_rate": 1.760175624587748e-05,
      "loss": 0.4432,
      "step": 8110
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2857556757597914,
      "learning_rate": 1.760111176426229e-05,
      "loss": 0.1932,
      "step": 8111
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3643268594497452,
      "learning_rate": 1.7600467207864862e-05,
      "loss": 0.3169,
      "step": 8112
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0749325057851162,
      "learning_rate": 1.7599822576691537e-05,
      "loss": 0.605,
      "step": 8113
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0319734437718375,
      "learning_rate": 1.7599177870748653e-05,
      "loss": 0.525,
      "step": 8114
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26579212920205025,
      "learning_rate": 1.7598533090042558e-05,
      "loss": 0.1958,
      "step": 8115
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6756167098356388,
      "learning_rate": 1.7597888234579595e-05,
      "loss": 0.4847,
      "step": 8116
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.45011883387876284,
      "learning_rate": 1.759724330436611e-05,
      "loss": 0.2647,
      "step": 8117
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5117162101559655,
      "learning_rate": 1.7596598299408442e-05,
      "loss": 0.3634,
      "step": 8118
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.32513786154304924,
      "learning_rate": 1.7595953219712944e-05,
      "loss": 0.1883,
      "step": 8119
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3069134410257085,
      "learning_rate": 1.7595308065285955e-05,
      "loss": 0.2046,
      "step": 8120
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2972168364470568,
      "learning_rate": 1.7594662836133824e-05,
      "loss": 0.1381,
      "step": 8121
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.808599474926739,
      "learning_rate": 1.7594017532262905e-05,
      "loss": 0.4303,
      "step": 8122
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4239429699198706,
      "learning_rate": 1.7593372153679542e-05,
      "loss": 0.3177,
      "step": 8123
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28194241010165666,
      "learning_rate": 1.7592726700390087e-05,
      "loss": 0.2285,
      "step": 8124
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7473060225220476,
      "learning_rate": 1.7592081172400887e-05,
      "loss": 0.5531,
      "step": 8125
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0673585872482536,
      "learning_rate": 1.7591435569718292e-05,
      "loss": 0.049,
      "step": 8126
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8609422191319052,
      "learning_rate": 1.759078989234866e-05,
      "loss": 0.5516,
      "step": 8127
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2582489680755007,
      "learning_rate": 1.7590144140298336e-05,
      "loss": 0.1451,
      "step": 8128
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5797085342587427,
      "learning_rate": 1.758949831357368e-05,
      "loss": 0.3836,
      "step": 8129
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26724780641386264,
      "learning_rate": 1.7588852412181037e-05,
      "loss": 0.2127,
      "step": 8130
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0012479443839084,
      "learning_rate": 1.758820643612677e-05,
      "loss": 0.6648,
      "step": 8131
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6829669260116805,
      "learning_rate": 1.758756038541723e-05,
      "loss": 0.3381,
      "step": 8132
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0142833462671772,
      "learning_rate": 1.7586914260058777e-05,
      "loss": 0.4969,
      "step": 8133
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3442300307599112,
      "learning_rate": 1.7586268060057764e-05,
      "loss": 0.2587,
      "step": 8134
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3283909852610653,
      "learning_rate": 1.758562178542055e-05,
      "loss": 0.2659,
      "step": 8135
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3457789120539798,
      "learning_rate": 1.758497543615349e-05,
      "loss": 0.6469,
      "step": 8136
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3279861875345224,
      "learning_rate": 1.758432901226295e-05,
      "loss": 0.0763,
      "step": 8137
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4030054421860116,
      "learning_rate": 1.758368251375528e-05,
      "loss": 0.3237,
      "step": 8138
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28316887623718967,
      "learning_rate": 1.758303594063685e-05,
      "loss": 0.1586,
      "step": 8139
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8882158219199581,
      "learning_rate": 1.7582389292914018e-05,
      "loss": 0.523,
      "step": 8140
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.30806135120741046,
      "learning_rate": 1.7581742570593146e-05,
      "loss": 0.2486,
      "step": 8141
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41602829638408256,
      "learning_rate": 1.7581095773680595e-05,
      "loss": 0.2822,
      "step": 8142
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28018309469047453,
      "learning_rate": 1.7580448902182724e-05,
      "loss": 0.2118,
      "step": 8143
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.590563157337964,
      "learning_rate": 1.757980195610591e-05,
      "loss": 0.616,
      "step": 8144
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6523085777129665,
      "learning_rate": 1.7579154935456506e-05,
      "loss": 0.0391,
      "step": 8145
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4764368657952784,
      "learning_rate": 1.7578507840240882e-05,
      "loss": 0.272,
      "step": 8146
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2746802046804477,
      "learning_rate": 1.7577860670465406e-05,
      "loss": 0.2377,
      "step": 8147
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3950880927335472,
      "learning_rate": 1.7577213426136442e-05,
      "loss": 0.2865,
      "step": 8148
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8337161925707623,
      "learning_rate": 1.757656610726036e-05,
      "loss": 0.5343,
      "step": 8149
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6681888328796434,
      "learning_rate": 1.7575918713843526e-05,
      "loss": 0.3451,
      "step": 8150
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38157238666051135,
      "learning_rate": 1.757527124589231e-05,
      "loss": 0.2859,
      "step": 8151
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40361255696868426,
      "learning_rate": 1.7574623703413086e-05,
      "loss": 0.1094,
      "step": 8152
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.48284042583117404,
      "learning_rate": 1.757397608641222e-05,
      "loss": 0.3682,
      "step": 8153
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.41671483362205425,
      "learning_rate": 1.7573328394896084e-05,
      "loss": 0.261,
      "step": 8154
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7296468540583041,
      "learning_rate": 1.7572680628871054e-05,
      "loss": 0.3572,
      "step": 8155
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38648414715507134,
      "learning_rate": 1.75720327883435e-05,
      "loss": 0.2483,
      "step": 8156
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0820840327588468,
      "learning_rate": 1.7571384873319794e-05,
      "loss": 0.6315,
      "step": 8157
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9367037402598782,
      "learning_rate": 1.7570736883806313e-05,
      "loss": 0.4268,
      "step": 8158
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.35299618167281843,
      "learning_rate": 1.757008881980943e-05,
      "loss": 0.3123,
      "step": 8159
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3217407996472597,
      "learning_rate": 1.7569440681335526e-05,
      "loss": 0.081,
      "step": 8160
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.24250269242135827,
      "learning_rate": 1.7568792468390973e-05,
      "loss": 0.1838,
      "step": 8161
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44969159565826994,
      "learning_rate": 1.7568144180982148e-05,
      "loss": 0.3136,
      "step": 8162
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9044604434823147,
      "learning_rate": 1.7567495819115426e-05,
      "loss": 0.3524,
      "step": 8163
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.473181197790938,
      "learning_rate": 1.75668473827972e-05,
      "loss": 0.8877,
      "step": 8164
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3295210652645846,
      "learning_rate": 1.7566198872033832e-05,
      "loss": 0.2114,
      "step": 8165
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.37605448119487545,
      "learning_rate": 1.7565550286831712e-05,
      "loss": 0.3691,
      "step": 8166
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7870773929822023,
      "learning_rate": 1.7564901627197216e-05,
      "loss": 0.4527,
      "step": 8167
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.8866705322336697,
      "learning_rate": 1.7564252893136733e-05,
      "loss": 0.8721,
      "step": 8168
    },
    {
      "epoch": 0.25,
      "grad_norm": 4.014371861297298,
      "learning_rate": 1.756360408465664e-05,
      "loss": 0.3806,
      "step": 8169
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4356457410609039,
      "learning_rate": 1.756295520176332e-05,
      "loss": 0.2778,
      "step": 8170
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2629327588183358,
      "learning_rate": 1.756230624446316e-05,
      "loss": 0.2163,
      "step": 8171
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.94064698518061,
      "learning_rate": 1.756165721276254e-05,
      "loss": 0.9336,
      "step": 8172
    },
    {
      "epoch": 0.25,
      "grad_norm": 3.5297973580103843,
      "learning_rate": 1.756100810666785e-05,
      "loss": 0.2686,
      "step": 8173
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.44816831382109135,
      "learning_rate": 1.7560358926185476e-05,
      "loss": 0.2606,
      "step": 8174
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7044013594765519,
      "learning_rate": 1.7559709671321802e-05,
      "loss": 0.4974,
      "step": 8175
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9342244486040331,
      "learning_rate": 1.7559060342083217e-05,
      "loss": 0.4216,
      "step": 8176
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.476362311234635,
      "learning_rate": 1.7558410938476105e-05,
      "loss": 0.3855,
      "step": 8177
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8788252662837642,
      "learning_rate": 1.7557761460506863e-05,
      "loss": 0.2404,
      "step": 8178
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5948704313989107,
      "learning_rate": 1.7557111908181877e-05,
      "loss": 0.3991,
      "step": 8179
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5487519742639154,
      "learning_rate": 1.7556462281507537e-05,
      "loss": 0.1189,
      "step": 8180
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4768769010258924,
      "learning_rate": 1.7555812580490234e-05,
      "loss": 0.7594,
      "step": 8181
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5410723743501694,
      "learning_rate": 1.7555162805136364e-05,
      "loss": 0.2952,
      "step": 8182
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.531214252361991,
      "learning_rate": 1.7554512955452313e-05,
      "loss": 0.3348,
      "step": 8183
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43953120013088387,
      "learning_rate": 1.755386303144448e-05,
      "loss": 0.2662,
      "step": 8184
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8266319082709745,
      "learning_rate": 1.7553213033119255e-05,
      "loss": 0.5676,
      "step": 8185
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0832748037627724,
      "learning_rate": 1.7552562960483034e-05,
      "loss": 0.6609,
      "step": 8186
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6879516969383679,
      "learning_rate": 1.7551912813542213e-05,
      "loss": 0.0948,
      "step": 8187
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6312249521725731,
      "learning_rate": 1.755126259230319e-05,
      "loss": 0.3364,
      "step": 8188
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3643514960935123,
      "learning_rate": 1.7550612296772366e-05,
      "loss": 0.1975,
      "step": 8189
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1680725461864967,
      "learning_rate": 1.7549961926956126e-05,
      "loss": 0.6934,
      "step": 8190
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7446657596233385,
      "learning_rate": 1.754931148286088e-05,
      "loss": 0.3802,
      "step": 8191
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6222202687105336,
      "learning_rate": 1.7548660964493023e-05,
      "loss": 0.2621,
      "step": 8192
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4329735379993621,
      "learning_rate": 1.7548010371858956e-05,
      "loss": 0.3162,
      "step": 8193
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6948431346740843,
      "learning_rate": 1.754735970496508e-05,
      "loss": 0.3935,
      "step": 8194
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5050757224292607,
      "learning_rate": 1.7546708963817792e-05,
      "loss": 0.2987,
      "step": 8195
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8958373695308309,
      "learning_rate": 1.7546058148423503e-05,
      "loss": 0.4263,
      "step": 8196
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4121601299091356,
      "learning_rate": 1.754540725878861e-05,
      "loss": 0.2731,
      "step": 8197
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4746417256682405,
      "learning_rate": 1.7544756294919513e-05,
      "loss": 0.1987,
      "step": 8198
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0406590306169463,
      "learning_rate": 1.7544105256822626e-05,
      "loss": 0.3729,
      "step": 8199
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7740730333985704,
      "learning_rate": 1.754345414450435e-05,
      "loss": 0.2946,
      "step": 8200
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4044233391456914,
      "learning_rate": 1.7542802957971085e-05,
      "loss": 0.2977,
      "step": 8201
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5227219211007338,
      "learning_rate": 1.7542151697229247e-05,
      "loss": 0.3208,
      "step": 8202
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9366312536698767,
      "learning_rate": 1.7541500362285236e-05,
      "loss": 0.8049,
      "step": 8203
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0704105158032198,
      "learning_rate": 1.754084895314546e-05,
      "loss": 0.6749,
      "step": 8204
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7832901128232233,
      "learning_rate": 1.7540197469816335e-05,
      "loss": 0.3038,
      "step": 8205
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40772130903556614,
      "learning_rate": 1.7539545912304266e-05,
      "loss": 0.2411,
      "step": 8206
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40397734260515356,
      "learning_rate": 1.7538894280615662e-05,
      "loss": 0.3132,
      "step": 8207
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.38826674650330306,
      "learning_rate": 1.753824257475694e-05,
      "loss": 0.1909,
      "step": 8208
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8448638915801154,
      "learning_rate": 1.7537590794734503e-05,
      "loss": 0.5395,
      "step": 8209
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3443067349135538,
      "learning_rate": 1.7536938940554772e-05,
      "loss": 0.1683,
      "step": 8210
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5635832189140979,
      "learning_rate": 1.753628701222415e-05,
      "loss": 0.3437,
      "step": 8211
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4380402188139035,
      "learning_rate": 1.753563500974906e-05,
      "loss": 0.2876,
      "step": 8212
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.440494962359766,
      "learning_rate": 1.753498293313591e-05,
      "loss": 0.3071,
      "step": 8213
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7515917998301652,
      "learning_rate": 1.753433078239112e-05,
      "loss": 0.2833,
      "step": 8214
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3788635660869964,
      "learning_rate": 1.7533678557521105e-05,
      "loss": 0.2544,
      "step": 8215
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3510011874203372,
      "learning_rate": 1.753302625853228e-05,
      "loss": 0.1781,
      "step": 8216
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8442451598912408,
      "learning_rate": 1.753237388543107e-05,
      "loss": 0.4194,
      "step": 8217
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5033521702969065,
      "learning_rate": 1.753172143822388e-05,
      "loss": 0.3477,
      "step": 8218
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.31890153555513645,
      "learning_rate": 1.753106891691714e-05,
      "loss": 0.2297,
      "step": 8219
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6172383357244436,
      "learning_rate": 1.753041632151727e-05,
      "loss": 0.4177,
      "step": 8220
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9088449324683954,
      "learning_rate": 1.7529763652030677e-05,
      "loss": 0.4898,
      "step": 8221
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.679360356665925,
      "learning_rate": 1.7529110908463797e-05,
      "loss": 0.7626,
      "step": 8222
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5123363663529543,
      "learning_rate": 1.7528458090823045e-05,
      "loss": 0.1674,
      "step": 8223
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3832018090061748,
      "learning_rate": 1.7527805199114846e-05,
      "loss": 0.2921,
      "step": 8224
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43723069105839896,
      "learning_rate": 1.7527152233345623e-05,
      "loss": 0.2802,
      "step": 8225
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8687252570655142,
      "learning_rate": 1.7526499193521797e-05,
      "loss": 0.3996,
      "step": 8226
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3931098188680657,
      "learning_rate": 1.75258460796498e-05,
      "loss": 0.1832,
      "step": 8227
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.34345571060064534,
      "learning_rate": 1.7525192891736047e-05,
      "loss": 0.2002,
      "step": 8228
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.3904587272170477,
      "learning_rate": 1.7524539629786975e-05,
      "loss": 0.7359,
      "step": 8229
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4644349587069691,
      "learning_rate": 1.7523886293809006e-05,
      "loss": 0.2751,
      "step": 8230
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5721977156729469,
      "learning_rate": 1.7523232883808566e-05,
      "loss": 0.4223,
      "step": 8231
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5463922640471589,
      "learning_rate": 1.7522579399792087e-05,
      "loss": 0.1887,
      "step": 8232
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40132288124805154,
      "learning_rate": 1.7521925841765992e-05,
      "loss": 0.2906,
      "step": 8233
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9093391459047353,
      "learning_rate": 1.7521272209736723e-05,
      "loss": 0.3852,
      "step": 8234
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6567053510313967,
      "learning_rate": 1.7520618503710696e-05,
      "loss": 0.4496,
      "step": 8235
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3719460520831129,
      "learning_rate": 1.7519964723694355e-05,
      "loss": 0.2105,
      "step": 8236
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5130331440903764,
      "learning_rate": 1.7519310869694122e-05,
      "loss": 0.2988,
      "step": 8237
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3318056347331882,
      "learning_rate": 1.751865694171644e-05,
      "loss": 0.166,
      "step": 8238
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1182133516552655,
      "learning_rate": 1.751800293976773e-05,
      "loss": 0.4909,
      "step": 8239
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.5195630768593753,
      "learning_rate": 1.7517348863854436e-05,
      "loss": 0.8657,
      "step": 8240
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0470287544391998,
      "learning_rate": 1.7516694713982992e-05,
      "loss": 0.3858,
      "step": 8241
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4008539895433175,
      "learning_rate": 1.7516040490159832e-05,
      "loss": 0.2546,
      "step": 8242
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2680277542213489,
      "learning_rate": 1.751538619239139e-05,
      "loss": 0.2597,
      "step": 8243
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7315887598807101,
      "learning_rate": 1.7514731820684112e-05,
      "loss": 0.5371,
      "step": 8244
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.334642601314767,
      "learning_rate": 1.7514077375044426e-05,
      "loss": 0.1175,
      "step": 8245
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4152449173905864,
      "learning_rate": 1.751342285547877e-05,
      "loss": 0.2075,
      "step": 8246
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43655292779627314,
      "learning_rate": 1.7512768261993594e-05,
      "loss": 0.2609,
      "step": 8247
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.37921581505311375,
      "learning_rate": 1.751211359459533e-05,
      "loss": 0.3133,
      "step": 8248
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4048367630489902,
      "learning_rate": 1.751145885329042e-05,
      "loss": 0.5285,
      "step": 8249
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8215312747889528,
      "learning_rate": 1.7510804038085308e-05,
      "loss": 0.5248,
      "step": 8250
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3123606693426211,
      "learning_rate": 1.7510149148986432e-05,
      "loss": 0.204,
      "step": 8251
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6890047638155056,
      "learning_rate": 1.750949418600024e-05,
      "loss": 0.3987,
      "step": 8252
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.0343449308164536,
      "learning_rate": 1.750883914913317e-05,
      "loss": 0.7617,
      "step": 8253
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39015481359927007,
      "learning_rate": 1.7508184038391673e-05,
      "loss": 0.2588,
      "step": 8254
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5556607874609836,
      "learning_rate": 1.750752885378219e-05,
      "loss": 0.1898,
      "step": 8255
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3387647225910612,
      "learning_rate": 1.7506873595311164e-05,
      "loss": 0.1992,
      "step": 8256
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.715560340736656,
      "learning_rate": 1.750621826298505e-05,
      "loss": 0.8261,
      "step": 8257
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2120831490331525,
      "learning_rate": 1.750556285681029e-05,
      "loss": 0.4758,
      "step": 8258
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.635504150411379,
      "learning_rate": 1.7504907376793328e-05,
      "loss": 0.4238,
      "step": 8259
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.29490235682960386,
      "learning_rate": 1.7504251822940627e-05,
      "loss": 0.2383,
      "step": 8260
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5488314302836067,
      "learning_rate": 1.750359619525862e-05,
      "loss": 0.3804,
      "step": 8261
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5325555024314848,
      "learning_rate": 1.7502940493753764e-05,
      "loss": 0.3883,
      "step": 8262
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6353847335549347,
      "learning_rate": 1.7502284718432515e-05,
      "loss": 0.2961,
      "step": 8263
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3356484482564496,
      "learning_rate": 1.750162886930132e-05,
      "loss": 0.1341,
      "step": 8264
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6142525225801629,
      "learning_rate": 1.7500972946366633e-05,
      "loss": 0.4065,
      "step": 8265
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.43629855847013016,
      "learning_rate": 1.7500316949634905e-05,
      "loss": 0.3087,
      "step": 8266
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5963642013613795,
      "learning_rate": 1.7499660879112587e-05,
      "loss": 0.3516,
      "step": 8267
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9826951467544274,
      "learning_rate": 1.749900473480614e-05,
      "loss": 0.5202,
      "step": 8268
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3704504284270499,
      "learning_rate": 1.7498348516722017e-05,
      "loss": 0.1873,
      "step": 8269
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5038348775165412,
      "learning_rate": 1.7497692224866676e-05,
      "loss": 0.3869,
      "step": 8270
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39660116843142024,
      "learning_rate": 1.749703585924657e-05,
      "loss": 0.265,
      "step": 8271
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7106937362404282,
      "learning_rate": 1.7496379419868157e-05,
      "loss": 0.3665,
      "step": 8272
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2689855669917528,
      "learning_rate": 1.7495722906737897e-05,
      "loss": 0.074,
      "step": 8273
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.373844417922149,
      "learning_rate": 1.749506631986225e-05,
      "loss": 0.2805,
      "step": 8274
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.23275150303481132,
      "learning_rate": 1.749440965924768e-05,
      "loss": 0.1542,
      "step": 8275
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7790982475173586,
      "learning_rate": 1.7493752924900632e-05,
      "loss": 0.4849,
      "step": 8276
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.48156566396490075,
      "learning_rate": 1.749309611682758e-05,
      "loss": 0.3174,
      "step": 8277
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4536176329887553,
      "learning_rate": 1.7492439235034986e-05,
      "loss": 0.289,
      "step": 8278
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.381934355213447,
      "learning_rate": 1.7491782279529307e-05,
      "loss": 0.2737,
      "step": 8279
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9505016934520476,
      "learning_rate": 1.7491125250317007e-05,
      "loss": 0.2918,
      "step": 8280
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9592311012579714,
      "learning_rate": 1.7490468147404557e-05,
      "loss": 0.6091,
      "step": 8281
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.32219953421748565,
      "learning_rate": 1.748981097079841e-05,
      "loss": 0.1108,
      "step": 8282
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39611159971264515,
      "learning_rate": 1.7489153720505043e-05,
      "loss": 0.3452,
      "step": 8283
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4216887163629315,
      "learning_rate": 1.7488496396530915e-05,
      "loss": 0.284,
      "step": 8284
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4817242056592361,
      "learning_rate": 1.7487838998882498e-05,
      "loss": 0.3814,
      "step": 8285
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8210400382545587,
      "learning_rate": 1.7487181527566254e-05,
      "loss": 0.4476,
      "step": 8286
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.449065868646513,
      "learning_rate": 1.748652398258866e-05,
      "loss": 0.2726,
      "step": 8287
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.26212310726224647,
      "learning_rate": 1.748586636395617e-05,
      "loss": 0.153,
      "step": 8288
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.9292607394327579,
      "learning_rate": 1.7485208671675272e-05,
      "loss": 0.8146,
      "step": 8289
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.31081746345229216,
      "learning_rate": 1.7484550905752427e-05,
      "loss": 0.2731,
      "step": 8290
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7003951722142971,
      "learning_rate": 1.7483893066194105e-05,
      "loss": 0.3831,
      "step": 8291
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.4077394395326036,
      "learning_rate": 1.7483235153006783e-05,
      "loss": 0.2456,
      "step": 8292
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3750939568730368,
      "learning_rate": 1.748257716619693e-05,
      "loss": 0.2187,
      "step": 8293
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7363469953034916,
      "learning_rate": 1.7481919105771018e-05,
      "loss": 0.4888,
      "step": 8294
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3361052625612154,
      "learning_rate": 1.748126097173553e-05,
      "loss": 0.2667,
      "step": 8295
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7657788239119273,
      "learning_rate": 1.7480602764096928e-05,
      "loss": 0.3635,
      "step": 8296
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.25022594627943756,
      "learning_rate": 1.7479944482861698e-05,
      "loss": 0.1862,
      "step": 8297
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0536981357423159,
      "learning_rate": 1.7479286128036313e-05,
      "loss": 0.5895,
      "step": 8298
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.0780482812688426,
      "learning_rate": 1.7478627699627252e-05,
      "loss": 0.2898,
      "step": 8299
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.6661948303995855,
      "learning_rate": 1.7477969197640986e-05,
      "loss": 0.8872,
      "step": 8300
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2913962005579635,
      "learning_rate": 1.7477310622084002e-05,
      "loss": 0.1983,
      "step": 8301
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3793586911116958,
      "learning_rate": 1.7476651972962776e-05,
      "loss": 0.3398,
      "step": 8302
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.8437032183258827,
      "learning_rate": 1.7475993250283786e-05,
      "loss": 0.4307,
      "step": 8303
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.4470659675019757,
      "learning_rate": 1.7475334454053517e-05,
      "loss": 0.7824,
      "step": 8304
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.28048633600038736,
      "learning_rate": 1.7474675584278444e-05,
      "loss": 0.0996,
      "step": 8305
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.27292192175566093,
      "learning_rate": 1.7474016640965054e-05,
      "loss": 0.19,
      "step": 8306
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.7583462693852145,
      "learning_rate": 1.7473357624119832e-05,
      "loss": 0.7479,
      "step": 8307
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.32182623072947697,
      "learning_rate": 1.7472698533749257e-05,
      "loss": 0.2803,
      "step": 8308
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.9913467362110024,
      "learning_rate": 1.7472039369859818e-05,
      "loss": 0.5094,
      "step": 8309
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3054238375990284,
      "learning_rate": 1.7471380132457988e-05,
      "loss": 0.2579,
      "step": 8310
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.86308037574049,
      "learning_rate": 1.7470720821550273e-05,
      "loss": 0.6252,
      "step": 8311
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.7038210868580796,
      "learning_rate": 1.747006143714314e-05,
      "loss": 0.446,
      "step": 8312
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.40205865606463337,
      "learning_rate": 1.746940197924309e-05,
      "loss": 0.3577,
      "step": 8313
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.34126693708297645,
      "learning_rate": 1.74687424478566e-05,
      "loss": 0.1874,
      "step": 8314
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5978767122923688,
      "learning_rate": 1.746808284299017e-05,
      "loss": 0.406,
      "step": 8315
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3381220393123559,
      "learning_rate": 1.746742316465028e-05,
      "loss": 0.0995,
      "step": 8316
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.2990403502919166,
      "learning_rate": 1.7466763412843427e-05,
      "loss": 0.6929,
      "step": 8317
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5730727415364906,
      "learning_rate": 1.7466103587576097e-05,
      "loss": 0.3353,
      "step": 8318
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.39684948441338996,
      "learning_rate": 1.746544368885478e-05,
      "loss": 0.1925,
      "step": 8319
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.31970936113664145,
      "learning_rate": 1.7464783716685977e-05,
      "loss": 0.3091,
      "step": 8320
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.6415872806257885,
      "learning_rate": 1.746412367107617e-05,
      "loss": 0.4349,
      "step": 8321
    },
    {
      "epoch": 0.25,
      "grad_norm": 1.1596266452580806,
      "learning_rate": 1.7463463552031865e-05,
      "loss": 0.8241,
      "step": 8322
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2648895665583006,
      "learning_rate": 1.7462803359559543e-05,
      "loss": 0.077,
      "step": 8323
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.3105698589412988,
      "learning_rate": 1.7462143093665707e-05,
      "loss": 0.2392,
      "step": 8324
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.2801457953724885,
      "learning_rate": 1.7461482754356856e-05,
      "loss": 0.1672,
      "step": 8325
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.5033842595313408,
      "learning_rate": 1.746082234163948e-05,
      "loss": 0.399,
      "step": 8326
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6278580599948926,
      "learning_rate": 1.746016185552008e-05,
      "loss": 0.3581,
      "step": 8327
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4200661493153005,
      "learning_rate": 1.7459501296005154e-05,
      "loss": 0.2828,
      "step": 8328
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4221086259011289,
      "learning_rate": 1.74588406631012e-05,
      "loss": 0.3297,
      "step": 8329
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.475549180809396,
      "learning_rate": 1.7458179956814713e-05,
      "loss": 0.8607,
      "step": 8330
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3184045319928204,
      "learning_rate": 1.74575191771522e-05,
      "loss": 0.2633,
      "step": 8331
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41412286360158246,
      "learning_rate": 1.745685832412016e-05,
      "loss": 0.1784,
      "step": 8332
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3766631909970262,
      "learning_rate": 1.7456197397725092e-05,
      "loss": 0.3197,
      "step": 8333
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20485957318864856,
      "learning_rate": 1.74555363979735e-05,
      "loss": 0.0746,
      "step": 8334
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9158648108750327,
      "learning_rate": 1.7454875324871894e-05,
      "loss": 0.6141,
      "step": 8335
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5709905731610956,
      "learning_rate": 1.7454214178426763e-05,
      "loss": 0.3409,
      "step": 8336
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37167285516579684,
      "learning_rate": 1.745355295864463e-05,
      "loss": 0.2833,
      "step": 8337
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30835746389822505,
      "learning_rate": 1.745289166553198e-05,
      "loss": 0.2393,
      "step": 8338
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5547579458069434,
      "learning_rate": 1.7452230299095337e-05,
      "loss": 0.8916,
      "step": 8339
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8998978222293378,
      "learning_rate": 1.7451568859341196e-05,
      "loss": 0.5058,
      "step": 8340
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7015574899446376,
      "learning_rate": 1.7450907346276073e-05,
      "loss": 0.3956,
      "step": 8341
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36775035678361556,
      "learning_rate": 1.7450245759906468e-05,
      "loss": 0.2334,
      "step": 8342
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3421849664067176,
      "learning_rate": 1.7449584100238895e-05,
      "loss": 0.2124,
      "step": 8343
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.32986967123000716,
      "learning_rate": 1.744892236727986e-05,
      "loss": 0.2818,
      "step": 8344
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6841835510829537,
      "learning_rate": 1.744826056103588e-05,
      "loss": 0.3576,
      "step": 8345
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9501577980934349,
      "learning_rate": 1.744759868151346e-05,
      "loss": 0.4846,
      "step": 8346
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39595437282285406,
      "learning_rate": 1.7446936728719113e-05,
      "loss": 0.2507,
      "step": 8347
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4370723711773226,
      "learning_rate": 1.744627470265935e-05,
      "loss": 0.9477,
      "step": 8348
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30023567831002007,
      "learning_rate": 1.744561260334069e-05,
      "loss": 0.261,
      "step": 8349
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9195548181442497,
      "learning_rate": 1.7444950430769644e-05,
      "loss": 0.8225,
      "step": 8350
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.34503660250185497,
      "learning_rate": 1.7444288184952723e-05,
      "loss": 0.2063,
      "step": 8351
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3336216516928903,
      "learning_rate": 1.7443625865896447e-05,
      "loss": 0.1952,
      "step": 8352
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6158963701864104,
      "learning_rate": 1.744296347360733e-05,
      "loss": 0.3906,
      "step": 8353
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8088447097107043,
      "learning_rate": 1.744230100809189e-05,
      "loss": 1.0103,
      "step": 8354
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2829040501328685,
      "learning_rate": 1.7441638469356643e-05,
      "loss": 0.2065,
      "step": 8355
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.373156604941761,
      "learning_rate": 1.7440975857408108e-05,
      "loss": 0.33,
      "step": 8356
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9807859426927739,
      "learning_rate": 1.74403131722528e-05,
      "loss": 0.4539,
      "step": 8357
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.958352056307255,
      "learning_rate": 1.7439650413897247e-05,
      "loss": 0.507,
      "step": 8358
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9638008028273797,
      "learning_rate": 1.743898758234797e-05,
      "loss": 0.458,
      "step": 8359
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.25940503664185055,
      "learning_rate": 1.743832467761148e-05,
      "loss": 0.1785,
      "step": 8360
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4415939412888301,
      "learning_rate": 1.7437661699694304e-05,
      "loss": 0.3922,
      "step": 8361
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.404007678854191,
      "learning_rate": 1.7436998648602967e-05,
      "loss": 0.3188,
      "step": 8362
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4387452134323058,
      "learning_rate": 1.743633552434399e-05,
      "loss": 0.3523,
      "step": 8363
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4015428434189938,
      "learning_rate": 1.7435672326923896e-05,
      "loss": 0.2044,
      "step": 8364
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.566454299067355,
      "learning_rate": 1.743500905634921e-05,
      "loss": 0.3787,
      "step": 8365
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.09709286582112,
      "learning_rate": 1.743434571262646e-05,
      "loss": 0.5884,
      "step": 8366
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3528620230972118,
      "learning_rate": 1.743368229576217e-05,
      "loss": 0.3299,
      "step": 8367
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5188905167432792,
      "learning_rate": 1.743301880576287e-05,
      "loss": 0.1397,
      "step": 8368
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6189170885149521,
      "learning_rate": 1.743235524263508e-05,
      "loss": 0.394,
      "step": 8369
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41623375680204167,
      "learning_rate": 1.7431691606385337e-05,
      "loss": 0.2693,
      "step": 8370
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39758803279164145,
      "learning_rate": 1.7431027897020167e-05,
      "loss": 0.2577,
      "step": 8371
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3216443077756162,
      "learning_rate": 1.74303641145461e-05,
      "loss": 0.2603,
      "step": 8372
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5927297538840578,
      "learning_rate": 1.7429700258969665e-05,
      "loss": 0.0798,
      "step": 8373
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4470816967789066,
      "learning_rate": 1.7429036330297395e-05,
      "loss": 0.3421,
      "step": 8374
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9601930132506953,
      "learning_rate": 1.7428372328535817e-05,
      "loss": 0.5198,
      "step": 8375
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3191972960035494,
      "learning_rate": 1.742770825369147e-05,
      "loss": 0.6776,
      "step": 8376
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6843053013203886,
      "learning_rate": 1.7427044105770886e-05,
      "loss": 0.336,
      "step": 8377
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.44814167044667447,
      "learning_rate": 1.7426379884780604e-05,
      "loss": 0.3251,
      "step": 8378
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2828125177056346,
      "learning_rate": 1.7425715590727143e-05,
      "loss": 0.2665,
      "step": 8379
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6479718090975638,
      "learning_rate": 1.7425051223617053e-05,
      "loss": 0.4988,
      "step": 8380
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30284355565522825,
      "learning_rate": 1.7424386783456867e-05,
      "loss": 0.113,
      "step": 8381
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43566511682599907,
      "learning_rate": 1.742372227025312e-05,
      "loss": 0.2258,
      "step": 8382
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3227187379698991,
      "learning_rate": 1.7423057684012348e-05,
      "loss": 0.2275,
      "step": 8383
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0820094138889254,
      "learning_rate": 1.7422393024741096e-05,
      "loss": 0.6071,
      "step": 8384
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36537933062430594,
      "learning_rate": 1.7421728292445897e-05,
      "loss": 0.3031,
      "step": 8385
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.66817808545797,
      "learning_rate": 1.742106348713329e-05,
      "loss": 0.2612,
      "step": 8386
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3828580496223754,
      "learning_rate": 1.7420398608809825e-05,
      "loss": 0.2965,
      "step": 8387
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6437578913666355,
      "learning_rate": 1.7419733657482033e-05,
      "loss": 0.3642,
      "step": 8388
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4276855600935852,
      "learning_rate": 1.741906863315646e-05,
      "loss": 0.7104,
      "step": 8389
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.240196948373592,
      "learning_rate": 1.7418403535839653e-05,
      "loss": 0.163,
      "step": 8390
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3781433753680938,
      "learning_rate": 1.7417738365538147e-05,
      "loss": 0.291,
      "step": 8391
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.376951836171906,
      "learning_rate": 1.741707312225849e-05,
      "loss": 0.2293,
      "step": 8392
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1397605394298365,
      "learning_rate": 1.7416407806007226e-05,
      "loss": 0.7184,
      "step": 8393
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8241122617601334,
      "learning_rate": 1.7415742416790905e-05,
      "loss": 0.4535,
      "step": 8394
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7208741947676668,
      "learning_rate": 1.7415076954616065e-05,
      "loss": 0.4739,
      "step": 8395
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.33942066522257314,
      "learning_rate": 1.7414411419489263e-05,
      "loss": 0.0756,
      "step": 8396
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2568837365523488,
      "learning_rate": 1.7413745811417044e-05,
      "loss": 0.2563,
      "step": 8397
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.822578910832048,
      "learning_rate": 1.7413080130405946e-05,
      "loss": 0.5645,
      "step": 8398
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.22693823599465726,
      "learning_rate": 1.741241437646253e-05,
      "loss": 0.1174,
      "step": 8399
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8701463919140203,
      "learning_rate": 1.7411748549593346e-05,
      "loss": 0.4046,
      "step": 8400
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37794598998045537,
      "learning_rate": 1.7411082649804938e-05,
      "loss": 0.2368,
      "step": 8401
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4649307699050695,
      "learning_rate": 1.741041667710386e-05,
      "loss": 0.7976,
      "step": 8402
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2950023803432716,
      "learning_rate": 1.7409750631496666e-05,
      "loss": 0.2597,
      "step": 8403
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8736503694157225,
      "learning_rate": 1.7409084512989904e-05,
      "loss": 0.5169,
      "step": 8404
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36017655552511785,
      "learning_rate": 1.7408418321590136e-05,
      "loss": 0.2307,
      "step": 8405
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6398885165498021,
      "learning_rate": 1.7407752057303905e-05,
      "loss": 0.3897,
      "step": 8406
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0872035777182325,
      "learning_rate": 1.7407085720137776e-05,
      "loss": 0.3333,
      "step": 8407
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.46260936434842825,
      "learning_rate": 1.7406419310098296e-05,
      "loss": 0.2679,
      "step": 8408
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24382419728259885,
      "learning_rate": 1.7405752827192028e-05,
      "loss": 0.1983,
      "step": 8409
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.31317775090456096,
      "learning_rate": 1.740508627142553e-05,
      "loss": 0.2261,
      "step": 8410
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5110628170418139,
      "learning_rate": 1.7404419642805353e-05,
      "loss": 0.9393,
      "step": 8411
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8245422632596467,
      "learning_rate": 1.740375294133806e-05,
      "loss": 0.391,
      "step": 8412
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7980779495934691,
      "learning_rate": 1.740308616703021e-05,
      "loss": 0.4919,
      "step": 8413
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2877459718868792,
      "learning_rate": 1.7402419319888362e-05,
      "loss": 0.2068,
      "step": 8414
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.383815861613435,
      "learning_rate": 1.7401752399919076e-05,
      "loss": 0.3064,
      "step": 8415
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0343989776561038,
      "learning_rate": 1.7401085407128918e-05,
      "loss": 0.3647,
      "step": 8416
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.940377616078426,
      "learning_rate": 1.7400418341524443e-05,
      "loss": 0.8689,
      "step": 8417
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21971046248888543,
      "learning_rate": 1.7399751203112217e-05,
      "loss": 0.096,
      "step": 8418
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3961271312799431,
      "learning_rate": 1.7399083991898808e-05,
      "loss": 0.31,
      "step": 8419
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9176085132327436,
      "learning_rate": 1.739841670789077e-05,
      "loss": 0.6044,
      "step": 8420
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40594658577085563,
      "learning_rate": 1.7397749351094677e-05,
      "loss": 0.3367,
      "step": 8421
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5364220874442851,
      "learning_rate": 1.7397081921517092e-05,
      "loss": 0.2993,
      "step": 8422
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36637401094665856,
      "learning_rate": 1.739641441916458e-05,
      "loss": 0.0754,
      "step": 8423
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43162498557326234,
      "learning_rate": 1.739574684404371e-05,
      "loss": 0.3323,
      "step": 8424
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9504783676173555,
      "learning_rate": 1.7395079196161048e-05,
      "loss": 0.3873,
      "step": 8425
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38147974305218085,
      "learning_rate": 1.739441147552317e-05,
      "loss": 0.3388,
      "step": 8426
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2220438954411963,
      "learning_rate": 1.739374368213663e-05,
      "loss": 0.0712,
      "step": 8427
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4190172415889315,
      "learning_rate": 1.739307581600801e-05,
      "loss": 0.3284,
      "step": 8428
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41899784114864824,
      "learning_rate": 1.7392407877143876e-05,
      "loss": 0.3226,
      "step": 8429
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6471557319528404,
      "learning_rate": 1.7391739865550807e-05,
      "loss": 0.5513,
      "step": 8430
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7982833569881537,
      "learning_rate": 1.739107178123536e-05,
      "loss": 0.4809,
      "step": 8431
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3657220126426507,
      "learning_rate": 1.7390403624204126e-05,
      "loss": 0.1704,
      "step": 8432
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24623793506222802,
      "learning_rate": 1.7389735394463663e-05,
      "loss": 0.2324,
      "step": 8433
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3027673686361794,
      "learning_rate": 1.7389067092020554e-05,
      "loss": 0.6513,
      "step": 8434
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2008593234878617,
      "learning_rate": 1.738839871688137e-05,
      "loss": 0.0984,
      "step": 8435
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2712090292848944,
      "learning_rate": 1.738773026905269e-05,
      "loss": 0.0992,
      "step": 8436
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39331250590466854,
      "learning_rate": 1.7387061748541086e-05,
      "loss": 0.2998,
      "step": 8437
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38247676063489944,
      "learning_rate": 1.738639315535314e-05,
      "loss": 0.2986,
      "step": 8438
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0426997956641197,
      "learning_rate": 1.7385724489495432e-05,
      "loss": 0.5056,
      "step": 8439
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8247089858272719,
      "learning_rate": 1.738505575097453e-05,
      "loss": 0.453,
      "step": 8440
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3973247847399334,
      "learning_rate": 1.738438693979702e-05,
      "loss": 0.1852,
      "step": 8441
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3979013432030682,
      "learning_rate": 1.7383718055969484e-05,
      "loss": 0.276,
      "step": 8442
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9588935198073598,
      "learning_rate": 1.73830490994985e-05,
      "loss": 0.8059,
      "step": 8443
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3483999916517531,
      "learning_rate": 1.738238007039065e-05,
      "loss": 0.2737,
      "step": 8444
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0090833288695198,
      "learning_rate": 1.7381710968652513e-05,
      "loss": 0.65,
      "step": 8445
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3228216386817651,
      "learning_rate": 1.7381041794290676e-05,
      "loss": 0.1726,
      "step": 8446
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6134441707137841,
      "learning_rate": 1.7380372547311722e-05,
      "loss": 0.3879,
      "step": 8447
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2341985549831178,
      "learning_rate": 1.7379703227722237e-05,
      "loss": 0.4391,
      "step": 8448
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4727287459463914,
      "learning_rate": 1.7379033835528798e-05,
      "loss": 0.286,
      "step": 8449
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3005743818202551,
      "learning_rate": 1.7378364370737997e-05,
      "loss": 0.1996,
      "step": 8450
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.33462763714686733,
      "learning_rate": 1.7377694833356423e-05,
      "loss": 0.2515,
      "step": 8451
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.524974732652785,
      "learning_rate": 1.7377025223390657e-05,
      "loss": 0.3203,
      "step": 8452
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2913159505539995,
      "learning_rate": 1.737635554084729e-05,
      "loss": 0.3178,
      "step": 8453
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9086816572203964,
      "learning_rate": 1.7375685785732908e-05,
      "loss": 0.5096,
      "step": 8454
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.32296737666698755,
      "learning_rate": 1.7375015958054104e-05,
      "loss": 0.1807,
      "step": 8455
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.334916745844342,
      "learning_rate": 1.7374346057817466e-05,
      "loss": 0.3198,
      "step": 8456
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4094129748274108,
      "learning_rate": 1.7373676085029586e-05,
      "loss": 0.3328,
      "step": 8457
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6492060777717248,
      "learning_rate": 1.7373006039697054e-05,
      "loss": 0.8787,
      "step": 8458
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21635422584187197,
      "learning_rate": 1.7372335921826463e-05,
      "loss": 0.0899,
      "step": 8459
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3590761227673211,
      "learning_rate": 1.7371665731424405e-05,
      "loss": 0.2829,
      "step": 8460
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0456017151517274,
      "learning_rate": 1.7370995468497477e-05,
      "loss": 0.3984,
      "step": 8461
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.482830330226446,
      "learning_rate": 1.7370325133052264e-05,
      "loss": 0.3622,
      "step": 8462
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.38266750915352815,
      "learning_rate": 1.736965472509537e-05,
      "loss": 0.3073,
      "step": 8463
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.28721632485837256,
      "learning_rate": 1.736898424463339e-05,
      "loss": 0.2092,
      "step": 8464
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6236125288646567,
      "learning_rate": 1.7368313691672915e-05,
      "loss": 0.482,
      "step": 8465
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1807693735011313,
      "learning_rate": 1.736764306622055e-05,
      "loss": 0.5278,
      "step": 8466
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4310581208087882,
      "learning_rate": 1.7366972368282883e-05,
      "loss": 0.8611,
      "step": 8467
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30710753743544716,
      "learning_rate": 1.736630159786652e-05,
      "loss": 0.2276,
      "step": 8468
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3242656442099113,
      "learning_rate": 1.7365630754978058e-05,
      "loss": 0.2825,
      "step": 8469
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2927582432472671,
      "learning_rate": 1.7364959839624098e-05,
      "loss": 0.0744,
      "step": 8470
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1392457818881283,
      "learning_rate": 1.7364288851811236e-05,
      "loss": 0.5636,
      "step": 8471
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6581412257717096,
      "learning_rate": 1.736361779154608e-05,
      "loss": 0.3948,
      "step": 8472
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42224198148728803,
      "learning_rate": 1.736294665883523e-05,
      "loss": 0.3211,
      "step": 8473
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4764564042896882,
      "learning_rate": 1.7362275453685286e-05,
      "loss": 0.2449,
      "step": 8474
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3593763305675696,
      "learning_rate": 1.7361604176102857e-05,
      "loss": 0.3462,
      "step": 8475
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.97452910779099,
      "learning_rate": 1.736093282609454e-05,
      "loss": 0.532,
      "step": 8476
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.31236710975542203,
      "learning_rate": 1.736026140366694e-05,
      "loss": 0.0745,
      "step": 8477
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4307852790195711,
      "learning_rate": 1.7359589908826676e-05,
      "loss": 0.3435,
      "step": 8478
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.23101808843335817,
      "learning_rate": 1.735891834158034e-05,
      "loss": 0.0722,
      "step": 8479
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3761182608843479,
      "learning_rate": 1.7358246701934545e-05,
      "loss": 0.3465,
      "step": 8480
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8700052701285501,
      "learning_rate": 1.7357574989895896e-05,
      "loss": 0.3876,
      "step": 8481
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6257989225078954,
      "learning_rate": 1.7356903205471004e-05,
      "loss": 0.377,
      "step": 8482
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3702374431997493,
      "learning_rate": 1.7356231348666476e-05,
      "loss": 0.2413,
      "step": 8483
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.4605902418048768,
      "learning_rate": 1.735555941948893e-05,
      "loss": 0.9214,
      "step": 8484
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3878180308888686,
      "learning_rate": 1.7354887417944962e-05,
      "loss": 0.3992,
      "step": 8485
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43167622335912165,
      "learning_rate": 1.7354215344041197e-05,
      "loss": 0.3331,
      "step": 8486
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3176047733086085,
      "learning_rate": 1.735354319778424e-05,
      "loss": 0.1983,
      "step": 8487
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.31391039717502944,
      "learning_rate": 1.7352870979180705e-05,
      "loss": 0.1713,
      "step": 8488
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8933036179576873,
      "learning_rate": 1.7352198688237205e-05,
      "loss": 0.4114,
      "step": 8489
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9151055693283154,
      "learning_rate": 1.7351526324960357e-05,
      "loss": 0.4071,
      "step": 8490
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6251949129005976,
      "learning_rate": 1.7350853889356773e-05,
      "loss": 0.4211,
      "step": 8491
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.25564812000709847,
      "learning_rate": 1.7350181381433068e-05,
      "loss": 0.2558,
      "step": 8492
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.8378573698752327,
      "learning_rate": 1.7349508801195863e-05,
      "loss": 0.9543,
      "step": 8493
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1567876855056138,
      "learning_rate": 1.7348836148651772e-05,
      "loss": 0.5391,
      "step": 8494
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.707008905170276,
      "learning_rate": 1.7348163423807412e-05,
      "loss": 0.4828,
      "step": 8495
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24156273153082797,
      "learning_rate": 1.73474906266694e-05,
      "loss": 0.1715,
      "step": 8496
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4133428683471825,
      "learning_rate": 1.734681775724436e-05,
      "loss": 0.2612,
      "step": 8497
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35727212128061897,
      "learning_rate": 1.734614481553891e-05,
      "loss": 0.3006,
      "step": 8498
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8276600323770785,
      "learning_rate": 1.734547180155967e-05,
      "loss": 0.5419,
      "step": 8499
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.33387997000934755,
      "learning_rate": 1.734479871531326e-05,
      "loss": 0.1825,
      "step": 8500
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5366533847379477,
      "learning_rate": 1.7344125556806304e-05,
      "loss": 0.3775,
      "step": 8501
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2775891848738123,
      "learning_rate": 1.7343452326045427e-05,
      "loss": 0.5865,
      "step": 8502
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.27240588697749835,
      "learning_rate": 1.7342779023037245e-05,
      "loss": 0.2547,
      "step": 8503
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8777776185031009,
      "learning_rate": 1.734210564778839e-05,
      "loss": 0.4627,
      "step": 8504
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3779826765068603,
      "learning_rate": 1.734143220030548e-05,
      "loss": 0.1989,
      "step": 8505
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.32478136307205374,
      "learning_rate": 1.7340758680595147e-05,
      "loss": 0.2779,
      "step": 8506
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3794687975280703,
      "learning_rate": 1.7340085088664017e-05,
      "loss": 0.2856,
      "step": 8507
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1263083113025205,
      "learning_rate": 1.7339411424518716e-05,
      "loss": 0.7097,
      "step": 8508
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.32622806133087545,
      "learning_rate": 1.7338737688165865e-05,
      "loss": 0.1391,
      "step": 8509
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3110855852042722,
      "learning_rate": 1.73380638796121e-05,
      "loss": 0.2903,
      "step": 8510
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4253675217567557,
      "learning_rate": 1.7337389998864048e-05,
      "loss": 0.2692,
      "step": 8511
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.737642031171737,
      "learning_rate": 1.733671604592834e-05,
      "loss": 0.8951,
      "step": 8512
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7998786989519969,
      "learning_rate": 1.7336042020811602e-05,
      "loss": 0.3757,
      "step": 8513
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4710766128802344,
      "learning_rate": 1.7335367923520473e-05,
      "loss": 0.3208,
      "step": 8514
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40631503710862393,
      "learning_rate": 1.7334693754061577e-05,
      "loss": 0.3476,
      "step": 8515
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.34733426389348987,
      "learning_rate": 1.7334019512441552e-05,
      "loss": 0.2654,
      "step": 8516
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3683531840878993,
      "learning_rate": 1.733334519866703e-05,
      "loss": 0.2139,
      "step": 8517
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3225455729219406,
      "learning_rate": 1.7332670812744642e-05,
      "loss": 0.0836,
      "step": 8518
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4057395773296951,
      "learning_rate": 1.7331996354681032e-05,
      "loss": 0.3326,
      "step": 8519
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.9006477657683054,
      "learning_rate": 1.7331321824482825e-05,
      "loss": 0.5193,
      "step": 8520
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4964005260187421,
      "learning_rate": 1.7330647222156662e-05,
      "loss": 0.3418,
      "step": 8521
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40254482272545034,
      "learning_rate": 1.7329972547709178e-05,
      "loss": 0.3026,
      "step": 8522
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36491016406089916,
      "learning_rate": 1.732929780114702e-05,
      "loss": 0.2527,
      "step": 8523
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4382744350706664,
      "learning_rate": 1.7328622982476808e-05,
      "loss": 0.2873,
      "step": 8524
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7733874695448275,
      "learning_rate": 1.7327948091705196e-05,
      "loss": 0.5297,
      "step": 8525
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24889793208878117,
      "learning_rate": 1.7327273128838823e-05,
      "loss": 0.0986,
      "step": 8526
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3755559184341901,
      "learning_rate": 1.732659809388432e-05,
      "loss": 0.2847,
      "step": 8527
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3869240105006348,
      "learning_rate": 1.732592298684834e-05,
      "loss": 0.2338,
      "step": 8528
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4098625575827651,
      "learning_rate": 1.7325247807737513e-05,
      "loss": 0.2815,
      "step": 8529
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2481942770669254,
      "learning_rate": 1.7324572556558492e-05,
      "loss": 0.7124,
      "step": 8530
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6040753371778731,
      "learning_rate": 1.7323897233317916e-05,
      "loss": 0.3864,
      "step": 8531
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7277640784715055,
      "learning_rate": 1.732322183802243e-05,
      "loss": 0.3841,
      "step": 8532
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3785399710807949,
      "learning_rate": 1.732254637067867e-05,
      "loss": 0.2734,
      "step": 8533
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3626342481545226,
      "learning_rate": 1.73218708312933e-05,
      "loss": 0.3128,
      "step": 8534
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.1756461147845684,
      "learning_rate": 1.7321195219872947e-05,
      "loss": 0.0747,
      "step": 8535
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.6410494618454161,
      "learning_rate": 1.732051953642427e-05,
      "loss": 0.8661,
      "step": 8536
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2825121451415918,
      "learning_rate": 1.731984378095392e-05,
      "loss": 0.1977,
      "step": 8537
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5931893189170871,
      "learning_rate": 1.7319167953468527e-05,
      "loss": 0.9086,
      "step": 8538
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3617916080850984,
      "learning_rate": 1.7318492053974757e-05,
      "loss": 0.2491,
      "step": 8539
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7607957462956442,
      "learning_rate": 1.7317816082479253e-05,
      "loss": 0.5194,
      "step": 8540
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.372299572708866,
      "learning_rate": 1.7317140038988666e-05,
      "loss": 0.231,
      "step": 8541
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39354757144286495,
      "learning_rate": 1.731646392350965e-05,
      "loss": 0.2957,
      "step": 8542
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1141785126158312,
      "learning_rate": 1.731578773604885e-05,
      "loss": 0.6473,
      "step": 8543
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2112668252805851,
      "learning_rate": 1.731511147661293e-05,
      "loss": 0.0982,
      "step": 8544
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3415553707999979,
      "learning_rate": 1.731443514520853e-05,
      "loss": 0.3027,
      "step": 8545
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3095692385627292,
      "learning_rate": 1.7313758741842313e-05,
      "loss": 0.2184,
      "step": 8546
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.62239236209798,
      "learning_rate": 1.7313082266520932e-05,
      "loss": 0.8806,
      "step": 8547
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7937316617624731,
      "learning_rate": 1.7312405719251038e-05,
      "loss": 0.438,
      "step": 8548
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8591102286518125,
      "learning_rate": 1.7311729100039292e-05,
      "loss": 0.4661,
      "step": 8549
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.33500221536223396,
      "learning_rate": 1.7311052408892348e-05,
      "loss": 0.1788,
      "step": 8550
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5877928915935795,
      "learning_rate": 1.7310375645816867e-05,
      "loss": 0.3892,
      "step": 8551
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35572145792559323,
      "learning_rate": 1.7309698810819503e-05,
      "loss": 0.276,
      "step": 8552
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4892141928762955,
      "learning_rate": 1.7309021903906917e-05,
      "loss": 0.3055,
      "step": 8553
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2523745017127965,
      "learning_rate": 1.730834492508577e-05,
      "loss": 0.1233,
      "step": 8554
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3723096541278316,
      "learning_rate": 1.7307667874362723e-05,
      "loss": 0.197,
      "step": 8555
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5763526534101613,
      "learning_rate": 1.7306990751744432e-05,
      "loss": 0.4177,
      "step": 8556
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4482254974117129,
      "learning_rate": 1.730631355723756e-05,
      "loss": 0.3092,
      "step": 8557
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0311682178656705,
      "learning_rate": 1.7305636290848774e-05,
      "loss": 0.5636,
      "step": 8558
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.25280213110907,
      "learning_rate": 1.7304958952584734e-05,
      "loss": 0.0754,
      "step": 8559
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4036385552474766,
      "learning_rate": 1.7304281542452104e-05,
      "loss": 0.2816,
      "step": 8560
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.3630734121250945,
      "learning_rate": 1.7303604060457545e-05,
      "loss": 0.2499,
      "step": 8561
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.31272756359055626,
      "learning_rate": 1.7302926506607734e-05,
      "loss": 0.248,
      "step": 8562
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40825391140054723,
      "learning_rate": 1.730224888090932e-05,
      "loss": 0.2577,
      "step": 8563
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41573740289806255,
      "learning_rate": 1.7301571183368987e-05,
      "loss": 0.293,
      "step": 8564
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4209986213262875,
      "learning_rate": 1.730089341399339e-05,
      "loss": 0.3278,
      "step": 8565
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6378923407919264,
      "learning_rate": 1.7300215572789197e-05,
      "loss": 0.4936,
      "step": 8566
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30563146378178807,
      "learning_rate": 1.7299537659763087e-05,
      "loss": 0.1052,
      "step": 8567
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4327496845376529,
      "learning_rate": 1.7298859674921723e-05,
      "loss": 0.0876,
      "step": 8568
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.31195399123289724,
      "learning_rate": 1.7298181618271775e-05,
      "loss": 0.2944,
      "step": 8569
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4216492258985633,
      "learning_rate": 1.7297503489819915e-05,
      "loss": 0.2435,
      "step": 8570
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0287163005059303,
      "learning_rate": 1.7296825289572818e-05,
      "loss": 0.6297,
      "step": 8571
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4827283446705275,
      "learning_rate": 1.7296147017537147e-05,
      "loss": 0.2345,
      "step": 8572
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37388277151336696,
      "learning_rate": 1.7295468673719584e-05,
      "loss": 0.2981,
      "step": 8573
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6672791836656434,
      "learning_rate": 1.7294790258126797e-05,
      "loss": 0.4274,
      "step": 8574
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4391753058360023,
      "learning_rate": 1.7294111770765468e-05,
      "loss": 0.3844,
      "step": 8575
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.27867490873485506,
      "learning_rate": 1.7293433211642264e-05,
      "loss": 0.2288,
      "step": 8576
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.43290893858509977,
      "learning_rate": 1.7292754580763863e-05,
      "loss": 0.2233,
      "step": 8577
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.34087366439735106,
      "learning_rate": 1.7292075878136944e-05,
      "loss": 0.2354,
      "step": 8578
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0206483911332562,
      "learning_rate": 1.7291397103768184e-05,
      "loss": 0.5922,
      "step": 8579
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4129490228572204,
      "learning_rate": 1.7290718257664258e-05,
      "loss": 0.2847,
      "step": 8580
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36839617049466045,
      "learning_rate": 1.729003933983185e-05,
      "loss": 0.2638,
      "step": 8581
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.878103776132351,
      "learning_rate": 1.7289360350277636e-05,
      "loss": 0.438,
      "step": 8582
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36344941077614495,
      "learning_rate": 1.7288681289008293e-05,
      "loss": 0.2984,
      "step": 8583
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7835762549791887,
      "learning_rate": 1.728800215603051e-05,
      "loss": 0.5267,
      "step": 8584
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2814463415539832,
      "learning_rate": 1.7287322951350957e-05,
      "loss": 0.0943,
      "step": 8585
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5218228056694031,
      "learning_rate": 1.7286643674976326e-05,
      "loss": 0.3686,
      "step": 8586
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.28828357502340424,
      "learning_rate": 1.72859643269133e-05,
      "loss": 0.1906,
      "step": 8587
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4969290640440944,
      "learning_rate": 1.7285284907168554e-05,
      "loss": 0.3782,
      "step": 8588
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0469539288585463,
      "learning_rate": 1.7284605415748786e-05,
      "loss": 0.3998,
      "step": 8589
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8812221639689122,
      "learning_rate": 1.728392585266067e-05,
      "loss": 0.635,
      "step": 8590
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3291079037261743,
      "learning_rate": 1.728324621791089e-05,
      "loss": 0.2259,
      "step": 8591
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4378584685599454,
      "learning_rate": 1.728256651150614e-05,
      "loss": 0.3956,
      "step": 8592
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.32045203102830083,
      "learning_rate": 1.7281886733453103e-05,
      "loss": 0.2784,
      "step": 8593
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1292317891990693,
      "learning_rate": 1.7281206883758473e-05,
      "loss": 0.4111,
      "step": 8594
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3450999767395046,
      "learning_rate": 1.7280526962428927e-05,
      "loss": 0.1855,
      "step": 8595
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3090678648075698,
      "learning_rate": 1.727984696947117e-05,
      "loss": 0.2339,
      "step": 8596
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.9234274082715717,
      "learning_rate": 1.7279166904891873e-05,
      "loss": 0.8976,
      "step": 8597
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1269037678474625,
      "learning_rate": 1.7278486768697743e-05,
      "loss": 0.3722,
      "step": 8598
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3358209912406416,
      "learning_rate": 1.7277806560895466e-05,
      "loss": 0.3548,
      "step": 8599
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.30715113723437004,
      "learning_rate": 1.7277126281491727e-05,
      "loss": 0.2338,
      "step": 8600
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4831473037931707,
      "learning_rate": 1.727644593049323e-05,
      "loss": 0.396,
      "step": 8601
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7844583661537661,
      "learning_rate": 1.7275765507906667e-05,
      "loss": 0.4032,
      "step": 8602
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.419622411385066,
      "learning_rate": 1.727508501373872e-05,
      "loss": 0.2352,
      "step": 8603
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.21892578664576406,
      "learning_rate": 1.7274404447996098e-05,
      "loss": 0.1505,
      "step": 8604
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40768053835851875,
      "learning_rate": 1.7273723810685493e-05,
      "loss": 0.3386,
      "step": 8605
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42303966728799425,
      "learning_rate": 1.7273043101813597e-05,
      "loss": 0.2828,
      "step": 8606
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.7485028289198237,
      "learning_rate": 1.727236232138711e-05,
      "loss": 0.3832,
      "step": 8607
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.668921887304968,
      "learning_rate": 1.7271681469412734e-05,
      "loss": 0.4605,
      "step": 8608
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.28309293792895274,
      "learning_rate": 1.7271000545897157e-05,
      "loss": 0.0738,
      "step": 8609
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4147839714261499,
      "learning_rate": 1.7270319550847086e-05,
      "loss": 0.308,
      "step": 8610
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3120945390332602,
      "learning_rate": 1.726963848426922e-05,
      "loss": 0.2674,
      "step": 8611
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.2351251601352857,
      "learning_rate": 1.726895734617026e-05,
      "loss": 0.8085,
      "step": 8612
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.20153765030464912,
      "learning_rate": 1.72682761365569e-05,
      "loss": 0.0704,
      "step": 8613
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.42284335517671573,
      "learning_rate": 1.7267594855435854e-05,
      "loss": 0.2883,
      "step": 8614
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4914705241177192,
      "learning_rate": 1.7266913502813816e-05,
      "loss": 0.2497,
      "step": 8615
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.561101560837514,
      "learning_rate": 1.726623207869749e-05,
      "loss": 0.4044,
      "step": 8616
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39284166409882476,
      "learning_rate": 1.7265550583093587e-05,
      "loss": 0.2758,
      "step": 8617
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5282791895973432,
      "learning_rate": 1.7264869016008804e-05,
      "loss": 0.3649,
      "step": 8618
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4119221408329162,
      "learning_rate": 1.726418737744985e-05,
      "loss": 0.266,
      "step": 8619
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1028677967596427,
      "learning_rate": 1.7263505667423424e-05,
      "loss": 0.5958,
      "step": 8620
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.817734020081985,
      "learning_rate": 1.726282388593625e-05,
      "loss": 0.9055,
      "step": 8621
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3073377137637675,
      "learning_rate": 1.7262142032995017e-05,
      "loss": 0.2521,
      "step": 8622
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.37831887365173794,
      "learning_rate": 1.726146010860644e-05,
      "loss": 0.295,
      "step": 8623
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.24194660119132827,
      "learning_rate": 1.7260778112777235e-05,
      "loss": 0.0964,
      "step": 8624
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.8264472887914098,
      "learning_rate": 1.7260096045514102e-05,
      "loss": 0.5902,
      "step": 8625
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6452346912729241,
      "learning_rate": 1.7259413906823758e-05,
      "loss": 0.405,
      "step": 8626
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6071588809816537,
      "learning_rate": 1.725873169671291e-05,
      "loss": 0.3277,
      "step": 8627
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.374356280031809,
      "learning_rate": 1.725804941518827e-05,
      "loss": 0.2528,
      "step": 8628
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39048562273734677,
      "learning_rate": 1.7257367062256554e-05,
      "loss": 0.3301,
      "step": 8629
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.139921392703104,
      "learning_rate": 1.725668463792447e-05,
      "loss": 0.4767,
      "step": 8630
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.34532967981746804,
      "learning_rate": 1.7256002142198737e-05,
      "loss": 0.2354,
      "step": 8631
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3089925886824362,
      "learning_rate": 1.7255319575086066e-05,
      "loss": 0.1754,
      "step": 8632
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.6129173917844324,
      "learning_rate": 1.725463693659317e-05,
      "loss": 0.4035,
      "step": 8633
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.504641987344786,
      "learning_rate": 1.7253954226726778e-05,
      "loss": 0.3366,
      "step": 8634
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.39900884963689165,
      "learning_rate": 1.725327144549359e-05,
      "loss": 0.2934,
      "step": 8635
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0386772223178886,
      "learning_rate": 1.725258859290033e-05,
      "loss": 0.5241,
      "step": 8636
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.35022522324708477,
      "learning_rate": 1.7251905668953727e-05,
      "loss": 0.2304,
      "step": 8637
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.5296647881137888,
      "learning_rate": 1.725122267366048e-05,
      "loss": 0.9549,
      "step": 8638
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.0953195791894925,
      "learning_rate": 1.725053960702732e-05,
      "loss": 0.347,
      "step": 8639
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5069442955615525,
      "learning_rate": 1.724985646906097e-05,
      "loss": 0.3594,
      "step": 8640
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.2809632120283642,
      "learning_rate": 1.7249173259768142e-05,
      "loss": 0.2169,
      "step": 8641
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.33086107580695445,
      "learning_rate": 1.724848997915556e-05,
      "loss": 0.2868,
      "step": 8642
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.438942911611652,
      "learning_rate": 1.7247806627229958e-05,
      "loss": 0.2743,
      "step": 8643
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.957374172833417,
      "learning_rate": 1.7247123203998045e-05,
      "loss": 0.8272,
      "step": 8644
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.4773168690177139,
      "learning_rate": 1.7246439709466548e-05,
      "loss": 0.2203,
      "step": 8645
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.40747949108902337,
      "learning_rate": 1.7245756143642188e-05,
      "loss": 0.2993,
      "step": 8646
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.36485137980725957,
      "learning_rate": 1.7245072506531705e-05,
      "loss": 0.273,
      "step": 8647
    },
    {
      "epoch": 0.26,
      "grad_norm": 1.1795209950375598,
      "learning_rate": 1.7244388798141807e-05,
      "loss": 0.3196,
      "step": 8648
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.45938028439582185,
      "learning_rate": 1.7243705018479233e-05,
      "loss": 0.2753,
      "step": 8649
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3057926788551393,
      "learning_rate": 1.72430211675507e-05,
      "loss": 0.1966,
      "step": 8650
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.5328476365541738,
      "learning_rate": 1.7242337245362945e-05,
      "loss": 0.4501,
      "step": 8651
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.41223750057729786,
      "learning_rate": 1.7241653251922695e-05,
      "loss": 0.3002,
      "step": 8652
    },
    {
      "epoch": 0.26,
      "grad_norm": 0.3321840578542064,
      "learning_rate": 1.7240969187236672e-05,
      "loss": 0.2675,
      "step": 8653
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4051160945219435,
      "learning_rate": 1.7240285051311617e-05,
      "loss": 0.1813,
      "step": 8654
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37782053023928286,
      "learning_rate": 1.7239600844154255e-05,
      "loss": 0.3232,
      "step": 8655
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0819783949334163,
      "learning_rate": 1.7238916565771316e-05,
      "loss": 0.488,
      "step": 8656
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.014798626087744,
      "learning_rate": 1.723823221616953e-05,
      "loss": 0.941,
      "step": 8657
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35226742943228084,
      "learning_rate": 1.723754779535564e-05,
      "loss": 0.2729,
      "step": 8658
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7687401963182657,
      "learning_rate": 1.7236863303336373e-05,
      "loss": 0.3295,
      "step": 8659
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3902511661671176,
      "learning_rate": 1.7236178740118462e-05,
      "loss": 0.2594,
      "step": 8660
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3589183313837415,
      "learning_rate": 1.7235494105708643e-05,
      "loss": 0.2418,
      "step": 8661
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48934013429662004,
      "learning_rate": 1.7234809400113652e-05,
      "loss": 0.2833,
      "step": 8662
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4618191276919265,
      "learning_rate": 1.7234124623340223e-05,
      "loss": 0.2784,
      "step": 8663
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4395838636122838,
      "learning_rate": 1.72334397753951e-05,
      "loss": 0.2839,
      "step": 8664
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36579170734957045,
      "learning_rate": 1.7232754856285017e-05,
      "loss": 0.2771,
      "step": 8665
    },
    {
      "epoch": 0.27,
      "grad_norm": 2.290081401525725,
      "learning_rate": 1.723206986601671e-05,
      "loss": 1.0519,
      "step": 8666
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7271232824808009,
      "learning_rate": 1.7231384804596917e-05,
      "loss": 0.3635,
      "step": 8667
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6949408003475508,
      "learning_rate": 1.7230699672032384e-05,
      "loss": 0.3831,
      "step": 8668
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42299872302347175,
      "learning_rate": 1.723001446832985e-05,
      "loss": 0.2698,
      "step": 8669
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5071569024548175,
      "learning_rate": 1.722932919349605e-05,
      "loss": 0.3211,
      "step": 8670
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2916491181569651,
      "learning_rate": 1.7228643847537732e-05,
      "loss": 0.0739,
      "step": 8671
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5451128814585783,
      "learning_rate": 1.7227958430461643e-05,
      "loss": 0.2845,
      "step": 8672
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2670922380031869,
      "learning_rate": 1.7227272942274515e-05,
      "loss": 0.1861,
      "step": 8673
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8299259087776525,
      "learning_rate": 1.7226587382983097e-05,
      "loss": 0.4646,
      "step": 8674
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.4321636185808488,
      "learning_rate": 1.7225901752594137e-05,
      "loss": 0.6739,
      "step": 8675
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3294315081997109,
      "learning_rate": 1.722521605111438e-05,
      "loss": 0.2848,
      "step": 8676
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5889648410378181,
      "learning_rate": 1.7224530278550565e-05,
      "loss": 0.38,
      "step": 8677
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3475378181285287,
      "learning_rate": 1.7223844434909446e-05,
      "loss": 0.2684,
      "step": 8678
    },
    {
      "epoch": 0.27,
      "grad_norm": 3.058739901807539,
      "learning_rate": 1.722315852019777e-05,
      "loss": 0.8215,
      "step": 8679
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32902436306454197,
      "learning_rate": 1.7222472534422282e-05,
      "loss": 0.0741,
      "step": 8680
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0414505674899386,
      "learning_rate": 1.7221786477589732e-05,
      "loss": 0.5831,
      "step": 8681
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3106425892739687,
      "learning_rate": 1.7221100349706873e-05,
      "loss": 0.2369,
      "step": 8682
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4823172190647378,
      "learning_rate": 1.722041415078045e-05,
      "loss": 0.3261,
      "step": 8683
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8342283738008804,
      "learning_rate": 1.721972788081722e-05,
      "loss": 0.4205,
      "step": 8684
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6498352181400318,
      "learning_rate": 1.721904153982393e-05,
      "loss": 0.4753,
      "step": 8685
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39290723436206587,
      "learning_rate": 1.7218355127807332e-05,
      "loss": 0.1437,
      "step": 8686
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.439045419837592,
      "learning_rate": 1.7217668644774184e-05,
      "loss": 0.2495,
      "step": 8687
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3899064879360479,
      "learning_rate": 1.7216982090731234e-05,
      "loss": 0.3282,
      "step": 8688
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.1999929653184493,
      "learning_rate": 1.721629546568524e-05,
      "loss": 0.0738,
      "step": 8689
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.546560859266331,
      "learning_rate": 1.721560876964296e-05,
      "loss": 0.9334,
      "step": 8690
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32164318572438927,
      "learning_rate": 1.7214922002611145e-05,
      "loss": 0.1744,
      "step": 8691
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5343323754685556,
      "learning_rate": 1.7214235164596555e-05,
      "loss": 0.372,
      "step": 8692
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7142430796982623,
      "learning_rate": 1.721354825560594e-05,
      "loss": 0.3735,
      "step": 8693
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4924649786602037,
      "learning_rate": 1.7212861275646065e-05,
      "loss": 0.416,
      "step": 8694
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48482841780961483,
      "learning_rate": 1.7212174224723692e-05,
      "loss": 0.1783,
      "step": 8695
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36313367532099283,
      "learning_rate": 1.7211487102845576e-05,
      "loss": 0.274,
      "step": 8696
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8729625080527772,
      "learning_rate": 1.721079991001847e-05,
      "loss": 0.2941,
      "step": 8697
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5045585378688455,
      "learning_rate": 1.721011264624915e-05,
      "loss": 0.2724,
      "step": 8698
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.25273139679668055,
      "learning_rate": 1.7209425311544363e-05,
      "loss": 0.1835,
      "step": 8699
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.289103066040285,
      "learning_rate": 1.7208737905910883e-05,
      "loss": 0.2085,
      "step": 8700
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4742061715853578,
      "learning_rate": 1.7208050429355463e-05,
      "loss": 0.4011,
      "step": 8701
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5934119645634155,
      "learning_rate": 1.720736288188487e-05,
      "loss": 0.395,
      "step": 8702
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6326238399359593,
      "learning_rate": 1.7206675263505876e-05,
      "loss": 0.5057,
      "step": 8703
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3846067800376684,
      "learning_rate": 1.7205987574225233e-05,
      "loss": 0.1408,
      "step": 8704
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42154103241722085,
      "learning_rate": 1.7205299814049715e-05,
      "loss": 0.3413,
      "step": 8705
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31857851451652947,
      "learning_rate": 1.720461198298609e-05,
      "loss": 0.2516,
      "step": 8706
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9717562209581193,
      "learning_rate": 1.720392408104112e-05,
      "loss": 0.7092,
      "step": 8707
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31908283986949176,
      "learning_rate": 1.720323610822157e-05,
      "loss": 0.1681,
      "step": 8708
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41449724765208057,
      "learning_rate": 1.720254806453422e-05,
      "loss": 0.2825,
      "step": 8709
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.485398016744125,
      "learning_rate": 1.7201859949985828e-05,
      "loss": 0.2873,
      "step": 8710
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7135729034189184,
      "learning_rate": 1.7201171764583168e-05,
      "loss": 0.5103,
      "step": 8711
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2387940187712916,
      "learning_rate": 1.720048350833301e-05,
      "loss": 0.2096,
      "step": 8712
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39305496316023725,
      "learning_rate": 1.719979518124213e-05,
      "loss": 0.139,
      "step": 8713
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42765307775983147,
      "learning_rate": 1.7199106783317292e-05,
      "loss": 0.3549,
      "step": 8714
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1063771556682087,
      "learning_rate": 1.7198418314565276e-05,
      "loss": 0.0755,
      "step": 8715
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5015092659299711,
      "learning_rate": 1.7197729774992848e-05,
      "loss": 0.279,
      "step": 8716
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31598373825298726,
      "learning_rate": 1.7197041164606786e-05,
      "loss": 0.2652,
      "step": 8717
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6948666132128193,
      "learning_rate": 1.719635248341387e-05,
      "loss": 0.4225,
      "step": 8718
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3714620117761926,
      "learning_rate": 1.7195663731420865e-05,
      "loss": 0.2851,
      "step": 8719
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8785591032219867,
      "learning_rate": 1.719497490863456e-05,
      "loss": 0.5386,
      "step": 8720
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3060256209713702,
      "learning_rate": 1.7194286015061717e-05,
      "loss": 0.1458,
      "step": 8721
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3250697204615862,
      "learning_rate": 1.7193597050709124e-05,
      "loss": 0.7343,
      "step": 8722
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31651842667885177,
      "learning_rate": 1.7192908015583554e-05,
      "loss": 0.2079,
      "step": 8723
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3472229934773599,
      "learning_rate": 1.719221890969179e-05,
      "loss": 0.3474,
      "step": 8724
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8691769934053599,
      "learning_rate": 1.7191529733040607e-05,
      "loss": 0.2752,
      "step": 8725
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8222035463246348,
      "learning_rate": 1.7190840485636793e-05,
      "loss": 0.4156,
      "step": 8726
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5003309020992465,
      "learning_rate": 1.719015116748712e-05,
      "loss": 0.364,
      "step": 8727
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3139537675891904,
      "learning_rate": 1.718946177859838e-05,
      "loss": 0.2848,
      "step": 8728
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7359159074116686,
      "learning_rate": 1.7188772318977345e-05,
      "loss": 0.9152,
      "step": 8729
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23079233547982983,
      "learning_rate": 1.71880827886308e-05,
      "loss": 0.1977,
      "step": 8730
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3722843208929422,
      "learning_rate": 1.7187393187565534e-05,
      "loss": 0.2326,
      "step": 8731
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32082977033026827,
      "learning_rate": 1.7186703515788327e-05,
      "loss": 0.219,
      "step": 8732
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8531607584196876,
      "learning_rate": 1.7186013773305973e-05,
      "loss": 0.8828,
      "step": 8733
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9650028678464189,
      "learning_rate": 1.7185323960125245e-05,
      "loss": 0.4241,
      "step": 8734
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.34802022846818675,
      "learning_rate": 1.7184634076252936e-05,
      "loss": 0.3276,
      "step": 8735
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3453559515892918,
      "learning_rate": 1.7183944121695833e-05,
      "loss": 0.2267,
      "step": 8736
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5629684112463756,
      "learning_rate": 1.7183254096460725e-05,
      "loss": 0.3858,
      "step": 8737
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7617560894378654,
      "learning_rate": 1.71825640005544e-05,
      "loss": 0.4173,
      "step": 8738
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.18737800449228456,
      "learning_rate": 1.718187383398364e-05,
      "loss": 0.0934,
      "step": 8739
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9316025112216813,
      "learning_rate": 1.718118359675525e-05,
      "loss": 0.3862,
      "step": 8740
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38926949177069825,
      "learning_rate": 1.7180493288876013e-05,
      "loss": 0.1846,
      "step": 8741
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3226950300871564,
      "learning_rate": 1.717980291035272e-05,
      "loss": 0.284,
      "step": 8742
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7637259214142917,
      "learning_rate": 1.717911246119216e-05,
      "loss": 0.382,
      "step": 8743
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7276189752369912,
      "learning_rate": 1.717842194140113e-05,
      "loss": 0.4906,
      "step": 8744
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3185327178628847,
      "learning_rate": 1.7177731350986423e-05,
      "loss": 0.0753,
      "step": 8745
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3848441822924753,
      "learning_rate": 1.7177040689954835e-05,
      "loss": 0.2929,
      "step": 8746
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32328764722556735,
      "learning_rate": 1.717634995831316e-05,
      "loss": 0.2722,
      "step": 8747
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.8047613416150676,
      "learning_rate": 1.717565915606819e-05,
      "loss": 0.9096,
      "step": 8748
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23148274002617578,
      "learning_rate": 1.7174968283226726e-05,
      "loss": 0.0999,
      "step": 8749
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42550145349101487,
      "learning_rate": 1.717427733979556e-05,
      "loss": 0.2623,
      "step": 8750
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4501479046908317,
      "learning_rate": 1.7173586325781495e-05,
      "loss": 0.2482,
      "step": 8751
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1248829506473967,
      "learning_rate": 1.7172895241191326e-05,
      "loss": 0.3978,
      "step": 8752
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36778005390031293,
      "learning_rate": 1.7172204086031858e-05,
      "loss": 0.3183,
      "step": 8753
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5107516051070032,
      "learning_rate": 1.7171512860309883e-05,
      "loss": 0.273,
      "step": 8754
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4064261671916408,
      "learning_rate": 1.7170821564032205e-05,
      "loss": 0.3224,
      "step": 8755
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9529287627512949,
      "learning_rate": 1.7170130197205623e-05,
      "loss": 0.5076,
      "step": 8756
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0882547368157034,
      "learning_rate": 1.7169438759836946e-05,
      "loss": 0.7206,
      "step": 8757
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.24534180196252486,
      "learning_rate": 1.7168747251932964e-05,
      "loss": 0.0764,
      "step": 8758
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3901578297028434,
      "learning_rate": 1.7168055673500494e-05,
      "loss": 0.3075,
      "step": 8759
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3548785480380522,
      "learning_rate": 1.7167364024546333e-05,
      "loss": 0.2526,
      "step": 8760
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.296887743220097,
      "learning_rate": 1.7166672305077283e-05,
      "loss": 0.5727,
      "step": 8761
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7990997288790752,
      "learning_rate": 1.7165980515100153e-05,
      "loss": 0.3835,
      "step": 8762
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9119011515947663,
      "learning_rate": 1.7165288654621753e-05,
      "loss": 0.3394,
      "step": 8763
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37467980409553503,
      "learning_rate": 1.7164596723648884e-05,
      "loss": 0.2408,
      "step": 8764
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.43860350173054685,
      "learning_rate": 1.7163904722188352e-05,
      "loss": 0.2959,
      "step": 8765
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4380454717375551,
      "learning_rate": 1.716321265024697e-05,
      "loss": 0.3329,
      "step": 8766
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21649752786959214,
      "learning_rate": 1.7162520507831544e-05,
      "loss": 0.0953,
      "step": 8767
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1132691800236645,
      "learning_rate": 1.7161828294948885e-05,
      "loss": 0.4803,
      "step": 8768
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35291985215626254,
      "learning_rate": 1.7161136011605804e-05,
      "loss": 0.2858,
      "step": 8769
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7140108760095754,
      "learning_rate": 1.716044365780911e-05,
      "loss": 0.5256,
      "step": 8770
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3640675027552676,
      "learning_rate": 1.7159751233565612e-05,
      "loss": 0.2693,
      "step": 8771
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5903351287401857,
      "learning_rate": 1.7159058738882132e-05,
      "loss": 0.8147,
      "step": 8772
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3038711434264899,
      "learning_rate": 1.7158366173765474e-05,
      "loss": 0.1865,
      "step": 8773
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3997628756660287,
      "learning_rate": 1.715767353822245e-05,
      "loss": 0.9594,
      "step": 8774
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9121165385192712,
      "learning_rate": 1.7156980832259883e-05,
      "loss": 0.2857,
      "step": 8775
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3479137808126192,
      "learning_rate": 1.715628805588458e-05,
      "loss": 0.1995,
      "step": 8776
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.27778928201412484,
      "learning_rate": 1.7155595209103363e-05,
      "loss": 0.2011,
      "step": 8777
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3454879806708938,
      "learning_rate": 1.7154902291923047e-05,
      "loss": 0.2911,
      "step": 8778
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6913828628218168,
      "learning_rate": 1.7154209304350448e-05,
      "loss": 0.4895,
      "step": 8779
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6345473300098701,
      "learning_rate": 1.715351624639238e-05,
      "loss": 0.3731,
      "step": 8780
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8429591034269761,
      "learning_rate": 1.7152823118055667e-05,
      "loss": 0.4362,
      "step": 8781
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32708242960115236,
      "learning_rate": 1.715212991934713e-05,
      "loss": 0.2224,
      "step": 8782
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5346671807049763,
      "learning_rate": 1.7151436650273582e-05,
      "loss": 0.3777,
      "step": 8783
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4252794147359707,
      "learning_rate": 1.7150743310841846e-05,
      "loss": 0.2753,
      "step": 8784
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4587954585196293,
      "learning_rate": 1.7150049901058746e-05,
      "loss": 0.2535,
      "step": 8785
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3681462801843087,
      "learning_rate": 1.7149356420931104e-05,
      "loss": 0.2246,
      "step": 8786
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.575741496809222,
      "learning_rate": 1.7148662870465738e-05,
      "loss": 0.39,
      "step": 8787
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4433558595288412,
      "learning_rate": 1.7147969249669478e-05,
      "loss": 0.2788,
      "step": 8788
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3413269976188689,
      "learning_rate": 1.714727555854914e-05,
      "loss": 0.3037,
      "step": 8789
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.364535629664709,
      "learning_rate": 1.7146581797111554e-05,
      "loss": 0.1804,
      "step": 8790
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3674893602920219,
      "learning_rate": 1.7145887965363547e-05,
      "loss": 0.24,
      "step": 8791
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3802588816294998,
      "learning_rate": 1.7145194063311942e-05,
      "loss": 0.8294,
      "step": 8792
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3463041904635422,
      "learning_rate": 1.714450009096357e-05,
      "loss": 0.0813,
      "step": 8793
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35989924030862,
      "learning_rate": 1.7143806048325252e-05,
      "loss": 0.3471,
      "step": 8794
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3578329348042829,
      "learning_rate": 1.714311193540382e-05,
      "loss": 0.079,
      "step": 8795
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3660678288725244,
      "learning_rate": 1.71424177522061e-05,
      "loss": 0.3342,
      "step": 8796
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.309277212127784,
      "learning_rate": 1.7141723498738927e-05,
      "loss": 0.1744,
      "step": 8797
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5793907848698376,
      "learning_rate": 1.7141029175009126e-05,
      "loss": 0.8283,
      "step": 8798
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.941065368015206,
      "learning_rate": 1.7140334781023536e-05,
      "loss": 0.5051,
      "step": 8799
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.39774193076929953,
      "learning_rate": 1.7139640316788978e-05,
      "loss": 0.2771,
      "step": 8800
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30125178704985733,
      "learning_rate": 1.713894578231229e-05,
      "loss": 0.2568,
      "step": 8801
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1158218534355278,
      "learning_rate": 1.7138251177600303e-05,
      "loss": 0.7079,
      "step": 8802
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.759614533517773,
      "learning_rate": 1.7137556502659854e-05,
      "loss": 0.3653,
      "step": 8803
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3880955378954337,
      "learning_rate": 1.7136861757497775e-05,
      "loss": 0.2648,
      "step": 8804
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3783593278668611,
      "learning_rate": 1.71361669421209e-05,
      "loss": 0.328,
      "step": 8805
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.19341484999379724,
      "learning_rate": 1.713547205653607e-05,
      "loss": 0.0904,
      "step": 8806
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.34816018305886165,
      "learning_rate": 1.7134777100750115e-05,
      "loss": 0.3313,
      "step": 8807
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5283550683528053,
      "learning_rate": 1.713408207476988e-05,
      "loss": 0.2886,
      "step": 8808
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3695950021519774,
      "learning_rate": 1.7133386978602195e-05,
      "loss": 0.2885,
      "step": 8809
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8388305339225327,
      "learning_rate": 1.7132691812253898e-05,
      "loss": 0.4611,
      "step": 8810
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.302487318425445,
      "learning_rate": 1.713199657573184e-05,
      "loss": 0.6942,
      "step": 8811
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.29969407875949544,
      "learning_rate": 1.7131301269042847e-05,
      "loss": 0.2729,
      "step": 8812
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5709672443965247,
      "learning_rate": 1.7130605892193764e-05,
      "loss": 0.388,
      "step": 8813
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3795253463294947,
      "learning_rate": 1.7129910445191438e-05,
      "loss": 0.2419,
      "step": 8814
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9221980063277515,
      "learning_rate": 1.7129214928042704e-05,
      "loss": 0.4841,
      "step": 8815
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3058274225746801,
      "learning_rate": 1.712851934075441e-05,
      "loss": 0.1137,
      "step": 8816
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5046346422804332,
      "learning_rate": 1.7127823683333397e-05,
      "loss": 0.2825,
      "step": 8817
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9956110901840175,
      "learning_rate": 1.712712795578651e-05,
      "loss": 0.493,
      "step": 8818
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2576219825278808,
      "learning_rate": 1.712643215812059e-05,
      "loss": 0.2474,
      "step": 8819
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8173382405571782,
      "learning_rate": 1.712573629034249e-05,
      "loss": 0.5304,
      "step": 8820
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7833714974240721,
      "learning_rate": 1.712504035245905e-05,
      "loss": 0.3913,
      "step": 8821
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9609388429192647,
      "learning_rate": 1.7124344344477117e-05,
      "loss": 0.4642,
      "step": 8822
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31647230537503945,
      "learning_rate": 1.7123648266403542e-05,
      "loss": 0.2067,
      "step": 8823
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5649074654843508,
      "learning_rate": 1.712295211824517e-05,
      "loss": 0.3776,
      "step": 8824
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.240906096490145,
      "learning_rate": 1.712225590000885e-05,
      "loss": 0.1994,
      "step": 8825
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.5804303935025896,
      "learning_rate": 1.7121559611701438e-05,
      "loss": 0.8751,
      "step": 8826
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3371970929917636,
      "learning_rate": 1.7120863253329773e-05,
      "loss": 0.2006,
      "step": 8827
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6015689514943885,
      "learning_rate": 1.7120166824900714e-05,
      "loss": 0.3628,
      "step": 8828
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9363831631752851,
      "learning_rate": 1.7119470326421113e-05,
      "loss": 0.444,
      "step": 8829
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5241989427524902,
      "learning_rate": 1.711877375789782e-05,
      "loss": 0.384,
      "step": 8830
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3555869210581336,
      "learning_rate": 1.7118077119337688e-05,
      "loss": 0.2694,
      "step": 8831
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2883159176975463,
      "learning_rate": 1.711738041074757e-05,
      "loss": 0.2229,
      "step": 8832
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.235816246458999,
      "learning_rate": 1.7116683632134324e-05,
      "loss": 0.6926,
      "step": 8833
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.22858501307074905,
      "learning_rate": 1.71159867835048e-05,
      "loss": 0.0981,
      "step": 8834
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2927162062172246,
      "learning_rate": 1.7115289864865856e-05,
      "loss": 0.8908,
      "step": 8835
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2866835621357484,
      "learning_rate": 1.711459287622435e-05,
      "loss": 0.223,
      "step": 8836
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3519534317038587,
      "learning_rate": 1.711389581758714e-05,
      "loss": 0.3095,
      "step": 8837
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.323920244872995,
      "learning_rate": 1.711319868896108e-05,
      "loss": 0.2543,
      "step": 8838
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6150729781813136,
      "learning_rate": 1.7112501490353037e-05,
      "loss": 0.52,
      "step": 8839
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45020561942336784,
      "learning_rate": 1.7111804221769857e-05,
      "loss": 0.192,
      "step": 8840
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42321226377306376,
      "learning_rate": 1.7111106883218407e-05,
      "loss": 0.3264,
      "step": 8841
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3339692480934886,
      "learning_rate": 1.711040947470555e-05,
      "loss": 0.3143,
      "step": 8842
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2709434129772469,
      "learning_rate": 1.7109711996238143e-05,
      "loss": 0.2754,
      "step": 8843
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8388009446216262,
      "learning_rate": 1.710901444782305e-05,
      "loss": 0.6324,
      "step": 8844
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36514974129745026,
      "learning_rate": 1.7108316829467138e-05,
      "loss": 0.0776,
      "step": 8845
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3708906679255709,
      "learning_rate": 1.7107619141177262e-05,
      "loss": 0.3257,
      "step": 8846
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3714424119219551,
      "learning_rate": 1.7106921382960293e-05,
      "loss": 0.251,
      "step": 8847
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3726719030160798,
      "learning_rate": 1.710622355482309e-05,
      "loss": 0.3321,
      "step": 8848
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0744028683052116,
      "learning_rate": 1.7105525656772524e-05,
      "loss": 0.0867,
      "step": 8849
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.41616070403725386,
      "learning_rate": 1.7104827688815456e-05,
      "loss": 0.2902,
      "step": 8850
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4345730256586618,
      "learning_rate": 1.7104129650958756e-05,
      "loss": 0.2691,
      "step": 8851
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.3064822291270362,
      "learning_rate": 1.7103431543209294e-05,
      "loss": 0.7962,
      "step": 8852
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.032448686990701,
      "learning_rate": 1.7102733365573933e-05,
      "loss": 0.5217,
      "step": 8853
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4058211363355786,
      "learning_rate": 1.7102035118059543e-05,
      "loss": 0.2934,
      "step": 8854
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.28488147651458806,
      "learning_rate": 1.7101336800672995e-05,
      "loss": 0.2576,
      "step": 8855
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6675318297389127,
      "learning_rate": 1.710063841342116e-05,
      "loss": 0.4587,
      "step": 8856
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.26636527795385995,
      "learning_rate": 1.7099939956310906e-05,
      "loss": 0.0717,
      "step": 8857
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5925323669807387,
      "learning_rate": 1.7099241429349108e-05,
      "loss": 0.1805,
      "step": 8858
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3628303962312373,
      "learning_rate": 1.7098542832542637e-05,
      "loss": 0.2813,
      "step": 8859
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.29274447166980905,
      "learning_rate": 1.7097844165898367e-05,
      "loss": 0.2584,
      "step": 8860
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.7207239384082436,
      "learning_rate": 1.709714542942317e-05,
      "loss": 0.8837,
      "step": 8861
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5901748759328577,
      "learning_rate": 1.7096446623123918e-05,
      "loss": 0.4494,
      "step": 8862
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6119432051758398,
      "learning_rate": 1.709574774700749e-05,
      "loss": 0.3836,
      "step": 8863
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35670798268496834,
      "learning_rate": 1.7095048801080762e-05,
      "loss": 0.2681,
      "step": 8864
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8377146929411358,
      "learning_rate": 1.709434978535061e-05,
      "loss": 0.6133,
      "step": 8865
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21331212388873122,
      "learning_rate": 1.7093650699823912e-05,
      "loss": 0.1742,
      "step": 8866
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.007433452209179,
      "learning_rate": 1.709295154450754e-05,
      "loss": 0.5946,
      "step": 8867
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3418399194886049,
      "learning_rate": 1.7092252319408377e-05,
      "loss": 0.1876,
      "step": 8868
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5995911749773369,
      "learning_rate": 1.7091553024533305e-05,
      "loss": 0.4142,
      "step": 8869
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9269730811113562,
      "learning_rate": 1.70908536598892e-05,
      "loss": 0.4884,
      "step": 8870
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.399644407504227,
      "learning_rate": 1.7090154225482944e-05,
      "loss": 0.2898,
      "step": 8871
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6784026052313687,
      "learning_rate": 1.7089454721321417e-05,
      "loss": 0.2891,
      "step": 8872
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3583089164782006,
      "learning_rate": 1.7088755147411504e-05,
      "loss": 0.257,
      "step": 8873
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8819735364795619,
      "learning_rate": 1.7088055503760082e-05,
      "loss": 0.558,
      "step": 8874
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21008990202380654,
      "learning_rate": 1.7087355790374038e-05,
      "loss": 0.1085,
      "step": 8875
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.900724992189026,
      "learning_rate": 1.7086656007260255e-05,
      "loss": 0.9122,
      "step": 8876
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30658879584656495,
      "learning_rate": 1.708595615442562e-05,
      "loss": 0.1733,
      "step": 8877
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.37283572746157617,
      "learning_rate": 1.7085256231877018e-05,
      "loss": 0.3118,
      "step": 8878
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.45512406488028945,
      "learning_rate": 1.7084556239621333e-05,
      "loss": 0.2867,
      "step": 8879
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.92151149875122,
      "learning_rate": 1.708385617766545e-05,
      "loss": 0.5603,
      "step": 8880
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5014209637417725,
      "learning_rate": 1.7083156046016263e-05,
      "loss": 0.3276,
      "step": 8881
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36119384814966343,
      "learning_rate": 1.7082455844680653e-05,
      "loss": 0.3034,
      "step": 8882
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0545348056965833,
      "learning_rate": 1.7081755573665515e-05,
      "loss": 0.4677,
      "step": 8883
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.29934178304288617,
      "learning_rate": 1.7081055232977735e-05,
      "loss": 0.2475,
      "step": 8884
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32633015114500236,
      "learning_rate": 1.70803548226242e-05,
      "loss": 0.2183,
      "step": 8885
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2623843391093502,
      "learning_rate": 1.7079654342611806e-05,
      "loss": 0.188,
      "step": 8886
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9165898067047151,
      "learning_rate": 1.7078953792947446e-05,
      "loss": 0.6004,
      "step": 8887
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0649038796125303,
      "learning_rate": 1.7078253173638004e-05,
      "loss": 0.4035,
      "step": 8888
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7142351093130834,
      "learning_rate": 1.7077552484690385e-05,
      "loss": 0.5073,
      "step": 8889
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2864258092431856,
      "learning_rate": 1.707685172611147e-05,
      "loss": 0.2303,
      "step": 8890
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4082610478416349,
      "learning_rate": 1.7076150897908163e-05,
      "loss": 0.3311,
      "step": 8891
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0295752797023376,
      "learning_rate": 1.7075450000087352e-05,
      "loss": 0.5265,
      "step": 8892
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5089497875053023,
      "learning_rate": 1.707474903265594e-05,
      "loss": 0.3186,
      "step": 8893
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.21637991765816394,
      "learning_rate": 1.7074047995620813e-05,
      "loss": 0.0713,
      "step": 8894
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.40485525924085475,
      "learning_rate": 1.7073346888988878e-05,
      "loss": 0.3023,
      "step": 8895
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3303219987508055,
      "learning_rate": 1.707264571276703e-05,
      "loss": 0.2413,
      "step": 8896
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5306090947117987,
      "learning_rate": 1.7071944466962164e-05,
      "loss": 0.3638,
      "step": 8897
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8819541424755888,
      "learning_rate": 1.7071243151581182e-05,
      "loss": 0.6204,
      "step": 8898
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.34298032037614923,
      "learning_rate": 1.7070541766630985e-05,
      "loss": 0.1628,
      "step": 8899
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3215205711067629,
      "learning_rate": 1.7069840312118468e-05,
      "loss": 0.2477,
      "step": 8900
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8875675305197678,
      "learning_rate": 1.706913878805054e-05,
      "loss": 0.4714,
      "step": 8901
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.339338360088818,
      "learning_rate": 1.7068437194434094e-05,
      "loss": 0.318,
      "step": 8902
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.17654161161031112,
      "learning_rate": 1.706773553127604e-05,
      "loss": 0.0753,
      "step": 8903
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6273120505686675,
      "learning_rate": 1.7067033798583277e-05,
      "loss": 0.3558,
      "step": 8904
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.35991782968409897,
      "learning_rate": 1.706633199636271e-05,
      "loss": 0.2676,
      "step": 8905
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6587487530053876,
      "learning_rate": 1.7065630124621244e-05,
      "loss": 0.5357,
      "step": 8906
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6746072172441189,
      "learning_rate": 1.706492818336578e-05,
      "loss": 0.4307,
      "step": 8907
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.34254004461440596,
      "learning_rate": 1.7064226172603237e-05,
      "loss": 0.334,
      "step": 8908
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3021360998630366,
      "learning_rate": 1.7063524092340505e-05,
      "loss": 0.2022,
      "step": 8909
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8760177676337199,
      "learning_rate": 1.70628219425845e-05,
      "loss": 0.4786,
      "step": 8910
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6911207019400087,
      "learning_rate": 1.706211972334213e-05,
      "loss": 0.7809,
      "step": 8911
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23247768822475373,
      "learning_rate": 1.7061417434620305e-05,
      "loss": 0.1175,
      "step": 8912
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.38650009807601776,
      "learning_rate": 1.7060715076425928e-05,
      "loss": 0.2891,
      "step": 8913
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2767403388605953,
      "learning_rate": 1.7060012648765915e-05,
      "loss": 0.2651,
      "step": 8914
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8073908798815401,
      "learning_rate": 1.7059310151647173e-05,
      "loss": 0.5317,
      "step": 8915
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2519854442152545,
      "learning_rate": 1.7058607585076613e-05,
      "loss": 0.4059,
      "step": 8916
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9254947356779469,
      "learning_rate": 1.705790494906115e-05,
      "loss": 0.568,
      "step": 8917
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.32901982805335894,
      "learning_rate": 1.70572022436077e-05,
      "loss": 0.2344,
      "step": 8918
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.9944494672482547,
      "learning_rate": 1.705649946872317e-05,
      "loss": 0.9553,
      "step": 8919
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3404825471028024,
      "learning_rate": 1.7055796624414477e-05,
      "loss": 0.2712,
      "step": 8920
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3210284504355096,
      "learning_rate": 1.7055093710688533e-05,
      "loss": 0.2247,
      "step": 8921
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.36293321621082725,
      "learning_rate": 1.705439072755226e-05,
      "loss": 0.2314,
      "step": 8922
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.48209546072395726,
      "learning_rate": 1.705368767501256e-05,
      "loss": 0.2993,
      "step": 8923
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7940016662168718,
      "learning_rate": 1.7052984553076367e-05,
      "loss": 0.5169,
      "step": 8924
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3185974216049705,
      "learning_rate": 1.705228136175059e-05,
      "loss": 0.267,
      "step": 8925
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.721990684049916,
      "learning_rate": 1.7051578101042148e-05,
      "loss": 0.9862,
      "step": 8926
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30537235249163397,
      "learning_rate": 1.7050874770957962e-05,
      "loss": 0.1673,
      "step": 8927
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5734692869076317,
      "learning_rate": 1.705017137150495e-05,
      "loss": 0.4144,
      "step": 8928
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.1198006081674536,
      "learning_rate": 1.704946790269003e-05,
      "loss": 0.2804,
      "step": 8929
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4224900396947946,
      "learning_rate": 1.7048764364520127e-05,
      "loss": 0.3137,
      "step": 8930
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4345069005951669,
      "learning_rate": 1.7048060757002162e-05,
      "loss": 0.2273,
      "step": 8931
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3412734075863163,
      "learning_rate": 1.7047357080143053e-05,
      "loss": 0.2994,
      "step": 8932
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3452954333269084,
      "learning_rate": 1.7046653333949724e-05,
      "loss": 0.2541,
      "step": 8933
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.2505253654523283,
      "learning_rate": 1.7045949518429107e-05,
      "loss": 0.6143,
      "step": 8934
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9479987582691475,
      "learning_rate": 1.704524563358812e-05,
      "loss": 0.5823,
      "step": 8935
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3185974031020155,
      "learning_rate": 1.7044541679433684e-05,
      "loss": 0.1919,
      "step": 8936
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4470144763596534,
      "learning_rate": 1.7043837655972732e-05,
      "loss": 0.3615,
      "step": 8937
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4307895486630805,
      "learning_rate": 1.7043133563212183e-05,
      "loss": 0.2723,
      "step": 8938
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9127799556194589,
      "learning_rate": 1.7042429401158972e-05,
      "loss": 0.5804,
      "step": 8939
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42750381824041417,
      "learning_rate": 1.7041725169820024e-05,
      "loss": 0.2695,
      "step": 8940
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3576446795844761,
      "learning_rate": 1.7041020869202266e-05,
      "loss": 0.3123,
      "step": 8941
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.23186475085207217,
      "learning_rate": 1.7040316499312628e-05,
      "loss": 0.0726,
      "step": 8942
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5015449721337483,
      "learning_rate": 1.7039612060158036e-05,
      "loss": 0.3744,
      "step": 8943
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4277407295280716,
      "learning_rate": 1.703890755174543e-05,
      "loss": 0.3104,
      "step": 8944
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.42457755474386233,
      "learning_rate": 1.703820297408173e-05,
      "loss": 0.3018,
      "step": 8945
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5055747550791397,
      "learning_rate": 1.7037498327173878e-05,
      "loss": 0.262,
      "step": 8946
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0806735796140472,
      "learning_rate": 1.7036793611028802e-05,
      "loss": 0.6036,
      "step": 8947
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5836954513473122,
      "learning_rate": 1.703608882565343e-05,
      "loss": 0.3848,
      "step": 8948
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2602669021915694,
      "learning_rate": 1.7035383971054705e-05,
      "loss": 0.2053,
      "step": 8949
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6146139806439384,
      "learning_rate": 1.703467904723956e-05,
      "loss": 0.3998,
      "step": 8950
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2691805083232117,
      "learning_rate": 1.7033974054214922e-05,
      "loss": 0.1742,
      "step": 8951
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.47264753589066416,
      "learning_rate": 1.7033268991987736e-05,
      "loss": 0.2771,
      "step": 8952
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.8916552713610022,
      "learning_rate": 1.7032563860564938e-05,
      "loss": 0.588,
      "step": 8953
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6602452480183609,
      "learning_rate": 1.7031858659953456e-05,
      "loss": 0.3663,
      "step": 8954
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4141966231997194,
      "learning_rate": 1.7031153390160237e-05,
      "loss": 0.2626,
      "step": 8955
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4622726826548961,
      "learning_rate": 1.703044805119222e-05,
      "loss": 0.3493,
      "step": 8956
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.6198825857451172,
      "learning_rate": 1.702974264305634e-05,
      "loss": 0.378,
      "step": 8957
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7731281439447313,
      "learning_rate": 1.702903716575954e-05,
      "loss": 0.5156,
      "step": 8958
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30045318940655036,
      "learning_rate": 1.7028331619308755e-05,
      "loss": 0.2066,
      "step": 8959
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3074949637034735,
      "learning_rate": 1.7027626003710937e-05,
      "loss": 0.1725,
      "step": 8960
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.30243349025664906,
      "learning_rate": 1.7026920318973017e-05,
      "loss": 0.2415,
      "step": 8961
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.0244318658100062,
      "learning_rate": 1.7026214565101945e-05,
      "loss": 0.5993,
      "step": 8962
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4058613830419625,
      "learning_rate": 1.7025508742104664e-05,
      "loss": 0.2814,
      "step": 8963
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.4745452798871859,
      "learning_rate": 1.7024802849988112e-05,
      "loss": 0.2789,
      "step": 8964
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.9244281309372276,
      "learning_rate": 1.702409688875924e-05,
      "loss": 0.5917,
      "step": 8965
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7327960744061291,
      "learning_rate": 1.7023390858424998e-05,
      "loss": 0.3762,
      "step": 8966
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3329984480317004,
      "learning_rate": 1.7022684758992317e-05,
      "loss": 0.2882,
      "step": 8967
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.31683168933932243,
      "learning_rate": 1.7021978590468155e-05,
      "loss": 0.2243,
      "step": 8968
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.6729044829939252,
      "learning_rate": 1.7021272352859458e-05,
      "loss": 0.7711,
      "step": 8969
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.2253342552640251,
      "learning_rate": 1.7020566046173173e-05,
      "loss": 0.1155,
      "step": 8970
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.439610242206411,
      "learning_rate": 1.701985967041625e-05,
      "loss": 0.9101,
      "step": 8971
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3104783175986948,
      "learning_rate": 1.7019153225595637e-05,
      "loss": 0.2004,
      "step": 8972
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.551614924651168,
      "learning_rate": 1.7018446711718284e-05,
      "loss": 0.3656,
      "step": 8973
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.34424663632357144,
      "learning_rate": 1.7017740128791144e-05,
      "loss": 0.2975,
      "step": 8974
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.7665031800153456,
      "learning_rate": 1.701703347682117e-05,
      "loss": 0.4108,
      "step": 8975
    },
    {
      "epoch": 0.27,
      "grad_norm": 1.226105529920471,
      "learning_rate": 1.7016326755815305e-05,
      "loss": 0.5095,
      "step": 8976
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.33625659198984087,
      "learning_rate": 1.701561996578051e-05,
      "loss": 0.1858,
      "step": 8977
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.5250675531697555,
      "learning_rate": 1.7014913106723743e-05,
      "loss": 0.399,
      "step": 8978
    },
    {
      "epoch": 0.27,
      "grad_norm": 0.3253637337294852,
      "learning_rate": 1.7014206178651948e-05,
      "loss": 0.2174,
      "step": 8979
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2914664038240855,
      "learning_rate": 1.7013499181572086e-05,
      "loss": 0.8891,
      "step": 8980
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21174017712145957,
      "learning_rate": 1.701279211549111e-05,
      "loss": 0.0723,
      "step": 8981
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.297898969673289,
      "learning_rate": 1.701208498041598e-05,
      "loss": 0.2767,
      "step": 8982
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7604131065021407,
      "learning_rate": 1.7011377776353653e-05,
      "loss": 0.4081,
      "step": 8983
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5417992414023608,
      "learning_rate": 1.701067050331108e-05,
      "loss": 0.374,
      "step": 8984
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44308734776547376,
      "learning_rate": 1.7009963161295223e-05,
      "loss": 0.2813,
      "step": 8985
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3992914408355675,
      "learning_rate": 1.7009255750313047e-05,
      "loss": 0.2716,
      "step": 8986
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42604287444753736,
      "learning_rate": 1.70085482703715e-05,
      "loss": 0.2464,
      "step": 8987
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.26279446976179,
      "learning_rate": 1.7007840721477554e-05,
      "loss": 0.1077,
      "step": 8988
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3806607911300006,
      "learning_rate": 1.7007133103638166e-05,
      "loss": 0.8951,
      "step": 8989
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5350002741338304,
      "learning_rate": 1.7006425416860294e-05,
      "loss": 0.2771,
      "step": 8990
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3531637936529865,
      "learning_rate": 1.70057176611509e-05,
      "loss": 0.3044,
      "step": 8991
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3938242843876358,
      "learning_rate": 1.7005009836516957e-05,
      "loss": 0.2776,
      "step": 8992
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39299109585920255,
      "learning_rate": 1.700430194296542e-05,
      "loss": 0.3108,
      "step": 8993
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5462618114768305,
      "learning_rate": 1.7003593980503254e-05,
      "loss": 0.1836,
      "step": 8994
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3680813533367189,
      "learning_rate": 1.7002885949137428e-05,
      "loss": 0.2895,
      "step": 8995
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8639791251097325,
      "learning_rate": 1.7002177848874903e-05,
      "loss": 0.3503,
      "step": 8996
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3457324747698543,
      "learning_rate": 1.700146967972265e-05,
      "loss": 0.3192,
      "step": 8997
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7164100360652055,
      "learning_rate": 1.7000761441687636e-05,
      "loss": 0.4709,
      "step": 8998
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5492209952925224,
      "learning_rate": 1.7000053134776825e-05,
      "loss": 0.3636,
      "step": 8999
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3435593778464047,
      "learning_rate": 1.6999344758997193e-05,
      "loss": 0.2386,
      "step": 9000
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3559970317978549,
      "learning_rate": 1.6998636314355697e-05,
      "loss": 0.2474,
      "step": 9001
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37209552752539143,
      "learning_rate": 1.6997927800859317e-05,
      "loss": 0.2792,
      "step": 9002
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3961909998267642,
      "learning_rate": 1.699721921851502e-05,
      "loss": 0.2211,
      "step": 9003
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9484181556911969,
      "learning_rate": 1.699651056732978e-05,
      "loss": 0.3692,
      "step": 9004
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35119067763473755,
      "learning_rate": 1.6995801847310568e-05,
      "loss": 0.2658,
      "step": 9005
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6701814905543966,
      "learning_rate": 1.6995093058464352e-05,
      "loss": 0.7938,
      "step": 9006
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5562401777981247,
      "learning_rate": 1.699438420079811e-05,
      "loss": 0.4112,
      "step": 9007
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5208267369043552,
      "learning_rate": 1.6993675274318814e-05,
      "loss": 0.3753,
      "step": 9008
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2909210326134254,
      "learning_rate": 1.699296627903344e-05,
      "loss": 0.2068,
      "step": 9009
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4816069538155151,
      "learning_rate": 1.699225721494896e-05,
      "loss": 0.3698,
      "step": 9010
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17972749045812914,
      "learning_rate": 1.699154808207236e-05,
      "loss": 0.0719,
      "step": 9011
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.8423480652822126,
      "learning_rate": 1.6990838880410606e-05,
      "loss": 0.8822,
      "step": 9012
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3209533998117327,
      "learning_rate": 1.6990129609970676e-05,
      "loss": 0.1996,
      "step": 9013
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5508452783129083,
      "learning_rate": 1.6989420270759552e-05,
      "loss": 0.3827,
      "step": 9014
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2889106203009574,
      "learning_rate": 1.6988710862784216e-05,
      "loss": 0.2519,
      "step": 9015
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7113772283447389,
      "learning_rate": 1.6988001386051637e-05,
      "loss": 0.3812,
      "step": 9016
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5676360931038043,
      "learning_rate": 1.6987291840568803e-05,
      "loss": 0.4167,
      "step": 9017
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29885559747997836,
      "learning_rate": 1.6986582226342692e-05,
      "loss": 0.2265,
      "step": 9018
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3534242113501411,
      "learning_rate": 1.698587254338029e-05,
      "loss": 0.8554,
      "step": 9019
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21378011695167412,
      "learning_rate": 1.698516279168857e-05,
      "loss": 0.1432,
      "step": 9020
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.47778031143911576,
      "learning_rate": 1.6984452971274525e-05,
      "loss": 0.3762,
      "step": 9021
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2857863784913821,
      "learning_rate": 1.698374308214513e-05,
      "loss": 0.163,
      "step": 9022
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5526786372295281,
      "learning_rate": 1.6983033124307374e-05,
      "loss": 0.3442,
      "step": 9023
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8785575390151378,
      "learning_rate": 1.6982323097768237e-05,
      "loss": 0.4106,
      "step": 9024
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7952732483351042,
      "learning_rate": 1.6981613002534715e-05,
      "loss": 0.4754,
      "step": 9025
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2940675881444353,
      "learning_rate": 1.6980902838613783e-05,
      "loss": 0.2123,
      "step": 9026
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5824420018443943,
      "learning_rate": 1.6980192606012428e-05,
      "loss": 0.3842,
      "step": 9027
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3386501564270979,
      "learning_rate": 1.697948230473765e-05,
      "loss": 0.2693,
      "step": 9028
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.17152983147655312,
      "learning_rate": 1.6978771934796424e-05,
      "loss": 0.074,
      "step": 9029
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1207550261594097,
      "learning_rate": 1.6978061496195742e-05,
      "loss": 0.6547,
      "step": 9030
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.31987256106517953,
      "learning_rate": 1.6977350988942595e-05,
      "loss": 0.0789,
      "step": 9031
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41308067781570107,
      "learning_rate": 1.6976640413043974e-05,
      "loss": 0.3149,
      "step": 9032
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.32786382604356007,
      "learning_rate": 1.6975929768506868e-05,
      "loss": 0.2947,
      "step": 9033
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8177012139433225,
      "learning_rate": 1.6975219055338274e-05,
      "loss": 0.4928,
      "step": 9034
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9090880605027462,
      "learning_rate": 1.6974508273545174e-05,
      "loss": 0.4988,
      "step": 9035
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4076096816568313,
      "learning_rate": 1.697379742313457e-05,
      "loss": 0.2834,
      "step": 9036
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4076998891995421,
      "learning_rate": 1.697308650411345e-05,
      "loss": 0.2682,
      "step": 9037
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.270584817562777,
      "learning_rate": 1.6972375516488814e-05,
      "loss": 0.2669,
      "step": 9038
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2379942965855735,
      "learning_rate": 1.697166446026765e-05,
      "loss": 0.1089,
      "step": 9039
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6250590039097202,
      "learning_rate": 1.697095333545696e-05,
      "loss": 0.3527,
      "step": 9040
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3436282870023666,
      "learning_rate": 1.6970242142063736e-05,
      "loss": 0.2102,
      "step": 9041
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7318456267291088,
      "learning_rate": 1.6969530880094977e-05,
      "loss": 0.4045,
      "step": 9042
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.909979959069282,
      "learning_rate": 1.6968819549557674e-05,
      "loss": 0.5439,
      "step": 9043
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.30551279768771683,
      "learning_rate": 1.6968108150458838e-05,
      "loss": 0.2418,
      "step": 9044
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35514942253738924,
      "learning_rate": 1.696739668280546e-05,
      "loss": 0.2909,
      "step": 9045
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9557077799291888,
      "learning_rate": 1.696668514660454e-05,
      "loss": 0.465,
      "step": 9046
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7078886902178865,
      "learning_rate": 1.696597354186308e-05,
      "loss": 0.9521,
      "step": 9047
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.22437086206659135,
      "learning_rate": 1.696526186858808e-05,
      "loss": 0.1148,
      "step": 9048
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5961890732596751,
      "learning_rate": 1.696455012678654e-05,
      "loss": 0.3456,
      "step": 9049
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38773527382458217,
      "learning_rate": 1.6963838316465467e-05,
      "loss": 0.2689,
      "step": 9050
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.34118865043933366,
      "learning_rate": 1.696312643763186e-05,
      "loss": 0.3133,
      "step": 9051
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8801226886708778,
      "learning_rate": 1.6962414490292723e-05,
      "loss": 0.4192,
      "step": 9052
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3865772289212066,
      "learning_rate": 1.6961702474455063e-05,
      "loss": 0.8291,
      "step": 9053
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.25365427356521936,
      "learning_rate": 1.6960990390125884e-05,
      "loss": 0.0757,
      "step": 9054
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.379821154775452,
      "learning_rate": 1.6960278237312188e-05,
      "loss": 0.2537,
      "step": 9055
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3450421994201378,
      "learning_rate": 1.6959566016020988e-05,
      "loss": 0.3145,
      "step": 9056
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2530519719266335,
      "learning_rate": 1.6958853726259286e-05,
      "loss": 0.0711,
      "step": 9057
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9432038969773471,
      "learning_rate": 1.6958141368034093e-05,
      "loss": 0.4729,
      "step": 9058
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3307541615033568,
      "learning_rate": 1.695742894135241e-05,
      "loss": 0.2524,
      "step": 9059
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.833731583857205,
      "learning_rate": 1.6956716446221256e-05,
      "loss": 0.521,
      "step": 9060
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43566647847567064,
      "learning_rate": 1.6956003882647637e-05,
      "loss": 0.322,
      "step": 9061
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.48478763760272947,
      "learning_rate": 1.695529125063856e-05,
      "loss": 0.3367,
      "step": 9062
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3410245689524289,
      "learning_rate": 1.695457855020104e-05,
      "loss": 0.2011,
      "step": 9063
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5480937846442919,
      "learning_rate": 1.695386578134209e-05,
      "loss": 0.4394,
      "step": 9064
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29800318331857256,
      "learning_rate": 1.6953152944068717e-05,
      "loss": 0.1545,
      "step": 9065
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5286960179100612,
      "learning_rate": 1.695244003838794e-05,
      "loss": 0.2236,
      "step": 9066
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5889743541717584,
      "learning_rate": 1.6951727064306775e-05,
      "loss": 0.3335,
      "step": 9067
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2797419102412227,
      "learning_rate": 1.6951014021832223e-05,
      "loss": 0.2387,
      "step": 9068
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44397212091228005,
      "learning_rate": 1.6950300910971313e-05,
      "loss": 0.3936,
      "step": 9069
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9619835265259723,
      "learning_rate": 1.6949587731731054e-05,
      "loss": 0.334,
      "step": 9070
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4108823411367326,
      "learning_rate": 1.6948874484118464e-05,
      "loss": 0.8672,
      "step": 9071
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.30282173553145547,
      "learning_rate": 1.694816116814056e-05,
      "loss": 0.2017,
      "step": 9072
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0319663606502028,
      "learning_rate": 1.6947447783804362e-05,
      "loss": 0.6029,
      "step": 9073
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4283710090528573,
      "learning_rate": 1.6946734331116886e-05,
      "loss": 0.2663,
      "step": 9074
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4464181618101928,
      "learning_rate": 1.694602081008515e-05,
      "loss": 0.3928,
      "step": 9075
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4547003458710954,
      "learning_rate": 1.6945307220716177e-05,
      "loss": 0.2756,
      "step": 9076
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39030739217372185,
      "learning_rate": 1.6944593563016987e-05,
      "loss": 0.3007,
      "step": 9077
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.27728088915633203,
      "learning_rate": 1.6943879836994602e-05,
      "loss": 0.133,
      "step": 9078
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1145203101489418,
      "learning_rate": 1.6943166042656035e-05,
      "loss": 0.6338,
      "step": 9079
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3268902676631926,
      "learning_rate": 1.6942452180008326e-05,
      "loss": 0.2829,
      "step": 9080
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.32000557335826907,
      "learning_rate": 1.6941738249058483e-05,
      "loss": 0.0759,
      "step": 9081
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4062427670596812,
      "learning_rate": 1.6941024249813535e-05,
      "loss": 0.3116,
      "step": 9082
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9450411808193894,
      "learning_rate": 1.6940310182280503e-05,
      "loss": 0.3278,
      "step": 9083
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6884104135272665,
      "learning_rate": 1.6939596046466424e-05,
      "loss": 0.4335,
      "step": 9084
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4144149231324844,
      "learning_rate": 1.6938881842378307e-05,
      "loss": 0.2874,
      "step": 9085
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4128460878523025,
      "learning_rate": 1.693816757002319e-05,
      "loss": 0.2755,
      "step": 9086
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24208876652480424,
      "learning_rate": 1.69374532294081e-05,
      "loss": 0.1751,
      "step": 9087
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4691849321819497,
      "learning_rate": 1.6936738820540065e-05,
      "loss": 0.8356,
      "step": 9088
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0001047579052684,
      "learning_rate": 1.6936024343426104e-05,
      "loss": 0.5202,
      "step": 9089
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6310631606759621,
      "learning_rate": 1.693530979807326e-05,
      "loss": 0.3742,
      "step": 9090
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3837664610999615,
      "learning_rate": 1.6934595184488548e-05,
      "loss": 0.23,
      "step": 9091
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3806547043287389,
      "learning_rate": 1.6933880502679013e-05,
      "loss": 0.3378,
      "step": 9092
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7596605655051997,
      "learning_rate": 1.6933165752651678e-05,
      "loss": 0.3656,
      "step": 9093
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4871468035665277,
      "learning_rate": 1.6932450934413582e-05,
      "loss": 0.2962,
      "step": 9094
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3339339884551093,
      "learning_rate": 1.6931736047971746e-05,
      "loss": 0.2747,
      "step": 9095
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.15846557946969314,
      "learning_rate": 1.6931021093333207e-05,
      "loss": 0.0918,
      "step": 9096
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.3418857450381876,
      "learning_rate": 1.693030607050501e-05,
      "loss": 0.928,
      "step": 9097
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3202880612110003,
      "learning_rate": 1.692959097949418e-05,
      "loss": 0.265,
      "step": 9098
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.644027249100461,
      "learning_rate": 1.692887582030775e-05,
      "loss": 0.3593,
      "step": 9099
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.34085185271695384,
      "learning_rate": 1.6928160592952766e-05,
      "loss": 0.2595,
      "step": 9100
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.093284971826776,
      "learning_rate": 1.6927445297436253e-05,
      "loss": 0.6831,
      "step": 9101
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7804063982536773,
      "learning_rate": 1.6926729933765253e-05,
      "loss": 0.4006,
      "step": 9102
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45079249326249055,
      "learning_rate": 1.6926014501946807e-05,
      "loss": 0.3604,
      "step": 9103
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.28192902099174494,
      "learning_rate": 1.6925299001987952e-05,
      "loss": 0.2046,
      "step": 9104
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.49493403623124904,
      "learning_rate": 1.6924583433895728e-05,
      "loss": 0.3429,
      "step": 9105
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21249188246530795,
      "learning_rate": 1.6923867797677173e-05,
      "loss": 0.1117,
      "step": 9106
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.833276543024151,
      "learning_rate": 1.6923152093339324e-05,
      "loss": 0.4734,
      "step": 9107
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7908263681512991,
      "learning_rate": 1.692243632088923e-05,
      "loss": 0.344,
      "step": 9108
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3576510187597052,
      "learning_rate": 1.6921720480333928e-05,
      "loss": 0.2343,
      "step": 9109
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3622849613699596,
      "learning_rate": 1.6921004571680466e-05,
      "loss": 0.3429,
      "step": 9110
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9281138089547599,
      "learning_rate": 1.692028859493588e-05,
      "loss": 0.4415,
      "step": 9111
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1092831050244316,
      "learning_rate": 1.6919572550107218e-05,
      "loss": 0.7026,
      "step": 9112
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3151996048632277,
      "learning_rate": 1.691885643720153e-05,
      "loss": 0.1662,
      "step": 9113
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2988800493886443,
      "learning_rate": 1.691814025622585e-05,
      "loss": 0.2314,
      "step": 9114
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.48100572330391267,
      "learning_rate": 1.6917424007187228e-05,
      "loss": 0.3079,
      "step": 9115
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.49996450581492263,
      "learning_rate": 1.6916707690092717e-05,
      "loss": 0.3555,
      "step": 9116
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.275427536554289,
      "learning_rate": 1.6915991304949358e-05,
      "loss": 0.1292,
      "step": 9117
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3984356569001293,
      "learning_rate": 1.69152748517642e-05,
      "loss": 0.2974,
      "step": 9118
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.776564330175693,
      "learning_rate": 1.69145583305443e-05,
      "loss": 0.4223,
      "step": 9119
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4927184600338464,
      "learning_rate": 1.691384174129669e-05,
      "loss": 0.0239,
      "step": 9120
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.32382078326692015,
      "learning_rate": 1.6913125084028434e-05,
      "loss": 0.304,
      "step": 9121
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2679982476011576,
      "learning_rate": 1.6912408358746577e-05,
      "loss": 0.202,
      "step": 9122
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3615680580373166,
      "learning_rate": 1.6911691565458177e-05,
      "loss": 0.8071,
      "step": 9123
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1677490182824195,
      "learning_rate": 1.6910974704170277e-05,
      "loss": 0.5266,
      "step": 9124
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.49908416710993597,
      "learning_rate": 1.6910257774889937e-05,
      "loss": 0.2764,
      "step": 9125
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.48190363431562583,
      "learning_rate": 1.6909540777624202e-05,
      "loss": 0.2654,
      "step": 9126
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39801805032301807,
      "learning_rate": 1.6908823712380136e-05,
      "loss": 0.3452,
      "step": 9127
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.33553638710840883,
      "learning_rate": 1.6908106579164786e-05,
      "loss": 0.2473,
      "step": 9128
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44878119547824186,
      "learning_rate": 1.6907389377985214e-05,
      "loss": 0.3339,
      "step": 9129
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9089981185795981,
      "learning_rate": 1.690667210884847e-05,
      "loss": 0.2611,
      "step": 9130
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4069868720552398,
      "learning_rate": 1.690595477176161e-05,
      "loss": 0.2739,
      "step": 9131
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4166457313811797,
      "learning_rate": 1.69052373667317e-05,
      "loss": 0.2023,
      "step": 9132
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3890584872804005,
      "learning_rate": 1.690451989376579e-05,
      "loss": 0.2766,
      "step": 9133
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3194811418079401,
      "learning_rate": 1.6903802352870942e-05,
      "loss": 0.2444,
      "step": 9134
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.506697781346332,
      "learning_rate": 1.6903084744054212e-05,
      "loss": 0.2276,
      "step": 9135
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42472666456165786,
      "learning_rate": 1.6902367067322666e-05,
      "loss": 0.3039,
      "step": 9136
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45252078291345815,
      "learning_rate": 1.6901649322683363e-05,
      "loss": 0.2756,
      "step": 9137
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.386991593602929,
      "learning_rate": 1.690093151014336e-05,
      "loss": 0.5996,
      "step": 9138
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3539929344672571,
      "learning_rate": 1.6900213629709724e-05,
      "loss": 0.2397,
      "step": 9139
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8390584211046926,
      "learning_rate": 1.6899495681389515e-05,
      "loss": 0.3557,
      "step": 9140
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3453157746741334,
      "learning_rate": 1.6898777665189795e-05,
      "loss": 0.2347,
      "step": 9141
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4005359557015618,
      "learning_rate": 1.6898059581117637e-05,
      "loss": 0.8414,
      "step": 9142
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8903427102180279,
      "learning_rate": 1.6897341429180092e-05,
      "loss": 0.3986,
      "step": 9143
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.624901834439531,
      "learning_rate": 1.6896623209384237e-05,
      "loss": 0.5253,
      "step": 9144
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29955554538482304,
      "learning_rate": 1.6895904921737133e-05,
      "loss": 0.215,
      "step": 9145
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.23837028113840272,
      "learning_rate": 1.6895186566245844e-05,
      "loss": 0.1953,
      "step": 9146
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.46781657063205484,
      "learning_rate": 1.689446814291744e-05,
      "loss": 0.2546,
      "step": 9147
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9402378971812939,
      "learning_rate": 1.6893749651758996e-05,
      "loss": 0.3922,
      "step": 9148
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5894939094373495,
      "learning_rate": 1.6893031092777572e-05,
      "loss": 0.3432,
      "step": 9149
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37400357436456727,
      "learning_rate": 1.6892312465980238e-05,
      "loss": 0.2629,
      "step": 9150
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.735238036761152,
      "learning_rate": 1.6891593771374067e-05,
      "loss": 0.8466,
      "step": 9151
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4324057295976516,
      "learning_rate": 1.689087500896613e-05,
      "loss": 0.3065,
      "step": 9152
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8660240245544844,
      "learning_rate": 1.6890156178763497e-05,
      "loss": 0.4456,
      "step": 9153
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3005072192735339,
      "learning_rate": 1.688943728077324e-05,
      "loss": 0.2305,
      "step": 9154
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.6903097893406203,
      "learning_rate": 1.688871831500243e-05,
      "loss": 0.8582,
      "step": 9155
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2166490064946146,
      "learning_rate": 1.6887999281458148e-05,
      "loss": 0.0728,
      "step": 9156
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38168916988061785,
      "learning_rate": 1.6887280180147457e-05,
      "loss": 0.315,
      "step": 9157
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3358050706706736,
      "learning_rate": 1.688656101107744e-05,
      "loss": 0.1802,
      "step": 9158
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4785908972362979,
      "learning_rate": 1.688584177425517e-05,
      "loss": 0.2257,
      "step": 9159
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4150181373397257,
      "learning_rate": 1.6885122469687723e-05,
      "loss": 0.8769,
      "step": 9160
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9593283782502015,
      "learning_rate": 1.6884403097382174e-05,
      "loss": 0.4213,
      "step": 9161
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3999722377338376,
      "learning_rate": 1.6883683657345603e-05,
      "loss": 0.3237,
      "step": 9162
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35696585463963226,
      "learning_rate": 1.688296414958509e-05,
      "loss": 0.1892,
      "step": 9163
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4164601046606526,
      "learning_rate": 1.6882244574107706e-05,
      "loss": 0.3368,
      "step": 9164
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.26334400151820714,
      "learning_rate": 1.6881524930920536e-05,
      "loss": 0.099,
      "step": 9165
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3804014481531692,
      "learning_rate": 1.6880805220030662e-05,
      "loss": 0.7931,
      "step": 9166
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4087204880844259,
      "learning_rate": 1.6880085441445163e-05,
      "loss": 0.1391,
      "step": 9167
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.45507283280604166,
      "learning_rate": 1.6879365595171116e-05,
      "loss": 0.3519,
      "step": 9168
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3561558677713132,
      "learning_rate": 1.687864568121561e-05,
      "loss": 0.2817,
      "step": 9169
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8704548358812476,
      "learning_rate": 1.687792569958572e-05,
      "loss": 0.5425,
      "step": 9170
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0089335155977666,
      "learning_rate": 1.6877205650288537e-05,
      "loss": 0.4344,
      "step": 9171
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29079881299458643,
      "learning_rate": 1.687648553333114e-05,
      "loss": 0.2028,
      "step": 9172
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.1001935804562746,
      "learning_rate": 1.6875765348720618e-05,
      "loss": 0.6879,
      "step": 9173
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.24574828243420993,
      "learning_rate": 1.6875045096464052e-05,
      "loss": 0.0763,
      "step": 9174
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36548607374849257,
      "learning_rate": 1.687432477656853e-05,
      "loss": 0.3209,
      "step": 9175
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.381069196341618,
      "learning_rate": 1.6873604389041143e-05,
      "loss": 0.0687,
      "step": 9176
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3984050251969884,
      "learning_rate": 1.6872883933888972e-05,
      "loss": 0.3091,
      "step": 9177
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7971235900533029,
      "learning_rate": 1.6872163411119106e-05,
      "loss": 0.4001,
      "step": 9178
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7744473369388575,
      "learning_rate": 1.6871442820738636e-05,
      "loss": 0.509,
      "step": 9179
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2886063255152848,
      "learning_rate": 1.6870722162754648e-05,
      "loss": 0.2558,
      "step": 9180
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4060980611090651,
      "learning_rate": 1.687000143717424e-05,
      "loss": 0.2737,
      "step": 9181
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4310010827593105,
      "learning_rate": 1.6869280644004493e-05,
      "loss": 0.2296,
      "step": 9182
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4610409387974432,
      "learning_rate": 1.6868559783252502e-05,
      "loss": 0.309,
      "step": 9183
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2873058672245046,
      "learning_rate": 1.6867838854925365e-05,
      "loss": 0.1489,
      "step": 9184
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3450724157610996,
      "learning_rate": 1.6867117859030166e-05,
      "loss": 0.1764,
      "step": 9185
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6397439144290349,
      "learning_rate": 1.6866396795574002e-05,
      "loss": 0.3984,
      "step": 9186
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3187863974041776,
      "learning_rate": 1.6865675664563966e-05,
      "loss": 0.2913,
      "step": 9187
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8128497613070684,
      "learning_rate": 1.6864954466007156e-05,
      "loss": 0.5505,
      "step": 9188
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.42260229758850115,
      "learning_rate": 1.6864233199910666e-05,
      "loss": 0.2907,
      "step": 9189
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8422434700732985,
      "learning_rate": 1.6863511866281587e-05,
      "loss": 0.3513,
      "step": 9190
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38792723150078995,
      "learning_rate": 1.6862790465127025e-05,
      "loss": 0.2398,
      "step": 9191
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.38859342546599657,
      "learning_rate": 1.6862068996454067e-05,
      "loss": 0.1417,
      "step": 9192
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3163329048238626,
      "learning_rate": 1.6861347460269815e-05,
      "loss": 0.2452,
      "step": 9193
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7770450665423668,
      "learning_rate": 1.6860625856581378e-05,
      "loss": 0.6049,
      "step": 9194
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2974657573578175,
      "learning_rate": 1.685990418539584e-05,
      "loss": 0.2003,
      "step": 9195
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7060277324600557,
      "learning_rate": 1.6859182446720308e-05,
      "loss": 0.4533,
      "step": 9196
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8899295149330371,
      "learning_rate": 1.6858460640561882e-05,
      "loss": 0.4553,
      "step": 9197
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4044720982910037,
      "learning_rate": 1.6857738766927666e-05,
      "loss": 0.7751,
      "step": 9198
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2424937122646534,
      "learning_rate": 1.6857016825824755e-05,
      "loss": 0.1974,
      "step": 9199
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.32346173555151153,
      "learning_rate": 1.6856294817260258e-05,
      "loss": 0.2543,
      "step": 9200
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4976721647274963,
      "learning_rate": 1.685557274124128e-05,
      "loss": 0.2499,
      "step": 9201
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.30686090900824686,
      "learning_rate": 1.685485059777492e-05,
      "loss": 0.1991,
      "step": 9202
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.564654691298761,
      "learning_rate": 1.685412838686828e-05,
      "loss": 0.42,
      "step": 9203
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.34167930110245587,
      "learning_rate": 1.6853406108528475e-05,
      "loss": 0.2433,
      "step": 9204
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4973435344883747,
      "learning_rate": 1.6852683762762604e-05,
      "loss": 0.377,
      "step": 9205
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43701649409299226,
      "learning_rate": 1.6851961349577776e-05,
      "loss": 0.3044,
      "step": 9206
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3731967447439997,
      "learning_rate": 1.6851238868981095e-05,
      "loss": 0.9287,
      "step": 9207
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.30747498585227284,
      "learning_rate": 1.6850516320979674e-05,
      "loss": 0.1765,
      "step": 9208
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5739235162376934,
      "learning_rate": 1.6849793705580617e-05,
      "loss": 0.3765,
      "step": 9209
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.26291725453063197,
      "learning_rate": 1.6849071022791036e-05,
      "loss": 0.1115,
      "step": 9210
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.26745375826555967,
      "learning_rate": 1.6848348272618046e-05,
      "loss": 0.2661,
      "step": 9211
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5755646723378336,
      "learning_rate": 1.6847625455068745e-05,
      "loss": 0.2693,
      "step": 9212
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3539367313704977,
      "learning_rate": 1.6846902570150252e-05,
      "loss": 0.2407,
      "step": 9213
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7720297467849515,
      "learning_rate": 1.684617961786968e-05,
      "loss": 0.4933,
      "step": 9214
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.849778310173705,
      "learning_rate": 1.6845456598234146e-05,
      "loss": 0.5063,
      "step": 9215
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36586425246948207,
      "learning_rate": 1.684473351125075e-05,
      "loss": 0.3023,
      "step": 9216
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.27467210681995685,
      "learning_rate": 1.684401035692662e-05,
      "loss": 0.0721,
      "step": 9217
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.37187065125698254,
      "learning_rate": 1.684328713526886e-05,
      "loss": 0.2924,
      "step": 9218
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9023354605269677,
      "learning_rate": 1.684256384628459e-05,
      "loss": 0.3507,
      "step": 9219
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6405695734752738,
      "learning_rate": 1.6841840489980924e-05,
      "loss": 0.4396,
      "step": 9220
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35385483941949725,
      "learning_rate": 1.6841117066364984e-05,
      "loss": 0.1824,
      "step": 9221
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4290240653915181,
      "learning_rate": 1.6840393575443877e-05,
      "loss": 0.3238,
      "step": 9222
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.21826821446388756,
      "learning_rate": 1.6839670017224732e-05,
      "loss": 0.2122,
      "step": 9223
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.4122826372244004,
      "learning_rate": 1.683894639171466e-05,
      "loss": 0.8274,
      "step": 9224
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9614460807641952,
      "learning_rate": 1.6838222698920787e-05,
      "loss": 0.4925,
      "step": 9225
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29353764953228934,
      "learning_rate": 1.6837498938850228e-05,
      "loss": 0.0715,
      "step": 9226
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.400062852207277,
      "learning_rate": 1.6836775111510105e-05,
      "loss": 0.3397,
      "step": 9227
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4527505769472098,
      "learning_rate": 1.6836051216907535e-05,
      "loss": 0.3098,
      "step": 9228
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4124732675115653,
      "learning_rate": 1.6835327255049645e-05,
      "loss": 0.359,
      "step": 9229
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5679633100684321,
      "learning_rate": 1.6834603225943556e-05,
      "loss": 0.3222,
      "step": 9230
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.36600278202312697,
      "learning_rate": 1.6833879129596394e-05,
      "loss": 0.3088,
      "step": 9231
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2097043828371543,
      "learning_rate": 1.683315496601528e-05,
      "loss": 0.077,
      "step": 9232
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.232862259319797,
      "learning_rate": 1.683243073520734e-05,
      "loss": 0.7865,
      "step": 9233
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.30122271535282763,
      "learning_rate": 1.6831706437179694e-05,
      "loss": 0.2618,
      "step": 9234
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6088765194078742,
      "learning_rate": 1.6830982071939477e-05,
      "loss": 0.2507,
      "step": 9235
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35761457481984094,
      "learning_rate": 1.6830257639493808e-05,
      "loss": 0.2444,
      "step": 9236
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5672637627881478,
      "learning_rate": 1.6829533139849818e-05,
      "loss": 0.7089,
      "step": 9237
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8256601362640632,
      "learning_rate": 1.6828808573014634e-05,
      "loss": 0.4075,
      "step": 9238
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2165164516524413,
      "learning_rate": 1.682808393899538e-05,
      "loss": 0.0775,
      "step": 9239
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0778077566280342,
      "learning_rate": 1.6827359237799196e-05,
      "loss": 0.481,
      "step": 9240
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.27199627556434197,
      "learning_rate": 1.68266344694332e-05,
      "loss": 0.2469,
      "step": 9241
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.511016256212096,
      "learning_rate": 1.6825909633904532e-05,
      "loss": 0.3415,
      "step": 9242
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.914541458761132,
      "learning_rate": 1.682518473122032e-05,
      "loss": 0.4452,
      "step": 9243
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.7670913046646561,
      "learning_rate": 1.6824459761387687e-05,
      "loss": 0.4524,
      "step": 9244
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3236178598539158,
      "learning_rate": 1.6823734724413777e-05,
      "loss": 0.2334,
      "step": 9245
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4852525530353445,
      "learning_rate": 1.6823009620305723e-05,
      "loss": 0.3869,
      "step": 9246
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6437605582274437,
      "learning_rate": 1.6822284449070654e-05,
      "loss": 0.3216,
      "step": 9247
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.9703810173894893,
      "learning_rate": 1.6821559210715705e-05,
      "loss": 0.8362,
      "step": 9248
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3408581264984008,
      "learning_rate": 1.682083390524801e-05,
      "loss": 0.1751,
      "step": 9249
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.28355936573882873,
      "learning_rate": 1.682010853267471e-05,
      "loss": 0.2112,
      "step": 9250
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9291414013069612,
      "learning_rate": 1.6819383093002937e-05,
      "loss": 0.6391,
      "step": 9251
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3489445613865244,
      "learning_rate": 1.6818657586239825e-05,
      "loss": 0.2554,
      "step": 9252
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7740344345683839,
      "learning_rate": 1.6817932012392524e-05,
      "loss": 0.4452,
      "step": 9253
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.30911602679476125,
      "learning_rate": 1.6817206371468158e-05,
      "loss": 0.2361,
      "step": 9254
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7144194194880966,
      "learning_rate": 1.6816480663473877e-05,
      "loss": 0.5766,
      "step": 9255
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7070235501338543,
      "learning_rate": 1.6815754888416812e-05,
      "loss": 0.3704,
      "step": 9256
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5442788283626225,
      "learning_rate": 1.6815029046304113e-05,
      "loss": 0.6829,
      "step": 9257
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.27339959592274915,
      "learning_rate": 1.681430313714291e-05,
      "loss": 0.2014,
      "step": 9258
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3759747990226787,
      "learning_rate": 1.6813577160940354e-05,
      "loss": 0.3241,
      "step": 9259
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2556494077854993,
      "learning_rate": 1.6812851117703582e-05,
      "loss": 0.1327,
      "step": 9260
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.44413767321857867,
      "learning_rate": 1.681212500743974e-05,
      "loss": 0.282,
      "step": 9261
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4657377935743905,
      "learning_rate": 1.6811398830155973e-05,
      "loss": 0.2219,
      "step": 9262
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.458559813670279,
      "learning_rate": 1.681067258585942e-05,
      "loss": 0.32,
      "step": 9263
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.43834330073213523,
      "learning_rate": 1.6809946274557228e-05,
      "loss": 0.3027,
      "step": 9264
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4373337741524414,
      "learning_rate": 1.6809219896256546e-05,
      "loss": 0.2843,
      "step": 9265
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9544515320164273,
      "learning_rate": 1.680849345096452e-05,
      "loss": 0.5451,
      "step": 9266
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.41434915717942733,
      "learning_rate": 1.680776693868829e-05,
      "loss": 0.0708,
      "step": 9267
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29000542010981145,
      "learning_rate": 1.6807040359435016e-05,
      "loss": 0.2427,
      "step": 9268
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.2072046371279246,
      "learning_rate": 1.6806313713211832e-05,
      "loss": 0.6557,
      "step": 9269
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.27753998378617,
      "learning_rate": 1.6805587000025897e-05,
      "loss": 0.2457,
      "step": 9270
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.8838352251163271,
      "learning_rate": 1.680486021988436e-05,
      "loss": 0.3719,
      "step": 9271
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3981370619639158,
      "learning_rate": 1.6804133372794366e-05,
      "loss": 0.2667,
      "step": 9272
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5099614192256811,
      "learning_rate": 1.680340645876307e-05,
      "loss": 0.3095,
      "step": 9273
    },
    {
      "epoch": 0.28,
      "grad_norm": 2.0127456069188034,
      "learning_rate": 1.680267947779762e-05,
      "loss": 0.8318,
      "step": 9274
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9696446760890837,
      "learning_rate": 1.6801952429905174e-05,
      "loss": 0.3466,
      "step": 9275
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.39791114504449954,
      "learning_rate": 1.6801225315092885e-05,
      "loss": 0.3127,
      "step": 9276
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.29516199891613215,
      "learning_rate": 1.68004981333679e-05,
      "loss": 0.2611,
      "step": 9277
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5634948361637533,
      "learning_rate": 1.6799770884737375e-05,
      "loss": 0.6817,
      "step": 9278
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.9209881346083298,
      "learning_rate": 1.679904356920847e-05,
      "loss": 0.5524,
      "step": 9279
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5605818134358299,
      "learning_rate": 1.6798316186788334e-05,
      "loss": 0.3295,
      "step": 9280
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3994241089311609,
      "learning_rate": 1.6797588737484127e-05,
      "loss": 0.2919,
      "step": 9281
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.19704550777499433,
      "learning_rate": 1.6796861221303006e-05,
      "loss": 0.1362,
      "step": 9282
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.5111846718226966,
      "learning_rate": 1.6796133638252126e-05,
      "loss": 0.3872,
      "step": 9283
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.0337564085133155,
      "learning_rate": 1.679540598833865e-05,
      "loss": 0.357,
      "step": 9284
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6582751637108809,
      "learning_rate": 1.6794678271569734e-05,
      "loss": 0.3422,
      "step": 9285
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.35046592068542154,
      "learning_rate": 1.6793950487952535e-05,
      "loss": 0.2508,
      "step": 9286
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5468927330617204,
      "learning_rate": 1.679322263749422e-05,
      "loss": 0.913,
      "step": 9287
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3333980449862742,
      "learning_rate": 1.6792494720201936e-05,
      "loss": 0.2842,
      "step": 9288
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.7608431910063439,
      "learning_rate": 1.6791766736082864e-05,
      "loss": 0.4678,
      "step": 9289
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3386508039481464,
      "learning_rate": 1.679103868514415e-05,
      "loss": 0.1721,
      "step": 9290
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2516234690626341,
      "learning_rate": 1.6790310567392966e-05,
      "loss": 0.1485,
      "step": 9291
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4945093103808694,
      "learning_rate": 1.6789582382836473e-05,
      "loss": 0.3318,
      "step": 9292
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.33863540589610325,
      "learning_rate": 1.6788854131481833e-05,
      "loss": 0.2741,
      "step": 9293
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.863210210556946,
      "learning_rate": 1.6788125813336213e-05,
      "loss": 0.4543,
      "step": 9294
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3207321497531212,
      "learning_rate": 1.6787397428406774e-05,
      "loss": 0.2059,
      "step": 9295
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.5362952885623107,
      "learning_rate": 1.678666897670069e-05,
      "loss": 0.9571,
      "step": 9296
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.6710004858741152,
      "learning_rate": 1.678594045822512e-05,
      "loss": 0.3857,
      "step": 9297
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.796362294199033,
      "learning_rate": 1.678521187298724e-05,
      "loss": 0.875,
      "step": 9298
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3181447314582427,
      "learning_rate": 1.678448322099421e-05,
      "loss": 0.1738,
      "step": 9299
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.28265421803556007,
      "learning_rate": 1.67837545022532e-05,
      "loss": 0.2722,
      "step": 9300
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.2687516536689657,
      "learning_rate": 1.6783025716771383e-05,
      "loss": 0.2034,
      "step": 9301
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.3432037499628737,
      "learning_rate": 1.6782296864555924e-05,
      "loss": 0.7447,
      "step": 9302
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.47477732064565115,
      "learning_rate": 1.6781567945614004e-05,
      "loss": 0.1664,
      "step": 9303
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.3187925213531273,
      "learning_rate": 1.678083895995278e-05,
      "loss": 0.2373,
      "step": 9304
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.4098500954736838,
      "learning_rate": 1.6780109907579435e-05,
      "loss": 0.3786,
      "step": 9305
    },
    {
      "epoch": 0.28,
      "grad_norm": 0.40886157381010485,
      "learning_rate": 1.6779380788501137e-05,
      "loss": 0.2712,
      "step": 9306
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8221374208090765,
      "learning_rate": 1.677865160272506e-05,
      "loss": 0.5791,
      "step": 9307
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2644885955138503,
      "learning_rate": 1.6777922350258377e-05,
      "loss": 0.1888,
      "step": 9308
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5093008541173296,
      "learning_rate": 1.6777193031108264e-05,
      "loss": 0.2705,
      "step": 9309
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.30795234311646147,
      "learning_rate": 1.6776463645281896e-05,
      "loss": 0.124,
      "step": 9310
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5261035165190289,
      "learning_rate": 1.677573419278645e-05,
      "loss": 0.3743,
      "step": 9311
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33410211252052197,
      "learning_rate": 1.67750046736291e-05,
      "loss": 0.1877,
      "step": 9312
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42800171293518774,
      "learning_rate": 1.6774275087817023e-05,
      "loss": 0.3205,
      "step": 9313
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8061948984364253,
      "learning_rate": 1.6773545435357402e-05,
      "loss": 0.3921,
      "step": 9314
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5918603762137392,
      "learning_rate": 1.6772815716257414e-05,
      "loss": 0.4386,
      "step": 9315
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7134021315400958,
      "learning_rate": 1.677208593052423e-05,
      "loss": 0.4469,
      "step": 9316
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3560328414871484,
      "learning_rate": 1.677135607816504e-05,
      "loss": 0.1764,
      "step": 9317
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3686453876148818,
      "learning_rate": 1.677062615918702e-05,
      "loss": 0.3106,
      "step": 9318
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.258828570661806,
      "learning_rate": 1.676989617359736e-05,
      "loss": 0.1335,
      "step": 9319
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.894540554869727,
      "learning_rate": 1.6769166121403222e-05,
      "loss": 0.8683,
      "step": 9320
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9166149751387785,
      "learning_rate": 1.6768436002611808e-05,
      "loss": 0.4045,
      "step": 9321
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45648692229880883,
      "learning_rate": 1.676770581723029e-05,
      "loss": 0.3108,
      "step": 9322
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4496399220019267,
      "learning_rate": 1.6766975565265854e-05,
      "loss": 0.323,
      "step": 9323
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3582938713081765,
      "learning_rate": 1.676624524672569e-05,
      "loss": 0.3396,
      "step": 9324
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.008118502786457,
      "learning_rate": 1.6765514861616977e-05,
      "loss": 0.5102,
      "step": 9325
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7103265612336929,
      "learning_rate": 1.67647844099469e-05,
      "loss": 0.405,
      "step": 9326
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3373811103438286,
      "learning_rate": 1.6764053891722652e-05,
      "loss": 0.2628,
      "step": 9327
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2673809620830444,
      "learning_rate": 1.6763323306951413e-05,
      "loss": 0.1657,
      "step": 9328
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.46812484774379637,
      "learning_rate": 1.6762592655640373e-05,
      "loss": 0.2595,
      "step": 9329
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41269604403840204,
      "learning_rate": 1.6761861937796725e-05,
      "loss": 0.2671,
      "step": 9330
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.36632821673234806,
      "learning_rate": 1.676113115342765e-05,
      "loss": 0.3116,
      "step": 9331
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9406984522108903,
      "learning_rate": 1.6760400302540348e-05,
      "loss": 0.4219,
      "step": 9332
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8297982301171707,
      "learning_rate": 1.6759669385142002e-05,
      "loss": 0.9155,
      "step": 9333
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8017279289711542,
      "learning_rate": 1.6758938401239802e-05,
      "loss": 0.4428,
      "step": 9334
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38375940295990685,
      "learning_rate": 1.675820735084094e-05,
      "loss": 0.2764,
      "step": 9335
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.29959009140615456,
      "learning_rate": 1.6757476233952615e-05,
      "loss": 0.2551,
      "step": 9336
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7803513382103593,
      "learning_rate": 1.675674505058201e-05,
      "loss": 0.9261,
      "step": 9337
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2536285159878965,
      "learning_rate": 1.675601380073633e-05,
      "loss": 0.0746,
      "step": 9338
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6417980240137131,
      "learning_rate": 1.675528248442276e-05,
      "loss": 0.4294,
      "step": 9339
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3256835243452297,
      "learning_rate": 1.6754551101648497e-05,
      "loss": 0.2399,
      "step": 9340
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9039903652416288,
      "learning_rate": 1.6753819652420742e-05,
      "loss": 0.6098,
      "step": 9341
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33935400070044236,
      "learning_rate": 1.6753088136746684e-05,
      "loss": 0.2689,
      "step": 9342
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8685678659939403,
      "learning_rate": 1.6752356554633524e-05,
      "loss": 0.4711,
      "step": 9343
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6561275573725024,
      "learning_rate": 1.6751624906088457e-05,
      "loss": 0.3463,
      "step": 9344
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3825424783959192,
      "learning_rate": 1.6750893191118683e-05,
      "loss": 0.2507,
      "step": 9345
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3195294871513739,
      "learning_rate": 1.67501614097314e-05,
      "loss": 0.2215,
      "step": 9346
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34570481994759555,
      "learning_rate": 1.674942956193381e-05,
      "loss": 0.2645,
      "step": 9347
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8101713848705278,
      "learning_rate": 1.6748697647733108e-05,
      "loss": 0.5376,
      "step": 9348
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3229923192913546,
      "learning_rate": 1.67479656671365e-05,
      "loss": 0.1784,
      "step": 9349
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4469395306841094,
      "learning_rate": 1.6747233620151183e-05,
      "loss": 0.3968,
      "step": 9350
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8764711606138166,
      "learning_rate": 1.674650150678436e-05,
      "loss": 0.6386,
      "step": 9351
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3725944566078134,
      "learning_rate": 1.6745769327043243e-05,
      "loss": 0.7339,
      "step": 9352
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.28341442128937433,
      "learning_rate": 1.6745037080935022e-05,
      "loss": 0.1654,
      "step": 9353
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31577188486846125,
      "learning_rate": 1.674430476846691e-05,
      "loss": 0.2785,
      "step": 9354
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2272210095295744,
      "learning_rate": 1.6743572389646106e-05,
      "loss": 0.0708,
      "step": 9355
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7729081486896662,
      "learning_rate": 1.6742839944479817e-05,
      "loss": 0.3826,
      "step": 9356
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41235985209509934,
      "learning_rate": 1.6742107432975253e-05,
      "loss": 0.3177,
      "step": 9357
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2969411421699476,
      "learning_rate": 1.6741374855139616e-05,
      "loss": 0.1928,
      "step": 9358
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4689465046902645,
      "learning_rate": 1.6740642210980117e-05,
      "loss": 0.4052,
      "step": 9359
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4299412491285371,
      "learning_rate": 1.673990950050396e-05,
      "loss": 0.2609,
      "step": 9360
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2710136253458517,
      "learning_rate": 1.6739176723718355e-05,
      "loss": 0.776,
      "step": 9361
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5405529779751209,
      "learning_rate": 1.673844388063051e-05,
      "loss": 0.1823,
      "step": 9362
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.414140099051515,
      "learning_rate": 1.673771097124764e-05,
      "loss": 0.3281,
      "step": 9363
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.27673738056909336,
      "learning_rate": 1.6736977995576956e-05,
      "loss": 0.0743,
      "step": 9364
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42688024321126655,
      "learning_rate": 1.673624495362566e-05,
      "loss": 0.3696,
      "step": 9365
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3535068526554479,
      "learning_rate": 1.6735511845400968e-05,
      "loss": 0.2543,
      "step": 9366
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3791908017064451,
      "learning_rate": 1.67347786709101e-05,
      "loss": 0.2594,
      "step": 9367
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5125544665828411,
      "learning_rate": 1.673404543016026e-05,
      "loss": 0.2358,
      "step": 9368
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8746099773571325,
      "learning_rate": 1.6733312123158666e-05,
      "loss": 0.6598,
      "step": 9369
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4712006928732005,
      "learning_rate": 1.6732578749912535e-05,
      "loss": 0.383,
      "step": 9370
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32074395833297986,
      "learning_rate": 1.6731845310429072e-05,
      "loss": 0.1903,
      "step": 9371
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41285204235635337,
      "learning_rate": 1.6731111804715506e-05,
      "loss": 0.3295,
      "step": 9372
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6620137929302996,
      "learning_rate": 1.6730378232779045e-05,
      "loss": 0.3533,
      "step": 9373
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5766685785682418,
      "learning_rate": 1.6729644594626907e-05,
      "loss": 0.4551,
      "step": 9374
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2279372588151104,
      "learning_rate": 1.6728910890266314e-05,
      "loss": 0.1024,
      "step": 9375
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6466282409387244,
      "learning_rate": 1.672817711970448e-05,
      "loss": 0.3097,
      "step": 9376
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.29211913837633574,
      "learning_rate": 1.6727443282948625e-05,
      "loss": 0.2388,
      "step": 9377
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4775733765730835,
      "learning_rate": 1.6726709380005968e-05,
      "loss": 0.3848,
      "step": 9378
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7197457439182444,
      "learning_rate": 1.6725975410883733e-05,
      "loss": 0.3533,
      "step": 9379
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.189470139188369,
      "learning_rate": 1.672524137558914e-05,
      "loss": 0.734,
      "step": 9380
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.28598389402786734,
      "learning_rate": 1.6724507274129408e-05,
      "loss": 0.1999,
      "step": 9381
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7978710337838407,
      "learning_rate": 1.672377310651176e-05,
      "loss": 0.4168,
      "step": 9382
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.49842953296919806,
      "learning_rate": 1.672303887274342e-05,
      "loss": 0.3803,
      "step": 9383
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4003545564715138,
      "learning_rate": 1.6722304572831613e-05,
      "loss": 0.2715,
      "step": 9384
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43493530685358434,
      "learning_rate": 1.672157020678356e-05,
      "loss": 0.3012,
      "step": 9385
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.29756559232195,
      "learning_rate": 1.672083577460649e-05,
      "loss": 0.1939,
      "step": 9386
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9928024879051622,
      "learning_rate": 1.6720101276307625e-05,
      "loss": 0.771,
      "step": 9387
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2717165106528491,
      "learning_rate": 1.671936671189419e-05,
      "loss": 0.1934,
      "step": 9388
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4415390876416986,
      "learning_rate": 1.6718632081373415e-05,
      "loss": 0.3379,
      "step": 9389
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2884671701343386,
      "learning_rate": 1.671789738475253e-05,
      "loss": 0.2129,
      "step": 9390
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2526940068152708,
      "learning_rate": 1.6717162622038756e-05,
      "loss": 0.5199,
      "step": 9391
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5871572250002113,
      "learning_rate": 1.6716427793239327e-05,
      "loss": 0.3828,
      "step": 9392
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.40812637200790364,
      "learning_rate": 1.671569289836147e-05,
      "loss": 0.22,
      "step": 9393
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3383071848185383,
      "learning_rate": 1.671495793741242e-05,
      "loss": 0.204,
      "step": 9394
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5971467672602521,
      "learning_rate": 1.6714222910399404e-05,
      "loss": 0.3922,
      "step": 9395
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34645207038780634,
      "learning_rate": 1.6713487817329653e-05,
      "loss": 0.2593,
      "step": 9396
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31625037135242556,
      "learning_rate": 1.67127526582104e-05,
      "loss": 0.1726,
      "step": 9397
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9252710589097731,
      "learning_rate": 1.6712017433048876e-05,
      "loss": 0.4442,
      "step": 9398
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3914925383699401,
      "learning_rate": 1.6711282141852317e-05,
      "loss": 0.1942,
      "step": 9399
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5650371698177452,
      "learning_rate": 1.6710546784627955e-05,
      "loss": 0.348,
      "step": 9400
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3085549590719583,
      "learning_rate": 1.670981136138303e-05,
      "loss": 0.2872,
      "step": 9401
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7601725048453845,
      "learning_rate": 1.6709075872124767e-05,
      "loss": 0.8524,
      "step": 9402
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3208293359911852,
      "learning_rate": 1.6708340316860413e-05,
      "loss": 0.0733,
      "step": 9403
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3816628654893855,
      "learning_rate": 1.67076046955972e-05,
      "loss": 0.3272,
      "step": 9404
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22378945422815147,
      "learning_rate": 1.670686900834236e-05,
      "loss": 0.1515,
      "step": 9405
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45881572673152127,
      "learning_rate": 1.6706133255103138e-05,
      "loss": 0.3276,
      "step": 9406
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32893695597318495,
      "learning_rate": 1.6705397435886773e-05,
      "loss": 0.2919,
      "step": 9407
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38769314650259806,
      "learning_rate": 1.6704661550700496e-05,
      "loss": 0.2656,
      "step": 9408
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.49533531239040757,
      "learning_rate": 1.670392559955156e-05,
      "loss": 0.318,
      "step": 9409
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1027690589054961,
      "learning_rate": 1.6703189582447193e-05,
      "loss": 0.3599,
      "step": 9410
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5045761213361182,
      "learning_rate": 1.670245349939464e-05,
      "loss": 0.641,
      "step": 9411
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.30590232968815423,
      "learning_rate": 1.6701717350401145e-05,
      "loss": 0.2054,
      "step": 9412
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.36068475748489087,
      "learning_rate": 1.670098113547395e-05,
      "loss": 0.323,
      "step": 9413
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7923741067474607,
      "learning_rate": 1.67002448546203e-05,
      "loss": 0.5078,
      "step": 9414
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2735631541926123,
      "learning_rate": 1.6699508507847433e-05,
      "loss": 0.1817,
      "step": 9415
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5599006864864108,
      "learning_rate": 1.66987720951626e-05,
      "loss": 0.3508,
      "step": 9416
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3761213365700857,
      "learning_rate": 1.6698035616573044e-05,
      "loss": 0.3055,
      "step": 9417
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7820037819529887,
      "learning_rate": 1.6697299072086005e-05,
      "loss": 0.3504,
      "step": 9418
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38517716221725234,
      "learning_rate": 1.669656246170874e-05,
      "loss": 0.3589,
      "step": 9419
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2213125414146848,
      "learning_rate": 1.6695825785448484e-05,
      "loss": 0.3214,
      "step": 9420
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6038725306718835,
      "learning_rate": 1.669508904331249e-05,
      "loss": 0.3843,
      "step": 9421
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3560768785499123,
      "learning_rate": 1.6694352235308012e-05,
      "loss": 0.2319,
      "step": 9422
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8100669625730177,
      "learning_rate": 1.669361536144229e-05,
      "loss": 0.5111,
      "step": 9423
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5414903525214754,
      "learning_rate": 1.669287842172258e-05,
      "loss": 0.3893,
      "step": 9424
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.40250610978113727,
      "learning_rate": 1.6692141416156135e-05,
      "loss": 0.2793,
      "step": 9425
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9862769179310844,
      "learning_rate": 1.6691404344750194e-05,
      "loss": 0.4901,
      "step": 9426
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2845086983178878,
      "learning_rate": 1.6690667207512016e-05,
      "loss": 0.1481,
      "step": 9427
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9533294928976466,
      "learning_rate": 1.668993000444885e-05,
      "loss": 0.7992,
      "step": 9428
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4189154729955989,
      "learning_rate": 1.6689192735567956e-05,
      "loss": 0.3093,
      "step": 9429
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5206695056883655,
      "learning_rate": 1.668845540087658e-05,
      "loss": 0.3906,
      "step": 9430
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.28019934418482534,
      "learning_rate": 1.668771800038198e-05,
      "loss": 0.2,
      "step": 9431
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3908987731875728,
      "learning_rate": 1.6686980534091414e-05,
      "loss": 0.9175,
      "step": 9432
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6349924055270901,
      "learning_rate": 1.6686243002012127e-05,
      "loss": 0.3885,
      "step": 9433
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9885284877917895,
      "learning_rate": 1.6685505404151385e-05,
      "loss": 0.648,
      "step": 9434
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31774437974791553,
      "learning_rate": 1.668476774051644e-05,
      "loss": 0.2005,
      "step": 9435
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.1895385235143839,
      "learning_rate": 1.6684030011114546e-05,
      "loss": 0.1554,
      "step": 9436
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5366412202257019,
      "learning_rate": 1.668329221595297e-05,
      "loss": 0.3597,
      "step": 9437
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8320866771989895,
      "learning_rate": 1.6682554355038965e-05,
      "loss": 0.0446,
      "step": 9438
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9656273450520871,
      "learning_rate": 1.6681816428379792e-05,
      "loss": 0.5709,
      "step": 9439
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31503162461136786,
      "learning_rate": 1.668107843598271e-05,
      "loss": 0.2238,
      "step": 9440
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.116259085793358,
      "learning_rate": 1.668034037785498e-05,
      "loss": 0.5923,
      "step": 9441
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4361937871844905,
      "learning_rate": 1.6679602254003864e-05,
      "loss": 0.3046,
      "step": 9442
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.44464656438608247,
      "learning_rate": 1.667886406443662e-05,
      "loss": 0.3778,
      "step": 9443
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3105755346625949,
      "learning_rate": 1.667812580916052e-05,
      "loss": 0.195,
      "step": 9444
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7123700637450938,
      "learning_rate": 1.6677387488182816e-05,
      "loss": 0.4293,
      "step": 9445
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.24221872568823039,
      "learning_rate": 1.667664910151078e-05,
      "loss": 0.0992,
      "step": 9446
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5804953632762326,
      "learning_rate": 1.667591064915167e-05,
      "loss": 0.714,
      "step": 9447
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2948413156090173,
      "learning_rate": 1.6675172131112754e-05,
      "loss": 0.2173,
      "step": 9448
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33449888305001085,
      "learning_rate": 1.6674433547401298e-05,
      "loss": 0.2334,
      "step": 9449
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.482859399037128,
      "learning_rate": 1.667369489802457e-05,
      "loss": 0.3852,
      "step": 9450
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7694119938399695,
      "learning_rate": 1.6672956182989838e-05,
      "loss": 0.4092,
      "step": 9451
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.9541909352856992,
      "learning_rate": 1.6672217402304366e-05,
      "loss": 0.8872,
      "step": 9452
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2911992456570553,
      "learning_rate": 1.6671478555975423e-05,
      "loss": 0.0747,
      "step": 9453
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.29270857481710394,
      "learning_rate": 1.6670739644010276e-05,
      "loss": 0.2416,
      "step": 9454
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25236761753879844,
      "learning_rate": 1.6670000666416203e-05,
      "loss": 0.2166,
      "step": 9455
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9645077611534772,
      "learning_rate": 1.6669261623200462e-05,
      "loss": 0.6164,
      "step": 9456
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0799584343924695,
      "learning_rate": 1.6668522514370337e-05,
      "loss": 0.3901,
      "step": 9457
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4317101975627504,
      "learning_rate": 1.666778333993309e-05,
      "loss": 0.3016,
      "step": 9458
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4295159348880152,
      "learning_rate": 1.6667044099895998e-05,
      "loss": 0.316,
      "step": 9459
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41626902836655433,
      "learning_rate": 1.6666304794266333e-05,
      "loss": 0.3483,
      "step": 9460
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0493167924808795,
      "learning_rate": 1.6665565423051364e-05,
      "loss": 0.4974,
      "step": 9461
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42734241488547664,
      "learning_rate": 1.6664825986258372e-05,
      "loss": 0.0716,
      "step": 9462
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3051644345813455,
      "learning_rate": 1.666408648389463e-05,
      "loss": 0.2636,
      "step": 9463
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3425536420691151,
      "learning_rate": 1.666334691596741e-05,
      "loss": 0.1499,
      "step": 9464
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3488748978479423,
      "learning_rate": 1.6662607282483995e-05,
      "loss": 0.678,
      "step": 9465
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.30765716007389043,
      "learning_rate": 1.666186758345165e-05,
      "loss": 0.2552,
      "step": 9466
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37703819563028407,
      "learning_rate": 1.6661127818877663e-05,
      "loss": 0.2815,
      "step": 9467
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7275104967793081,
      "learning_rate": 1.666038798876931e-05,
      "loss": 0.3865,
      "step": 9468
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.871441155019135,
      "learning_rate": 1.6659648093133868e-05,
      "loss": 0.5533,
      "step": 9469
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1161483729790944,
      "learning_rate": 1.665890813197862e-05,
      "loss": 0.501,
      "step": 9470
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7921432317138464,
      "learning_rate": 1.6658168105310836e-05,
      "loss": 0.3177,
      "step": 9471
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.384660724205866,
      "learning_rate": 1.6657428013137807e-05,
      "loss": 0.2292,
      "step": 9472
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4358336837638683,
      "learning_rate": 1.6656687855466807e-05,
      "loss": 0.2445,
      "step": 9473
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2635164183364023,
      "learning_rate": 1.6655947632305128e-05,
      "loss": 0.1428,
      "step": 9474
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6546107412508466,
      "learning_rate": 1.665520734366004e-05,
      "loss": 0.3422,
      "step": 9475
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.114747163618711,
      "learning_rate": 1.6654466989538836e-05,
      "loss": 0.5053,
      "step": 9476
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3292268856462903,
      "learning_rate": 1.6653726569948796e-05,
      "loss": 0.2653,
      "step": 9477
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37935451943140225,
      "learning_rate": 1.6652986084897204e-05,
      "loss": 0.3415,
      "step": 9478
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5081632301991224,
      "learning_rate": 1.6652245534391344e-05,
      "loss": 0.4463,
      "step": 9479
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0027330605393376,
      "learning_rate": 1.6651504918438505e-05,
      "loss": 0.5268,
      "step": 9480
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3192461779721762,
      "learning_rate": 1.6650764237045973e-05,
      "loss": 0.2169,
      "step": 9481
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31042530832873233,
      "learning_rate": 1.665002349022103e-05,
      "loss": 0.2014,
      "step": 9482
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37282044911607676,
      "learning_rate": 1.6649282677970972e-05,
      "loss": 0.318,
      "step": 9483
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5195865303304678,
      "learning_rate": 1.6648541800303085e-05,
      "loss": 0.4168,
      "step": 9484
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3461223830999629,
      "learning_rate": 1.664780085722465e-05,
      "loss": 0.2246,
      "step": 9485
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.544489942007544,
      "learning_rate": 1.6647059848742968e-05,
      "loss": 0.4588,
      "step": 9486
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9287388946975056,
      "learning_rate": 1.6646318774865323e-05,
      "loss": 0.6571,
      "step": 9487
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.14598687694448,
      "learning_rate": 1.6645577635599005e-05,
      "loss": 0.0969,
      "step": 9488
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7400845771288942,
      "learning_rate": 1.6644836430951307e-05,
      "loss": 0.4342,
      "step": 9489
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25966124877655744,
      "learning_rate": 1.6644095160929523e-05,
      "loss": 0.2212,
      "step": 9490
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25764636864045115,
      "learning_rate": 1.6643353825540945e-05,
      "loss": 0.2193,
      "step": 9491
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5857547370811065,
      "learning_rate": 1.664261242479287e-05,
      "loss": 0.3781,
      "step": 9492
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5296709458026548,
      "learning_rate": 1.664187095869258e-05,
      "loss": 0.4622,
      "step": 9493
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2979146949916381,
      "learning_rate": 1.664112942724738e-05,
      "loss": 0.1952,
      "step": 9494
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7903456817339523,
      "learning_rate": 1.6640387830464568e-05,
      "loss": 0.9401,
      "step": 9495
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32803806822493187,
      "learning_rate": 1.6639646168351432e-05,
      "loss": 0.2613,
      "step": 9496
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4088589229512836,
      "learning_rate": 1.6638904440915274e-05,
      "loss": 0.8109,
      "step": 9497
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4276438006958825,
      "learning_rate": 1.663816264816339e-05,
      "loss": 0.0732,
      "step": 9498
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4276904752338928,
      "learning_rate": 1.663742079010308e-05,
      "loss": 0.3513,
      "step": 9499
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3236371323243782,
      "learning_rate": 1.6636678866741635e-05,
      "loss": 0.1128,
      "step": 9500
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.46024588522604104,
      "learning_rate": 1.663593687808636e-05,
      "loss": 0.3291,
      "step": 9501
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31479604249361937,
      "learning_rate": 1.6635194824144555e-05,
      "loss": 0.2799,
      "step": 9502
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4272376325176475,
      "learning_rate": 1.663445270492352e-05,
      "loss": 0.0737,
      "step": 9503
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.380681628515514,
      "learning_rate": 1.663371052043056e-05,
      "loss": 0.3081,
      "step": 9504
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8063478040890877,
      "learning_rate": 1.663296827067297e-05,
      "loss": 0.6369,
      "step": 9505
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8629758707616186,
      "learning_rate": 1.6632225955658053e-05,
      "loss": 0.5713,
      "step": 9506
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3909314162848485,
      "learning_rate": 1.6631483575393122e-05,
      "loss": 0.2228,
      "step": 9507
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4101599418059564,
      "learning_rate": 1.663074112988547e-05,
      "loss": 0.2936,
      "step": 9508
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33409824552679196,
      "learning_rate": 1.66299986191424e-05,
      "loss": 0.2788,
      "step": 9509
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.713415644387562,
      "learning_rate": 1.6629256043171233e-05,
      "loss": 0.4727,
      "step": 9510
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2784426017972477,
      "learning_rate": 1.6628513401979252e-05,
      "loss": 0.0736,
      "step": 9511
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8283968581531456,
      "learning_rate": 1.662777069557378e-05,
      "loss": 0.3787,
      "step": 9512
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32712549912300953,
      "learning_rate": 1.6627027923962122e-05,
      "loss": 0.2632,
      "step": 9513
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3057546133717422,
      "learning_rate": 1.662628508715158e-05,
      "loss": 0.2749,
      "step": 9514
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.5052583889823572,
      "learning_rate": 1.6625542185149463e-05,
      "loss": 0.6478,
      "step": 9515
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5242797796165506,
      "learning_rate": 1.6624799217963083e-05,
      "loss": 0.1678,
      "step": 9516
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.35245639222873276,
      "learning_rate": 1.662405618559975e-05,
      "loss": 0.3098,
      "step": 9517
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6150972617434968,
      "learning_rate": 1.6623313088066774e-05,
      "loss": 0.3573,
      "step": 9518
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8583372288073551,
      "learning_rate": 1.662256992537146e-05,
      "loss": 0.4371,
      "step": 9519
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2612519341760607,
      "learning_rate": 1.6621826697521127e-05,
      "loss": 0.2032,
      "step": 9520
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4000487125320583,
      "learning_rate": 1.6621083404523084e-05,
      "loss": 0.207,
      "step": 9521
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33714431013393953,
      "learning_rate": 1.6620340046384642e-05,
      "loss": 0.234,
      "step": 9522
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4648136742946325,
      "learning_rate": 1.6619596623113118e-05,
      "loss": 0.9744,
      "step": 9523
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.3198372705534291,
      "learning_rate": 1.6618853134715826e-05,
      "loss": 0.3853,
      "step": 9524
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4266621452896047,
      "learning_rate": 1.6618109581200074e-05,
      "loss": 0.3352,
      "step": 9525
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31768649560436985,
      "learning_rate": 1.6617365962573187e-05,
      "loss": 0.2114,
      "step": 9526
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4324284143393936,
      "learning_rate": 1.6616622278842473e-05,
      "loss": 0.3216,
      "step": 9527
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7377788106190393,
      "learning_rate": 1.6615878530015255e-05,
      "loss": 0.519,
      "step": 9528
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.30788404849887013,
      "learning_rate": 1.6615134716098844e-05,
      "loss": 0.1665,
      "step": 9529
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7820286313966853,
      "learning_rate": 1.6614390837100565e-05,
      "loss": 0.4161,
      "step": 9530
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3482973932362064,
      "learning_rate": 1.6613646893027726e-05,
      "loss": 0.256,
      "step": 9531
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33066428117941776,
      "learning_rate": 1.661290288388766e-05,
      "loss": 0.3413,
      "step": 9532
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.23317101092141837,
      "learning_rate": 1.6612158809687676e-05,
      "loss": 0.1012,
      "step": 9533
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9262552749795636,
      "learning_rate": 1.6611414670435095e-05,
      "loss": 0.6164,
      "step": 9534
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3145490452543163,
      "learning_rate": 1.6610670466137245e-05,
      "loss": 0.2258,
      "step": 9535
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43298567761905526,
      "learning_rate": 1.6609926196801444e-05,
      "loss": 0.3634,
      "step": 9536
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2935057686233021,
      "learning_rate": 1.6609181862435013e-05,
      "loss": 0.2387,
      "step": 9537
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.2972451821242004,
      "learning_rate": 1.6608437463045275e-05,
      "loss": 0.6279,
      "step": 9538
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2702974627103093,
      "learning_rate": 1.6607692998639554e-05,
      "loss": 0.1292,
      "step": 9539
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3061256563640889,
      "learning_rate": 1.660694846922518e-05,
      "loss": 0.2324,
      "step": 9540
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.43144165283068703,
      "learning_rate": 1.6606203874809467e-05,
      "loss": 0.3036,
      "step": 9541
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.013564969838194,
      "learning_rate": 1.660545921539975e-05,
      "loss": 0.2858,
      "step": 9542
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34379602114734237,
      "learning_rate": 1.660471449100335e-05,
      "loss": 0.3162,
      "step": 9543
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2940192549394352,
      "learning_rate": 1.66039697016276e-05,
      "loss": 0.2096,
      "step": 9544
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9237074028014463,
      "learning_rate": 1.6603224847279816e-05,
      "loss": 0.4878,
      "step": 9545
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8539455576052106,
      "learning_rate": 1.660247992796734e-05,
      "loss": 0.03,
      "step": 9546
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1030170026800084,
      "learning_rate": 1.660173494369749e-05,
      "loss": 0.5828,
      "step": 9547
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.37186591469262864,
      "learning_rate": 1.66009898944776e-05,
      "loss": 0.1689,
      "step": 9548
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.39526841616732183,
      "learning_rate": 1.6600244780315e-05,
      "loss": 0.2944,
      "step": 9549
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32509820252420957,
      "learning_rate": 1.659949960121702e-05,
      "loss": 0.258,
      "step": 9550
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.41027345003127924,
      "learning_rate": 1.659875435719099e-05,
      "loss": 0.2205,
      "step": 9551
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.805765569198155,
      "learning_rate": 1.6598009048244246e-05,
      "loss": 0.3991,
      "step": 9552
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.45756015892658114,
      "learning_rate": 1.6597263674384116e-05,
      "loss": 0.1825,
      "step": 9553
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6346477473837843,
      "learning_rate": 1.6596518235617934e-05,
      "loss": 0.3989,
      "step": 9554
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3225621262774873,
      "learning_rate": 1.659577273195304e-05,
      "loss": 0.2591,
      "step": 9555
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8012024330742,
      "learning_rate": 1.659502716339676e-05,
      "loss": 0.9147,
      "step": 9556
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5940178037152175,
      "learning_rate": 1.6594281529956432e-05,
      "loss": 0.1295,
      "step": 9557
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.39330333295544884,
      "learning_rate": 1.6593535831639393e-05,
      "loss": 0.3074,
      "step": 9558
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.27436361469770854,
      "learning_rate": 1.659279006845298e-05,
      "loss": 0.1596,
      "step": 9559
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7779078552879619,
      "learning_rate": 1.659204424040453e-05,
      "loss": 0.5048,
      "step": 9560
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31221410972892455,
      "learning_rate": 1.6591298347501378e-05,
      "loss": 0.2465,
      "step": 9561
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5850384475520336,
      "learning_rate": 1.6590552389750866e-05,
      "loss": 0.3865,
      "step": 9562
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25690887369796206,
      "learning_rate": 1.6589806367160327e-05,
      "loss": 0.1919,
      "step": 9563
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8505439712858252,
      "learning_rate": 1.6589060279737107e-05,
      "loss": 0.595,
      "step": 9564
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9467635524653157,
      "learning_rate": 1.658831412748854e-05,
      "loss": 0.0652,
      "step": 9565
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5759466461402606,
      "learning_rate": 1.6587567910421976e-05,
      "loss": 0.2843,
      "step": 9566
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.38102469525173477,
      "learning_rate": 1.6586821628544752e-05,
      "loss": 0.2904,
      "step": 9567
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3344012682105654,
      "learning_rate": 1.6586075281864205e-05,
      "loss": 0.2778,
      "step": 9568
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9430995615323662,
      "learning_rate": 1.6585328870387684e-05,
      "loss": 0.6015,
      "step": 9569
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5315072851510505,
      "learning_rate": 1.6584582394122534e-05,
      "loss": 0.2737,
      "step": 9570
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4611180421051092,
      "learning_rate": 1.658383585307609e-05,
      "loss": 0.2074,
      "step": 9571
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32599486776659603,
      "learning_rate": 1.6583089247255708e-05,
      "loss": 0.2482,
      "step": 9572
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.355055500731627,
      "learning_rate": 1.6582342576668725e-05,
      "loss": 0.7072,
      "step": 9573
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3166677077083047,
      "learning_rate": 1.658159584132249e-05,
      "loss": 0.2634,
      "step": 9574
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8131612659768418,
      "learning_rate": 1.658084904122435e-05,
      "loss": 0.4827,
      "step": 9575
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3686292389868505,
      "learning_rate": 1.6580102176381648e-05,
      "loss": 0.2382,
      "step": 9576
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7167234939070557,
      "learning_rate": 1.6579355246801744e-05,
      "loss": 0.911,
      "step": 9577
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.8129068087951703,
      "learning_rate": 1.6578608252491973e-05,
      "loss": 0.3918,
      "step": 9578
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3392335630267796,
      "learning_rate": 1.657786119345969e-05,
      "loss": 0.3511,
      "step": 9579
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3276196896517309,
      "learning_rate": 1.6577114069712244e-05,
      "loss": 0.1715,
      "step": 9580
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31426833270549803,
      "learning_rate": 1.6576366881256985e-05,
      "loss": 0.1629,
      "step": 9581
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.42194715883009626,
      "learning_rate": 1.6575619628101263e-05,
      "loss": 0.2035,
      "step": 9582
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.0853811751950777,
      "learning_rate": 1.6574872310252434e-05,
      "loss": 0.4733,
      "step": 9583
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5526604409916441,
      "learning_rate": 1.6574124927717847e-05,
      "loss": 0.4087,
      "step": 9584
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3499920046635511,
      "learning_rate": 1.6573377480504857e-05,
      "loss": 0.1757,
      "step": 9585
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.36456609583507815,
      "learning_rate": 1.6572629968620817e-05,
      "loss": 0.3293,
      "step": 9586
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6268149541814423,
      "learning_rate": 1.6571882392073076e-05,
      "loss": 0.3962,
      "step": 9587
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.4898503312676636,
      "learning_rate": 1.6571134750869e-05,
      "loss": 0.8126,
      "step": 9588
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22902059093002933,
      "learning_rate": 1.6570387045015933e-05,
      "loss": 0.0722,
      "step": 9589
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.34358037247487855,
      "learning_rate": 1.656963927452124e-05,
      "loss": 0.2627,
      "step": 9590
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.33413040637380526,
      "learning_rate": 1.656889143939227e-05,
      "loss": 0.2606,
      "step": 9591
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.7345413729907153,
      "learning_rate": 1.656814353963639e-05,
      "loss": 0.7026,
      "step": 9592
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9350333444626793,
      "learning_rate": 1.656739557526095e-05,
      "loss": 0.5923,
      "step": 9593
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.31778315199755486,
      "learning_rate": 1.656664754627331e-05,
      "loss": 0.2084,
      "step": 9594
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7137668044507086,
      "learning_rate": 1.6565899452680832e-05,
      "loss": 0.46,
      "step": 9595
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6945471320211333,
      "learning_rate": 1.6565151294490873e-05,
      "loss": 0.3862,
      "step": 9596
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3317070997173432,
      "learning_rate": 1.6564403071710797e-05,
      "loss": 0.3042,
      "step": 9597
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3725371784621213,
      "learning_rate": 1.6563654784347965e-05,
      "loss": 0.1346,
      "step": 9598
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.27382946687350784,
      "learning_rate": 1.6562906432409737e-05,
      "loss": 0.2547,
      "step": 9599
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.22252260022323095,
      "learning_rate": 1.6562158015903474e-05,
      "loss": 0.0759,
      "step": 9600
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.6141429129193525,
      "learning_rate": 1.656140953483654e-05,
      "loss": 0.8845,
      "step": 9601
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3657025065697399,
      "learning_rate": 1.6560660989216304e-05,
      "loss": 0.3007,
      "step": 9602
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3850834650182277,
      "learning_rate": 1.6559912379050124e-05,
      "loss": 0.275,
      "step": 9603
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.46843365880839827,
      "learning_rate": 1.655916370434537e-05,
      "loss": 0.2653,
      "step": 9604
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.741812929175144,
      "learning_rate": 1.6558414965109403e-05,
      "loss": 0.5031,
      "step": 9605
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9473775155213475,
      "learning_rate": 1.655766616134959e-05,
      "loss": 0.392,
      "step": 9606
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.25637176046688254,
      "learning_rate": 1.6556917293073302e-05,
      "loss": 0.1338,
      "step": 9607
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5367847541721512,
      "learning_rate": 1.6556168360287905e-05,
      "loss": 0.3485,
      "step": 9608
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2452986299515768,
      "learning_rate": 1.6555419363000763e-05,
      "loss": 0.2126,
      "step": 9609
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1172054098158064,
      "learning_rate": 1.6554670301219252e-05,
      "loss": 0.6306,
      "step": 9610
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7084351175976011,
      "learning_rate": 1.6553921174950737e-05,
      "loss": 0.3811,
      "step": 9611
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6432429979498852,
      "learning_rate": 1.6553171984202588e-05,
      "loss": 0.3824,
      "step": 9612
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3387025276532711,
      "learning_rate": 1.6552422728982177e-05,
      "loss": 0.2709,
      "step": 9613
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.1934808797591459,
      "learning_rate": 1.6551673409296873e-05,
      "loss": 0.7255,
      "step": 9614
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32254324204681795,
      "learning_rate": 1.6550924025154053e-05,
      "loss": 0.2509,
      "step": 9615
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.738736627495235,
      "learning_rate": 1.6550174576561085e-05,
      "loss": 0.8327,
      "step": 9616
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.28317187625467855,
      "learning_rate": 1.6549425063525346e-05,
      "loss": 0.2026,
      "step": 9617
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.3471609666169092,
      "learning_rate": 1.6548675486054208e-05,
      "loss": 0.1937,
      "step": 9618
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7132890098347374,
      "learning_rate": 1.6547925844155047e-05,
      "loss": 0.41,
      "step": 9619
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2810102407030721,
      "learning_rate": 1.6547176137835236e-05,
      "loss": 0.2793,
      "step": 9620
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4226942389697432,
      "learning_rate": 1.6546426367102147e-05,
      "loss": 0.3001,
      "step": 9621
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.4370912621181822,
      "learning_rate": 1.654567653196317e-05,
      "loss": 0.3109,
      "step": 9622
    },
    {
      "epoch": 0.29,
      "grad_norm": 1.8574725945704038,
      "learning_rate": 1.6544926632425667e-05,
      "loss": 0.7974,
      "step": 9623
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.7918402218830705,
      "learning_rate": 1.6544176668497024e-05,
      "loss": 0.3954,
      "step": 9624
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9733403139572501,
      "learning_rate": 1.6543426640184618e-05,
      "loss": 0.5587,
      "step": 9625
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2573455733739721,
      "learning_rate": 1.6542676547495827e-05,
      "loss": 0.1733,
      "step": 9626
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5166448193807197,
      "learning_rate": 1.6541926390438034e-05,
      "loss": 0.3869,
      "step": 9627
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.2515117566617355,
      "learning_rate": 1.6541176169018612e-05,
      "loss": 0.1933,
      "step": 9628
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.6853421287094812,
      "learning_rate": 1.6540425883244948e-05,
      "loss": 0.5167,
      "step": 9629
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.32685069896537633,
      "learning_rate": 1.6539675533124423e-05,
      "loss": 0.1932,
      "step": 9630
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.5789906930537786,
      "learning_rate": 1.653892511866442e-05,
      "loss": 0.3895,
      "step": 9631
    },
    {
      "epoch": 0.29,
      "grad_norm": 0.9151569916335893,
      "learning_rate": 1.6538174639872316e-05,
      "loss": 0.5109,
      "step": 9632
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.30286361349443497,
      "learning_rate": 1.6537424096755504e-05,
      "loss": 0.2451,
      "step": 9633
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6730256990331178,
      "learning_rate": 1.653667348932136e-05,
      "loss": 0.3138,
      "step": 9634
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3599611889834981,
      "learning_rate": 1.6535922817577267e-05,
      "loss": 0.2279,
      "step": 9635
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29013891258281976,
      "learning_rate": 1.6535172081530618e-05,
      "loss": 0.1883,
      "step": 9636
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6800603319974959,
      "learning_rate": 1.65344212811888e-05,
      "loss": 0.3854,
      "step": 9637
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33177474017738695,
      "learning_rate": 1.653367041655919e-05,
      "loss": 0.31,
      "step": 9638
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5007288218062986,
      "learning_rate": 1.653291948764918e-05,
      "loss": 0.0787,
      "step": 9639
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40446503530922095,
      "learning_rate": 1.6532168494466162e-05,
      "loss": 0.3138,
      "step": 9640
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1346046048672789,
      "learning_rate": 1.653141743701752e-05,
      "loss": 0.5967,
      "step": 9641
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.146025597142766,
      "learning_rate": 1.6530666315310645e-05,
      "loss": 0.6851,
      "step": 9642
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1451929885467447,
      "learning_rate": 1.6529915129352922e-05,
      "loss": 0.3423,
      "step": 9643
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41574603466098264,
      "learning_rate": 1.6529163879151752e-05,
      "loss": 0.2749,
      "step": 9644
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24859678978822006,
      "learning_rate": 1.6528412564714513e-05,
      "loss": 0.2018,
      "step": 9645
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7915338751907016,
      "learning_rate": 1.6527661186048606e-05,
      "loss": 0.3946,
      "step": 9646
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4166572739833662,
      "learning_rate": 1.652690974316142e-05,
      "loss": 0.2766,
      "step": 9647
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3376202488007468,
      "learning_rate": 1.6526158236060347e-05,
      "loss": 0.1687,
      "step": 9648
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41558966020710336,
      "learning_rate": 1.6525406664752784e-05,
      "loss": 0.3043,
      "step": 9649
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.216495459216468,
      "learning_rate": 1.652465502924612e-05,
      "loss": 0.3897,
      "step": 9650
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3721902261855902,
      "learning_rate": 1.6523903329547758e-05,
      "loss": 0.3206,
      "step": 9651
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3754032054624295,
      "learning_rate": 1.6523151565665086e-05,
      "loss": 0.0619,
      "step": 9652
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36556794899292766,
      "learning_rate": 1.65223997376055e-05,
      "loss": 0.2812,
      "step": 9653
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0063962357396496,
      "learning_rate": 1.65216478453764e-05,
      "loss": 0.3714,
      "step": 9654
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7607793981769135,
      "learning_rate": 1.6520895888985184e-05,
      "loss": 0.4863,
      "step": 9655
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23558950117846192,
      "learning_rate": 1.6520143868439247e-05,
      "loss": 0.1816,
      "step": 9656
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6888928096919482,
      "learning_rate": 1.651939178374599e-05,
      "loss": 0.4073,
      "step": 9657
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36877630215427454,
      "learning_rate": 1.6518639634912815e-05,
      "loss": 0.235,
      "step": 9658
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1169682366833458,
      "learning_rate": 1.6517887421947116e-05,
      "loss": 0.5501,
      "step": 9659
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1174684547637739,
      "learning_rate": 1.6517135144856295e-05,
      "loss": 0.6268,
      "step": 9660
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33268742878580015,
      "learning_rate": 1.651638280364775e-05,
      "loss": 0.2581,
      "step": 9661
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0246766527261146,
      "learning_rate": 1.6515630398328896e-05,
      "loss": 0.4422,
      "step": 9662
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2934017226340593,
      "learning_rate": 1.6514877928907122e-05,
      "loss": 0.2683,
      "step": 9663
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.772172572807782,
      "learning_rate": 1.6514125395389833e-05,
      "loss": 0.4857,
      "step": 9664
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.32118286546818064,
      "learning_rate": 1.6513372797784436e-05,
      "loss": 0.1765,
      "step": 9665
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.906808082597096,
      "learning_rate": 1.6512620136098337e-05,
      "loss": 0.4423,
      "step": 9666
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.32455033588767535,
      "learning_rate": 1.6511867410338935e-05,
      "loss": 0.2348,
      "step": 9667
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.323498134984144,
      "learning_rate": 1.6511114620513637e-05,
      "loss": 0.2683,
      "step": 9668
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41429743203088903,
      "learning_rate": 1.6510361766629853e-05,
      "loss": 0.2663,
      "step": 9669
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8010102615753127,
      "learning_rate": 1.6509608848694992e-05,
      "loss": 0.8754,
      "step": 9670
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3347855403407284,
      "learning_rate": 1.6508855866716452e-05,
      "loss": 0.2327,
      "step": 9671
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36150257724485546,
      "learning_rate": 1.6508102820701645e-05,
      "loss": 0.3388,
      "step": 9672
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7492248917562474,
      "learning_rate": 1.6507349710657987e-05,
      "loss": 0.6363,
      "step": 9673
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3244935573570275,
      "learning_rate": 1.650659653659288e-05,
      "loss": 0.2482,
      "step": 9674
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9121508580380222,
      "learning_rate": 1.650584329851373e-05,
      "loss": 0.5354,
      "step": 9675
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3029033784942933,
      "learning_rate": 1.6505089996427955e-05,
      "loss": 0.2188,
      "step": 9676
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.551736394504739,
      "learning_rate": 1.650433663034297e-05,
      "loss": 0.2928,
      "step": 9677
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2788281673489433,
      "learning_rate": 1.6503583200266174e-05,
      "loss": 0.1145,
      "step": 9678
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.44590200917714384,
      "learning_rate": 1.6502829706204992e-05,
      "loss": 0.3807,
      "step": 9679
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.28164922386557545,
      "learning_rate": 1.6502076148166825e-05,
      "loss": 0.2132,
      "step": 9680
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5101243183596653,
      "learning_rate": 1.6501322526159097e-05,
      "loss": 0.3612,
      "step": 9681
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6256254579691937,
      "learning_rate": 1.650056884018922e-05,
      "loss": 0.3954,
      "step": 9682
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2203312577306675,
      "learning_rate": 1.6499815090264607e-05,
      "loss": 0.6658,
      "step": 9683
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39952304759228774,
      "learning_rate": 1.6499061276392674e-05,
      "loss": 0.1688,
      "step": 9684
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37341865648270045,
      "learning_rate": 1.6498307398580837e-05,
      "loss": 0.2333,
      "step": 9685
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33251004361753705,
      "learning_rate": 1.6497553456836514e-05,
      "loss": 0.2557,
      "step": 9686
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3402691268650784,
      "learning_rate": 1.6496799451167124e-05,
      "loss": 0.2088,
      "step": 9687
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5865594980064657,
      "learning_rate": 1.649604538158008e-05,
      "loss": 0.4711,
      "step": 9688
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4793419758944193,
      "learning_rate": 1.6495291248082808e-05,
      "loss": 0.0753,
      "step": 9689
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38828378484534387,
      "learning_rate": 1.649453705068272e-05,
      "loss": 0.2807,
      "step": 9690
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.681968014777172,
      "learning_rate": 1.649378278938724e-05,
      "loss": 0.3528,
      "step": 9691
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33098756215626396,
      "learning_rate": 1.6493028464203792e-05,
      "loss": 0.3149,
      "step": 9692
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9173516724265776,
      "learning_rate": 1.6492274075139788e-05,
      "loss": 0.4975,
      "step": 9693
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.430565087745628,
      "learning_rate": 1.6491519622202656e-05,
      "loss": 0.292,
      "step": 9694
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.27176379926362615,
      "learning_rate": 1.649076510539982e-05,
      "loss": 0.1846,
      "step": 9695
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.47230446813293175,
      "learning_rate": 1.6490010524738704e-05,
      "loss": 0.3235,
      "step": 9696
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7698779415026034,
      "learning_rate": 1.6489255880226723e-05,
      "loss": 0.3729,
      "step": 9697
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.27724305467922056,
      "learning_rate": 1.6488501171871312e-05,
      "loss": 0.2067,
      "step": 9698
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38400866316214155,
      "learning_rate": 1.648774639967989e-05,
      "loss": 0.3395,
      "step": 9699
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7367693835148932,
      "learning_rate": 1.6486991563659886e-05,
      "loss": 0.0702,
      "step": 9700
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3850302020872056,
      "learning_rate": 1.6486236663818718e-05,
      "loss": 0.6696,
      "step": 9701
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5729826633474047,
      "learning_rate": 1.6485481700163826e-05,
      "loss": 0.2796,
      "step": 9702
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.351997379143732,
      "learning_rate": 1.648472667270263e-05,
      "loss": 0.2767,
      "step": 9703
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.31647003199416124,
      "learning_rate": 1.648397158144256e-05,
      "loss": 0.2415,
      "step": 9704
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5082374027514124,
      "learning_rate": 1.648321642639104e-05,
      "loss": 0.2782,
      "step": 9705
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7154233407104857,
      "learning_rate": 1.6482461207555503e-05,
      "loss": 0.2839,
      "step": 9706
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.69114456097258,
      "learning_rate": 1.6481705924943385e-05,
      "loss": 0.2696,
      "step": 9707
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.359506698080836,
      "learning_rate": 1.6480950578562107e-05,
      "loss": 0.2397,
      "step": 9708
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3488098582799921,
      "learning_rate": 1.648019516841911e-05,
      "loss": 0.6969,
      "step": 9709
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33077545916092727,
      "learning_rate": 1.6479439694521816e-05,
      "loss": 0.2714,
      "step": 9710
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48651487685352246,
      "learning_rate": 1.647868415687766e-05,
      "loss": 0.2841,
      "step": 9711
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.083799032002917,
      "learning_rate": 1.6477928555494083e-05,
      "loss": 0.5127,
      "step": 9712
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37586847479170543,
      "learning_rate": 1.6477172890378512e-05,
      "loss": 0.2803,
      "step": 9713
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.794182751900558,
      "learning_rate": 1.647641716153838e-05,
      "loss": 0.4616,
      "step": 9714
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24449792923764574,
      "learning_rate": 1.6475661368981127e-05,
      "loss": 0.2087,
      "step": 9715
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8507959690417134,
      "learning_rate": 1.6474905512714187e-05,
      "loss": 0.4066,
      "step": 9716
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23974221502378731,
      "learning_rate": 1.6474149592744996e-05,
      "loss": 0.1626,
      "step": 9717
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3527163801030777,
      "learning_rate": 1.647339360908099e-05,
      "loss": 0.7362,
      "step": 9718
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8603989270142012,
      "learning_rate": 1.647263756172961e-05,
      "loss": 0.4622,
      "step": 9719
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.671525371700062,
      "learning_rate": 1.647188145069829e-05,
      "loss": 0.9045,
      "step": 9720
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3157071294101892,
      "learning_rate": 1.647112527599447e-05,
      "loss": 0.1875,
      "step": 9721
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29012737227180857,
      "learning_rate": 1.6470369037625592e-05,
      "loss": 0.2937,
      "step": 9722
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5605929735956613,
      "learning_rate": 1.6469612735599094e-05,
      "loss": 0.3846,
      "step": 9723
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0645774672782244,
      "learning_rate": 1.6468856369922416e-05,
      "loss": 0.2851,
      "step": 9724
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2834283912790384,
      "learning_rate": 1.6468099940603003e-05,
      "loss": 0.1548,
      "step": 9725
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3178669299124444,
      "learning_rate": 1.646734344764829e-05,
      "loss": 0.225,
      "step": 9726
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48025831196009494,
      "learning_rate": 1.646658689106573e-05,
      "loss": 0.3857,
      "step": 9727
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.40995766813340334,
      "learning_rate": 1.6465830270862755e-05,
      "loss": 0.2507,
      "step": 9728
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5263439693509806,
      "learning_rate": 1.6465073587046818e-05,
      "loss": 0.8942,
      "step": 9729
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.27225669973767547,
      "learning_rate": 1.6464316839625357e-05,
      "loss": 0.1962,
      "step": 9730
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.644575997610186,
      "learning_rate": 1.646356002860582e-05,
      "loss": 0.4689,
      "step": 9731
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7942671253586097,
      "learning_rate": 1.6462803153995658e-05,
      "loss": 0.3848,
      "step": 9732
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4692553413377178,
      "learning_rate": 1.6462046215802306e-05,
      "loss": 0.3426,
      "step": 9733
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2214123644688224,
      "learning_rate": 1.646128921403322e-05,
      "loss": 0.1503,
      "step": 9734
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3128134799834813,
      "learning_rate": 1.646053214869584e-05,
      "loss": 0.2638,
      "step": 9735
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2736779094663389,
      "learning_rate": 1.6459775019797623e-05,
      "loss": 0.0748,
      "step": 9736
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8445553187574582,
      "learning_rate": 1.645901782734601e-05,
      "loss": 0.4504,
      "step": 9737
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8121977575686834,
      "learning_rate": 1.645826057134846e-05,
      "loss": 0.4651,
      "step": 9738
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.25662587919749474,
      "learning_rate": 1.6457503251812417e-05,
      "loss": 0.0764,
      "step": 9739
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3435275650335652,
      "learning_rate": 1.6456745868745327e-05,
      "loss": 0.3032,
      "step": 9740
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39291235204005176,
      "learning_rate": 1.645598842215465e-05,
      "loss": 0.3043,
      "step": 9741
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9294700198808907,
      "learning_rate": 1.645523091204784e-05,
      "loss": 0.6148,
      "step": 9742
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.18080521156278448,
      "learning_rate": 1.6454473338432335e-05,
      "loss": 0.0736,
      "step": 9743
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4123316677549291,
      "learning_rate": 1.6453715701315602e-05,
      "loss": 0.303,
      "step": 9744
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3084993366201611,
      "learning_rate": 1.645295800070509e-05,
      "loss": 0.24,
      "step": 9745
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.46452468117288714,
      "learning_rate": 1.645220023660825e-05,
      "loss": 0.4141,
      "step": 9746
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.807358601824422,
      "learning_rate": 1.6451442409032547e-05,
      "loss": 0.406,
      "step": 9747
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5346478752934523,
      "learning_rate": 1.645068451798543e-05,
      "loss": 0.3336,
      "step": 9748
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3012789989582763,
      "learning_rate": 1.644992656347435e-05,
      "loss": 0.2775,
      "step": 9749
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4962491185150175,
      "learning_rate": 1.6449168545506776e-05,
      "loss": 0.5348,
      "step": 9750
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.50062938213586,
      "learning_rate": 1.644841046409016e-05,
      "loss": 0.361,
      "step": 9751
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24887099164219767,
      "learning_rate": 1.644765231923196e-05,
      "loss": 0.1864,
      "step": 9752
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3776210483375961,
      "learning_rate": 1.6446894110939632e-05,
      "loss": 0.3319,
      "step": 9753
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.28443900143236167,
      "learning_rate": 1.6446135839220638e-05,
      "loss": 0.1068,
      "step": 9754
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6467769137000259,
      "learning_rate": 1.6445377504082443e-05,
      "loss": 0.8019,
      "step": 9755
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7443298903753738,
      "learning_rate": 1.64446191055325e-05,
      "loss": 0.3952,
      "step": 9756
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.34389570045321705,
      "learning_rate": 1.644386064357827e-05,
      "loss": 0.2718,
      "step": 9757
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.28599370435398697,
      "learning_rate": 1.6443102118227226e-05,
      "loss": 0.2578,
      "step": 9758
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6834995186996795,
      "learning_rate": 1.644234352948682e-05,
      "loss": 0.8662,
      "step": 9759
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2254613058950699,
      "learning_rate": 1.6441584877364518e-05,
      "loss": 0.371,
      "step": 9760
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3593782913920298,
      "learning_rate": 1.6440826161867788e-05,
      "loss": 0.2226,
      "step": 9761
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3231693068122815,
      "learning_rate": 1.6440067383004086e-05,
      "loss": 0.2084,
      "step": 9762
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48123786522388307,
      "learning_rate": 1.6439308540780886e-05,
      "loss": 0.2769,
      "step": 9763
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3215845573611402,
      "learning_rate": 1.6438549635205646e-05,
      "loss": 0.2815,
      "step": 9764
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6362406555126996,
      "learning_rate": 1.643779066628584e-05,
      "loss": 0.4011,
      "step": 9765
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6822333925533374,
      "learning_rate": 1.643703163402893e-05,
      "loss": 0.3877,
      "step": 9766
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33362286320878015,
      "learning_rate": 1.6436272538442382e-05,
      "loss": 0.2513,
      "step": 9767
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.1198289602980356,
      "learning_rate": 1.643551337953367e-05,
      "loss": 0.7655,
      "step": 9768
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3058414338401155,
      "learning_rate": 1.643475415731026e-05,
      "loss": 0.2673,
      "step": 9769
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3166190829629039,
      "learning_rate": 1.6433994871779617e-05,
      "loss": 0.6963,
      "step": 9770
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3272282360101209,
      "learning_rate": 1.6433235522949217e-05,
      "loss": 0.1658,
      "step": 9771
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26571572040022395,
      "learning_rate": 1.6432476110826533e-05,
      "loss": 0.1864,
      "step": 9772
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5838935550975269,
      "learning_rate": 1.643171663541903e-05,
      "loss": 0.3905,
      "step": 9773
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8133520751689483,
      "learning_rate": 1.6430957096734183e-05,
      "loss": 0.5345,
      "step": 9774
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33414785926120827,
      "learning_rate": 1.6430197494779462e-05,
      "loss": 0.2125,
      "step": 9775
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3421755817003918,
      "learning_rate": 1.6429437829562344e-05,
      "loss": 0.2969,
      "step": 9776
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9774827079936189,
      "learning_rate": 1.64286781010903e-05,
      "loss": 0.5401,
      "step": 9777
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0023315724674915,
      "learning_rate": 1.6427918309370807e-05,
      "loss": 0.3691,
      "step": 9778
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4769904514672766,
      "learning_rate": 1.6427158454411336e-05,
      "loss": 0.2614,
      "step": 9779
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.28251840537823625,
      "learning_rate": 1.6426398536219368e-05,
      "loss": 0.1982,
      "step": 9780
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3057199775896885,
      "learning_rate": 1.6425638554802374e-05,
      "loss": 0.2779,
      "step": 9781
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41422483302233876,
      "learning_rate": 1.6424878510167838e-05,
      "loss": 0.3168,
      "step": 9782
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7181563539511604,
      "learning_rate": 1.6424118402323228e-05,
      "loss": 0.5516,
      "step": 9783
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3761949345875796,
      "learning_rate": 1.6423358231276032e-05,
      "loss": 0.1755,
      "step": 9784
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4054732399784564,
      "learning_rate": 1.642259799703372e-05,
      "loss": 0.3025,
      "step": 9785
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7814099733437343,
      "learning_rate": 1.6421837699603777e-05,
      "loss": 0.408,
      "step": 9786
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.35387069783328656,
      "learning_rate": 1.6421077338993683e-05,
      "loss": 0.2942,
      "step": 9787
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8776102710042941,
      "learning_rate": 1.6420316915210917e-05,
      "loss": 0.3688,
      "step": 9788
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4062462266367006,
      "learning_rate": 1.641955642826296e-05,
      "loss": 0.2818,
      "step": 9789
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4670091206211423,
      "learning_rate": 1.6418795878157294e-05,
      "loss": 0.2855,
      "step": 9790
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4846440829987312,
      "learning_rate": 1.6418035264901403e-05,
      "loss": 0.2742,
      "step": 9791
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3545243027002872,
      "learning_rate": 1.641727458850277e-05,
      "loss": 0.2743,
      "step": 9792
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3233150210361744,
      "learning_rate": 1.641651384896888e-05,
      "loss": 0.204,
      "step": 9793
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3979097445721659,
      "learning_rate": 1.6415753046307208e-05,
      "loss": 0.3291,
      "step": 9794
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9246662165083345,
      "learning_rate": 1.641499218052525e-05,
      "loss": 0.5933,
      "step": 9795
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2108822328137052,
      "learning_rate": 1.641423125163049e-05,
      "loss": 0.7659,
      "step": 9796
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7072531539927732,
      "learning_rate": 1.6413470259630414e-05,
      "loss": 0.3775,
      "step": 9797
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6137008015756046,
      "learning_rate": 1.64127092045325e-05,
      "loss": 0.3413,
      "step": 9798
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29394410843078117,
      "learning_rate": 1.641194808634425e-05,
      "loss": 0.2614,
      "step": 9799
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4107185804030759,
      "learning_rate": 1.641118690507314e-05,
      "loss": 0.3776,
      "step": 9800
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.1665698812929149,
      "learning_rate": 1.6410425660726668e-05,
      "loss": 0.0747,
      "step": 9801
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5014848045323743,
      "learning_rate": 1.6409664353312318e-05,
      "loss": 0.8582,
      "step": 9802
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.30760153771673676,
      "learning_rate": 1.640890298283758e-05,
      "loss": 0.1967,
      "step": 9803
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39320710464644815,
      "learning_rate": 1.6408141549309945e-05,
      "loss": 0.2913,
      "step": 9804
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5462827919211255,
      "learning_rate": 1.640738005273691e-05,
      "loss": 0.3756,
      "step": 9805
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6163034106364076,
      "learning_rate": 1.6406618493125955e-05,
      "loss": 0.3514,
      "step": 9806
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41756576715045896,
      "learning_rate": 1.640585687048458e-05,
      "loss": 0.2926,
      "step": 9807
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.32922779835321503,
      "learning_rate": 1.6405095184820278e-05,
      "loss": 0.3274,
      "step": 9808
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.7377880332088016,
      "learning_rate": 1.6404333436140546e-05,
      "loss": 0.8216,
      "step": 9809
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26911486826308206,
      "learning_rate": 1.640357162445287e-05,
      "loss": 0.1996,
      "step": 9810
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.45291164951651824,
      "learning_rate": 1.640280974976475e-05,
      "loss": 0.3472,
      "step": 9811
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3064489125221783,
      "learning_rate": 1.6402047812083682e-05,
      "loss": 0.2256,
      "step": 9812
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5848786471644265,
      "learning_rate": 1.640128581141716e-05,
      "loss": 0.9089,
      "step": 9813
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2537625358076295,
      "learning_rate": 1.6400523747772683e-05,
      "loss": 0.1129,
      "step": 9814
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6458472257442311,
      "learning_rate": 1.6399761621157746e-05,
      "loss": 0.4488,
      "step": 9815
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.30688721494801907,
      "learning_rate": 1.639899943157985e-05,
      "loss": 0.2141,
      "step": 9816
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.35556123982698934,
      "learning_rate": 1.639823717904649e-05,
      "loss": 0.2416,
      "step": 9817
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5391222146851202,
      "learning_rate": 1.639747486356517e-05,
      "loss": 0.3926,
      "step": 9818
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2047202124758606,
      "learning_rate": 1.6396712485143384e-05,
      "loss": 0.1212,
      "step": 9819
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9066184027951307,
      "learning_rate": 1.639595004378864e-05,
      "loss": 0.5486,
      "step": 9820
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9339063298567519,
      "learning_rate": 1.639518753950843e-05,
      "loss": 0.2046,
      "step": 9821
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3255666478890968,
      "learning_rate": 1.6394424972310264e-05,
      "loss": 0.2582,
      "step": 9822
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.325019708137522,
      "learning_rate": 1.639366234220164e-05,
      "loss": 0.2852,
      "step": 9823
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8052922222761544,
      "learning_rate": 1.6392899649190063e-05,
      "loss": 0.5003,
      "step": 9824
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3577657827512504,
      "learning_rate": 1.6392136893283034e-05,
      "loss": 0.0799,
      "step": 9825
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37369942158309083,
      "learning_rate": 1.6391374074488063e-05,
      "loss": 0.3592,
      "step": 9826
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9538302844210829,
      "learning_rate": 1.6390611192812646e-05,
      "loss": 0.3215,
      "step": 9827
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5968553645499619,
      "learning_rate": 1.6389848248264294e-05,
      "loss": 0.4197,
      "step": 9828
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.38474059246972875,
      "learning_rate": 1.6389085240850516e-05,
      "loss": 0.2764,
      "step": 9829
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2594956364031962,
      "learning_rate": 1.6388322170578812e-05,
      "loss": 0.1959,
      "step": 9830
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.32757743442067044,
      "learning_rate": 1.638755903745669e-05,
      "loss": 0.1823,
      "step": 9831
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8500909677327914,
      "learning_rate": 1.6386795841491668e-05,
      "loss": 0.3471,
      "step": 9832
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6008120975899661,
      "learning_rate": 1.638603258269124e-05,
      "loss": 0.4885,
      "step": 9833
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.305652131539632,
      "learning_rate": 1.6385269261062923e-05,
      "loss": 0.186,
      "step": 9834
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3565971168655629,
      "learning_rate": 1.638450587661423e-05,
      "loss": 0.2919,
      "step": 9835
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9375532376950527,
      "learning_rate": 1.6383742429352665e-05,
      "loss": 0.2974,
      "step": 9836
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0665200306220841,
      "learning_rate": 1.638297891928574e-05,
      "loss": 0.69,
      "step": 9837
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.51636793607456,
      "learning_rate": 1.638221534642097e-05,
      "loss": 0.28,
      "step": 9838
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3683412848908192,
      "learning_rate": 1.6381451710765866e-05,
      "loss": 0.2815,
      "step": 9839
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.19659351731777802,
      "learning_rate": 1.638068801232794e-05,
      "loss": 0.0719,
      "step": 9840
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.34509720572558344,
      "learning_rate": 1.6379924251114702e-05,
      "loss": 0.3417,
      "step": 9841
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5610442185024825,
      "learning_rate": 1.6379160427133675e-05,
      "loss": 0.2863,
      "step": 9842
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3288406747127115,
      "learning_rate": 1.6378396540392368e-05,
      "loss": 0.1691,
      "step": 9843
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5253674353292919,
      "learning_rate": 1.6377632590898293e-05,
      "loss": 0.3648,
      "step": 9844
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.225447487310859,
      "learning_rate": 1.6376868578658976e-05,
      "loss": 0.3858,
      "step": 9845
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.35551850354144965,
      "learning_rate": 1.6376104503681923e-05,
      "loss": 0.3474,
      "step": 9846
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.41167787530590855,
      "learning_rate": 1.6375340365974657e-05,
      "loss": 0.2804,
      "step": 9847
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.862463834123328,
      "learning_rate": 1.6374576165544694e-05,
      "loss": 0.3867,
      "step": 9848
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26983110558971923,
      "learning_rate": 1.6373811902399553e-05,
      "loss": 0.1838,
      "step": 9849
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8921491202693773,
      "learning_rate": 1.6373047576546758e-05,
      "loss": 0.4474,
      "step": 9850
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42635553747163446,
      "learning_rate": 1.6372283187993822e-05,
      "loss": 0.1894,
      "step": 9851
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3423931931461474,
      "learning_rate": 1.637151873674826e-05,
      "loss": 0.3278,
      "step": 9852
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2955262810643111,
      "learning_rate": 1.6370754222817607e-05,
      "loss": 0.2091,
      "step": 9853
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.255167268585019,
      "learning_rate": 1.6369989646209378e-05,
      "loss": 0.6625,
      "step": 9854
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8613328303852035,
      "learning_rate": 1.6369225006931095e-05,
      "loss": 0.4749,
      "step": 9855
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.692529433720846,
      "learning_rate": 1.6368460304990277e-05,
      "loss": 0.465,
      "step": 9856
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3942151874685829,
      "learning_rate": 1.6367695540394453e-05,
      "loss": 0.2433,
      "step": 9857
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4214910391602781,
      "learning_rate": 1.6366930713151146e-05,
      "loss": 0.2931,
      "step": 9858
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5236945985362601,
      "learning_rate": 1.636616582326788e-05,
      "loss": 0.4144,
      "step": 9859
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3809788584696941,
      "learning_rate": 1.6365400870752175e-05,
      "loss": 0.1832,
      "step": 9860
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3858976436096755,
      "learning_rate": 1.6364635855611566e-05,
      "loss": 0.2026,
      "step": 9861
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3428208386384005,
      "learning_rate": 1.6363870777853577e-05,
      "loss": 0.2181,
      "step": 9862
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.8112430725443434,
      "learning_rate": 1.6363105637485727e-05,
      "loss": 0.768,
      "step": 9863
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3168928875200497,
      "learning_rate": 1.6362340434515556e-05,
      "loss": 0.2467,
      "step": 9864
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2890714195052482,
      "learning_rate": 1.6361575168950585e-05,
      "loss": 0.8217,
      "step": 9865
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.34659326711626537,
      "learning_rate": 1.6360809840798343e-05,
      "loss": 0.2292,
      "step": 9866
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5493774829084043,
      "learning_rate": 1.6360044450066362e-05,
      "loss": 0.3715,
      "step": 9867
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6446869103268741,
      "learning_rate": 1.635927899676217e-05,
      "loss": 0.3903,
      "step": 9868
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2553745703516021,
      "learning_rate": 1.6358513480893298e-05,
      "loss": 0.1626,
      "step": 9869
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3591889644277467,
      "learning_rate": 1.635774790246728e-05,
      "loss": 0.3059,
      "step": 9870
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3269620406372855,
      "learning_rate": 1.6356982261491644e-05,
      "loss": 0.2322,
      "step": 9871
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0169292994650787,
      "learning_rate": 1.6356216557973926e-05,
      "loss": 0.5849,
      "step": 9872
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8744169693849018,
      "learning_rate": 1.635545079192166e-05,
      "loss": 0.5105,
      "step": 9873
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7585160503489817,
      "learning_rate": 1.6354684963342372e-05,
      "loss": 0.5267,
      "step": 9874
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.23892407874861496,
      "learning_rate": 1.635391907224361e-05,
      "loss": 0.0735,
      "step": 9875
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36274495505976767,
      "learning_rate": 1.6353153118632896e-05,
      "loss": 0.3357,
      "step": 9876
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33597887836652635,
      "learning_rate": 1.6352387102517774e-05,
      "loss": 0.281,
      "step": 9877
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.48800190877832716,
      "learning_rate": 1.6351621023905776e-05,
      "loss": 0.2732,
      "step": 9878
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.20558619430537614,
      "learning_rate": 1.635085488280444e-05,
      "loss": 0.0767,
      "step": 9879
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3822976586561787,
      "learning_rate": 1.6350088679221306e-05,
      "loss": 0.2581,
      "step": 9880
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4382724570179211,
      "learning_rate": 1.6349322413163908e-05,
      "loss": 0.2456,
      "step": 9881
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.30694134192912564,
      "learning_rate": 1.634855608463979e-05,
      "loss": 0.2656,
      "step": 9882
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9401266302554535,
      "learning_rate": 1.6347789693656483e-05,
      "loss": 0.5227,
      "step": 9883
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4906617605481061,
      "learning_rate": 1.6347023240221536e-05,
      "loss": 0.2724,
      "step": 9884
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.39245202223176073,
      "learning_rate": 1.6346256724342487e-05,
      "loss": 0.3284,
      "step": 9885
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8617807123637001,
      "learning_rate": 1.6345490146026875e-05,
      "loss": 0.4119,
      "step": 9886
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.6779594851664956,
      "learning_rate": 1.634472350528224e-05,
      "loss": 0.7634,
      "step": 9887
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.25120012998489394,
      "learning_rate": 1.634395680211613e-05,
      "loss": 0.2077,
      "step": 9888
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3826943152638382,
      "learning_rate": 1.6343190036536088e-05,
      "loss": 0.2914,
      "step": 9889
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2656928896521156,
      "learning_rate": 1.634242320854965e-05,
      "loss": 0.0684,
      "step": 9890
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2116780227464867,
      "learning_rate": 1.634165631816437e-05,
      "loss": 0.7704,
      "step": 9891
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9710037241703326,
      "learning_rate": 1.6340889365387785e-05,
      "loss": 0.4039,
      "step": 9892
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7318255779004713,
      "learning_rate": 1.6340122350227446e-05,
      "loss": 0.4284,
      "step": 9893
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3440918817872595,
      "learning_rate": 1.6339355272690895e-05,
      "loss": 0.2349,
      "step": 9894
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2910739935096455,
      "learning_rate": 1.6338588132785683e-05,
      "loss": 0.2709,
      "step": 9895
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2791587215853102,
      "learning_rate": 1.6337820930519357e-05,
      "loss": 0.6794,
      "step": 9896
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.24878434090483145,
      "learning_rate": 1.633705366589946e-05,
      "loss": 0.0959,
      "step": 9897
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9221256739402776,
      "learning_rate": 1.6336286338933545e-05,
      "loss": 0.4007,
      "step": 9898
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33943077668157484,
      "learning_rate": 1.633551894962916e-05,
      "loss": 0.2442,
      "step": 9899
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.36339792916430874,
      "learning_rate": 1.6334751497993856e-05,
      "loss": 0.3362,
      "step": 9900
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.864279991761935,
      "learning_rate": 1.633398398403518e-05,
      "loss": 0.3926,
      "step": 9901
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.0436633620252027,
      "learning_rate": 1.6333216407760687e-05,
      "loss": 0.5284,
      "step": 9902
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.329080803804036,
      "learning_rate": 1.6332448769177925e-05,
      "loss": 0.227,
      "step": 9903
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.169652352848653,
      "learning_rate": 1.6331681068294453e-05,
      "loss": 0.7211,
      "step": 9904
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2920474836103776,
      "learning_rate": 1.6330913305117817e-05,
      "loss": 0.2633,
      "step": 9905
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2023020394789994,
      "learning_rate": 1.6330145479655572e-05,
      "loss": 0.8006,
      "step": 9906
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2736712762595727,
      "learning_rate": 1.6329377591915274e-05,
      "loss": 0.1443,
      "step": 9907
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.31860657365774864,
      "learning_rate": 1.6328609641904477e-05,
      "loss": 0.1732,
      "step": 9908
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.6353295307983038,
      "learning_rate": 1.6327841629630735e-05,
      "loss": 0.466,
      "step": 9909
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7389710724906264,
      "learning_rate": 1.63270735551016e-05,
      "loss": 0.4059,
      "step": 9910
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.42738955014888336,
      "learning_rate": 1.6326305418324643e-05,
      "loss": 0.3354,
      "step": 9911
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29670656474470214,
      "learning_rate": 1.6325537219307404e-05,
      "loss": 0.2383,
      "step": 9912
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.606757465719981,
      "learning_rate": 1.632476895805745e-05,
      "loss": 0.823,
      "step": 9913
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7937215589571579,
      "learning_rate": 1.632400063458234e-05,
      "loss": 0.0433,
      "step": 9914
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26803840294666026,
      "learning_rate": 1.632323224888963e-05,
      "loss": 0.1487,
      "step": 9915
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3099474898507763,
      "learning_rate": 1.632246380098688e-05,
      "loss": 0.2074,
      "step": 9916
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9798192544852524,
      "learning_rate": 1.632169529088165e-05,
      "loss": 0.5419,
      "step": 9917
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.35543421926706614,
      "learning_rate": 1.6320926718581508e-05,
      "loss": 0.283,
      "step": 9918
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8179418737028148,
      "learning_rate": 1.6320158084094e-05,
      "loss": 0.5124,
      "step": 9919
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.27442760502123037,
      "learning_rate": 1.6319389387426702e-05,
      "loss": 0.163,
      "step": 9920
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33438073867793017,
      "learning_rate": 1.6318620628587173e-05,
      "loss": 0.2353,
      "step": 9921
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.2701490433709999,
      "learning_rate": 1.631785180758297e-05,
      "loss": 0.6593,
      "step": 9922
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4229476498908925,
      "learning_rate": 1.631708292442167e-05,
      "loss": 0.2729,
      "step": 9923
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.35467556439702025,
      "learning_rate": 1.631631397911082e-05,
      "loss": 0.2783,
      "step": 9924
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.33161287043740517,
      "learning_rate": 1.6315544971658003e-05,
      "loss": 0.0748,
      "step": 9925
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.37316645622524214,
      "learning_rate": 1.631477590207077e-05,
      "loss": 0.2603,
      "step": 9926
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.800520398097059,
      "learning_rate": 1.63140067703567e-05,
      "loss": 0.4098,
      "step": 9927
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.4022948336587027,
      "learning_rate": 1.631323757652335e-05,
      "loss": 0.9555,
      "step": 9928
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.31128397528947777,
      "learning_rate": 1.631246832057829e-05,
      "loss": 0.2475,
      "step": 9929
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3936258538578464,
      "learning_rate": 1.6311699002529094e-05,
      "loss": 0.2658,
      "step": 9930
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.7072019430169009,
      "learning_rate": 1.6310929622383326e-05,
      "loss": 0.2841,
      "step": 9931
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.060329201540452,
      "learning_rate": 1.6310160180148552e-05,
      "loss": 0.5888,
      "step": 9932
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5265826354512088,
      "learning_rate": 1.6309390675832348e-05,
      "loss": 0.0284,
      "step": 9933
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.29945120864100666,
      "learning_rate": 1.6308621109442284e-05,
      "loss": 0.2424,
      "step": 9934
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4371263955361159,
      "learning_rate": 1.6307851480985927e-05,
      "loss": 0.3963,
      "step": 9935
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.22660957870870493,
      "learning_rate": 1.630708179047085e-05,
      "loss": 0.2277,
      "step": 9936
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.666022682972071,
      "learning_rate": 1.630631203790463e-05,
      "loss": 0.8656,
      "step": 9937
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.4887139226662167,
      "learning_rate": 1.6305542223294836e-05,
      "loss": 0.2909,
      "step": 9938
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3574508157763857,
      "learning_rate": 1.6304772346649044e-05,
      "loss": 0.2996,
      "step": 9939
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26676058297658933,
      "learning_rate": 1.630400240797483e-05,
      "loss": 0.1134,
      "step": 9940
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.9807849631977762,
      "learning_rate": 1.6303232407279758e-05,
      "loss": 0.6541,
      "step": 9941
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26858875237390334,
      "learning_rate": 1.6302462344571415e-05,
      "loss": 0.2691,
      "step": 9942
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5273453602037412,
      "learning_rate": 1.6301692219857378e-05,
      "loss": 0.4076,
      "step": 9943
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.31754936655023014,
      "learning_rate": 1.6300922033145214e-05,
      "loss": 0.2584,
      "step": 9944
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8217638567759695,
      "learning_rate": 1.6300151784442508e-05,
      "loss": 0.554,
      "step": 9945
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.26000444958128316,
      "learning_rate": 1.6299381473756837e-05,
      "loss": 0.1475,
      "step": 9946
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2959239015416507,
      "learning_rate": 1.6298611101095774e-05,
      "loss": 0.2906,
      "step": 9947
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3212739696655125,
      "learning_rate": 1.629784066646691e-05,
      "loss": 0.2146,
      "step": 9948
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.27077838903189394,
      "learning_rate": 1.629707016987781e-05,
      "loss": 0.0987,
      "step": 9949
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.3697315136997887,
      "learning_rate": 1.6296299611336068e-05,
      "loss": 0.7494,
      "step": 9950
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.5126691095739069,
      "learning_rate": 1.6295528990849252e-05,
      "loss": 0.3803,
      "step": 9951
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.086785141980013,
      "learning_rate": 1.629475830842495e-05,
      "loss": 0.4918,
      "step": 9952
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.31614786833271274,
      "learning_rate": 1.629398756407075e-05,
      "loss": 0.2292,
      "step": 9953
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3453235490968876,
      "learning_rate": 1.6293216757794227e-05,
      "loss": 0.3073,
      "step": 9954
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3007757367163501,
      "learning_rate": 1.629244588960297e-05,
      "loss": 0.1771,
      "step": 9955
    },
    {
      "epoch": 0.3,
      "grad_norm": 1.5941345178035253,
      "learning_rate": 1.6291674959504554e-05,
      "loss": 0.9505,
      "step": 9956
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.32376292430874914,
      "learning_rate": 1.6290903967506574e-05,
      "loss": 0.1941,
      "step": 9957
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.3637144495240645,
      "learning_rate": 1.6290132913616607e-05,
      "loss": 0.265,
      "step": 9958
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.2939645822629649,
      "learning_rate": 1.6289361797842244e-05,
      "loss": 0.2454,
      "step": 9959
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9972958536244886,
      "learning_rate": 1.628859062019107e-05,
      "loss": 0.6208,
      "step": 9960
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.463151054448686,
      "learning_rate": 1.6287819380670676e-05,
      "loss": 0.3244,
      "step": 9961
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3261950263052064,
      "learning_rate": 1.6287048079288643e-05,
      "loss": 0.2073,
      "step": 9962
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.45738256952286194,
      "learning_rate": 1.628627671605256e-05,
      "loss": 0.2751,
      "step": 9963
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9051657272487491,
      "learning_rate": 1.6285505290970023e-05,
      "loss": 0.4899,
      "step": 9964
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3767189828132606,
      "learning_rate": 1.6284733804048618e-05,
      "loss": 0.3238,
      "step": 9965
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29274178546470786,
      "learning_rate": 1.6283962255295928e-05,
      "loss": 0.1962,
      "step": 9966
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41686435646719266,
      "learning_rate": 1.628319064471955e-05,
      "loss": 0.2838,
      "step": 9967
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8161192485361667,
      "learning_rate": 1.628241897232708e-05,
      "loss": 0.3652,
      "step": 9968
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6498213354186725,
      "learning_rate": 1.62816472381261e-05,
      "loss": 0.4536,
      "step": 9969
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3416245342234563,
      "learning_rate": 1.628087544212421e-05,
      "loss": 0.1373,
      "step": 9970
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.32179788411080856,
      "learning_rate": 1.6280103584328997e-05,
      "loss": 0.3068,
      "step": 9971
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4590908113550105,
      "learning_rate": 1.6279331664748065e-05,
      "loss": 0.227,
      "step": 9972
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0665452707752918,
      "learning_rate": 1.6278559683388995e-05,
      "loss": 0.6266,
      "step": 9973
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29488364871039463,
      "learning_rate": 1.6277787640259394e-05,
      "loss": 0.1675,
      "step": 9974
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.28300996629042474,
      "learning_rate": 1.627701553536685e-05,
      "loss": 0.1641,
      "step": 9975
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2828892960594558,
      "learning_rate": 1.6276243368718964e-05,
      "loss": 0.2029,
      "step": 9976
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3077623046192395,
      "learning_rate": 1.6275471140323328e-05,
      "loss": 0.2666,
      "step": 9977
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8461414827796775,
      "learning_rate": 1.6274698850187545e-05,
      "loss": 0.5146,
      "step": 9978
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4034503509165909,
      "learning_rate": 1.6273926498319208e-05,
      "loss": 0.0801,
      "step": 9979
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35016994934634166,
      "learning_rate": 1.6273154084725917e-05,
      "loss": 0.2998,
      "step": 9980
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9796041953277405,
      "learning_rate": 1.6272381609415274e-05,
      "loss": 0.2836,
      "step": 9981
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1861528525801834,
      "learning_rate": 1.6271609072394876e-05,
      "loss": 0.7547,
      "step": 9982
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31338155732090595,
      "learning_rate": 1.6270836473672325e-05,
      "loss": 0.2521,
      "step": 9983
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5971382747126979,
      "learning_rate": 1.6270063813255222e-05,
      "loss": 0.3538,
      "step": 9984
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.27182063292044195,
      "learning_rate": 1.6269291091151163e-05,
      "loss": 0.1903,
      "step": 9985
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6994803367622996,
      "learning_rate": 1.6268518307367763e-05,
      "loss": 0.484,
      "step": 9986
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31674442670899516,
      "learning_rate": 1.6267745461912613e-05,
      "loss": 0.1752,
      "step": 9987
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4074469340218703,
      "learning_rate": 1.6266972554793318e-05,
      "loss": 0.2369,
      "step": 9988
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35683665303019063,
      "learning_rate": 1.626619958601749e-05,
      "loss": 0.2717,
      "step": 9989
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4103609023729149,
      "learning_rate": 1.6265426555592728e-05,
      "loss": 0.2948,
      "step": 9990
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8678543097638567,
      "learning_rate": 1.6264653463526632e-05,
      "loss": 0.8852,
      "step": 9991
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9126010751464899,
      "learning_rate": 1.6263880309826824e-05,
      "loss": 0.5587,
      "step": 9992
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5369252103141383,
      "learning_rate": 1.626310709450089e-05,
      "loss": 0.3161,
      "step": 9993
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34025071949373104,
      "learning_rate": 1.6262333817556454e-05,
      "loss": 0.2837,
      "step": 9994
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.48737657103346715,
      "learning_rate": 1.6261560479001117e-05,
      "loss": 0.391,
      "step": 9995
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.22362762846323045,
      "learning_rate": 1.6260787078842487e-05,
      "loss": 0.1689,
      "step": 9996
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9331080441766487,
      "learning_rate": 1.6260013617088174e-05,
      "loss": 0.4911,
      "step": 9997
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.300322620411484,
      "learning_rate": 1.6259240093745786e-05,
      "loss": 0.2136,
      "step": 9998
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7460738509815503,
      "learning_rate": 1.625846650882293e-05,
      "loss": 0.801,
      "step": 9999
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2437361312620079,
      "learning_rate": 1.625769286232723e-05,
      "loss": 0.557,
      "step": 10000
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31350781765933133,
      "learning_rate": 1.625691915426628e-05,
      "loss": 0.2691,
      "step": 10001
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7627224044732956,
      "learning_rate": 1.625614538464771e-05,
      "loss": 0.2883,
      "step": 10002
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31160836158711525,
      "learning_rate": 1.6255371553479118e-05,
      "loss": 0.2658,
      "step": 10003
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5725018796159532,
      "learning_rate": 1.6254597660768122e-05,
      "loss": 0.4712,
      "step": 10004
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.20656262289158892,
      "learning_rate": 1.6253823706522334e-05,
      "loss": 0.095,
      "step": 10005
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4840946857090279,
      "learning_rate": 1.6253049690749377e-05,
      "loss": 0.3578,
      "step": 10006
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2942950380235673,
      "learning_rate": 1.6252275613456855e-05,
      "loss": 0.1686,
      "step": 10007
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34553450727367996,
      "learning_rate": 1.625150147465239e-05,
      "loss": 0.3051,
      "step": 10008
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9918772601679903,
      "learning_rate": 1.6250727274343594e-05,
      "loss": 0.5226,
      "step": 10009
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7951338312804148,
      "learning_rate": 1.624995301253809e-05,
      "loss": 0.6218,
      "step": 10010
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3320678531413539,
      "learning_rate": 1.624917868924349e-05,
      "loss": 0.2033,
      "step": 10011
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34571689239909376,
      "learning_rate": 1.6248404304467412e-05,
      "loss": 0.3297,
      "step": 10012
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.32120362481232145,
      "learning_rate": 1.6247629858217478e-05,
      "loss": 0.2738,
      "step": 10013
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2637216277240583,
      "learning_rate": 1.6246855350501304e-05,
      "loss": 0.0739,
      "step": 10014
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4259372512070767,
      "learning_rate": 1.6246080781326513e-05,
      "loss": 0.266,
      "step": 10015
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29931228292977324,
      "learning_rate": 1.624530615070073e-05,
      "loss": 0.1921,
      "step": 10016
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5783988770606687,
      "learning_rate": 1.624453145863156e-05,
      "loss": 0.8247,
      "step": 10017
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9645328069088143,
      "learning_rate": 1.624375670512664e-05,
      "loss": 0.5011,
      "step": 10018
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4095630428802657,
      "learning_rate": 1.6242981890193587e-05,
      "loss": 0.332,
      "step": 10019
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5527036793589799,
      "learning_rate": 1.6242207013840025e-05,
      "loss": 0.2697,
      "step": 10020
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3893548424748718,
      "learning_rate": 1.6241432076073575e-05,
      "loss": 0.3376,
      "step": 10021
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0747763424592456,
      "learning_rate": 1.624065707690186e-05,
      "loss": 0.3542,
      "step": 10022
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46533856817735064,
      "learning_rate": 1.6239882016332507e-05,
      "loss": 0.3115,
      "step": 10023
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.26790195475469253,
      "learning_rate": 1.6239106894373146e-05,
      "loss": 0.2053,
      "step": 10024
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4012518374956076,
      "learning_rate": 1.6238331711031397e-05,
      "loss": 0.3107,
      "step": 10025
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42084533157193116,
      "learning_rate": 1.623755646631489e-05,
      "loss": 0.2373,
      "step": 10026
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.054883450212891,
      "learning_rate": 1.6236781160231245e-05,
      "loss": 0.5795,
      "step": 10027
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7677984831122716,
      "learning_rate": 1.6236005792788095e-05,
      "loss": 0.4723,
      "step": 10028
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5564956747575187,
      "learning_rate": 1.6235230363993072e-05,
      "loss": 0.2101,
      "step": 10029
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3826873260548635,
      "learning_rate": 1.62344548738538e-05,
      "loss": 0.318,
      "step": 10030
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.28638824958030773,
      "learning_rate": 1.623367932237791e-05,
      "loss": 0.246,
      "step": 10031
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1480709381854046,
      "learning_rate": 1.623290370957303e-05,
      "loss": 0.6387,
      "step": 10032
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18048632376387388,
      "learning_rate": 1.623212803544679e-05,
      "loss": 0.1009,
      "step": 10033
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.692090426862907,
      "learning_rate": 1.623135230000683e-05,
      "loss": 0.3585,
      "step": 10034
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.33831326943482964,
      "learning_rate": 1.6230576503260774e-05,
      "loss": 0.2446,
      "step": 10035
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4893050807026389,
      "learning_rate": 1.6229800645216253e-05,
      "loss": 0.3991,
      "step": 10036
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.49615097145498743,
      "learning_rate": 1.6229024725880906e-05,
      "loss": 0.3141,
      "step": 10037
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9443971381029856,
      "learning_rate": 1.622824874526236e-05,
      "loss": 0.6003,
      "step": 10038
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2959409344273699,
      "learning_rate": 1.622747270336826e-05,
      "loss": 0.1957,
      "step": 10039
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9080434521533317,
      "learning_rate": 1.622669660020623e-05,
      "loss": 0.3829,
      "step": 10040
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.8897541304926369,
      "learning_rate": 1.622592043578391e-05,
      "loss": 0.8785,
      "step": 10041
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.19112974340839406,
      "learning_rate": 1.6225144210108937e-05,
      "loss": 0.1681,
      "step": 10042
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4171700011865395,
      "learning_rate": 1.6224367923188947e-05,
      "loss": 0.2872,
      "step": 10043
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4503120602673835,
      "learning_rate": 1.6223591575031576e-05,
      "loss": 0.2575,
      "step": 10044
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9599384898096948,
      "learning_rate": 1.6222815165644466e-05,
      "loss": 0.6318,
      "step": 10045
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7136782380196341,
      "learning_rate": 1.6222038695035252e-05,
      "loss": 0.4101,
      "step": 10046
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.838431219757667,
      "learning_rate": 1.6221262163211573e-05,
      "loss": 0.5398,
      "step": 10047
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.33311756137852244,
      "learning_rate": 1.622048557018107e-05,
      "loss": 0.2232,
      "step": 10048
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4606204589188692,
      "learning_rate": 1.6219708915951385e-05,
      "loss": 0.3381,
      "step": 10049
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3849029950344831,
      "learning_rate": 1.6218932200530157e-05,
      "loss": 0.2693,
      "step": 10050
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2896285455074227,
      "learning_rate": 1.6218155423925025e-05,
      "loss": 0.1865,
      "step": 10051
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30989227261111185,
      "learning_rate": 1.6217378586143633e-05,
      "loss": 0.1726,
      "step": 10052
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42990421650803545,
      "learning_rate": 1.6216601687193628e-05,
      "loss": 0.2503,
      "step": 10053
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4835317543853505,
      "learning_rate": 1.6215824727082647e-05,
      "loss": 0.406,
      "step": 10054
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.48111206316081434,
      "learning_rate": 1.621504770581834e-05,
      "loss": 0.3583,
      "step": 10055
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8798883901083725,
      "learning_rate": 1.6214270623408348e-05,
      "loss": 0.6016,
      "step": 10056
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3492089518482501,
      "learning_rate": 1.6213493479860317e-05,
      "loss": 0.1882,
      "step": 10057
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5480559093141361,
      "learning_rate": 1.621271627518189e-05,
      "loss": 0.4257,
      "step": 10058
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31209429847292275,
      "learning_rate": 1.6211939009380717e-05,
      "loss": 0.132,
      "step": 10059
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2574052555995597,
      "learning_rate": 1.6211161682464446e-05,
      "loss": 0.264,
      "step": 10060
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2806712071182556,
      "learning_rate": 1.6210384294440724e-05,
      "loss": 0.0723,
      "step": 10061
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4086768417174611,
      "learning_rate": 1.6209606845317195e-05,
      "loss": 0.3364,
      "step": 10062
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39891154282346164,
      "learning_rate": 1.620882933510151e-05,
      "loss": 0.3912,
      "step": 10063
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2995559424539842,
      "learning_rate": 1.6208051763801322e-05,
      "loss": 0.8389,
      "step": 10064
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7178575488622171,
      "learning_rate": 1.6207274131424276e-05,
      "loss": 0.388,
      "step": 10065
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30850164724301476,
      "learning_rate": 1.6206496437978027e-05,
      "loss": 0.2057,
      "step": 10066
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34807586324638934,
      "learning_rate": 1.620571868347022e-05,
      "loss": 0.3294,
      "step": 10067
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8904008653669212,
      "learning_rate": 1.620494086790851e-05,
      "loss": 0.2928,
      "step": 10068
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4716509497690328,
      "learning_rate": 1.6204162991300553e-05,
      "loss": 0.278,
      "step": 10069
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.41518788982868404,
      "learning_rate": 1.6203385053654e-05,
      "loss": 0.274,
      "step": 10070
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35207000099440716,
      "learning_rate": 1.6202607054976503e-05,
      "loss": 0.3117,
      "step": 10071
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38685771781687384,
      "learning_rate": 1.6201828995275715e-05,
      "loss": 0.2776,
      "step": 10072
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3355765806331273,
      "learning_rate": 1.6201050874559295e-05,
      "loss": 0.2652,
      "step": 10073
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.46661737108916457,
      "learning_rate": 1.6200272692834893e-05,
      "loss": 0.2911,
      "step": 10074
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35352895196389666,
      "learning_rate": 1.619949445011017e-05,
      "loss": 0.2684,
      "step": 10075
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8439686675380815,
      "learning_rate": 1.619871614639278e-05,
      "loss": 0.3509,
      "step": 10076
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.7955086773064997,
      "learning_rate": 1.6197937781690384e-05,
      "loss": 0.9595,
      "step": 10077
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.32396893315593533,
      "learning_rate": 1.6197159356010634e-05,
      "loss": 0.2595,
      "step": 10078
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37050282873148715,
      "learning_rate": 1.619638086936119e-05,
      "loss": 0.2346,
      "step": 10079
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5753358618692102,
      "learning_rate": 1.6195602321749714e-05,
      "loss": 0.3453,
      "step": 10080
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36835970636466625,
      "learning_rate": 1.619482371318386e-05,
      "loss": 0.2466,
      "step": 10081
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2047960888915288,
      "learning_rate": 1.6194045043671295e-05,
      "loss": 0.863,
      "step": 10082
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23776407649698336,
      "learning_rate": 1.619326631321968e-05,
      "loss": 0.1667,
      "step": 10083
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9410259515924458,
      "learning_rate": 1.619248752183667e-05,
      "loss": 0.4236,
      "step": 10084
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2949325892012961,
      "learning_rate": 1.6191708669529927e-05,
      "loss": 0.2598,
      "step": 10085
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.1846838024869446,
      "learning_rate": 1.619092975630712e-05,
      "loss": 0.7128,
      "step": 10086
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5799872562348066,
      "learning_rate": 1.6190150782175912e-05,
      "loss": 0.3607,
      "step": 10087
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0208220031608115,
      "learning_rate": 1.618937174714396e-05,
      "loss": 0.4114,
      "step": 10088
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31600347739260326,
      "learning_rate": 1.618859265121893e-05,
      "loss": 0.2293,
      "step": 10089
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31431049842772224,
      "learning_rate": 1.6187813494408496e-05,
      "loss": 0.2296,
      "step": 10090
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3980362243275202,
      "learning_rate": 1.618703427672031e-05,
      "loss": 0.2706,
      "step": 10091
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37273598306053934,
      "learning_rate": 1.618625499816205e-05,
      "loss": 0.1395,
      "step": 10092
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5002809404761319,
      "learning_rate": 1.6185475658741373e-05,
      "loss": 0.283,
      "step": 10093
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43194055072776855,
      "learning_rate": 1.618469625846595e-05,
      "loss": 0.2333,
      "step": 10094
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.003852188773741,
      "learning_rate": 1.6183916797343457e-05,
      "loss": 0.5244,
      "step": 10095
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29187477678782686,
      "learning_rate": 1.6183137275381554e-05,
      "loss": 0.2693,
      "step": 10096
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6388506229622785,
      "learning_rate": 1.6182357692587913e-05,
      "loss": 0.4436,
      "step": 10097
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3224813546635144,
      "learning_rate": 1.6181578048970198e-05,
      "loss": 0.2234,
      "step": 10098
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2291617974188027,
      "learning_rate": 1.618079834453609e-05,
      "loss": 0.5643,
      "step": 10099
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9517808307721121,
      "learning_rate": 1.6180018579293246e-05,
      "loss": 0.5922,
      "step": 10100
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3669562442999021,
      "learning_rate": 1.617923875324935e-05,
      "loss": 0.2892,
      "step": 10101
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.28799539372292465,
      "learning_rate": 1.617845886641207e-05,
      "loss": 0.2129,
      "step": 10102
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35430583705464685,
      "learning_rate": 1.617767891878908e-05,
      "loss": 0.2587,
      "step": 10103
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9794354689766773,
      "learning_rate": 1.6176898910388053e-05,
      "loss": 0.307,
      "step": 10104
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5388962122577584,
      "learning_rate": 1.617611884121666e-05,
      "loss": 0.3748,
      "step": 10105
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6073449244963284,
      "learning_rate": 1.6175338711282576e-05,
      "loss": 0.3674,
      "step": 10106
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3507558137266751,
      "learning_rate": 1.617455852059348e-05,
      "loss": 0.2387,
      "step": 10107
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44617736776348677,
      "learning_rate": 1.6173778269157046e-05,
      "loss": 0.3458,
      "step": 10108
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39431294332192235,
      "learning_rate": 1.617299795698095e-05,
      "loss": 0.289,
      "step": 10109
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3278759115147903,
      "learning_rate": 1.6172217584072867e-05,
      "loss": 0.2249,
      "step": 10110
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29126318115989036,
      "learning_rate": 1.6171437150440475e-05,
      "loss": 0.0747,
      "step": 10111
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40361826048400345,
      "learning_rate": 1.617065665609146e-05,
      "loss": 0.3134,
      "step": 10112
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39464913098897386,
      "learning_rate": 1.6169876101033487e-05,
      "loss": 0.3008,
      "step": 10113
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42438824237135947,
      "learning_rate": 1.616909548527425e-05,
      "loss": 0.3776,
      "step": 10114
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9066195580027661,
      "learning_rate": 1.6168314808821418e-05,
      "loss": 0.2917,
      "step": 10115
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3959414185834494,
      "learning_rate": 1.6167534071682674e-05,
      "loss": 0.308,
      "step": 10116
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.438859231272131,
      "learning_rate": 1.6166753273865702e-05,
      "loss": 0.2959,
      "step": 10117
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9237814376491238,
      "learning_rate": 1.616597241537818e-05,
      "loss": 0.5949,
      "step": 10118
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5494948083691189,
      "learning_rate": 1.6165191496227795e-05,
      "loss": 0.4028,
      "step": 10119
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.28209172474062455,
      "learning_rate": 1.6164410516422227e-05,
      "loss": 0.2161,
      "step": 10120
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24188143558439557,
      "learning_rate": 1.6163629475969156e-05,
      "loss": 0.1736,
      "step": 10121
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8642241439337967,
      "learning_rate": 1.616284837487628e-05,
      "loss": 0.369,
      "step": 10122
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8031878640538964,
      "learning_rate": 1.6162067213151266e-05,
      "loss": 0.5249,
      "step": 10123
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5537804085516839,
      "learning_rate": 1.616128599080181e-05,
      "loss": 0.1718,
      "step": 10124
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3887606363175585,
      "learning_rate": 1.6160504707835595e-05,
      "loss": 0.3166,
      "step": 10125
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34528622825160976,
      "learning_rate": 1.6159723364260305e-05,
      "loss": 0.2641,
      "step": 10126
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.5880518156748236,
      "learning_rate": 1.615894196008363e-05,
      "loss": 0.7375,
      "step": 10127
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.032732198426722,
      "learning_rate": 1.615816049531326e-05,
      "loss": 0.4621,
      "step": 10128
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3330031706356474,
      "learning_rate": 1.6157378969956878e-05,
      "loss": 0.1807,
      "step": 10129
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.35281735399524,
      "learning_rate": 1.6156597384022176e-05,
      "loss": 0.2393,
      "step": 10130
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0344327341685675,
      "learning_rate": 1.6155815737516845e-05,
      "loss": 0.437,
      "step": 10131
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.29088798293742546,
      "learning_rate": 1.6155034030448575e-05,
      "loss": 0.2706,
      "step": 10132
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6098387500496655,
      "learning_rate": 1.615425226282505e-05,
      "loss": 0.274,
      "step": 10133
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4120994613740694,
      "learning_rate": 1.615347043465397e-05,
      "loss": 0.5531,
      "step": 10134
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3592217815348341,
      "learning_rate": 1.615268854594302e-05,
      "loss": 0.2531,
      "step": 10135
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3354753814974503,
      "learning_rate": 1.61519065966999e-05,
      "loss": 0.8588,
      "step": 10136
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31743925968679165,
      "learning_rate": 1.6151124586932295e-05,
      "loss": 0.2569,
      "step": 10137
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8742329820884579,
      "learning_rate": 1.6150342516647904e-05,
      "loss": 0.4536,
      "step": 10138
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30090004995997177,
      "learning_rate": 1.614956038585442e-05,
      "loss": 0.2494,
      "step": 10139
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6860794656599609,
      "learning_rate": 1.6148778194559534e-05,
      "loss": 0.513,
      "step": 10140
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2131782580098741,
      "learning_rate": 1.614799594277095e-05,
      "loss": 0.0948,
      "step": 10141
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.182228783958301,
      "learning_rate": 1.6147213630496356e-05,
      "loss": 0.8352,
      "step": 10142
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.33378170073481117,
      "learning_rate": 1.614643125774345e-05,
      "loss": 0.1864,
      "step": 10143
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3317217891485893,
      "learning_rate": 1.6145648824519934e-05,
      "loss": 0.3048,
      "step": 10144
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8967722006230978,
      "learning_rate": 1.61448663308335e-05,
      "loss": 0.6495,
      "step": 10145
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6579177490198678,
      "learning_rate": 1.614408377669185e-05,
      "loss": 0.3813,
      "step": 10146
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5196796397163574,
      "learning_rate": 1.6143301162102683e-05,
      "loss": 0.4545,
      "step": 10147
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36276669730942623,
      "learning_rate": 1.61425184870737e-05,
      "loss": 0.2395,
      "step": 10148
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5262002410690669,
      "learning_rate": 1.6141735751612596e-05,
      "loss": 0.3574,
      "step": 10149
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.286286518626717,
      "learning_rate": 1.614095295572707e-05,
      "loss": 0.1742,
      "step": 10150
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38905552022338863,
      "learning_rate": 1.6140170099424835e-05,
      "loss": 0.2389,
      "step": 10151
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.27150990056151103,
      "learning_rate": 1.6139387182713587e-05,
      "loss": 0.1996,
      "step": 10152
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2378415353524583,
      "learning_rate": 1.613860420560102e-05,
      "loss": 0.8473,
      "step": 10153
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4313409220023354,
      "learning_rate": 1.6137821168094855e-05,
      "loss": 0.5252,
      "step": 10154
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38345765987616565,
      "learning_rate": 1.613703807020278e-05,
      "loss": 0.349,
      "step": 10155
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5885725544154339,
      "learning_rate": 1.6136254911932507e-05,
      "loss": 0.2762,
      "step": 10156
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3931694455447643,
      "learning_rate": 1.6135471693291736e-05,
      "loss": 0.2969,
      "step": 10157
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6068439631940356,
      "learning_rate": 1.613468841428818e-05,
      "loss": 0.384,
      "step": 10158
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.1747189758608561,
      "learning_rate": 1.613390507492954e-05,
      "loss": 0.073,
      "step": 10159
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5106631355726351,
      "learning_rate": 1.613312167522352e-05,
      "loss": 0.3918,
      "step": 10160
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2579922879704796,
      "learning_rate": 1.6132338215177833e-05,
      "loss": 0.0753,
      "step": 10161
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3398016370553681,
      "learning_rate": 1.6131554694800185e-05,
      "loss": 0.3231,
      "step": 10162
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3762454731725895,
      "learning_rate": 1.6130771114098284e-05,
      "loss": 0.6549,
      "step": 10163
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6680649804015112,
      "learning_rate": 1.612998747307984e-05,
      "loss": 0.4967,
      "step": 10164
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.673170556256316,
      "learning_rate": 1.6129203771752565e-05,
      "loss": 0.363,
      "step": 10165
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39039982897786596,
      "learning_rate": 1.6128420010124162e-05,
      "loss": 0.244,
      "step": 10166
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34846746604229095,
      "learning_rate": 1.612763618820235e-05,
      "loss": 0.2734,
      "step": 10167
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4062254032033733,
      "learning_rate": 1.612685230599483e-05,
      "loss": 0.2759,
      "step": 10168
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.21990081073264225,
      "learning_rate": 1.6126068363509326e-05,
      "loss": 0.0714,
      "step": 10169
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7120814807773679,
      "learning_rate": 1.6125284360753547e-05,
      "loss": 0.4108,
      "step": 10170
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42500382393487823,
      "learning_rate": 1.6124500297735206e-05,
      "loss": 0.2705,
      "step": 10171
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4583544744348682,
      "learning_rate": 1.612371617446201e-05,
      "loss": 0.4484,
      "step": 10172
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4823692269152424,
      "learning_rate": 1.6122931990941683e-05,
      "loss": 0.3505,
      "step": 10173
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3794250969718895,
      "learning_rate": 1.6122147747181937e-05,
      "loss": 0.2789,
      "step": 10174
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34898735264596614,
      "learning_rate": 1.6121363443190484e-05,
      "loss": 0.2952,
      "step": 10175
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0121003766270429,
      "learning_rate": 1.6120579078975047e-05,
      "loss": 0.0651,
      "step": 10176
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.096665391014745,
      "learning_rate": 1.6119794654543336e-05,
      "loss": 0.9472,
      "step": 10177
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2131161226351602,
      "learning_rate": 1.6119010169903076e-05,
      "loss": 0.149,
      "step": 10178
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36109918065813623,
      "learning_rate": 1.6118225625061973e-05,
      "loss": 0.2855,
      "step": 10179
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3641859947994503,
      "learning_rate": 1.611744102002776e-05,
      "loss": 0.2465,
      "step": 10180
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4724125951793205,
      "learning_rate": 1.6116656354808147e-05,
      "loss": 0.7737,
      "step": 10181
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.0315265973058156,
      "learning_rate": 1.6115871629410854e-05,
      "loss": 0.4242,
      "step": 10182
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8172084183879502,
      "learning_rate": 1.6115086843843607e-05,
      "loss": 0.5637,
      "step": 10183
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3032206863242367,
      "learning_rate": 1.6114301998114123e-05,
      "loss": 0.2162,
      "step": 10184
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4243484960211544,
      "learning_rate": 1.6113517092230125e-05,
      "loss": 0.2658,
      "step": 10185
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44658361989071194,
      "learning_rate": 1.611273212619933e-05,
      "loss": 0.3869,
      "step": 10186
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.18159800481051427,
      "learning_rate": 1.6111947100029467e-05,
      "loss": 0.1055,
      "step": 10187
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6787234790192218,
      "learning_rate": 1.6111162013728258e-05,
      "loss": 0.3991,
      "step": 10188
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.339390405672875,
      "learning_rate": 1.6110376867303427e-05,
      "loss": 0.2297,
      "step": 10189
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6999128289233386,
      "learning_rate": 1.6109591660762695e-05,
      "loss": 0.442,
      "step": 10190
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34809148322775935,
      "learning_rate": 1.6108806394113795e-05,
      "loss": 0.2853,
      "step": 10191
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4385961591173317,
      "learning_rate": 1.6108021067364442e-05,
      "loss": 0.8256,
      "step": 10192
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.34246563167308774,
      "learning_rate": 1.610723568052237e-05,
      "loss": 0.1931,
      "step": 10193
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5468335188536906,
      "learning_rate": 1.6106450233595304e-05,
      "loss": 0.3983,
      "step": 10194
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36577987614658547,
      "learning_rate": 1.610566472659097e-05,
      "loss": 0.12,
      "step": 10195
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39970124771900667,
      "learning_rate": 1.61048791595171e-05,
      "loss": 0.2489,
      "step": 10196
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3052212431676089,
      "learning_rate": 1.610409353238142e-05,
      "loss": 0.2599,
      "step": 10197
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31011217907177124,
      "learning_rate": 1.6103307845191655e-05,
      "loss": 0.2446,
      "step": 10198
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6076158990454708,
      "learning_rate": 1.6102522097955543e-05,
      "loss": 0.4256,
      "step": 10199
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.3550978602615726,
      "learning_rate": 1.610173629068081e-05,
      "loss": 0.5231,
      "step": 10200
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4687912395489149,
      "learning_rate": 1.610095042337519e-05,
      "loss": 0.6786,
      "step": 10201
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2907210410044799,
      "learning_rate": 1.6100164496046408e-05,
      "loss": 0.2075,
      "step": 10202
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3271681182109278,
      "learning_rate": 1.60993785087022e-05,
      "loss": 0.3168,
      "step": 10203
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2965642979405976,
      "learning_rate": 1.6098592461350305e-05,
      "loss": 0.3088,
      "step": 10204
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3275108215446399,
      "learning_rate": 1.6097806353998447e-05,
      "loss": 0.1709,
      "step": 10205
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5463863700916005,
      "learning_rate": 1.6097020186654366e-05,
      "loss": 0.2049,
      "step": 10206
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.36643293834345375,
      "learning_rate": 1.609623395932579e-05,
      "loss": 0.3239,
      "step": 10207
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6531584233893164,
      "learning_rate": 1.609544767202046e-05,
      "loss": 0.3658,
      "step": 10208
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3384296254481939,
      "learning_rate": 1.609466132474611e-05,
      "loss": 0.3156,
      "step": 10209
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8423311506209934,
      "learning_rate": 1.609387491751048e-05,
      "loss": 0.5696,
      "step": 10210
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.333153559822756,
      "learning_rate": 1.6093088450321296e-05,
      "loss": 0.1729,
      "step": 10211
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5539532720224598,
      "learning_rate": 1.609230192318631e-05,
      "loss": 0.3736,
      "step": 10212
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8528408217988468,
      "learning_rate": 1.609151533611325e-05,
      "loss": 0.3963,
      "step": 10213
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3035560507945973,
      "learning_rate": 1.609072868910986e-05,
      "loss": 0.2688,
      "step": 10214
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3170450070829141,
      "learning_rate": 1.6089941982183873e-05,
      "loss": 0.1891,
      "step": 10215
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38352810547678406,
      "learning_rate": 1.6089155215343035e-05,
      "loss": 0.3036,
      "step": 10216
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4606109110352815,
      "learning_rate": 1.6088368388595083e-05,
      "loss": 0.2727,
      "step": 10217
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.460445919451501,
      "learning_rate": 1.6087581501947763e-05,
      "loss": 0.8865,
      "step": 10218
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.841057966615726,
      "learning_rate": 1.608679455540881e-05,
      "loss": 0.389,
      "step": 10219
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6920019131339973,
      "learning_rate": 1.608600754898597e-05,
      "loss": 0.3899,
      "step": 10220
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.24995510853106662,
      "learning_rate": 1.6085220482686988e-05,
      "loss": 0.2531,
      "step": 10221
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.4688300798523841,
      "learning_rate": 1.60844333565196e-05,
      "loss": 0.6548,
      "step": 10222
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.631799338143344,
      "learning_rate": 1.608364617049156e-05,
      "loss": 0.3467,
      "step": 10223
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6634880967055506,
      "learning_rate": 1.60828589246106e-05,
      "loss": 0.3242,
      "step": 10224
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3390732349098744,
      "learning_rate": 1.6082071618884478e-05,
      "loss": 0.2932,
      "step": 10225
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.26952196130382033,
      "learning_rate": 1.6081284253320935e-05,
      "loss": 0.1814,
      "step": 10226
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8741380061935106,
      "learning_rate": 1.608049682792771e-05,
      "loss": 0.4015,
      "step": 10227
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5702604464317862,
      "learning_rate": 1.6079709342712565e-05,
      "loss": 0.1612,
      "step": 10228
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.38564143140302337,
      "learning_rate": 1.6078921797683233e-05,
      "loss": 0.2987,
      "step": 10229
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4190552419470364,
      "learning_rate": 1.6078134192847468e-05,
      "loss": 0.2389,
      "step": 10230
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.923041898129934,
      "learning_rate": 1.6077346528213024e-05,
      "loss": 0.7733,
      "step": 10231
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.47359659527510295,
      "learning_rate": 1.6076558803787636e-05,
      "loss": 0.3273,
      "step": 10232
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.44318384233426783,
      "learning_rate": 1.6075771019579072e-05,
      "loss": 0.3394,
      "step": 10233
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30258830195518865,
      "learning_rate": 1.6074983175595068e-05,
      "loss": 0.2507,
      "step": 10234
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2585606926413553,
      "learning_rate": 1.6074195271843382e-05,
      "loss": 0.6817,
      "step": 10235
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.27474873235560954,
      "learning_rate": 1.6073407308331767e-05,
      "loss": 0.1601,
      "step": 10236
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.265728503186176,
      "learning_rate": 1.607261928506797e-05,
      "loss": 0.2055,
      "step": 10237
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.39198188340340545,
      "learning_rate": 1.6071831202059746e-05,
      "loss": 0.2718,
      "step": 10238
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30536711555286106,
      "learning_rate": 1.607104305931485e-05,
      "loss": 0.2304,
      "step": 10239
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2728110427972774,
      "learning_rate": 1.6070254856841035e-05,
      "loss": 0.7816,
      "step": 10240
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6048036182609239,
      "learning_rate": 1.6069466594646056e-05,
      "loss": 0.3785,
      "step": 10241
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8134576702477433,
      "learning_rate": 1.6068678272737664e-05,
      "loss": 0.3483,
      "step": 10242
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37388378054674165,
      "learning_rate": 1.6067889891123623e-05,
      "loss": 0.2002,
      "step": 10243
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5407578551376494,
      "learning_rate": 1.6067101449811683e-05,
      "loss": 0.3783,
      "step": 10244
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3025369560331677,
      "learning_rate": 1.60663129488096e-05,
      "loss": 0.272,
      "step": 10245
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5068114497278625,
      "learning_rate": 1.6065524388125136e-05,
      "loss": 0.2947,
      "step": 10246
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2758426866662696,
      "learning_rate": 1.6064735767766047e-05,
      "loss": 0.0755,
      "step": 10247
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40595211525515273,
      "learning_rate": 1.606394708774009e-05,
      "loss": 0.3236,
      "step": 10248
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7448877771615153,
      "learning_rate": 1.606315834805503e-05,
      "loss": 0.38,
      "step": 10249
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.43092963668195017,
      "learning_rate": 1.6062369548718622e-05,
      "loss": 0.284,
      "step": 10250
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.47165107428485964,
      "learning_rate": 1.6061580689738625e-05,
      "loss": 0.3794,
      "step": 10251
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.2946873509672642,
      "learning_rate": 1.6060791771122802e-05,
      "loss": 0.2068,
      "step": 10252
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2503000005610219,
      "learning_rate": 1.606000279287892e-05,
      "loss": 0.8176,
      "step": 10253
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9679615395551169,
      "learning_rate": 1.6059213755014736e-05,
      "loss": 0.5027,
      "step": 10254
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.37406954456216956,
      "learning_rate": 1.605842465753801e-05,
      "loss": 0.1695,
      "step": 10255
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.30643598722838017,
      "learning_rate": 1.605763550045651e-05,
      "loss": 0.1906,
      "step": 10256
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.32640248228233343,
      "learning_rate": 1.6056846283778004e-05,
      "loss": 0.2853,
      "step": 10257
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.9145252989572731,
      "learning_rate": 1.6056057007510244e-05,
      "loss": 0.3625,
      "step": 10258
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6856773519411579,
      "learning_rate": 1.6055267671661005e-05,
      "loss": 0.4688,
      "step": 10259
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5512017738941171,
      "learning_rate": 1.605447827623805e-05,
      "loss": 0.185,
      "step": 10260
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3464926445463203,
      "learning_rate": 1.6053688821249145e-05,
      "loss": 0.2673,
      "step": 10261
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.2638300813927499,
      "learning_rate": 1.605289930670206e-05,
      "loss": 0.4142,
      "step": 10262
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3087788736892104,
      "learning_rate": 1.6052109732604556e-05,
      "loss": 0.2654,
      "step": 10263
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.42682556225820717,
      "learning_rate": 1.6051320098964407e-05,
      "loss": 0.2864,
      "step": 10264
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.23043980429755398,
      "learning_rate": 1.605053040578938e-05,
      "loss": 0.1373,
      "step": 10265
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6414159363327141,
      "learning_rate": 1.6049740653087243e-05,
      "loss": 0.4484,
      "step": 10266
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8782433433329959,
      "learning_rate": 1.604895084086577e-05,
      "loss": 0.4398,
      "step": 10267
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.32768800632527434,
      "learning_rate": 1.6048160969132723e-05,
      "loss": 0.3515,
      "step": 10268
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7919560436460807,
      "learning_rate": 1.6047371037895884e-05,
      "loss": 0.3483,
      "step": 10269
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.6902376391383332,
      "learning_rate": 1.6046581047163016e-05,
      "loss": 0.3027,
      "step": 10270
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3480482467679667,
      "learning_rate": 1.6045790996941893e-05,
      "loss": 0.2499,
      "step": 10271
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.308598623951589,
      "learning_rate": 1.6045000887240293e-05,
      "loss": 0.8978,
      "step": 10272
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5643770347778224,
      "learning_rate": 1.604421071806598e-05,
      "loss": 0.3647,
      "step": 10273
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3235996477740257,
      "learning_rate": 1.6043420489426734e-05,
      "loss": 0.3117,
      "step": 10274
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.264731980993492,
      "learning_rate": 1.6042630201330333e-05,
      "loss": 0.1982,
      "step": 10275
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.3820112443760946,
      "learning_rate": 1.6041839853784545e-05,
      "loss": 0.2746,
      "step": 10276
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.4753352614362085,
      "learning_rate": 1.6041049446797145e-05,
      "loss": 0.2609,
      "step": 10277
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.594925998379228,
      "learning_rate": 1.6040258980375914e-05,
      "loss": 0.0223,
      "step": 10278
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.40647054364367335,
      "learning_rate": 1.6039468454528634e-05,
      "loss": 0.2793,
      "step": 10279
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.27139384290630025,
      "learning_rate": 1.603867786926307e-05,
      "loss": 0.2536,
      "step": 10280
    },
    {
      "epoch": 0.31,
      "grad_norm": 1.6973625653678843,
      "learning_rate": 1.6037887224587007e-05,
      "loss": 0.9212,
      "step": 10281
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.8904394488652503,
      "learning_rate": 1.6037096520508226e-05,
      "loss": 0.4679,
      "step": 10282
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.5707011858238319,
      "learning_rate": 1.6036305757034502e-05,
      "loss": 0.4191,
      "step": 10283
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.31805248521819124,
      "learning_rate": 1.6035514934173614e-05,
      "loss": 0.2447,
      "step": 10284
    },
    {
      "epoch": 0.31,
      "grad_norm": 0.7829483014874011,
      "learning_rate": 1.6034724051933345e-05,
      "loss": 0.5452,
      "step": 10285
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24753877881520336,
      "learning_rate": 1.6033933110321474e-05,
      "loss": 0.1868,
      "step": 10286
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8931326118588443,
      "learning_rate": 1.6033142109345786e-05,
      "loss": 0.6458,
      "step": 10287
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3341860579158121,
      "learning_rate": 1.6032351049014058e-05,
      "loss": 0.1805,
      "step": 10288
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4278134148740906,
      "learning_rate": 1.6031559929334075e-05,
      "loss": 0.2538,
      "step": 10289
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3493488052682259,
      "learning_rate": 1.6030768750313624e-05,
      "loss": 0.8307,
      "step": 10290
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39264558568694136,
      "learning_rate": 1.602997751196048e-05,
      "loss": 0.3078,
      "step": 10291
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38268017181567654,
      "learning_rate": 1.602918621428244e-05,
      "loss": 0.2953,
      "step": 10292
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32785974334527196,
      "learning_rate": 1.6028394857287278e-05,
      "loss": 0.2339,
      "step": 10293
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7890774373869017,
      "learning_rate": 1.6027603440982785e-05,
      "loss": 0.5147,
      "step": 10294
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2102296683561182,
      "learning_rate": 1.602681196537675e-05,
      "loss": 0.0971,
      "step": 10295
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.00860652556256,
      "learning_rate": 1.602602043047695e-05,
      "loss": 0.6503,
      "step": 10296
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.23764393917572643,
      "learning_rate": 1.6025228836291177e-05,
      "loss": 0.0753,
      "step": 10297
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.382695858214615,
      "learning_rate": 1.6024437182827223e-05,
      "loss": 0.324,
      "step": 10298
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31057278597795884,
      "learning_rate": 1.6023645470092874e-05,
      "loss": 0.2693,
      "step": 10299
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6722426188723721,
      "learning_rate": 1.6022853698095922e-05,
      "loss": 0.4899,
      "step": 10300
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7814371306923633,
      "learning_rate": 1.602206186684415e-05,
      "loss": 0.3665,
      "step": 10301
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.30776125775993823,
      "learning_rate": 1.602126997634535e-05,
      "loss": 0.2005,
      "step": 10302
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4228387288069302,
      "learning_rate": 1.6020478026607317e-05,
      "loss": 0.2453,
      "step": 10303
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.23167450057799244,
      "learning_rate": 1.6019686017637842e-05,
      "loss": 0.1972,
      "step": 10304
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2842419473740465,
      "learning_rate": 1.601889394944471e-05,
      "loss": 0.6232,
      "step": 10305
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32310967289416925,
      "learning_rate": 1.6018101822035722e-05,
      "loss": 0.1238,
      "step": 10306
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4110724154100903,
      "learning_rate": 1.6017309635418667e-05,
      "loss": 0.3644,
      "step": 10307
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0920655107245099,
      "learning_rate": 1.6016517389601344e-05,
      "loss": 0.5148,
      "step": 10308
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7680235231491838,
      "learning_rate": 1.601572508459154e-05,
      "loss": 0.4365,
      "step": 10309
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29695502980963356,
      "learning_rate": 1.601493272039705e-05,
      "loss": 0.264,
      "step": 10310
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.34963192217294314,
      "learning_rate": 1.601414029702568e-05,
      "loss": 0.278,
      "step": 10311
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1133510881698179,
      "learning_rate": 1.6013347814485215e-05,
      "loss": 0.3109,
      "step": 10312
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45951357947694,
      "learning_rate": 1.6012555272783454e-05,
      "loss": 0.2832,
      "step": 10313
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2382347114115046,
      "learning_rate": 1.60117626719282e-05,
      "loss": 0.1035,
      "step": 10314
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3332075758973336,
      "learning_rate": 1.601097001192724e-05,
      "loss": 0.2039,
      "step": 10315
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5956338598080975,
      "learning_rate": 1.6010177292788385e-05,
      "loss": 0.3619,
      "step": 10316
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.316650603592301,
      "learning_rate": 1.6009384514519423e-05,
      "loss": 0.2855,
      "step": 10317
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.899565178742066,
      "learning_rate": 1.6008591677128163e-05,
      "loss": 0.5053,
      "step": 10318
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5065753827908516,
      "learning_rate": 1.60077987806224e-05,
      "loss": 0.2694,
      "step": 10319
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9167659612671755,
      "learning_rate": 1.6007005825009936e-05,
      "loss": 0.5192,
      "step": 10320
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3379519116121797,
      "learning_rate": 1.6006212810298574e-05,
      "loss": 0.2378,
      "step": 10321
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3200329387638765,
      "learning_rate": 1.600541973649611e-05,
      "loss": 0.299,
      "step": 10322
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29803741687572477,
      "learning_rate": 1.6004626603610353e-05,
      "loss": 0.1192,
      "step": 10323
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9636194797952179,
      "learning_rate": 1.6003833411649106e-05,
      "loss": 0.5806,
      "step": 10324
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3176318851800954,
      "learning_rate": 1.6003040160620166e-05,
      "loss": 0.2286,
      "step": 10325
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7598415288104038,
      "learning_rate": 1.6002246850531342e-05,
      "loss": 0.4675,
      "step": 10326
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31745022825297114,
      "learning_rate": 1.6001453481390444e-05,
      "loss": 0.2926,
      "step": 10327
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6607779354592949,
      "learning_rate": 1.6000660053205265e-05,
      "loss": 0.9147,
      "step": 10328
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3258168639823852,
      "learning_rate": 1.599986656598362e-05,
      "loss": 0.1766,
      "step": 10329
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.418419511059506,
      "learning_rate": 1.5999073019733313e-05,
      "loss": 0.2385,
      "step": 10330
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5635548883257073,
      "learning_rate": 1.5998279414462154e-05,
      "loss": 0.2781,
      "step": 10331
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25476175025066305,
      "learning_rate": 1.5997485750177948e-05,
      "loss": 0.1425,
      "step": 10332
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.434480854044699,
      "learning_rate": 1.5996692026888502e-05,
      "loss": 0.3285,
      "step": 10333
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3020928303805796,
      "learning_rate": 1.599589824460163e-05,
      "loss": 0.2223,
      "step": 10334
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7803530745469702,
      "learning_rate": 1.5995104403325136e-05,
      "loss": 0.5269,
      "step": 10335
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7061478194046378,
      "learning_rate": 1.599431050306683e-05,
      "loss": 0.402,
      "step": 10336
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1845205160407752,
      "learning_rate": 1.5993516543834528e-05,
      "loss": 0.7432,
      "step": 10337
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2988818839780128,
      "learning_rate": 1.5992722525636037e-05,
      "loss": 0.1658,
      "step": 10338
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5336335098735155,
      "learning_rate": 1.599192844847917e-05,
      "loss": 0.3407,
      "step": 10339
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.30002235404385474,
      "learning_rate": 1.599113431237174e-05,
      "loss": 0.232,
      "step": 10340
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3963368802846525,
      "learning_rate": 1.599034011732156e-05,
      "loss": 0.1949,
      "step": 10341
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44218043066108836,
      "learning_rate": 1.598954586333644e-05,
      "loss": 0.2534,
      "step": 10342
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32963347692737516,
      "learning_rate": 1.59887515504242e-05,
      "loss": 0.2233,
      "step": 10343
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8762674585031514,
      "learning_rate": 1.598795717859265e-05,
      "loss": 0.5057,
      "step": 10344
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2939544834757549,
      "learning_rate": 1.5987162747849605e-05,
      "loss": 0.2479,
      "step": 10345
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9828434687796462,
      "learning_rate": 1.5986368258202885e-05,
      "loss": 0.6061,
      "step": 10346
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2530373548856043,
      "learning_rate": 1.5985573709660305e-05,
      "loss": 0.077,
      "step": 10347
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3673216835228116,
      "learning_rate": 1.598477910222968e-05,
      "loss": 0.331,
      "step": 10348
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21621559789340578,
      "learning_rate": 1.598398443591883e-05,
      "loss": 0.0696,
      "step": 10349
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4447967060012601,
      "learning_rate": 1.598318971073557e-05,
      "loss": 0.2921,
      "step": 10350
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3428955888726182,
      "learning_rate": 1.5982394926687724e-05,
      "loss": 0.2665,
      "step": 10351
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.380462405331936,
      "learning_rate": 1.5981600083783105e-05,
      "loss": 0.2578,
      "step": 10352
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38892047887588027,
      "learning_rate": 1.598080518202954e-05,
      "loss": 0.3195,
      "step": 10353
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3020034113046113,
      "learning_rate": 1.5980010221434843e-05,
      "loss": 0.836,
      "step": 10354
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8472861216297737,
      "learning_rate": 1.597921520200684e-05,
      "loss": 0.3377,
      "step": 10355
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2996734594144719,
      "learning_rate": 1.5978420123753347e-05,
      "loss": 0.1665,
      "step": 10356
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5309967605593479,
      "learning_rate": 1.5977624986682192e-05,
      "loss": 0.4205,
      "step": 10357
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2716379611492963,
      "learning_rate": 1.5976829790801193e-05,
      "loss": 0.2494,
      "step": 10358
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4888916125007215,
      "learning_rate": 1.5976034536118178e-05,
      "loss": 0.2779,
      "step": 10359
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5119059653082972,
      "learning_rate": 1.597523922264097e-05,
      "loss": 0.3277,
      "step": 10360
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3523361086354919,
      "learning_rate": 1.597444385037739e-05,
      "loss": 0.2849,
      "step": 10361
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6331445509886742,
      "learning_rate": 1.597364841933527e-05,
      "loss": 0.4273,
      "step": 10362
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.49263575960882705,
      "learning_rate": 1.5972852929522426e-05,
      "loss": 0.3976,
      "step": 10363
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2511764878642623,
      "learning_rate": 1.5972057380946693e-05,
      "loss": 0.1765,
      "step": 10364
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8742920244362313,
      "learning_rate": 1.597126177361589e-05,
      "loss": 0.3071,
      "step": 10365
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4362003959750747,
      "learning_rate": 1.5970466107537852e-05,
      "loss": 0.2643,
      "step": 10366
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1818136297926913,
      "learning_rate": 1.5969670382720404e-05,
      "loss": 0.6819,
      "step": 10367
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6705718452293677,
      "learning_rate": 1.5968874599171376e-05,
      "loss": 0.3843,
      "step": 10368
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2722715596641395,
      "learning_rate": 1.596807875689859e-05,
      "loss": 0.2173,
      "step": 10369
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5258008703163287,
      "learning_rate": 1.596728285590989e-05,
      "loss": 0.3522,
      "step": 10370
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.28850975878182894,
      "learning_rate": 1.5966486896213092e-05,
      "loss": 0.2112,
      "step": 10371
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.289929493350683,
      "learning_rate": 1.596569087781603e-05,
      "loss": 0.8721,
      "step": 10372
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2649431115708881,
      "learning_rate": 1.5964894800726544e-05,
      "loss": 0.0711,
      "step": 10373
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.1852846845283778,
      "learning_rate": 1.5964098664952458e-05,
      "loss": 0.5682,
      "step": 10374
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3139634427688694,
      "learning_rate": 1.5963302470501607e-05,
      "loss": 0.2097,
      "step": 10375
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32264886101695467,
      "learning_rate": 1.5962506217381826e-05,
      "loss": 0.2922,
      "step": 10376
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5809455234552312,
      "learning_rate": 1.5961709905600943e-05,
      "loss": 0.3918,
      "step": 10377
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2037241291998777,
      "learning_rate": 1.59609135351668e-05,
      "loss": 0.6604,
      "step": 10378
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29483001959220323,
      "learning_rate": 1.5960117106087224e-05,
      "loss": 0.1713,
      "step": 10379
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5207975822878383,
      "learning_rate": 1.5959320618370057e-05,
      "loss": 0.4363,
      "step": 10380
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3034714993749331,
      "learning_rate": 1.5958524072023132e-05,
      "loss": 0.2636,
      "step": 10381
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2984649234642631,
      "learning_rate": 1.5957727467054288e-05,
      "loss": 0.0709,
      "step": 10382
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7763232770293735,
      "learning_rate": 1.5956930803471366e-05,
      "loss": 0.4491,
      "step": 10383
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31347447575007914,
      "learning_rate": 1.595613408128219e-05,
      "loss": 0.2167,
      "step": 10384
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6691522526447178,
      "learning_rate": 1.5955337300494612e-05,
      "loss": 0.5112,
      "step": 10385
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6690428555279145,
      "learning_rate": 1.5954540461116463e-05,
      "loss": 0.3663,
      "step": 10386
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35109184823129397,
      "learning_rate": 1.5953743563155585e-05,
      "loss": 0.3221,
      "step": 10387
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2967307970736954,
      "learning_rate": 1.595294660661982e-05,
      "loss": 0.2,
      "step": 10388
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.4984739194892482,
      "learning_rate": 1.5952149591517012e-05,
      "loss": 0.9894,
      "step": 10389
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9784366197708477,
      "learning_rate": 1.595135251785499e-05,
      "loss": 0.5124,
      "step": 10390
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3385077881104313,
      "learning_rate": 1.5950555385641612e-05,
      "loss": 0.1885,
      "step": 10391
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4217797486167712,
      "learning_rate": 1.5949758194884706e-05,
      "loss": 0.1796,
      "step": 10392
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31922505568092985,
      "learning_rate": 1.5948960945592125e-05,
      "loss": 0.2971,
      "step": 10393
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41485536477347046,
      "learning_rate": 1.5948163637771706e-05,
      "loss": 0.2935,
      "step": 10394
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7882868952769179,
      "learning_rate": 1.5947366271431297e-05,
      "loss": 0.403,
      "step": 10395
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6550397563180694,
      "learning_rate": 1.5946568846578744e-05,
      "loss": 0.8302,
      "step": 10396
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44061145532144586,
      "learning_rate": 1.5945771363221883e-05,
      "loss": 0.0722,
      "step": 10397
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37561217802848407,
      "learning_rate": 1.594497382136857e-05,
      "loss": 0.2984,
      "step": 10398
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.30093368212022303,
      "learning_rate": 1.5944176221026654e-05,
      "loss": 0.276,
      "step": 10399
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.28908430704108296,
      "learning_rate": 1.594337856220397e-05,
      "loss": 0.1973,
      "step": 10400
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35896600981439314,
      "learning_rate": 1.5942580844908376e-05,
      "loss": 0.0675,
      "step": 10401
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.340065213257828,
      "learning_rate": 1.5941783069147715e-05,
      "loss": 0.2905,
      "step": 10402
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7543586960909692,
      "learning_rate": 1.5940985234929837e-05,
      "loss": 0.4167,
      "step": 10403
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4216501645874854,
      "learning_rate": 1.594018734226259e-05,
      "loss": 0.3763,
      "step": 10404
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4173308670691338,
      "learning_rate": 1.5939389391153827e-05,
      "loss": 0.2549,
      "step": 10405
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6539401591800263,
      "learning_rate": 1.5938591381611394e-05,
      "loss": 0.3108,
      "step": 10406
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3564618121702372,
      "learning_rate": 1.5937793313643147e-05,
      "loss": 0.2606,
      "step": 10407
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6847232538210752,
      "learning_rate": 1.5936995187256938e-05,
      "loss": 0.4539,
      "step": 10408
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5096465500612672,
      "learning_rate": 1.5936197002460614e-05,
      "loss": 0.289,
      "step": 10409
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2505177090607871,
      "learning_rate": 1.5935398759262032e-05,
      "loss": 0.1979,
      "step": 10410
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3682010953215748,
      "learning_rate": 1.593460045766904e-05,
      "loss": 0.2899,
      "step": 10411
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35093223699680576,
      "learning_rate": 1.59338020976895e-05,
      "loss": 0.331,
      "step": 10412
    },
    {
      "epoch": 0.32,
      "grad_norm": 2.0303325329094144,
      "learning_rate": 1.5933003679331265e-05,
      "loss": 0.7943,
      "step": 10413
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8705896032122663,
      "learning_rate": 1.5932205202602183e-05,
      "loss": 0.3655,
      "step": 10414
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6321676147569507,
      "learning_rate": 1.5931406667510116e-05,
      "loss": 0.325,
      "step": 10415
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3626445088736829,
      "learning_rate": 1.5930608074062916e-05,
      "loss": 0.2594,
      "step": 10416
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5099526786339733,
      "learning_rate": 1.5929809422268444e-05,
      "loss": 0.4217,
      "step": 10417
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2462904827215338,
      "learning_rate": 1.5929010712134555e-05,
      "loss": 0.1922,
      "step": 10418
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6271751006999349,
      "learning_rate": 1.592821194366911e-05,
      "loss": 0.4378,
      "step": 10419
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3164914374409993,
      "learning_rate": 1.5927413116879967e-05,
      "loss": 0.2108,
      "step": 10420
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44031887603860054,
      "learning_rate": 1.5926614231774976e-05,
      "loss": 0.2482,
      "step": 10421
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6261001498071608,
      "learning_rate": 1.5925815288362013e-05,
      "loss": 0.8677,
      "step": 10422
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.31913401822826964,
      "learning_rate": 1.5925016286648924e-05,
      "loss": 0.2552,
      "step": 10423
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8660608615605102,
      "learning_rate": 1.5924217226643573e-05,
      "loss": 0.3469,
      "step": 10424
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35514234503312314,
      "learning_rate": 1.5923418108353827e-05,
      "loss": 0.252,
      "step": 10425
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2367573295015586,
      "learning_rate": 1.5922618931787547e-05,
      "loss": 0.7756,
      "step": 10426
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5597811536309918,
      "learning_rate": 1.592181969695259e-05,
      "loss": 0.3679,
      "step": 10427
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8016322957485993,
      "learning_rate": 1.5921020403856822e-05,
      "loss": 0.3582,
      "step": 10428
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2987262568621384,
      "learning_rate": 1.592022105250811e-05,
      "loss": 0.1651,
      "step": 10429
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45697543461201645,
      "learning_rate": 1.591942164291431e-05,
      "loss": 0.3576,
      "step": 10430
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.30272938584369335,
      "learning_rate": 1.5918622175083293e-05,
      "loss": 0.1007,
      "step": 10431
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.641639566753297,
      "learning_rate": 1.5917822649022928e-05,
      "loss": 0.7584,
      "step": 10432
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2784317441227277,
      "learning_rate": 1.5917023064741073e-05,
      "loss": 0.1286,
      "step": 10433
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35102677929802756,
      "learning_rate": 1.59162234222456e-05,
      "loss": 0.2507,
      "step": 10434
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45930000610038085,
      "learning_rate": 1.5915423721544372e-05,
      "loss": 0.3059,
      "step": 10435
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9141059675935301,
      "learning_rate": 1.591462396264526e-05,
      "loss": 0.3503,
      "step": 10436
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8374613180831111,
      "learning_rate": 1.5913824145556133e-05,
      "loss": 0.4386,
      "step": 10437
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.28170117086664964,
      "learning_rate": 1.5913024270284856e-05,
      "loss": 0.1957,
      "step": 10438
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4679415484157726,
      "learning_rate": 1.5912224336839303e-05,
      "loss": 0.2678,
      "step": 10439
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.20601420377248855,
      "learning_rate": 1.591142434522734e-05,
      "loss": 0.0725,
      "step": 10440
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3840444115089893,
      "learning_rate": 1.591062429545684e-05,
      "loss": 0.3177,
      "step": 10441
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2912969997653665,
      "learning_rate": 1.5909824187535673e-05,
      "loss": 0.1593,
      "step": 10442
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4115435455902664,
      "learning_rate": 1.590902402147171e-05,
      "loss": 0.3414,
      "step": 10443
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9835288227419768,
      "learning_rate": 1.5908223797272825e-05,
      "loss": 0.3906,
      "step": 10444
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8462255281826562,
      "learning_rate": 1.5907423514946893e-05,
      "loss": 0.4982,
      "step": 10445
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2964312483358562,
      "learning_rate": 1.5906623174501778e-05,
      "loss": 0.2703,
      "step": 10446
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32250866950204826,
      "learning_rate": 1.5905822775945368e-05,
      "loss": 0.1763,
      "step": 10447
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5273743521665754,
      "learning_rate": 1.5905022319285527e-05,
      "loss": 0.3922,
      "step": 10448
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2146660122699302,
      "learning_rate": 1.590422180453013e-05,
      "loss": 0.1171,
      "step": 10449
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1299146427737345,
      "learning_rate": 1.590342123168706e-05,
      "loss": 0.7823,
      "step": 10450
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.33819002444580504,
      "learning_rate": 1.590262060076419e-05,
      "loss": 0.1275,
      "step": 10451
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39944443554483533,
      "learning_rate": 1.5901819911769395e-05,
      "loss": 0.3446,
      "step": 10452
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32287966389938794,
      "learning_rate": 1.5901019164710555e-05,
      "loss": 0.2975,
      "step": 10453
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6982794786146638,
      "learning_rate": 1.5900218359595546e-05,
      "loss": 0.4766,
      "step": 10454
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9830087780465137,
      "learning_rate": 1.5899417496432247e-05,
      "loss": 0.0842,
      "step": 10455
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6474842012511581,
      "learning_rate": 1.5898616575228536e-05,
      "loss": 0.304,
      "step": 10456
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3083987024977686,
      "learning_rate": 1.5897815595992297e-05,
      "loss": 0.2198,
      "step": 10457
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.26443847432644246,
      "learning_rate": 1.5897014558731405e-05,
      "loss": 0.2122,
      "step": 10458
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40932949849860834,
      "learning_rate": 1.5896213463453746e-05,
      "loss": 0.2563,
      "step": 10459
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.46087416609369497,
      "learning_rate": 1.58954123101672e-05,
      "loss": 0.2875,
      "step": 10460
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3508409246550915,
      "learning_rate": 1.5894611098879643e-05,
      "loss": 0.3165,
      "step": 10461
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5794642829860199,
      "learning_rate": 1.5893809829598964e-05,
      "loss": 0.384,
      "step": 10462
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9586201165731281,
      "learning_rate": 1.5893008502333046e-05,
      "loss": 0.6041,
      "step": 10463
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29657929066853894,
      "learning_rate": 1.589220711708977e-05,
      "loss": 0.2726,
      "step": 10464
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4019290504665589,
      "learning_rate": 1.5891405673877023e-05,
      "loss": 0.2997,
      "step": 10465
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40531460877740505,
      "learning_rate": 1.5890604172702687e-05,
      "loss": 0.2455,
      "step": 10466
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.49486450147831135,
      "learning_rate": 1.5889802613574648e-05,
      "loss": 0.2559,
      "step": 10467
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.27749460103747414,
      "learning_rate": 1.5889000996500798e-05,
      "loss": 0.1017,
      "step": 10468
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8311728483460645,
      "learning_rate": 1.5888199321489013e-05,
      "loss": 0.4753,
      "step": 10469
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2650520613727233,
      "learning_rate": 1.5887397588547187e-05,
      "loss": 0.2252,
      "step": 10470
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4869638631045118,
      "learning_rate": 1.588659579768321e-05,
      "loss": 0.3861,
      "step": 10471
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6395201639462875,
      "learning_rate": 1.5885793948904963e-05,
      "loss": 0.3807,
      "step": 10472
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0614782711408635,
      "learning_rate": 1.5884992042220336e-05,
      "loss": 0.3935,
      "step": 10473
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4025333544506078,
      "learning_rate": 1.5884190077637226e-05,
      "loss": 0.2431,
      "step": 10474
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42641484014663866,
      "learning_rate": 1.5883388055163517e-05,
      "loss": 0.24,
      "step": 10475
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3353989417169314,
      "learning_rate": 1.5882585974807097e-05,
      "loss": 0.2417,
      "step": 10476
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.28919309462655085,
      "learning_rate": 1.5881783836575862e-05,
      "loss": 0.2167,
      "step": 10477
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8742022837123237,
      "learning_rate": 1.5880981640477702e-05,
      "loss": 0.4482,
      "step": 10478
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37639353310036017,
      "learning_rate": 1.588017938652051e-05,
      "loss": 0.1986,
      "step": 10479
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.43732700276719566,
      "learning_rate": 1.5879377074712185e-05,
      "loss": 0.3668,
      "step": 10480
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8070077058411368,
      "learning_rate": 1.5878574705060606e-05,
      "loss": 0.5606,
      "step": 10481
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3519080330637695,
      "learning_rate": 1.587777227757367e-05,
      "loss": 0.3276,
      "step": 10482
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2558159740984137,
      "learning_rate": 1.5876969792259286e-05,
      "loss": 0.0735,
      "step": 10483
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37097164384612374,
      "learning_rate": 1.5876167249125338e-05,
      "loss": 0.3231,
      "step": 10484
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2916696342131118,
      "learning_rate": 1.587536464817972e-05,
      "loss": 0.0722,
      "step": 10485
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.23130397353256035,
      "learning_rate": 1.587456198943033e-05,
      "loss": 0.1044,
      "step": 10486
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4607257398604869,
      "learning_rate": 1.5873759272885072e-05,
      "loss": 0.3565,
      "step": 10487
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2806824930289118,
      "learning_rate": 1.5872956498551834e-05,
      "loss": 0.1988,
      "step": 10488
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4558851087675135,
      "learning_rate": 1.5872153666438518e-05,
      "loss": 0.3723,
      "step": 10489
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0750354689140604,
      "learning_rate": 1.587135077655302e-05,
      "loss": 0.527,
      "step": 10490
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1782477533865248,
      "learning_rate": 1.5870547828903246e-05,
      "loss": 0.7051,
      "step": 10491
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.29535150088165674,
      "learning_rate": 1.5869744823497085e-05,
      "loss": 0.0689,
      "step": 10492
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40442042465641115,
      "learning_rate": 1.5868941760342447e-05,
      "loss": 0.2971,
      "step": 10493
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3886662298263242,
      "learning_rate": 1.5868138639447227e-05,
      "loss": 0.2577,
      "step": 10494
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2556444495837325,
      "learning_rate": 1.586733546081933e-05,
      "loss": 0.2024,
      "step": 10495
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7297210256994556,
      "learning_rate": 1.5866532224466654e-05,
      "loss": 0.334,
      "step": 10496
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.33404525522413236,
      "learning_rate": 1.5865728930397103e-05,
      "loss": 0.2754,
      "step": 10497
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7744852148638504,
      "learning_rate": 1.586492557861858e-05,
      "loss": 0.3723,
      "step": 10498
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7355472602516132,
      "learning_rate": 1.586412216913899e-05,
      "loss": 0.4965,
      "step": 10499
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.34462437392118683,
      "learning_rate": 1.5863318701966243e-05,
      "loss": 0.3219,
      "step": 10500
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32405711815739446,
      "learning_rate": 1.586251517710823e-05,
      "loss": 0.1869,
      "step": 10501
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5942093928860052,
      "learning_rate": 1.5861711594572867e-05,
      "loss": 0.3769,
      "step": 10502
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1972669919149943,
      "learning_rate": 1.5860907954368056e-05,
      "loss": 0.0789,
      "step": 10503
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2651718466764283,
      "learning_rate": 1.5860104256501704e-05,
      "loss": 0.4954,
      "step": 10504
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3011407054406,
      "learning_rate": 1.585930050098172e-05,
      "loss": 0.2059,
      "step": 10505
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9713028733710041,
      "learning_rate": 1.5858496687816005e-05,
      "loss": 0.4999,
      "step": 10506
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3017821592827925,
      "learning_rate": 1.5857692817012474e-05,
      "loss": 0.2522,
      "step": 10507
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3293094961203271,
      "learning_rate": 1.585688888857903e-05,
      "loss": 0.843,
      "step": 10508
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.27573149171250627,
      "learning_rate": 1.5856084902523592e-05,
      "loss": 0.1186,
      "step": 10509
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8149679853112556,
      "learning_rate": 1.5855280858854064e-05,
      "loss": 0.49,
      "step": 10510
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3449735957557162,
      "learning_rate": 1.5854476757578353e-05,
      "loss": 0.2291,
      "step": 10511
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.34408848721668295,
      "learning_rate": 1.5853672598704375e-05,
      "loss": 0.3055,
      "step": 10512
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9930621741575925,
      "learning_rate": 1.5852868382240037e-05,
      "loss": 0.3993,
      "step": 10513
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.42439068823775844,
      "learning_rate": 1.5852064108193258e-05,
      "loss": 0.2239,
      "step": 10514
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37942283326538223,
      "learning_rate": 1.5851259776571942e-05,
      "loss": 0.27,
      "step": 10515
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.45626803557119056,
      "learning_rate": 1.585045538738401e-05,
      "loss": 0.2994,
      "step": 10516
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4105486702750178,
      "learning_rate": 1.5849650940637374e-05,
      "loss": 0.2584,
      "step": 10517
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.341157094718942,
      "learning_rate": 1.5848846436339945e-05,
      "loss": 0.2547,
      "step": 10518
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7239785468820139,
      "learning_rate": 1.584804187449964e-05,
      "loss": 0.412,
      "step": 10519
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3423711446713258,
      "learning_rate": 1.584723725512438e-05,
      "loss": 0.2285,
      "step": 10520
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9129346077212653,
      "learning_rate": 1.584643257822207e-05,
      "loss": 0.6138,
      "step": 10521
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7375489972767209,
      "learning_rate": 1.5845627843800632e-05,
      "loss": 0.4155,
      "step": 10522
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.40606742215805114,
      "learning_rate": 1.584482305186799e-05,
      "loss": 0.3719,
      "step": 10523
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.24965292260753502,
      "learning_rate": 1.5844018202432054e-05,
      "loss": 0.1953,
      "step": 10524
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6050670601371128,
      "learning_rate": 1.5843213295500735e-05,
      "loss": 0.4025,
      "step": 10525
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9110085922139695,
      "learning_rate": 1.584240833108197e-05,
      "loss": 0.6446,
      "step": 10526
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.19547346016564052,
      "learning_rate": 1.5841603309183667e-05,
      "loss": 0.0716,
      "step": 10527
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6351469569367472,
      "learning_rate": 1.584079822981375e-05,
      "loss": 0.3596,
      "step": 10528
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.38821756037842364,
      "learning_rate": 1.583999309298014e-05,
      "loss": 0.2387,
      "step": 10529
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.34711982623856846,
      "learning_rate": 1.5839187898690753e-05,
      "loss": 0.3454,
      "step": 10530
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9765298651317328,
      "learning_rate": 1.583838264695351e-05,
      "loss": 0.3855,
      "step": 10531
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.3698349993482952,
      "learning_rate": 1.5837577337776342e-05,
      "loss": 0.7585,
      "step": 10532
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.25828222878923696,
      "learning_rate": 1.5836771971167172e-05,
      "loss": 0.0791,
      "step": 10533
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4007233805741597,
      "learning_rate": 1.583596654713391e-05,
      "loss": 0.3187,
      "step": 10534
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4435297142025844,
      "learning_rate": 1.5835161065684492e-05,
      "loss": 0.2857,
      "step": 10535
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.21974214447246904,
      "learning_rate": 1.583435552682684e-05,
      "loss": 0.1771,
      "step": 10536
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8641419866353175,
      "learning_rate": 1.583354993056888e-05,
      "loss": 0.0747,
      "step": 10537
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39109200853105863,
      "learning_rate": 1.5832744276918535e-05,
      "loss": 0.2427,
      "step": 10538
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4018658345664961,
      "learning_rate": 1.5831938565883732e-05,
      "loss": 0.3286,
      "step": 10539
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.051850384419171,
      "learning_rate": 1.5831132797472398e-05,
      "loss": 0.343,
      "step": 10540
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5287795048826139,
      "learning_rate": 1.583032697169246e-05,
      "loss": 0.3846,
      "step": 10541
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2591689722089943,
      "learning_rate": 1.5829521088551852e-05,
      "loss": 0.2059,
      "step": 10542
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5383782012062884,
      "learning_rate": 1.5828715148058497e-05,
      "loss": 0.4044,
      "step": 10543
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8954251017186955,
      "learning_rate": 1.5827909150220317e-05,
      "loss": 0.459,
      "step": 10544
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3356622979917197,
      "learning_rate": 1.5827103095045253e-05,
      "loss": 0.2096,
      "step": 10545
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5599888729409608,
      "learning_rate": 1.5826296982541234e-05,
      "loss": 0.3339,
      "step": 10546
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3620029688802392,
      "learning_rate": 1.5825490812716185e-05,
      "loss": 0.2893,
      "step": 10547
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3499342706958694,
      "learning_rate": 1.582468458557804e-05,
      "loss": 0.2714,
      "step": 10548
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4106465922454802,
      "learning_rate": 1.5823878301134737e-05,
      "loss": 0.3674,
      "step": 10549
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.714421772431557,
      "learning_rate": 1.5823071959394197e-05,
      "loss": 0.383,
      "step": 10550
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5993129028251926,
      "learning_rate": 1.582226556036436e-05,
      "loss": 0.356,
      "step": 10551
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3246240370375241,
      "learning_rate": 1.582145910405316e-05,
      "loss": 0.2217,
      "step": 10552
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4279674926580497,
      "learning_rate": 1.582065259046853e-05,
      "loss": 0.2974,
      "step": 10553
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.32273919893092884,
      "learning_rate": 1.5819846019618404e-05,
      "loss": 0.2556,
      "step": 10554
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.587244531105573,
      "learning_rate": 1.581903939151072e-05,
      "loss": 0.3375,
      "step": 10555
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8370720803903015,
      "learning_rate": 1.581823270615341e-05,
      "loss": 0.4061,
      "step": 10556
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2545418067742633,
      "learning_rate": 1.581742596355441e-05,
      "loss": 0.2088,
      "step": 10557
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.42665099992745,
      "learning_rate": 1.5816619163721662e-05,
      "loss": 0.6893,
      "step": 10558
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3555981975724867,
      "learning_rate": 1.58158123066631e-05,
      "loss": 0.2746,
      "step": 10559
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.908678830101858,
      "learning_rate": 1.5815005392386665e-05,
      "loss": 0.468,
      "step": 10560
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3131521535047258,
      "learning_rate": 1.581419842090029e-05,
      "loss": 0.2209,
      "step": 10561
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2243367861496925,
      "learning_rate": 1.581339139221192e-05,
      "loss": 0.698,
      "step": 10562
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3363530962982072,
      "learning_rate": 1.581258430632949e-05,
      "loss": 0.2231,
      "step": 10563
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8594982555183535,
      "learning_rate": 1.5811777163260946e-05,
      "loss": 0.4625,
      "step": 10564
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2961693288372886,
      "learning_rate": 1.5810969963014225e-05,
      "loss": 0.2203,
      "step": 10565
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.240016331032115,
      "learning_rate": 1.581016270559727e-05,
      "loss": 0.2142,
      "step": 10566
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.0802898652922268,
      "learning_rate": 1.5809355391018017e-05,
      "loss": 0.5899,
      "step": 10567
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.198005005083741,
      "learning_rate": 1.5808548019284424e-05,
      "loss": 0.3745,
      "step": 10568
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9111460073595514,
      "learning_rate": 1.5807740590404418e-05,
      "loss": 0.4732,
      "step": 10569
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.30596532822853034,
      "learning_rate": 1.5806933104385948e-05,
      "loss": 0.2117,
      "step": 10570
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3490950727414725,
      "learning_rate": 1.5806125561236963e-05,
      "loss": 0.2876,
      "step": 10571
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8892766712561027,
      "learning_rate": 1.5805317960965403e-05,
      "loss": 0.3292,
      "step": 10572
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6885598343228765,
      "learning_rate": 1.5804510303579212e-05,
      "loss": 0.4903,
      "step": 10573
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2808554906625055,
      "learning_rate": 1.5803702589086344e-05,
      "loss": 0.1997,
      "step": 10574
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5356259982746676,
      "learning_rate": 1.5802894817494738e-05,
      "loss": 0.3219,
      "step": 10575
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.22911128522410343,
      "learning_rate": 1.580208698881234e-05,
      "loss": 0.0749,
      "step": 10576
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3353355837399502,
      "learning_rate": 1.5801279103047104e-05,
      "loss": 0.31,
      "step": 10577
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3029622666778738,
      "learning_rate": 1.5800471160206978e-05,
      "loss": 0.1728,
      "step": 10578
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3258463822639178,
      "learning_rate": 1.5799663160299904e-05,
      "loss": 0.2207,
      "step": 10579
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.2307208600284694,
      "learning_rate": 1.5798855103333838e-05,
      "loss": 0.7512,
      "step": 10580
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.1652080272303509,
      "learning_rate": 1.5798046989316728e-05,
      "loss": 0.4363,
      "step": 10581
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44791807808263534,
      "learning_rate": 1.579723881825652e-05,
      "loss": 0.3686,
      "step": 10582
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2385137364335011,
      "learning_rate": 1.5796430590161172e-05,
      "loss": 0.0754,
      "step": 10583
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.33681456736960047,
      "learning_rate": 1.579562230503863e-05,
      "loss": 0.2923,
      "step": 10584
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3764762351889725,
      "learning_rate": 1.579481396289685e-05,
      "loss": 0.0923,
      "step": 10585
    },
    {
      "epoch": 0.32,
      "grad_norm": 1.6303384702954051,
      "learning_rate": 1.5794005563743783e-05,
      "loss": 0.9779,
      "step": 10586
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.8632270042531226,
      "learning_rate": 1.5793197107587386e-05,
      "loss": 0.5182,
      "step": 10587
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3899813890918665,
      "learning_rate": 1.579238859443561e-05,
      "loss": 0.2571,
      "step": 10588
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.33559858622416483,
      "learning_rate": 1.5791580024296403e-05,
      "loss": 0.2714,
      "step": 10589
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4273617723489627,
      "learning_rate": 1.579077139717773e-05,
      "loss": 0.3985,
      "step": 10590
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.7784176621958767,
      "learning_rate": 1.5789962713087544e-05,
      "loss": 0.3697,
      "step": 10591
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3222473300993492,
      "learning_rate": 1.5789153972033796e-05,
      "loss": 0.1675,
      "step": 10592
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3270967650249444,
      "learning_rate": 1.5788345174024446e-05,
      "loss": 0.2619,
      "step": 10593
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2863138814619625,
      "learning_rate": 1.5787536319067454e-05,
      "loss": 0.2063,
      "step": 10594
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4886947700426063,
      "learning_rate": 1.5786727407170777e-05,
      "loss": 0.3644,
      "step": 10595
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.5455159767632622,
      "learning_rate": 1.578591843834237e-05,
      "loss": 0.29,
      "step": 10596
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.35458995183071823,
      "learning_rate": 1.5785109412590193e-05,
      "loss": 0.2914,
      "step": 10597
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6202054816911433,
      "learning_rate": 1.5784300329922207e-05,
      "loss": 0.4183,
      "step": 10598
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6290602450361376,
      "learning_rate": 1.578349119034637e-05,
      "loss": 0.4915,
      "step": 10599
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.37092761619126546,
      "learning_rate": 1.5782681993870644e-05,
      "loss": 0.2529,
      "step": 10600
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.39779787431028873,
      "learning_rate": 1.5781872740502988e-05,
      "loss": 0.3125,
      "step": 10601
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3352239672045087,
      "learning_rate": 1.5781063430251367e-05,
      "loss": 0.2381,
      "step": 10602
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.44279382499064324,
      "learning_rate": 1.5780254063123744e-05,
      "loss": 0.255,
      "step": 10603
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.2686971435658265,
      "learning_rate": 1.5779444639128077e-05,
      "loss": 0.1351,
      "step": 10604
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.47456767510625475,
      "learning_rate": 1.5778635158272332e-05,
      "loss": 0.2374,
      "step": 10605
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.41670270614484634,
      "learning_rate": 1.5777825620564474e-05,
      "loss": 0.3242,
      "step": 10606
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.4088161034152301,
      "learning_rate": 1.5777016026012465e-05,
      "loss": 0.3315,
      "step": 10607
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.458198231858032,
      "learning_rate": 1.5776206374624275e-05,
      "loss": 0.3975,
      "step": 10608
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.9556669223260214,
      "learning_rate": 1.577539666640786e-05,
      "loss": 0.3238,
      "step": 10609
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.6451379167518144,
      "learning_rate": 1.5774586901371196e-05,
      "loss": 0.3385,
      "step": 10610
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.33025511011883185,
      "learning_rate": 1.5773777079522247e-05,
      "loss": 0.2232,
      "step": 10611
    },
    {
      "epoch": 0.32,
      "grad_norm": 0.3155898886068228,
      "learning_rate": 1.5772967200868977e-05,
      "loss": 0.1971,
      "step": 10612
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31924119031063136,
      "learning_rate": 1.5772157265419357e-05,
      "loss": 0.2803,
      "step": 10613
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5969417054637725,
      "learning_rate": 1.5771347273181357e-05,
      "loss": 0.9281,
      "step": 10614
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3196638126824037,
      "learning_rate": 1.5770537224162938e-05,
      "loss": 0.2148,
      "step": 10615
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4293149714265361,
      "learning_rate": 1.576972711837208e-05,
      "loss": 0.3962,
      "step": 10616
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6307227606067017,
      "learning_rate": 1.5768916955816746e-05,
      "loss": 0.4084,
      "step": 10617
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42764620651584934,
      "learning_rate": 1.5768106736504908e-05,
      "loss": 0.2519,
      "step": 10618
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4016892102882692,
      "learning_rate": 1.5767296460444538e-05,
      "loss": 0.2954,
      "step": 10619
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.29489697420489586,
      "learning_rate": 1.5766486127643614e-05,
      "loss": 0.2191,
      "step": 10620
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2387938492452032,
      "learning_rate": 1.5765675738110093e-05,
      "loss": 0.1328,
      "step": 10621
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7988946211161333,
      "learning_rate": 1.576486529185196e-05,
      "loss": 0.4622,
      "step": 10622
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8269942249237948,
      "learning_rate": 1.5764054788877184e-05,
      "loss": 0.4597,
      "step": 10623
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2879166586908265,
      "learning_rate": 1.5763244229193745e-05,
      "loss": 0.1994,
      "step": 10624
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3352180555060752,
      "learning_rate": 1.576243361280961e-05,
      "loss": 0.3429,
      "step": 10625
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8771155718765753,
      "learning_rate": 1.5761622939732757e-05,
      "loss": 0.486,
      "step": 10626
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4905493394658598,
      "learning_rate": 1.5760812209971163e-05,
      "loss": 0.7544,
      "step": 10627
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4240326330095446,
      "learning_rate": 1.57600014235328e-05,
      "loss": 0.1816,
      "step": 10628
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4042949218020417,
      "learning_rate": 1.575919058042565e-05,
      "loss": 0.3359,
      "step": 10629
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3299270707924102,
      "learning_rate": 1.575837968065769e-05,
      "loss": 0.1106,
      "step": 10630
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3409215788367946,
      "learning_rate": 1.5757568724236892e-05,
      "loss": 0.2873,
      "step": 10631
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6577184069183216,
      "learning_rate": 1.575675771117124e-05,
      "loss": 0.4835,
      "step": 10632
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3033429294457229,
      "learning_rate": 1.5755946641468712e-05,
      "loss": 0.1754,
      "step": 10633
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5353777668683598,
      "learning_rate": 1.5755135515137287e-05,
      "loss": 0.406,
      "step": 10634
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24973005778002122,
      "learning_rate": 1.575432433218494e-05,
      "loss": 0.1277,
      "step": 10635
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3471581704230338,
      "learning_rate": 1.5753513092619664e-05,
      "loss": 0.3065,
      "step": 10636
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4833209172737045,
      "learning_rate": 1.5752701796449432e-05,
      "loss": 0.0773,
      "step": 10637
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3274631917265453,
      "learning_rate": 1.5751890443682224e-05,
      "loss": 0.3,
      "step": 10638
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3005715129325594,
      "learning_rate": 1.5751079034326025e-05,
      "loss": 0.1126,
      "step": 10639
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7841992610848169,
      "learning_rate": 1.5750267568388822e-05,
      "loss": 0.497,
      "step": 10640
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.706589770727184,
      "learning_rate": 1.574945604587859e-05,
      "loss": 0.3752,
      "step": 10641
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6499708030156452,
      "learning_rate": 1.574864446680332e-05,
      "loss": 0.4305,
      "step": 10642
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2625779949418867,
      "learning_rate": 1.5747832831170994e-05,
      "loss": 0.244,
      "step": 10643
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40394685720098916,
      "learning_rate": 1.5747021138989597e-05,
      "loss": 0.1445,
      "step": 10644
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4416294113024544,
      "learning_rate": 1.5746209390267117e-05,
      "loss": 0.7294,
      "step": 10645
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5648024190229503,
      "learning_rate": 1.5745397585011533e-05,
      "loss": 0.1736,
      "step": 10646
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3554287281270294,
      "learning_rate": 1.5744585723230843e-05,
      "loss": 0.278,
      "step": 10647
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4111277089600404,
      "learning_rate": 1.5743773804933024e-05,
      "loss": 0.3251,
      "step": 10648
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5117793830073079,
      "learning_rate": 1.574296183012607e-05,
      "loss": 0.3871,
      "step": 10649
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8423407523343518,
      "learning_rate": 1.5742149798817972e-05,
      "loss": 0.4338,
      "step": 10650
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3808768557903327,
      "learning_rate": 1.5741337711016707e-05,
      "loss": 0.2525,
      "step": 10651
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4036349811951621,
      "learning_rate": 1.5740525566730278e-05,
      "loss": 0.2882,
      "step": 10652
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47427467204214513,
      "learning_rate": 1.573971336596667e-05,
      "loss": 0.2621,
      "step": 10653
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24037362777233795,
      "learning_rate": 1.573890110873387e-05,
      "loss": 0.2017,
      "step": 10654
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7236568672196257,
      "learning_rate": 1.573808879503987e-05,
      "loss": 0.461,
      "step": 10655
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3103849931419569,
      "learning_rate": 1.5737276424892672e-05,
      "loss": 0.2082,
      "step": 10656
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7594651957259524,
      "learning_rate": 1.5736463998300258e-05,
      "loss": 0.3285,
      "step": 10657
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6036565407109425,
      "learning_rate": 1.573565151527062e-05,
      "loss": 0.4834,
      "step": 10658
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3742575394423163,
      "learning_rate": 1.573483897581176e-05,
      "loss": 0.2921,
      "step": 10659
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38557542250366866,
      "learning_rate": 1.5734026379931665e-05,
      "loss": 0.2565,
      "step": 10660
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32309985453719736,
      "learning_rate": 1.573321372763833e-05,
      "loss": 0.2324,
      "step": 10661
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47765441448673074,
      "learning_rate": 1.5732401018939753e-05,
      "loss": 0.2949,
      "step": 10662
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.29339786279053626,
      "learning_rate": 1.573158825384393e-05,
      "loss": 0.1159,
      "step": 10663
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2568745513322042,
      "learning_rate": 1.5730775432358854e-05,
      "loss": 0.6992,
      "step": 10664
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32707886431950367,
      "learning_rate": 1.5729962554492524e-05,
      "loss": 0.1839,
      "step": 10665
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.49263875342892793,
      "learning_rate": 1.572914962025294e-05,
      "loss": 0.4018,
      "step": 10666
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32568177276845023,
      "learning_rate": 1.5728336629648092e-05,
      "loss": 0.2855,
      "step": 10667
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6631297997303438,
      "learning_rate": 1.5727523582685983e-05,
      "loss": 0.7905,
      "step": 10668
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3643070511085864,
      "learning_rate": 1.5726710479374617e-05,
      "loss": 0.1184,
      "step": 10669
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.33345520881188984,
      "learning_rate": 1.5725897319721988e-05,
      "loss": 0.2514,
      "step": 10670
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27536314230925996,
      "learning_rate": 1.5725084103736094e-05,
      "loss": 0.158,
      "step": 10671
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.30651288773369506,
      "learning_rate": 1.572427083142494e-05,
      "loss": 0.2409,
      "step": 10672
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2167101189999199,
      "learning_rate": 1.572345750279653e-05,
      "loss": 0.601,
      "step": 10673
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2823578675021654,
      "learning_rate": 1.572264411785886e-05,
      "loss": 0.1991,
      "step": 10674
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.626147284013267,
      "learning_rate": 1.572183067661993e-05,
      "loss": 0.4564,
      "step": 10675
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6525978340453505,
      "learning_rate": 1.5721017179087753e-05,
      "loss": 0.342,
      "step": 10676
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1882034067801692,
      "learning_rate": 1.5720203625270324e-05,
      "loss": 0.7622,
      "step": 10677
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.25988495871321615,
      "learning_rate": 1.571939001517565e-05,
      "loss": 0.2036,
      "step": 10678
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40788260796802284,
      "learning_rate": 1.5718576348811732e-05,
      "loss": 0.3263,
      "step": 10679
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.138096863147161,
      "learning_rate": 1.571776262618658e-05,
      "loss": 0.3821,
      "step": 10680
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2937016470266374,
      "learning_rate": 1.57169488473082e-05,
      "loss": 0.1778,
      "step": 10681
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.715651711105387,
      "learning_rate": 1.5716135012184595e-05,
      "loss": 0.3088,
      "step": 10682
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3202364076939007,
      "learning_rate": 1.571532112082377e-05,
      "loss": 0.2496,
      "step": 10683
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8597508767590002,
      "learning_rate": 1.5714507173233738e-05,
      "loss": 0.5775,
      "step": 10684
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3189063306584317,
      "learning_rate": 1.5713693169422503e-05,
      "loss": 0.2451,
      "step": 10685
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.471293327762981,
      "learning_rate": 1.5712879109398075e-05,
      "loss": 0.8909,
      "step": 10686
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.34200298682840424,
      "learning_rate": 1.571206499316846e-05,
      "loss": 0.1248,
      "step": 10687
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.39330405756473574,
      "learning_rate": 1.5711250820741674e-05,
      "loss": 0.3019,
      "step": 10688
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3689453217106408,
      "learning_rate": 1.5710436592125724e-05,
      "loss": 0.1727,
      "step": 10689
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27563842919309456,
      "learning_rate": 1.5709622307328615e-05,
      "loss": 0.2828,
      "step": 10690
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6771216513645955,
      "learning_rate": 1.570880796635837e-05,
      "loss": 0.3973,
      "step": 10691
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6072214368766755,
      "learning_rate": 1.570799356922299e-05,
      "loss": 0.3225,
      "step": 10692
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36101988983692174,
      "learning_rate": 1.570717911593049e-05,
      "loss": 0.2855,
      "step": 10693
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.669830095676474,
      "learning_rate": 1.5706364606488883e-05,
      "loss": 0.8181,
      "step": 10694
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38492230345864525,
      "learning_rate": 1.5705550040906186e-05,
      "loss": 0.2825,
      "step": 10695
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3369624405448377,
      "learning_rate": 1.570473541919041e-05,
      "loss": 0.2327,
      "step": 10696
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.334098163803501,
      "learning_rate": 1.5703920741349567e-05,
      "loss": 0.2627,
      "step": 10697
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7956991196761829,
      "learning_rate": 1.570310600739168e-05,
      "loss": 0.3994,
      "step": 10698
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3904651402327098,
      "learning_rate": 1.5702291217324753e-05,
      "loss": 0.1977,
      "step": 10699
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6208658907192997,
      "learning_rate": 1.5701476371156813e-05,
      "loss": 0.3723,
      "step": 10700
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3912526928906151,
      "learning_rate": 1.570066146889587e-05,
      "loss": 0.279,
      "step": 10701
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2006019032310489,
      "learning_rate": 1.5699846510549947e-05,
      "loss": 0.2088,
      "step": 10702
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.919208037467632,
      "learning_rate": 1.5699031496127056e-05,
      "loss": 0.8706,
      "step": 10703
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.85412210825004,
      "learning_rate": 1.5698216425635224e-05,
      "loss": 0.4532,
      "step": 10704
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8738699345022428,
      "learning_rate": 1.5697401299082454e-05,
      "loss": 0.5645,
      "step": 10705
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32851452015090876,
      "learning_rate": 1.569658611647678e-05,
      "loss": 0.204,
      "step": 10706
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.682625209997303,
      "learning_rate": 1.569577087782622e-05,
      "loss": 0.8503,
      "step": 10707
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3318599583876309,
      "learning_rate": 1.569495558313879e-05,
      "loss": 0.282,
      "step": 10708
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6733822351621238,
      "learning_rate": 1.569414023242251e-05,
      "loss": 0.4341,
      "step": 10709
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.21682559825102712,
      "learning_rate": 1.5693324825685406e-05,
      "loss": 0.1304,
      "step": 10710
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2490300947883818,
      "learning_rate": 1.5692509362935502e-05,
      "loss": 0.1388,
      "step": 10711
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1212085556505047,
      "learning_rate": 1.5691693844180816e-05,
      "loss": 0.686,
      "step": 10712
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.26975561671919895,
      "learning_rate": 1.5690878269429375e-05,
      "loss": 0.2348,
      "step": 10713
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6398765122230717,
      "learning_rate": 1.5690062638689195e-05,
      "loss": 0.3459,
      "step": 10714
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35839710538584196,
      "learning_rate": 1.5689246951968313e-05,
      "loss": 0.2412,
      "step": 10715
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1943302717732256,
      "learning_rate": 1.5688431209274745e-05,
      "loss": 0.7411,
      "step": 10716
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6735312681168145,
      "learning_rate": 1.568761541061652e-05,
      "loss": 0.3643,
      "step": 10717
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3824247320234973,
      "learning_rate": 1.568679955600166e-05,
      "loss": 0.2385,
      "step": 10718
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2426184175879152,
      "learning_rate": 1.56859836454382e-05,
      "loss": 0.0695,
      "step": 10719
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3055946703619946,
      "learning_rate": 1.568516767893416e-05,
      "loss": 0.2619,
      "step": 10720
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2567450555906433,
      "learning_rate": 1.5684351656497566e-05,
      "loss": 0.1749,
      "step": 10721
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.157359977939743,
      "learning_rate": 1.5683535578136452e-05,
      "loss": 0.6821,
      "step": 10722
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9885545946478647,
      "learning_rate": 1.5682719443858847e-05,
      "loss": 0.4936,
      "step": 10723
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2878213900015326,
      "learning_rate": 1.5681903253672776e-05,
      "loss": 0.1943,
      "step": 10724
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7306877263023476,
      "learning_rate": 1.5681087007586274e-05,
      "loss": 0.5199,
      "step": 10725
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3169255194084001,
      "learning_rate": 1.5680270705607367e-05,
      "loss": 0.2476,
      "step": 10726
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.75495119223924,
      "learning_rate": 1.5679454347744086e-05,
      "loss": 0.5151,
      "step": 10727
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40088121353371803,
      "learning_rate": 1.5678637934004465e-05,
      "loss": 0.1387,
      "step": 10728
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2722674062855634,
      "learning_rate": 1.5677821464396535e-05,
      "loss": 0.2267,
      "step": 10729
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.21565795512361757,
      "learning_rate": 1.5677004938928334e-05,
      "loss": 0.0668,
      "step": 10730
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.179315076747232,
      "learning_rate": 1.5676188357607886e-05,
      "loss": 0.7606,
      "step": 10731
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2678391704434987,
      "learning_rate": 1.5675371720443233e-05,
      "loss": 0.2271,
      "step": 10732
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35814760628887143,
      "learning_rate": 1.56745550274424e-05,
      "loss": 0.3126,
      "step": 10733
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.904100662893341,
      "learning_rate": 1.5673738278613432e-05,
      "loss": 0.4077,
      "step": 10734
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6375573808573015,
      "learning_rate": 1.5672921473964358e-05,
      "loss": 0.487,
      "step": 10735
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.032464169227967,
      "learning_rate": 1.5672104613503215e-05,
      "loss": 0.339,
      "step": 10736
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27208625718928053,
      "learning_rate": 1.5671287697238038e-05,
      "loss": 0.216,
      "step": 10737
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5506506171253895,
      "learning_rate": 1.5670470725176873e-05,
      "loss": 0.3993,
      "step": 10738
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23581906395293234,
      "learning_rate": 1.566965369732775e-05,
      "loss": 0.1491,
      "step": 10739
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3876909214735924,
      "learning_rate": 1.5668836613698704e-05,
      "loss": 0.8652,
      "step": 10740
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5337006035607373,
      "learning_rate": 1.566801947429778e-05,
      "loss": 0.3206,
      "step": 10741
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9535275949630111,
      "learning_rate": 1.5667202279133016e-05,
      "loss": 0.5243,
      "step": 10742
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9896449925963049,
      "learning_rate": 1.566638502821245e-05,
      "loss": 0.2939,
      "step": 10743
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3202513104776582,
      "learning_rate": 1.566556772154413e-05,
      "loss": 0.3191,
      "step": 10744
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1004160818215023,
      "learning_rate": 1.5664750359136084e-05,
      "loss": 0.2744,
      "step": 10745
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9329198803849548,
      "learning_rate": 1.566393294099636e-05,
      "loss": 0.4206,
      "step": 10746
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3395808399094057,
      "learning_rate": 1.5663115467133002e-05,
      "loss": 0.2314,
      "step": 10747
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2754272228080943,
      "learning_rate": 1.566229793755405e-05,
      "loss": 0.1609,
      "step": 10748
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3527017865499461,
      "learning_rate": 1.5661480352267546e-05,
      "loss": 0.2752,
      "step": 10749
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8490931556200002,
      "learning_rate": 1.5660662711281538e-05,
      "loss": 0.3982,
      "step": 10750
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4245575656721783,
      "learning_rate": 1.5659845014604066e-05,
      "loss": 0.3207,
      "step": 10751
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38004138442236657,
      "learning_rate": 1.5659027262243177e-05,
      "loss": 0.32,
      "step": 10752
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5992918017718483,
      "learning_rate": 1.5658209454206917e-05,
      "loss": 0.8365,
      "step": 10753
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.3619744527444264,
      "learning_rate": 1.565739159050333e-05,
      "loss": 0.2745,
      "step": 10754
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2989790540751211,
      "learning_rate": 1.5656573671140462e-05,
      "loss": 0.265,
      "step": 10755
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2967675834986206,
      "learning_rate": 1.565575569612636e-05,
      "loss": 0.2073,
      "step": 10756
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27101023489751536,
      "learning_rate": 1.5654937665469073e-05,
      "loss": 0.1771,
      "step": 10757
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8893185472215933,
      "learning_rate": 1.5654119579176652e-05,
      "loss": 0.4596,
      "step": 10758
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6127291163051379,
      "learning_rate": 1.565330143725714e-05,
      "loss": 0.4651,
      "step": 10759
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31399808790559225,
      "learning_rate": 1.5652483239718588e-05,
      "loss": 0.229,
      "step": 10760
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5546626995731911,
      "learning_rate": 1.5651664986569045e-05,
      "loss": 0.3922,
      "step": 10761
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.312969756865313,
      "learning_rate": 1.5650846677816566e-05,
      "loss": 0.2737,
      "step": 10762
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9728221832269095,
      "learning_rate": 1.565002831346919e-05,
      "loss": 0.3538,
      "step": 10763
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6337987869281955,
      "learning_rate": 1.5649209893534985e-05,
      "loss": 0.2824,
      "step": 10764
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.319860995416702,
      "learning_rate": 1.564839141802199e-05,
      "loss": 0.2256,
      "step": 10765
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31114381976607275,
      "learning_rate": 1.564757288693826e-05,
      "loss": 0.207,
      "step": 10766
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2935853144641728,
      "learning_rate": 1.5646754300291857e-05,
      "loss": 0.2642,
      "step": 10767
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6315421933369652,
      "learning_rate": 1.564593565809082e-05,
      "loss": 0.4743,
      "step": 10768
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38377316951143364,
      "learning_rate": 1.5645116960343215e-05,
      "loss": 0.0751,
      "step": 10769
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.336636435788124,
      "learning_rate": 1.564429820705709e-05,
      "loss": 0.3137,
      "step": 10770
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.086362128650075,
      "learning_rate": 1.56434793982405e-05,
      "loss": 0.4654,
      "step": 10771
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8587891268730485,
      "learning_rate": 1.56426605339015e-05,
      "loss": 0.5628,
      "step": 10772
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.28925738763686476,
      "learning_rate": 1.5641841614048155e-05,
      "loss": 0.237,
      "step": 10773
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3907103033700596,
      "learning_rate": 1.5641022638688516e-05,
      "loss": 0.2921,
      "step": 10774
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2845049403391711,
      "learning_rate": 1.5640203607830634e-05,
      "loss": 0.1492,
      "step": 10775
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6653827357563062,
      "learning_rate": 1.5639384521482577e-05,
      "loss": 0.3914,
      "step": 10776
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7286852289303687,
      "learning_rate": 1.5638565379652402e-05,
      "loss": 0.4896,
      "step": 10777
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3109235462936408,
      "learning_rate": 1.5637746182348162e-05,
      "loss": 0.193,
      "step": 10778
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36943538155711164,
      "learning_rate": 1.563692692957792e-05,
      "loss": 0.2895,
      "step": 10779
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2609226859190044,
      "learning_rate": 1.563610762134974e-05,
      "loss": 0.1978,
      "step": 10780
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.9421126529691395,
      "learning_rate": 1.5635288257671674e-05,
      "loss": 0.8105,
      "step": 10781
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5517047895983127,
      "learning_rate": 1.563446883855179e-05,
      "loss": 0.1814,
      "step": 10782
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35445676197741105,
      "learning_rate": 1.5633649363998148e-05,
      "loss": 0.2887,
      "step": 10783
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4587701074700498,
      "learning_rate": 1.5632829834018808e-05,
      "loss": 0.2124,
      "step": 10784
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47394184513748966,
      "learning_rate": 1.5632010248621835e-05,
      "loss": 0.3786,
      "step": 10785
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5201871124809693,
      "learning_rate": 1.563119060781529e-05,
      "loss": 0.3204,
      "step": 10786
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6519549000778119,
      "learning_rate": 1.5630370911607246e-05,
      "loss": 0.4132,
      "step": 10787
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3208943495193793,
      "learning_rate": 1.5629551160005753e-05,
      "loss": 0.2249,
      "step": 10788
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.26394045925431286,
      "learning_rate": 1.5628731353018887e-05,
      "loss": 0.1335,
      "step": 10789
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.172069928658507,
      "learning_rate": 1.562791149065471e-05,
      "loss": 0.6579,
      "step": 10790
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2570327161081266,
      "learning_rate": 1.5627091572921288e-05,
      "loss": 0.2166,
      "step": 10791
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5571336197834493,
      "learning_rate": 1.5626271599826687e-05,
      "loss": 0.3644,
      "step": 10792
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3690359300903372,
      "learning_rate": 1.5625451571378974e-05,
      "loss": 0.2392,
      "step": 10793
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6672282726313583,
      "learning_rate": 1.5624631487586216e-05,
      "loss": 0.4691,
      "step": 10794
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7030576115459903,
      "learning_rate": 1.5623811348456484e-05,
      "loss": 0.4551,
      "step": 10795
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9281110833851252,
      "learning_rate": 1.562299115399785e-05,
      "loss": 0.4881,
      "step": 10796
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2749412173001568,
      "learning_rate": 1.5622170904218373e-05,
      "loss": 0.2219,
      "step": 10797
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.328520584563723,
      "learning_rate": 1.562135059912613e-05,
      "loss": 0.2624,
      "step": 10798
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2593870914547395,
      "learning_rate": 1.5620530238729188e-05,
      "loss": 0.0728,
      "step": 10799
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.8397446306981227,
      "learning_rate": 1.5619709823035623e-05,
      "loss": 0.9792,
      "step": 10800
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3246354559697931,
      "learning_rate": 1.5618889352053504e-05,
      "loss": 0.1685,
      "step": 10801
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.546539248839929,
      "learning_rate": 1.56180688257909e-05,
      "loss": 0.3306,
      "step": 10802
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9130030356608283,
      "learning_rate": 1.561724824425589e-05,
      "loss": 0.3234,
      "step": 10803
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2132802314425197,
      "learning_rate": 1.561642760745654e-05,
      "loss": 0.4123,
      "step": 10804
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8488629452228144,
      "learning_rate": 1.561560691540093e-05,
      "loss": 0.5026,
      "step": 10805
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32125314049250236,
      "learning_rate": 1.561478616809713e-05,
      "loss": 0.2179,
      "step": 10806
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.30104975249721544,
      "learning_rate": 1.5613965365553217e-05,
      "loss": 0.2004,
      "step": 10807
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9623452098032009,
      "learning_rate": 1.5613144507777263e-05,
      "loss": 0.3367,
      "step": 10808
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.34717352854253997,
      "learning_rate": 1.561232359477735e-05,
      "loss": 0.3306,
      "step": 10809
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3019759595671525,
      "learning_rate": 1.5611502626561547e-05,
      "loss": 0.2048,
      "step": 10810
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7298186950613539,
      "learning_rate": 1.561068160313794e-05,
      "loss": 0.4321,
      "step": 10811
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7626513486145166,
      "learning_rate": 1.5609860524514593e-05,
      "loss": 0.3453,
      "step": 10812
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.45099968374112,
      "learning_rate": 1.56090393906996e-05,
      "loss": 0.8294,
      "step": 10813
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31274507255191164,
      "learning_rate": 1.560821820170103e-05,
      "loss": 0.1653,
      "step": 10814
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3546416169008938,
      "learning_rate": 1.5607396957526963e-05,
      "loss": 0.2209,
      "step": 10815
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3549042928065981,
      "learning_rate": 1.560657565818548e-05,
      "loss": 0.3065,
      "step": 10816
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.1717036042823227,
      "learning_rate": 1.560575430368466e-05,
      "loss": 0.0742,
      "step": 10817
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9855529334353286,
      "learning_rate": 1.5604932894032584e-05,
      "loss": 0.5487,
      "step": 10818
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2757539048591494,
      "learning_rate": 1.5604111429237336e-05,
      "loss": 0.0727,
      "step": 10819
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4034117381180424,
      "learning_rate": 1.5603289909306996e-05,
      "loss": 0.3231,
      "step": 10820
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3133881423247424,
      "learning_rate": 1.5602468334249644e-05,
      "loss": 0.275,
      "step": 10821
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2301894642626567,
      "learning_rate": 1.5601646704073368e-05,
      "loss": 0.7732,
      "step": 10822
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3763789623345838,
      "learning_rate": 1.5600825018786245e-05,
      "loss": 0.1534,
      "step": 10823
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3410194456308231,
      "learning_rate": 1.5600003278396366e-05,
      "loss": 0.2838,
      "step": 10824
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2320835553575659,
      "learning_rate": 1.5599181482911806e-05,
      "loss": 0.0987,
      "step": 10825
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32783290573257606,
      "learning_rate": 1.5598359632340664e-05,
      "loss": 0.2864,
      "step": 10826
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5296829354589627,
      "learning_rate": 1.5597537726691015e-05,
      "loss": 0.2892,
      "step": 10827
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3960443162875186,
      "learning_rate": 1.5596715765970946e-05,
      "loss": 0.2388,
      "step": 10828
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5330432160765839,
      "learning_rate": 1.5595893750188545e-05,
      "loss": 0.4255,
      "step": 10829
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8297407342432732,
      "learning_rate": 1.5595071679351903e-05,
      "loss": 0.3795,
      "step": 10830
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7067623598411612,
      "learning_rate": 1.5594249553469107e-05,
      "loss": 0.8864,
      "step": 10831
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.33438431849557226,
      "learning_rate": 1.559342737254824e-05,
      "loss": 0.2517,
      "step": 10832
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.33121116845561693,
      "learning_rate": 1.5592605136597392e-05,
      "loss": 0.2587,
      "step": 10833
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.269856575136128,
      "learning_rate": 1.5591782845624656e-05,
      "loss": 0.186,
      "step": 10834
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4931291506527427,
      "learning_rate": 1.5590960499638123e-05,
      "loss": 0.248,
      "step": 10835
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7586888209834791,
      "learning_rate": 1.5590138098645876e-05,
      "loss": 0.3807,
      "step": 10836
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6230305304350007,
      "learning_rate": 1.5589315642656014e-05,
      "loss": 0.3131,
      "step": 10837
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3511909940788423,
      "learning_rate": 1.5588493131676624e-05,
      "loss": 0.2326,
      "step": 10838
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.369389232788184,
      "learning_rate": 1.5587670565715804e-05,
      "loss": 0.3519,
      "step": 10839
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7932811247999192,
      "learning_rate": 1.5586847944781637e-05,
      "loss": 0.4518,
      "step": 10840
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2711137253919783,
      "learning_rate": 1.5586025268882225e-05,
      "loss": 0.1626,
      "step": 10841
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3843585481889966,
      "learning_rate": 1.5585202538025658e-05,
      "loss": 0.3127,
      "step": 10842
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8439216756497376,
      "learning_rate": 1.558437975222003e-05,
      "loss": 0.3618,
      "step": 10843
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3220316156723763,
      "learning_rate": 1.5583556911473436e-05,
      "loss": 0.268,
      "step": 10844
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.46023685623183375,
      "learning_rate": 1.5582734015793975e-05,
      "loss": 0.3266,
      "step": 10845
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6827351406482185,
      "learning_rate": 1.5581911065189737e-05,
      "loss": 0.3401,
      "step": 10846
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24944135267119166,
      "learning_rate": 1.5581088059668822e-05,
      "loss": 0.1867,
      "step": 10847
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.6253013815260202,
      "learning_rate": 1.5580264999239325e-05,
      "loss": 0.759,
      "step": 10848
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8583143305112446,
      "learning_rate": 1.5579441883909344e-05,
      "loss": 0.6381,
      "step": 10849
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35925059461587255,
      "learning_rate": 1.5578618713686983e-05,
      "loss": 0.3226,
      "step": 10850
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32385413334640656,
      "learning_rate": 1.557779548858033e-05,
      "loss": 0.1796,
      "step": 10851
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5519784327033141,
      "learning_rate": 1.5576972208597493e-05,
      "loss": 0.398,
      "step": 10852
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5823076858888485,
      "learning_rate": 1.5576148873746567e-05,
      "loss": 0.3674,
      "step": 10853
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.32219709775407984,
      "learning_rate": 1.557532548403565e-05,
      "loss": 0.1784,
      "step": 10854
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3662515529288048,
      "learning_rate": 1.557450203947285e-05,
      "loss": 0.2046,
      "step": 10855
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3218689504672189,
      "learning_rate": 1.5573678540066265e-05,
      "loss": 0.2399,
      "step": 10856
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3418382110590471,
      "learning_rate": 1.5572854985823997e-05,
      "loss": 0.337,
      "step": 10857
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8741982969503745,
      "learning_rate": 1.5572031376754147e-05,
      "loss": 0.5096,
      "step": 10858
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2898196712990861,
      "learning_rate": 1.557120771286482e-05,
      "loss": 0.9261,
      "step": 10859
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.27643930446354165,
      "learning_rate": 1.5570383994164115e-05,
      "loss": 0.2027,
      "step": 10860
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7014553959284672,
      "learning_rate": 1.5569560220660145e-05,
      "loss": 0.5148,
      "step": 10861
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4077552969462171,
      "learning_rate": 1.5568736392361007e-05,
      "loss": 0.2818,
      "step": 10862
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.39471820775440847,
      "learning_rate": 1.5567912509274802e-05,
      "loss": 0.3802,
      "step": 10863
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23451542739593786,
      "learning_rate": 1.556708857140965e-05,
      "loss": 0.0988,
      "step": 10864
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.26563731020890596,
      "learning_rate": 1.556626457877364e-05,
      "loss": 0.2364,
      "step": 10865
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.22554942850095,
      "learning_rate": 1.5565440531374892e-05,
      "loss": 0.1048,
      "step": 10866
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.1547872893067486,
      "learning_rate": 1.5564616429221512e-05,
      "loss": 0.6741,
      "step": 10867
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3591494616413213,
      "learning_rate": 1.5563792272321604e-05,
      "loss": 0.3466,
      "step": 10868
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31717929079346574,
      "learning_rate": 1.5562968060683276e-05,
      "loss": 0.1737,
      "step": 10869
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8061949867972676,
      "learning_rate": 1.5562143794314635e-05,
      "loss": 0.4286,
      "step": 10870
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6750348703135746,
      "learning_rate": 1.5561319473223794e-05,
      "loss": 0.369,
      "step": 10871
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.190621496812119,
      "learning_rate": 1.5560495097418863e-05,
      "loss": 0.6239,
      "step": 10872
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2637220059809261,
      "learning_rate": 1.5559670666907953e-05,
      "loss": 0.0774,
      "step": 10873
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38639321823497264,
      "learning_rate": 1.555884618169917e-05,
      "loss": 0.32,
      "step": 10874
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24141895192865148,
      "learning_rate": 1.555802164180063e-05,
      "loss": 0.2192,
      "step": 10875
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.4224646960978535,
      "learning_rate": 1.5557197047220446e-05,
      "loss": 0.8925,
      "step": 10876
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7803228844661988,
      "learning_rate": 1.5556372397966727e-05,
      "loss": 0.3887,
      "step": 10877
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5479438546974948,
      "learning_rate": 1.555554769404759e-05,
      "loss": 0.3758,
      "step": 10878
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38769241342162397,
      "learning_rate": 1.5554722935471148e-05,
      "loss": 0.2828,
      "step": 10879
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2953020998930516,
      "learning_rate": 1.555389812224551e-05,
      "loss": 0.2718,
      "step": 10880
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.7511775000314755,
      "learning_rate": 1.5553073254378798e-05,
      "loss": 0.6486,
      "step": 10881
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.237804234261668,
      "learning_rate": 1.5552248331879125e-05,
      "loss": 0.0777,
      "step": 10882
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.34371244581461796,
      "learning_rate": 1.5551423354754604e-05,
      "loss": 0.2785,
      "step": 10883
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2856152851485762,
      "learning_rate": 1.5550598323013352e-05,
      "loss": 0.1299,
      "step": 10884
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.5919770541726619,
      "learning_rate": 1.554977323666349e-05,
      "loss": 0.831,
      "step": 10885
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3201996163539571,
      "learning_rate": 1.554894809571313e-05,
      "loss": 0.3005,
      "step": 10886
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44601287862335637,
      "learning_rate": 1.5548122900170395e-05,
      "loss": 0.2853,
      "step": 10887
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.431678412204444,
      "learning_rate": 1.5547297650043402e-05,
      "loss": 0.3148,
      "step": 10888
    },
    {
      "epoch": 0.33,
      "grad_norm": 2.176510488594027,
      "learning_rate": 1.5546472345340267e-05,
      "loss": 0.9425,
      "step": 10889
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8390274196110049,
      "learning_rate": 1.554564698606911e-05,
      "loss": 0.388,
      "step": 10890
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.421558038931509,
      "learning_rate": 1.554482157223806e-05,
      "loss": 0.2341,
      "step": 10891
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.24181871568283705,
      "learning_rate": 1.554399610385523e-05,
      "loss": 0.2203,
      "step": 10892
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5168637994987627,
      "learning_rate": 1.5543170580928736e-05,
      "loss": 0.3609,
      "step": 10893
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23281029099891604,
      "learning_rate": 1.554234500346671e-05,
      "loss": 0.1547,
      "step": 10894
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6777543261432091,
      "learning_rate": 1.554151937147727e-05,
      "loss": 0.3863,
      "step": 10895
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40007659004375695,
      "learning_rate": 1.5540693684968537e-05,
      "loss": 0.2948,
      "step": 10896
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.37563321515113823,
      "learning_rate": 1.5539867943948637e-05,
      "loss": 0.2636,
      "step": 10897
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.46313311491138637,
      "learning_rate": 1.5539042148425695e-05,
      "loss": 0.3659,
      "step": 10898
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.40215990969227267,
      "learning_rate": 1.5538216298407834e-05,
      "loss": 0.2625,
      "step": 10899
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3735501386634476,
      "learning_rate": 1.553739039390318e-05,
      "loss": 0.2314,
      "step": 10900
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.38055987172420186,
      "learning_rate": 1.5536564434919852e-05,
      "loss": 0.1968,
      "step": 10901
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.30911202113654,
      "learning_rate": 1.5535738421465985e-05,
      "loss": 0.2484,
      "step": 10902
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36404108024599635,
      "learning_rate": 1.55349123535497e-05,
      "loss": 0.2943,
      "step": 10903
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44169980600939207,
      "learning_rate": 1.5534086231179126e-05,
      "loss": 0.3876,
      "step": 10904
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5133025128730149,
      "learning_rate": 1.5533260054362395e-05,
      "loss": 0.2186,
      "step": 10905
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36320820699803336,
      "learning_rate": 1.5532433823107625e-05,
      "loss": 0.3126,
      "step": 10906
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8126459803272869,
      "learning_rate": 1.5531607537422955e-05,
      "loss": 0.3205,
      "step": 10907
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9999253833202527,
      "learning_rate": 1.5530781197316506e-05,
      "loss": 0.3036,
      "step": 10908
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.478102407486345,
      "learning_rate": 1.5529954802796417e-05,
      "loss": 0.3606,
      "step": 10909
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2500880022498679,
      "learning_rate": 1.552912835387081e-05,
      "loss": 0.1928,
      "step": 10910
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3064203023898204,
      "learning_rate": 1.5528301850547817e-05,
      "loss": 0.2283,
      "step": 10911
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.7273008512365383,
      "learning_rate": 1.552747529283557e-05,
      "loss": 0.3636,
      "step": 10912
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6018796530190849,
      "learning_rate": 1.5526648680742205e-05,
      "loss": 0.4727,
      "step": 10913
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.34679464921638536,
      "learning_rate": 1.5525822014275856e-05,
      "loss": 0.168,
      "step": 10914
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.28890285389401876,
      "learning_rate": 1.552499529344465e-05,
      "loss": 0.2506,
      "step": 10915
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4000234239313747,
      "learning_rate": 1.5524168518256718e-05,
      "loss": 0.2324,
      "step": 10916
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.47010162031070146,
      "learning_rate": 1.5523341688720202e-05,
      "loss": 0.3622,
      "step": 10917
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6224947179784929,
      "learning_rate": 1.5522514804843233e-05,
      "loss": 0.1615,
      "step": 10918
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.35610585530090416,
      "learning_rate": 1.5521687866633946e-05,
      "loss": 0.2969,
      "step": 10919
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2973328201296045,
      "learning_rate": 1.5520860874100474e-05,
      "loss": 0.1099,
      "step": 10920
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.44611777850069323,
      "learning_rate": 1.552003382725096e-05,
      "loss": 0.307,
      "step": 10921
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.4486760821504176,
      "learning_rate": 1.551920672609353e-05,
      "loss": 0.3521,
      "step": 10922
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.31003036096027836,
      "learning_rate": 1.551837957063634e-05,
      "loss": 0.1918,
      "step": 10923
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.5472378850108276,
      "learning_rate": 1.5517552360887505e-05,
      "loss": 0.3853,
      "step": 10924
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.275484136161931,
      "learning_rate": 1.551672509685518e-05,
      "loss": 0.1325,
      "step": 10925
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.9182086986766331,
      "learning_rate": 1.5515897778547494e-05,
      "loss": 0.5251,
      "step": 10926
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.36371316009740967,
      "learning_rate": 1.5515070405972596e-05,
      "loss": 0.2485,
      "step": 10927
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6662400905351612,
      "learning_rate": 1.5514242979138618e-05,
      "loss": 0.3102,
      "step": 10928
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3429738585370152,
      "learning_rate": 1.5513415498053703e-05,
      "loss": 0.2401,
      "step": 10929
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.6040097570713231,
      "learning_rate": 1.5512587962725995e-05,
      "loss": 0.4579,
      "step": 10930
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.765322971774674,
      "learning_rate": 1.5511760373163628e-05,
      "loss": 0.4025,
      "step": 10931
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3631520559308392,
      "learning_rate": 1.5510932729374753e-05,
      "loss": 0.2111,
      "step": 10932
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.2827216912956028,
      "learning_rate": 1.5510105031367507e-05,
      "loss": 0.1903,
      "step": 10933
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.23129169040615996,
      "learning_rate": 1.5509277279150033e-05,
      "loss": 0.1863,
      "step": 10934
    },
    {
      "epoch": 0.33,
      "grad_norm": 1.2995143563567868,
      "learning_rate": 1.5508449472730476e-05,
      "loss": 0.7432,
      "step": 10935
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.8488146614181914,
      "learning_rate": 1.5507621612116987e-05,
      "loss": 0.3528,
      "step": 10936
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.3957540482823067,
      "learning_rate": 1.5506793697317698e-05,
      "loss": 0.2799,
      "step": 10937
    },
    {
      "epoch": 0.33,
      "grad_norm": 0.42357172318905734,
      "learning_rate": 1.550596572834076e-05,
      "loss": 0.3275,
      "step": 10938
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1476275455503653,
      "learning_rate": 1.550513770519432e-05,
      "loss": 0.5638,
      "step": 10939
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2969823274454294,
      "learning_rate": 1.5504309627886528e-05,
      "loss": 0.2574,
      "step": 10940
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8614147269307534,
      "learning_rate": 1.5503481496425522e-05,
      "loss": 0.56,
      "step": 10941
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30487914133475885,
      "learning_rate": 1.5502653310819456e-05,
      "loss": 0.2225,
      "step": 10942
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5096235937555573,
      "learning_rate": 1.5501825071076476e-05,
      "loss": 0.3061,
      "step": 10943
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24144401699988693,
      "learning_rate": 1.5500996777204732e-05,
      "loss": 0.1083,
      "step": 10944
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33108933553932246,
      "learning_rate": 1.550016842921237e-05,
      "loss": 0.2962,
      "step": 10945
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.29436984617647083,
      "learning_rate": 1.5499340027107543e-05,
      "loss": 0.172,
      "step": 10946
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45411271781978685,
      "learning_rate": 1.5498511570898398e-05,
      "loss": 0.2824,
      "step": 10947
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7522536509379969,
      "learning_rate": 1.5497683060593084e-05,
      "loss": 0.4623,
      "step": 10948
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8180292077521241,
      "learning_rate": 1.5496854496199762e-05,
      "loss": 0.6222,
      "step": 10949
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5667089110240409,
      "learning_rate": 1.5496025877726572e-05,
      "loss": 0.3459,
      "step": 10950
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3617625722298182,
      "learning_rate": 1.5495197205181672e-05,
      "loss": 0.2317,
      "step": 10951
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2899659597144691,
      "learning_rate": 1.5494368478573213e-05,
      "loss": 0.2592,
      "step": 10952
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2554141788831509,
      "learning_rate": 1.549353969790935e-05,
      "loss": 0.1117,
      "step": 10953
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.789346645408168,
      "learning_rate": 1.5492710863198235e-05,
      "loss": 0.4926,
      "step": 10954
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3932445826625796,
      "learning_rate": 1.5491881974448025e-05,
      "loss": 0.0776,
      "step": 10955
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33711675559915116,
      "learning_rate": 1.5491053031666872e-05,
      "loss": 0.2995,
      "step": 10956
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37752094180625834,
      "learning_rate": 1.549022403486293e-05,
      "loss": 0.286,
      "step": 10957
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.48050741989912865,
      "learning_rate": 1.548939498404436e-05,
      "loss": 0.403,
      "step": 10958
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.003564711206854,
      "learning_rate": 1.5488565879219315e-05,
      "loss": 0.4083,
      "step": 10959
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27535194091135473,
      "learning_rate": 1.5487736720395954e-05,
      "loss": 0.1952,
      "step": 10960
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2535891572711702,
      "learning_rate": 1.5486907507582435e-05,
      "loss": 0.1651,
      "step": 10961
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7566049481239017,
      "learning_rate": 1.5486078240786914e-05,
      "loss": 0.3683,
      "step": 10962
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33321240095222426,
      "learning_rate": 1.5485248920017545e-05,
      "loss": 0.3122,
      "step": 10963
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3060297069128391,
      "learning_rate": 1.5484419545282494e-05,
      "loss": 0.2202,
      "step": 10964
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43107383847714575,
      "learning_rate": 1.5483590116589922e-05,
      "loss": 0.3658,
      "step": 10965
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2326813810371564,
      "learning_rate": 1.5482760633947987e-05,
      "loss": 0.0753,
      "step": 10966
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2766870030971937,
      "learning_rate": 1.5481931097364845e-05,
      "loss": 0.8804,
      "step": 10967
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.47428262459224957,
      "learning_rate": 1.5481101506848665e-05,
      "loss": 0.2148,
      "step": 10968
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.369870440368484,
      "learning_rate": 1.54802718624076e-05,
      "loss": 0.303,
      "step": 10969
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32386348656473135,
      "learning_rate": 1.547944216404982e-05,
      "loss": 0.2438,
      "step": 10970
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3329324884072156,
      "learning_rate": 1.5478612411783482e-05,
      "loss": 0.2327,
      "step": 10971
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5586555435457785,
      "learning_rate": 1.5477782605616757e-05,
      "loss": 0.3868,
      "step": 10972
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4570200493065545,
      "learning_rate": 1.5476952745557804e-05,
      "loss": 0.2148,
      "step": 10973
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5406659563979512,
      "learning_rate": 1.5476122831614785e-05,
      "loss": 0.3961,
      "step": 10974
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3810959621128044,
      "learning_rate": 1.5475292863795867e-05,
      "loss": 0.2541,
      "step": 10975
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.477246047386268,
      "learning_rate": 1.5474462842109218e-05,
      "loss": 0.4126,
      "step": 10976
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45794646631754565,
      "learning_rate": 1.5473632766563002e-05,
      "loss": 0.2175,
      "step": 10977
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0232519866884822,
      "learning_rate": 1.5472802637165387e-05,
      "loss": 0.4434,
      "step": 10978
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2576983431359498,
      "learning_rate": 1.5471972453924537e-05,
      "loss": 0.1852,
      "step": 10979
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4583181461537763,
      "learning_rate": 1.5471142216848623e-05,
      "loss": 0.2912,
      "step": 10980
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.384115507188974,
      "learning_rate": 1.547031192594581e-05,
      "loss": 0.2952,
      "step": 10981
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9552949381388861,
      "learning_rate": 1.5469481581224274e-05,
      "loss": 0.5127,
      "step": 10982
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3046094508327165,
      "learning_rate": 1.546865118269217e-05,
      "loss": 0.2151,
      "step": 10983
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9640447504320336,
      "learning_rate": 1.546782073035768e-05,
      "loss": 0.5076,
      "step": 10984
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7597737807946322,
      "learning_rate": 1.5466990224228973e-05,
      "loss": 0.6246,
      "step": 10985
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.446213844998861,
      "learning_rate": 1.5466159664314214e-05,
      "loss": 0.2498,
      "step": 10986
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4152524715325414,
      "learning_rate": 1.5465329050621577e-05,
      "loss": 0.3075,
      "step": 10987
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.31435581592351103,
      "learning_rate": 1.5464498383159235e-05,
      "loss": 0.2503,
      "step": 10988
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6747645932011401,
      "learning_rate": 1.5463667661935363e-05,
      "loss": 0.4648,
      "step": 10989
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32558437156938425,
      "learning_rate": 1.5462836886958125e-05,
      "loss": 0.1567,
      "step": 10990
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41096078025361027,
      "learning_rate": 1.5462006058235702e-05,
      "loss": 0.2409,
      "step": 10991
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2851456383226356,
      "learning_rate": 1.5461175175776272e-05,
      "loss": 0.222,
      "step": 10992
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5013314918903029,
      "learning_rate": 1.5460344239587998e-05,
      "loss": 0.38,
      "step": 10993
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4108809846640008,
      "learning_rate": 1.545951324967906e-05,
      "loss": 0.2983,
      "step": 10994
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1871540236986964,
      "learning_rate": 1.545868220605764e-05,
      "loss": 0.786,
      "step": 10995
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2683774137748142,
      "learning_rate": 1.5457851108731904e-05,
      "loss": 0.1919,
      "step": 10996
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6838977150136734,
      "learning_rate": 1.5457019957710033e-05,
      "loss": 0.8832,
      "step": 10997
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7403736069533219,
      "learning_rate": 1.5456188753000208e-05,
      "loss": 0.3801,
      "step": 10998
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2558487068376344,
      "learning_rate": 1.5455357494610595e-05,
      "loss": 0.2247,
      "step": 10999
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37716792925706755,
      "learning_rate": 1.5454526182549387e-05,
      "loss": 0.1954,
      "step": 11000
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32198437933797186,
      "learning_rate": 1.5453694816824755e-05,
      "loss": 0.222,
      "step": 11001
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3623033757644665,
      "learning_rate": 1.5452863397444872e-05,
      "loss": 0.7681,
      "step": 11002
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9525793105151114,
      "learning_rate": 1.545203192441793e-05,
      "loss": 0.6509,
      "step": 11003
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3289329284676275,
      "learning_rate": 1.5451200397752105e-05,
      "loss": 0.3116,
      "step": 11004
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.26397474335614834,
      "learning_rate": 1.5450368817455576e-05,
      "loss": 0.0766,
      "step": 11005
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3793005521100116,
      "learning_rate": 1.544953718353652e-05,
      "loss": 0.304,
      "step": 11006
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6746855414381079,
      "learning_rate": 1.544870549600313e-05,
      "loss": 0.3843,
      "step": 11007
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.475871202864324,
      "learning_rate": 1.5447873754863583e-05,
      "loss": 0.3087,
      "step": 11008
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24403186689209222,
      "learning_rate": 1.5447041960126057e-05,
      "loss": 0.1148,
      "step": 11009
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36653786670447924,
      "learning_rate": 1.5446210111798742e-05,
      "loss": 0.2804,
      "step": 11010
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27989415379758353,
      "learning_rate": 1.5445378209889824e-05,
      "loss": 0.2484,
      "step": 11011
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1252320502954807,
      "learning_rate": 1.5444546254407477e-05,
      "loss": 0.6611,
      "step": 11012
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.074532164474512,
      "learning_rate": 1.5443714245359896e-05,
      "loss": 0.817,
      "step": 11013
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3343797519646456,
      "learning_rate": 1.5442882182755264e-05,
      "loss": 0.2135,
      "step": 11014
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40608569770877423,
      "learning_rate": 1.5442050066601763e-05,
      "loss": 0.3773,
      "step": 11015
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6445316728000337,
      "learning_rate": 1.5441217896907582e-05,
      "loss": 0.3085,
      "step": 11016
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3298898513343306,
      "learning_rate": 1.544038567368091e-05,
      "loss": 0.3115,
      "step": 11017
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2057756640953938,
      "learning_rate": 1.5439553396929937e-05,
      "loss": 0.0915,
      "step": 11018
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35461671457177557,
      "learning_rate": 1.5438721066662844e-05,
      "loss": 0.2687,
      "step": 11019
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9283844572927954,
      "learning_rate": 1.5437888682887825e-05,
      "loss": 0.3908,
      "step": 11020
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5554412068977477,
      "learning_rate": 1.5437056245613068e-05,
      "loss": 0.8326,
      "step": 11021
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4857361811443917,
      "learning_rate": 1.5436223754846763e-05,
      "loss": 0.3064,
      "step": 11022
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3357729250716934,
      "learning_rate": 1.54353912105971e-05,
      "loss": 0.2691,
      "step": 11023
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36184874541935563,
      "learning_rate": 1.543455861287227e-05,
      "loss": 0.2512,
      "step": 11024
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.870645191382963,
      "learning_rate": 1.5433725961680462e-05,
      "loss": 0.3692,
      "step": 11025
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.7420505505043697,
      "learning_rate": 1.543289325702987e-05,
      "loss": 0.8292,
      "step": 11026
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.15494360096295254,
      "learning_rate": 1.5432060498928693e-05,
      "loss": 0.0738,
      "step": 11027
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4087153528330345,
      "learning_rate": 1.543122768738511e-05,
      "loss": 0.3219,
      "step": 11028
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32513057979276055,
      "learning_rate": 1.5430394822407322e-05,
      "loss": 0.2564,
      "step": 11029
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3933193790094354,
      "learning_rate": 1.5429561904003525e-05,
      "loss": 0.9314,
      "step": 11030
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9054405262496305,
      "learning_rate": 1.542872893218191e-05,
      "loss": 0.407,
      "step": 11031
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9427252886374892,
      "learning_rate": 1.5427895906950677e-05,
      "loss": 0.4843,
      "step": 11032
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.31483649481963416,
      "learning_rate": 1.5427062828318015e-05,
      "loss": 0.242,
      "step": 11033
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6013832187550938,
      "learning_rate": 1.5426229696292126e-05,
      "loss": 0.6819,
      "step": 11034
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.29890261639837695,
      "learning_rate": 1.5425396510881202e-05,
      "loss": 0.2463,
      "step": 11035
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.1321465629050573,
      "learning_rate": 1.542456327209344e-05,
      "loss": 0.9937,
      "step": 11036
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.224073562233393,
      "learning_rate": 1.5423729979937044e-05,
      "loss": 0.1313,
      "step": 11037
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2526002517190357,
      "learning_rate": 1.5422896634420203e-05,
      "loss": 0.1852,
      "step": 11038
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.787261140581706,
      "learning_rate": 1.5422063235551123e-05,
      "loss": 0.5375,
      "step": 11039
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40794595523049554,
      "learning_rate": 1.5421229783337997e-05,
      "loss": 0.2982,
      "step": 11040
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36954122443921006,
      "learning_rate": 1.5420396277789037e-05,
      "loss": 0.3309,
      "step": 11041
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30669254744142477,
      "learning_rate": 1.5419562718912427e-05,
      "loss": 0.235,
      "step": 11042
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9968996433694375,
      "learning_rate": 1.5418729106716378e-05,
      "loss": 0.5773,
      "step": 11043
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0605638675086226,
      "learning_rate": 1.5417895441209094e-05,
      "loss": 0.3239,
      "step": 11044
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2990596222575464,
      "learning_rate": 1.5417061722398766e-05,
      "loss": 0.1955,
      "step": 11045
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2805206529170104,
      "learning_rate": 1.5416227950293606e-05,
      "loss": 0.2009,
      "step": 11046
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33548326163595016,
      "learning_rate": 1.541539412490181e-05,
      "loss": 0.3161,
      "step": 11047
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9081323229754986,
      "learning_rate": 1.5414560246231587e-05,
      "loss": 0.4992,
      "step": 11048
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8568216549995005,
      "learning_rate": 1.5413726314291137e-05,
      "loss": 0.472,
      "step": 11049
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6802036044017913,
      "learning_rate": 1.541289232908867e-05,
      "loss": 0.234,
      "step": 11050
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3928494017611304,
      "learning_rate": 1.5412058290632382e-05,
      "loss": 0.2325,
      "step": 11051
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4426159576966517,
      "learning_rate": 1.5411224198930487e-05,
      "loss": 0.7848,
      "step": 11052
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.330243973600448,
      "learning_rate": 1.5410390053991185e-05,
      "loss": 0.2594,
      "step": 11053
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4859090429093266,
      "learning_rate": 1.5409555855822685e-05,
      "loss": 0.2824,
      "step": 11054
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5390525474085467,
      "learning_rate": 1.54087216044332e-05,
      "loss": 0.0758,
      "step": 11055
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37067379560072766,
      "learning_rate": 1.5407887299830927e-05,
      "loss": 0.3175,
      "step": 11056
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0191559731007374,
      "learning_rate": 1.540705294202408e-05,
      "loss": 0.4188,
      "step": 11057
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3491517643274222,
      "learning_rate": 1.540621853102087e-05,
      "loss": 0.3413,
      "step": 11058
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3975886391270421,
      "learning_rate": 1.54053840668295e-05,
      "loss": 0.157,
      "step": 11059
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.355638479477576,
      "learning_rate": 1.5404549549458186e-05,
      "loss": 0.2931,
      "step": 11060
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24778986041197112,
      "learning_rate": 1.540371497891513e-05,
      "loss": 0.1048,
      "step": 11061
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6877215361367002,
      "learning_rate": 1.5402880355208552e-05,
      "loss": 0.9142,
      "step": 11062
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.31539719224811824,
      "learning_rate": 1.540204567834666e-05,
      "loss": 0.1129,
      "step": 11063
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30831428983184633,
      "learning_rate": 1.5401210948337662e-05,
      "loss": 0.1909,
      "step": 11064
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3172060834302762,
      "learning_rate": 1.5400376165189775e-05,
      "loss": 0.2954,
      "step": 11065
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9405436966464614,
      "learning_rate": 1.539954132891121e-05,
      "loss": 0.4161,
      "step": 11066
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6682484041497259,
      "learning_rate": 1.5398706439510182e-05,
      "loss": 0.8636,
      "step": 11067
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23072422037001533,
      "learning_rate": 1.5397871496994902e-05,
      "loss": 0.1185,
      "step": 11068
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35372664380517166,
      "learning_rate": 1.5397036501373588e-05,
      "loss": 0.2978,
      "step": 11069
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30008970106877675,
      "learning_rate": 1.5396201452654454e-05,
      "loss": 0.2747,
      "step": 11070
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2838215704464353,
      "learning_rate": 1.539536635084571e-05,
      "loss": 0.7486,
      "step": 11071
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7676560359228036,
      "learning_rate": 1.539453119595558e-05,
      "loss": 0.3351,
      "step": 11072
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6196010214480322,
      "learning_rate": 1.5393695987992275e-05,
      "loss": 0.3515,
      "step": 11073
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3337916895923301,
      "learning_rate": 1.5392860726964015e-05,
      "loss": 0.2803,
      "step": 11074
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8917494882573906,
      "learning_rate": 1.5392025412879017e-05,
      "loss": 0.5676,
      "step": 11075
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30239932543503695,
      "learning_rate": 1.53911900457455e-05,
      "loss": 0.2733,
      "step": 11076
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.926340825582588,
      "learning_rate": 1.539035462557168e-05,
      "loss": 0.5749,
      "step": 11077
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2169104206595309,
      "learning_rate": 1.5389519152365778e-05,
      "loss": 0.168,
      "step": 11078
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2712927880572351,
      "learning_rate": 1.538868362613601e-05,
      "loss": 0.0679,
      "step": 11079
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1877918675715984,
      "learning_rate": 1.5387848046890602e-05,
      "loss": 0.6671,
      "step": 11080
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3748977809293984,
      "learning_rate": 1.5387012414637776e-05,
      "loss": 0.2829,
      "step": 11081
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6110432162052359,
      "learning_rate": 1.5386176729385742e-05,
      "loss": 0.3306,
      "step": 11082
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.265788337751123,
      "learning_rate": 1.5385340991142738e-05,
      "loss": 0.234,
      "step": 11083
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6849298498414681,
      "learning_rate": 1.538450519991697e-05,
      "loss": 0.4927,
      "step": 11084
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9934157938730608,
      "learning_rate": 1.538366935571667e-05,
      "loss": 0.6345,
      "step": 11085
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.532967281337487,
      "learning_rate": 1.5382833458550056e-05,
      "loss": 0.2868,
      "step": 11086
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2854541878526702,
      "learning_rate": 1.538199750842536e-05,
      "loss": 0.1606,
      "step": 11087
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.26183113495652294,
      "learning_rate": 1.53811615053508e-05,
      "loss": 0.2425,
      "step": 11088
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3948429204496019,
      "learning_rate": 1.5380325449334605e-05,
      "loss": 0.2482,
      "step": 11089
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7039601123374946,
      "learning_rate": 1.5379489340384995e-05,
      "loss": 0.5126,
      "step": 11090
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3536050641666569,
      "learning_rate": 1.53786531785102e-05,
      "loss": 0.2006,
      "step": 11091
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3289790994571376,
      "learning_rate": 1.537781696371844e-05,
      "loss": 0.2713,
      "step": 11092
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.620852350404142,
      "learning_rate": 1.5376980696017954e-05,
      "loss": 0.8952,
      "step": 11093
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2980157741827963,
      "learning_rate": 1.5376144375416963e-05,
      "loss": 0.2385,
      "step": 11094
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.8615485639096054,
      "learning_rate": 1.537530800192369e-05,
      "loss": 0.8374,
      "step": 11095
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2526115395757512,
      "learning_rate": 1.5374471575546373e-05,
      "loss": 0.1906,
      "step": 11096
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33607212344558046,
      "learning_rate": 1.537363509629323e-05,
      "loss": 0.2034,
      "step": 11097
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5185229442007544,
      "learning_rate": 1.53727985641725e-05,
      "loss": 0.2858,
      "step": 11098
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7059182462854178,
      "learning_rate": 1.5371961979192413e-05,
      "loss": 0.4985,
      "step": 11099
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2492465349249182,
      "learning_rate": 1.5371125341361194e-05,
      "loss": 0.208,
      "step": 11100
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3459317403731565,
      "learning_rate": 1.5370288650687073e-05,
      "loss": 0.3241,
      "step": 11101
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7184547771549052,
      "learning_rate": 1.536945190717829e-05,
      "loss": 0.3381,
      "step": 11102
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.223601704620675,
      "learning_rate": 1.5368615110843072e-05,
      "loss": 0.9277,
      "step": 11103
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5454176528668838,
      "learning_rate": 1.536777826168965e-05,
      "loss": 0.2675,
      "step": 11104
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38655161512354214,
      "learning_rate": 1.536694135972626e-05,
      "loss": 0.2158,
      "step": 11105
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.29988757527131904,
      "learning_rate": 1.5366104404961136e-05,
      "loss": 0.1976,
      "step": 11106
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3235885432643391,
      "learning_rate": 1.536526739740251e-05,
      "loss": 0.2525,
      "step": 11107
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6884040917721247,
      "learning_rate": 1.5364430337058615e-05,
      "loss": 0.4465,
      "step": 11108
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35550173476224334,
      "learning_rate": 1.536359322393769e-05,
      "loss": 0.1991,
      "step": 11109
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38223016235344814,
      "learning_rate": 1.5362756058047975e-05,
      "loss": 0.3421,
      "step": 11110
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9899652692261872,
      "learning_rate": 1.5361918839397696e-05,
      "loss": 0.4731,
      "step": 11111
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3532970169806828,
      "learning_rate": 1.53610815679951e-05,
      "loss": 0.3292,
      "step": 11112
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1634135925675668,
      "learning_rate": 1.5360244243848418e-05,
      "loss": 0.3769,
      "step": 11113
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7329952490687506,
      "learning_rate": 1.5359406866965886e-05,
      "loss": 0.3823,
      "step": 11114
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2038666469224132,
      "learning_rate": 1.535856943735575e-05,
      "loss": 0.1471,
      "step": 11115
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7630047607402751,
      "learning_rate": 1.535773195502624e-05,
      "loss": 0.4934,
      "step": 11116
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4481160101191656,
      "learning_rate": 1.5356894419985605e-05,
      "loss": 0.3179,
      "step": 11117
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3616887192413619,
      "learning_rate": 1.5356056832242078e-05,
      "loss": 0.2423,
      "step": 11118
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35083625849605804,
      "learning_rate": 1.5355219191803903e-05,
      "loss": 0.3127,
      "step": 11119
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8460891970080344,
      "learning_rate": 1.5354381498679316e-05,
      "loss": 0.4455,
      "step": 11120
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6371936844174013,
      "learning_rate": 1.5353543752876564e-05,
      "loss": 0.8309,
      "step": 11121
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7500225706865542,
      "learning_rate": 1.5352705954403887e-05,
      "loss": 0.3821,
      "step": 11122
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3790470397072169,
      "learning_rate": 1.535186810326953e-05,
      "loss": 0.2802,
      "step": 11123
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2214168478261545,
      "learning_rate": 1.535103019948173e-05,
      "loss": 0.2137,
      "step": 11124
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.48186400948327834,
      "learning_rate": 1.5350192243048737e-05,
      "loss": 0.304,
      "step": 11125
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6823934720144009,
      "learning_rate": 1.534935423397879e-05,
      "loss": 0.3773,
      "step": 11126
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8803133577950332,
      "learning_rate": 1.5348516172280136e-05,
      "loss": 0.4853,
      "step": 11127
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3159801957701489,
      "learning_rate": 1.534767805796102e-05,
      "loss": 0.2126,
      "step": 11128
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.6107530109827346,
      "learning_rate": 1.534683989102969e-05,
      "loss": 0.7075,
      "step": 11129
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.31433895153502595,
      "learning_rate": 1.5346001671494386e-05,
      "loss": 0.2773,
      "step": 11130
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8110369232038643,
      "learning_rate": 1.5345163399363362e-05,
      "loss": 0.3134,
      "step": 11131
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.34105944310244857,
      "learning_rate": 1.534432507464486e-05,
      "loss": 0.2416,
      "step": 11132
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.39558106919466574,
      "learning_rate": 1.5343486697347126e-05,
      "loss": 0.2204,
      "step": 11133
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.658860837125151,
      "learning_rate": 1.5342648267478418e-05,
      "loss": 0.4906,
      "step": 11134
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23974608720910776,
      "learning_rate": 1.5341809785046975e-05,
      "loss": 0.2231,
      "step": 11135
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3258610405954249,
      "learning_rate": 1.534097125006105e-05,
      "loss": 0.1939,
      "step": 11136
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3440626613567514,
      "learning_rate": 1.5340132662528893e-05,
      "loss": 0.227,
      "step": 11137
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0018496666865242,
      "learning_rate": 1.5339294022458754e-05,
      "loss": 0.6734,
      "step": 11138
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8542665827213072,
      "learning_rate": 1.5338455329858884e-05,
      "loss": 0.5051,
      "step": 11139
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7174249027878629,
      "learning_rate": 1.533761658473753e-05,
      "loss": 0.5075,
      "step": 11140
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30124907682231167,
      "learning_rate": 1.5336777787102953e-05,
      "loss": 0.0771,
      "step": 11141
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3341736374589541,
      "learning_rate": 1.5335938936963398e-05,
      "loss": 0.2893,
      "step": 11142
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43953675565827,
      "learning_rate": 1.5335100034327116e-05,
      "loss": 0.3112,
      "step": 11143
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23677586835279266,
      "learning_rate": 1.533426107920237e-05,
      "loss": 0.0719,
      "step": 11144
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4890127202063928,
      "learning_rate": 1.533342207159741e-05,
      "loss": 0.2779,
      "step": 11145
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2795718260317195,
      "learning_rate": 1.5332583011520485e-05,
      "loss": 0.2003,
      "step": 11146
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3566982502354645,
      "learning_rate": 1.533174389897985e-05,
      "loss": 0.8659,
      "step": 11147
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32566176587733275,
      "learning_rate": 1.533090473398377e-05,
      "loss": 0.2464,
      "step": 11148
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2747620919923033,
      "learning_rate": 1.533006551654049e-05,
      "loss": 0.8047,
      "step": 11149
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4524160926260354,
      "learning_rate": 1.5329226246658278e-05,
      "loss": 0.215,
      "step": 11150
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36047886474745033,
      "learning_rate": 1.5328386924345384e-05,
      "loss": 0.3256,
      "step": 11151
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.639701556977438,
      "learning_rate": 1.532754754961006e-05,
      "loss": 0.3457,
      "step": 11152
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43239505972104186,
      "learning_rate": 1.5326708122460573e-05,
      "loss": 0.3533,
      "step": 11153
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.21219616258528778,
      "learning_rate": 1.532586864290518e-05,
      "loss": 0.1229,
      "step": 11154
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33178495389072216,
      "learning_rate": 1.5325029110952137e-05,
      "loss": 0.2722,
      "step": 11155
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8114721595060966,
      "learning_rate": 1.532418952660971e-05,
      "loss": 0.3219,
      "step": 11156
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8448900382618368,
      "learning_rate": 1.5323349889886147e-05,
      "loss": 0.579,
      "step": 11157
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0110634237934024,
      "learning_rate": 1.5322510200789724e-05,
      "loss": 0.4802,
      "step": 11158
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27523184163820497,
      "learning_rate": 1.5321670459328692e-05,
      "loss": 0.2128,
      "step": 11159
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5260234622697154,
      "learning_rate": 1.532083066551131e-05,
      "loss": 0.3598,
      "step": 11160
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36917617520227786,
      "learning_rate": 1.5319990819345847e-05,
      "loss": 0.3053,
      "step": 11161
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.45758291151195596,
      "learning_rate": 1.5319150920840566e-05,
      "loss": 0.2516,
      "step": 11162
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23839482205587825,
      "learning_rate": 1.5318310970003733e-05,
      "loss": 0.0988,
      "step": 11163
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.454095231123641,
      "learning_rate": 1.53174709668436e-05,
      "loss": 0.4488,
      "step": 11164
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4143715621061932,
      "learning_rate": 1.5316630911368437e-05,
      "loss": 0.2588,
      "step": 11165
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.36934950089611013,
      "learning_rate": 1.5315790803586512e-05,
      "loss": 0.3397,
      "step": 11166
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9971531792335999,
      "learning_rate": 1.5314950643506087e-05,
      "loss": 0.3858,
      "step": 11167
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9170557614904501,
      "learning_rate": 1.531411043113543e-05,
      "loss": 0.5244,
      "step": 11168
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2979887649768526,
      "learning_rate": 1.5313270166482807e-05,
      "loss": 0.2283,
      "step": 11169
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7592815739730819,
      "learning_rate": 1.531242984955648e-05,
      "loss": 0.3284,
      "step": 11170
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.312881955264634,
      "learning_rate": 1.5311589480364725e-05,
      "loss": 0.2808,
      "step": 11171
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.276265586752239,
      "learning_rate": 1.5310749058915803e-05,
      "loss": 0.0985,
      "step": 11172
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.41743178233406897,
      "learning_rate": 1.5309908585217983e-05,
      "loss": 0.2957,
      "step": 11173
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.43709299182521055,
      "learning_rate": 1.5309068059279538e-05,
      "loss": 0.2909,
      "step": 11174
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.5459921903946723,
      "learning_rate": 1.530822748110873e-05,
      "loss": 0.6866,
      "step": 11175
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.061886385273925,
      "learning_rate": 1.5307386850713838e-05,
      "loss": 0.4078,
      "step": 11176
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3915632597230542,
      "learning_rate": 1.5306546168103123e-05,
      "loss": 0.2844,
      "step": 11177
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3296122970588902,
      "learning_rate": 1.5305705433284866e-05,
      "loss": 0.241,
      "step": 11178
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2810503982201755,
      "learning_rate": 1.530486464626733e-05,
      "loss": 0.7349,
      "step": 11179
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0055820021219346,
      "learning_rate": 1.530402380705879e-05,
      "loss": 0.0508,
      "step": 11180
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5040081720629109,
      "learning_rate": 1.530318291566752e-05,
      "loss": 0.362,
      "step": 11181
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33474759349773,
      "learning_rate": 1.530234197210179e-05,
      "loss": 0.188,
      "step": 11182
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46288905103499356,
      "learning_rate": 1.5301500976369873e-05,
      "loss": 0.2614,
      "step": 11183
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3571620197079316,
      "learning_rate": 1.530065992848005e-05,
      "loss": 0.3075,
      "step": 11184
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7727303435059746,
      "learning_rate": 1.5299818828440586e-05,
      "loss": 0.385,
      "step": 11185
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32988397562414096,
      "learning_rate": 1.5298977676259762e-05,
      "loss": 0.2151,
      "step": 11186
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3511843662248823,
      "learning_rate": 1.5298136471945854e-05,
      "loss": 0.2221,
      "step": 11187
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0021158575130316,
      "learning_rate": 1.5297295215507137e-05,
      "loss": 0.6387,
      "step": 11188
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30670439645442915,
      "learning_rate": 1.529645390695188e-05,
      "loss": 0.2544,
      "step": 11189
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.46700820891104045,
      "learning_rate": 1.5295612546288376e-05,
      "loss": 0.3017,
      "step": 11190
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.22141232658407994,
      "learning_rate": 1.5294771133524887e-05,
      "loss": 0.0735,
      "step": 11191
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.38465510274549003,
      "learning_rate": 1.5293929668669698e-05,
      "loss": 0.3303,
      "step": 11192
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5968715142028053,
      "learning_rate": 1.5293088151731087e-05,
      "loss": 0.3666,
      "step": 11193
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4008645661716332,
      "learning_rate": 1.5292246582717334e-05,
      "loss": 0.3654,
      "step": 11194
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.255880037549198,
      "learning_rate": 1.529140496163672e-05,
      "loss": 0.1722,
      "step": 11195
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2958960667418434,
      "learning_rate": 1.529056328849752e-05,
      "loss": 0.1848,
      "step": 11196
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.328447320587647,
      "learning_rate": 1.5289721563308017e-05,
      "loss": 0.7796,
      "step": 11197
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8447863997150623,
      "learning_rate": 1.5288879786076493e-05,
      "loss": 0.3506,
      "step": 11198
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.1466794840148595,
      "learning_rate": 1.528803795681123e-05,
      "loss": 0.707,
      "step": 11199
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.33567532367945274,
      "learning_rate": 1.5287196075520513e-05,
      "loss": 0.1891,
      "step": 11200
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3447056749813004,
      "learning_rate": 1.5286354142212618e-05,
      "loss": 0.3236,
      "step": 11201
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3831435440049565,
      "learning_rate": 1.528551215689583e-05,
      "loss": 0.3054,
      "step": 11202
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.672208099783269,
      "learning_rate": 1.5284670119578434e-05,
      "loss": 0.4987,
      "step": 11203
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.23844896245173736,
      "learning_rate": 1.5283828030268716e-05,
      "loss": 0.086,
      "step": 11204
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35062136447211817,
      "learning_rate": 1.528298588897496e-05,
      "loss": 0.3195,
      "step": 11205
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3054166177755601,
      "learning_rate": 1.528214369570545e-05,
      "loss": 0.015,
      "step": 11206
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44876086925271613,
      "learning_rate": 1.5281301450468474e-05,
      "loss": 0.3825,
      "step": 11207
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3683064513621807,
      "learning_rate": 1.5280459153272314e-05,
      "loss": 0.2503,
      "step": 11208
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30227039879345674,
      "learning_rate": 1.527961680412526e-05,
      "loss": 0.184,
      "step": 11209
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5159147290797791,
      "learning_rate": 1.5278774403035595e-05,
      "loss": 0.3808,
      "step": 11210
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6677193372840885,
      "learning_rate": 1.5277931950011614e-05,
      "loss": 0.3612,
      "step": 11211
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.44410989192621175,
      "learning_rate": 1.5277089445061603e-05,
      "loss": 0.397,
      "step": 11212
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16831085852137456,
      "learning_rate": 1.527624688819385e-05,
      "loss": 0.1253,
      "step": 11213
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9277940332378872,
      "learning_rate": 1.5275404279416644e-05,
      "loss": 0.413,
      "step": 11214
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3168353660755869,
      "learning_rate": 1.5274561618738274e-05,
      "loss": 0.2317,
      "step": 11215
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.9240377083150395,
      "learning_rate": 1.527371890616703e-05,
      "loss": 0.8337,
      "step": 11216
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.5548922071121688,
      "learning_rate": 1.5272876141711206e-05,
      "loss": 0.3792,
      "step": 11217
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0048492448908934,
      "learning_rate": 1.5272033325379086e-05,
      "loss": 0.4528,
      "step": 11218
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2623211325955454,
      "learning_rate": 1.5271190457178976e-05,
      "loss": 0.2305,
      "step": 11219
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4158426546110854,
      "learning_rate": 1.5270347537119157e-05,
      "loss": 0.3908,
      "step": 11220
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7637256518082158,
      "learning_rate": 1.5269504565207922e-05,
      "loss": 0.5043,
      "step": 11221
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.16548076048348312,
      "learning_rate": 1.5268661541453565e-05,
      "loss": 0.0713,
      "step": 11222
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.40822952434866583,
      "learning_rate": 1.526781846586439e-05,
      "loss": 0.2928,
      "step": 11223
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3910179308207988,
      "learning_rate": 1.5266975338448677e-05,
      "loss": 0.2561,
      "step": 11224
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3543702979080434,
      "learning_rate": 1.526613215921473e-05,
      "loss": 0.3171,
      "step": 11225
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.3708048702120164,
      "learning_rate": 1.526528892817084e-05,
      "loss": 0.0836,
      "step": 11226
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.55182740604011,
      "learning_rate": 1.526444564532531e-05,
      "loss": 0.4193,
      "step": 11227
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2834258357362601,
      "learning_rate": 1.5263602310686426e-05,
      "loss": 0.2406,
      "step": 11228
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9843411659565412,
      "learning_rate": 1.5262758924262495e-05,
      "loss": 0.6076,
      "step": 11229
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8802445921969542,
      "learning_rate": 1.5261915486061808e-05,
      "loss": 0.511,
      "step": 11230
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3376680693461168,
      "learning_rate": 1.5261071996092663e-05,
      "loss": 0.3271,
      "step": 11231
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.27633800878883835,
      "learning_rate": 1.5260228454363364e-05,
      "loss": 0.1993,
      "step": 11232
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.24622058029840346,
      "learning_rate": 1.5259384860882205e-05,
      "loss": 0.1465,
      "step": 11233
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.075737117750154,
      "learning_rate": 1.5258541215657487e-05,
      "loss": 0.2904,
      "step": 11234
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6511829569831076,
      "learning_rate": 1.525769751869751e-05,
      "loss": 0.3729,
      "step": 11235
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3687053062500887,
      "learning_rate": 1.5256853770010577e-05,
      "loss": 0.282,
      "step": 11236
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.35544532519026917,
      "learning_rate": 1.5256009969604983e-05,
      "loss": 0.2349,
      "step": 11237
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4285493357654094,
      "learning_rate": 1.5255166117489038e-05,
      "loss": 0.3622,
      "step": 11238
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8535683145522202,
      "learning_rate": 1.5254322213671038e-05,
      "loss": 0.628,
      "step": 11239
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4050452892765467,
      "learning_rate": 1.5253478258159287e-05,
      "loss": 0.158,
      "step": 11240
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.30703545660970266,
      "learning_rate": 1.5252634250962084e-05,
      "loss": 0.0719,
      "step": 11241
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.4102320621557596,
      "learning_rate": 1.5251790192087743e-05,
      "loss": 0.3569,
      "step": 11242
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3411253383938233,
      "learning_rate": 1.525094608154456e-05,
      "loss": 0.2517,
      "step": 11243
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8918048661836265,
      "learning_rate": 1.5250101919340839e-05,
      "loss": 0.4904,
      "step": 11244
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6590102447224273,
      "learning_rate": 1.5249257705484893e-05,
      "loss": 0.2515,
      "step": 11245
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3566390362755947,
      "learning_rate": 1.5248413439985019e-05,
      "loss": 0.3094,
      "step": 11246
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.8799863127863439,
      "learning_rate": 1.5247569122849527e-05,
      "loss": 0.4508,
      "step": 11247
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3292944076841432,
      "learning_rate": 1.524672475408672e-05,
      "loss": 0.2806,
      "step": 11248
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.2201908886800383,
      "learning_rate": 1.5245880333704914e-05,
      "loss": 0.7417,
      "step": 11249
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3632560949743605,
      "learning_rate": 1.5245035861712409e-05,
      "loss": 0.1896,
      "step": 11250
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2727715225057192,
      "learning_rate": 1.5244191338117515e-05,
      "loss": 0.1829,
      "step": 11251
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.9968105727384617,
      "learning_rate": 1.5243346762928543e-05,
      "loss": 0.4465,
      "step": 11252
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.0195537669600736,
      "learning_rate": 1.5242502136153798e-05,
      "loss": 0.5359,
      "step": 11253
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2633358891203423,
      "learning_rate": 1.5241657457801593e-05,
      "loss": 0.2573,
      "step": 11254
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.37084846478070327,
      "learning_rate": 1.5240812727880236e-05,
      "loss": 0.3116,
      "step": 11255
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.6612406534244111,
      "learning_rate": 1.5239967946398041e-05,
      "loss": 0.4452,
      "step": 11256
    },
    {
      "epoch": 0.34,
      "grad_norm": 1.4714453542013648,
      "learning_rate": 1.5239123113363316e-05,
      "loss": 0.792,
      "step": 11257
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.2534287300933058,
      "learning_rate": 1.5238278228784374e-05,
      "loss": 0.1411,
      "step": 11258
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.350556466923105,
      "learning_rate": 1.5237433292669527e-05,
      "loss": 0.2371,
      "step": 11259
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.32565722026131916,
      "learning_rate": 1.5236588305027087e-05,
      "loss": 0.2232,
      "step": 11260
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3228977445091139,
      "learning_rate": 1.523574326586537e-05,
      "loss": 0.3209,
      "step": 11261
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.7489937413852058,
      "learning_rate": 1.5234898175192687e-05,
      "loss": 0.4302,
      "step": 11262
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.25883878332945015,
      "learning_rate": 1.5234053033017353e-05,
      "loss": 0.0692,
      "step": 11263
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.3517147723030113,
      "learning_rate": 1.5233207839347683e-05,
      "loss": 0.3166,
      "step": 11264
    },
    {
      "epoch": 0.34,
      "grad_norm": 0.703658526777675,
      "learning_rate": 1.5232362594191994e-05,
      "loss": 0.4501,
      "step": 11265
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4854034784682278,
      "learning_rate": 1.5231517297558598e-05,
      "loss": 0.3942,
      "step": 11266
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4200681330158501,
      "learning_rate": 1.5230671949455816e-05,
      "loss": 0.2413,
      "step": 11267
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5973915021730575,
      "learning_rate": 1.5229826549891962e-05,
      "loss": 0.3188,
      "step": 11268
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5186894640883265,
      "learning_rate": 1.5228981098875348e-05,
      "loss": 0.1824,
      "step": 11269
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5188884749128646,
      "learning_rate": 1.5228135596414304e-05,
      "loss": 0.461,
      "step": 11270
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.355199009834097,
      "learning_rate": 1.5227290042517137e-05,
      "loss": 0.2129,
      "step": 11271
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46501207870013267,
      "learning_rate": 1.5226444437192175e-05,
      "loss": 0.3213,
      "step": 11272
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2926320711834437,
      "learning_rate": 1.522559878044773e-05,
      "loss": 0.2261,
      "step": 11273
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5361217832672528,
      "learning_rate": 1.5224753072292124e-05,
      "loss": 0.3667,
      "step": 11274
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0039630021817976,
      "learning_rate": 1.522390731273368e-05,
      "loss": 0.5914,
      "step": 11275
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3556426764574292,
      "learning_rate": 1.5223061501780719e-05,
      "loss": 0.1057,
      "step": 11276
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5675952291512177,
      "learning_rate": 1.5222215639441556e-05,
      "loss": 0.3436,
      "step": 11277
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2715459354266918,
      "learning_rate": 1.5221369725724522e-05,
      "loss": 0.2741,
      "step": 11278
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4561589771354364,
      "learning_rate": 1.522052376063793e-05,
      "loss": 0.3891,
      "step": 11279
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25843021740145194,
      "learning_rate": 1.521967774419011e-05,
      "loss": 0.175,
      "step": 11280
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4417744040490464,
      "learning_rate": 1.521883167638938e-05,
      "loss": 0.3143,
      "step": 11281
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.28483493933756465,
      "learning_rate": 1.5217985557244068e-05,
      "loss": 0.1978,
      "step": 11282
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3192743397704203,
      "learning_rate": 1.5217139386762498e-05,
      "loss": 0.7869,
      "step": 11283
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39523886863658175,
      "learning_rate": 1.5216293164952993e-05,
      "loss": 0.2695,
      "step": 11284
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.45356211815761704,
      "learning_rate": 1.5215446891823881e-05,
      "loss": 0.3942,
      "step": 11285
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4303372278424846,
      "learning_rate": 1.5214600567383487e-05,
      "loss": 0.2077,
      "step": 11286
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3224073128237562,
      "learning_rate": 1.5213754191640133e-05,
      "loss": 0.3189,
      "step": 11287
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7433679857454681,
      "learning_rate": 1.5212907764602154e-05,
      "loss": 0.3406,
      "step": 11288
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18663784885956297,
      "learning_rate": 1.5212061286277873e-05,
      "loss": 0.0921,
      "step": 11289
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3307261207505583,
      "learning_rate": 1.5211214756675614e-05,
      "loss": 0.3097,
      "step": 11290
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2963046402635058,
      "learning_rate": 1.5210368175803716e-05,
      "loss": 0.1689,
      "step": 11291
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5370535313948701,
      "learning_rate": 1.5209521543670494e-05,
      "loss": 0.4065,
      "step": 11292
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.001342300020846,
      "learning_rate": 1.5208674860284288e-05,
      "loss": 0.5863,
      "step": 11293
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6560292229175239,
      "learning_rate": 1.5207828125653428e-05,
      "loss": 0.5483,
      "step": 11294
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5867120863064202,
      "learning_rate": 1.5206981339786237e-05,
      "loss": 0.2525,
      "step": 11295
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35370106053913253,
      "learning_rate": 1.5206134502691052e-05,
      "loss": 0.2806,
      "step": 11296
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3365604305439267,
      "learning_rate": 1.5205287614376201e-05,
      "loss": 0.276,
      "step": 11297
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4622422464994792,
      "learning_rate": 1.520444067485002e-05,
      "loss": 0.2898,
      "step": 11298
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2669951255260245,
      "learning_rate": 1.5203593684120836e-05,
      "loss": 0.1145,
      "step": 11299
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6272282764689208,
      "learning_rate": 1.5202746642196989e-05,
      "loss": 0.3815,
      "step": 11300
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3085970742032315,
      "learning_rate": 1.5201899549086807e-05,
      "loss": 0.2536,
      "step": 11301
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2798061353058983,
      "learning_rate": 1.5201052404798626e-05,
      "loss": 0.2445,
      "step": 11302
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2879718963437616,
      "learning_rate": 1.5200205209340779e-05,
      "loss": 0.6751,
      "step": 11303
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6906891630532543,
      "learning_rate": 1.5199357962721603e-05,
      "loss": 0.3118,
      "step": 11304
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.300584602257619,
      "learning_rate": 1.519851066494943e-05,
      "loss": 0.2364,
      "step": 11305
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7310633087465055,
      "learning_rate": 1.5197663316032603e-05,
      "loss": 0.365,
      "step": 11306
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4143860319957576,
      "learning_rate": 1.5196815915979453e-05,
      "loss": 0.2154,
      "step": 11307
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25851126963106325,
      "learning_rate": 1.5195968464798316e-05,
      "loss": 0.2055,
      "step": 11308
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8990784898963748,
      "learning_rate": 1.5195120962497532e-05,
      "loss": 0.3059,
      "step": 11309
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4757509948758816,
      "learning_rate": 1.5194273409085438e-05,
      "loss": 0.3119,
      "step": 11310
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.818899005297548,
      "learning_rate": 1.5193425804570377e-05,
      "loss": 0.8409,
      "step": 11311
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7684193525205627,
      "learning_rate": 1.5192578148960677e-05,
      "loss": 0.3714,
      "step": 11312
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6666173136799222,
      "learning_rate": 1.519173044226469e-05,
      "loss": 0.4304,
      "step": 11313
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2649159993080999,
      "learning_rate": 1.5190882684490751e-05,
      "loss": 0.2227,
      "step": 11314
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3797444673514681,
      "learning_rate": 1.5190034875647197e-05,
      "loss": 0.2522,
      "step": 11315
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9446326549889964,
      "learning_rate": 1.5189187015742373e-05,
      "loss": 0.652,
      "step": 11316
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21852553685541157,
      "learning_rate": 1.5188339104784621e-05,
      "loss": 0.1077,
      "step": 11317
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39937551895236706,
      "learning_rate": 1.518749114278228e-05,
      "loss": 0.2851,
      "step": 11318
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46429251993472165,
      "learning_rate": 1.5186643129743695e-05,
      "loss": 0.2414,
      "step": 11319
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3515578247320622,
      "learning_rate": 1.518579506567721e-05,
      "loss": 0.3323,
      "step": 11320
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0576801742357673,
      "learning_rate": 1.5184946950591167e-05,
      "loss": 0.4252,
      "step": 11321
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.872022289264479,
      "learning_rate": 1.5184098784493906e-05,
      "loss": 0.778,
      "step": 11322
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.31735705658791175,
      "learning_rate": 1.5183250567393777e-05,
      "loss": 0.1658,
      "step": 11323
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.50935301932745,
      "learning_rate": 1.5182402299299123e-05,
      "loss": 0.3813,
      "step": 11324
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38268000415675046,
      "learning_rate": 1.518155398021829e-05,
      "loss": 0.2624,
      "step": 11325
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.30008149923738775,
      "learning_rate": 1.5180705610159625e-05,
      "loss": 0.2153,
      "step": 11326
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25563870285581414,
      "learning_rate": 1.5179857189131475e-05,
      "loss": 0.1231,
      "step": 11327
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.338854413588033,
      "learning_rate": 1.5179008717142183e-05,
      "loss": 0.259,
      "step": 11328
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6803709703704658,
      "learning_rate": 1.51781601942001e-05,
      "loss": 0.5199,
      "step": 11329
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6313608221737207,
      "learning_rate": 1.5177311620313572e-05,
      "loss": 0.3647,
      "step": 11330
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.28590287126663205,
      "learning_rate": 1.517646299549095e-05,
      "loss": 0.2872,
      "step": 11331
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2658234676679993,
      "learning_rate": 1.5175614319740581e-05,
      "loss": 0.1888,
      "step": 11332
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4348200741032543,
      "learning_rate": 1.5174765593070818e-05,
      "loss": 0.3597,
      "step": 11333
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38894005133736065,
      "learning_rate": 1.5173916815490007e-05,
      "loss": 0.0262,
      "step": 11334
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4265601264051104,
      "learning_rate": 1.5173067987006497e-05,
      "loss": 0.2406,
      "step": 11335
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36433067082691,
      "learning_rate": 1.5172219107628645e-05,
      "loss": 0.1669,
      "step": 11336
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37911964368365425,
      "learning_rate": 1.5171370177364799e-05,
      "loss": 0.3415,
      "step": 11337
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33156858154005014,
      "learning_rate": 1.517052119622331e-05,
      "loss": 0.2455,
      "step": 11338
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7021698018955852,
      "learning_rate": 1.5169672164212534e-05,
      "loss": 0.4899,
      "step": 11339
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.17936202607756088,
      "learning_rate": 1.5168823081340822e-05,
      "loss": 0.0979,
      "step": 11340
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3251682132517678,
      "learning_rate": 1.5167973947616528e-05,
      "loss": 0.2156,
      "step": 11341
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6575399186319972,
      "learning_rate": 1.5167124763048002e-05,
      "loss": 0.9055,
      "step": 11342
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36338558256740344,
      "learning_rate": 1.5166275527643608e-05,
      "loss": 0.2221,
      "step": 11343
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4693300670846654,
      "learning_rate": 1.5165426241411693e-05,
      "loss": 0.3957,
      "step": 11344
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3168378993683423,
      "learning_rate": 1.5164576904360615e-05,
      "loss": 0.1216,
      "step": 11345
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37529840535812686,
      "learning_rate": 1.516372751649873e-05,
      "loss": 0.3281,
      "step": 11346
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6292576353855933,
      "learning_rate": 1.5162878077834396e-05,
      "loss": 0.38,
      "step": 11347
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4676802074214971,
      "learning_rate": 1.5162028588375965e-05,
      "loss": 0.3267,
      "step": 11348
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2520752984624377,
      "learning_rate": 1.51611790481318e-05,
      "loss": 0.2005,
      "step": 11349
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6423655825338996,
      "learning_rate": 1.5160329457110258e-05,
      "loss": 0.3499,
      "step": 11350
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34078288331393414,
      "learning_rate": 1.5159479815319697e-05,
      "loss": 0.2379,
      "step": 11351
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9479574568537238,
      "learning_rate": 1.5158630122768476e-05,
      "loss": 0.5455,
      "step": 11352
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9235391870113626,
      "learning_rate": 1.5157780379464954e-05,
      "loss": 0.5036,
      "step": 11353
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.533799709914852,
      "learning_rate": 1.5156930585417492e-05,
      "loss": 0.2888,
      "step": 11354
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36839417626910415,
      "learning_rate": 1.515608074063445e-05,
      "loss": 0.3016,
      "step": 11355
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2923628769613283,
      "learning_rate": 1.515523084512419e-05,
      "loss": 0.2903,
      "step": 11356
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3959119298132578,
      "learning_rate": 1.5154380898895067e-05,
      "loss": 0.9679,
      "step": 11357
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.20761969755165174,
      "learning_rate": 1.5153530901955455e-05,
      "loss": 0.0726,
      "step": 11358
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43835136439510125,
      "learning_rate": 1.5152680854313709e-05,
      "loss": 0.3092,
      "step": 11359
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4200112815211736,
      "learning_rate": 1.5151830755978191e-05,
      "loss": 0.2449,
      "step": 11360
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.245725265598771,
      "learning_rate": 1.5150980606957269e-05,
      "loss": 0.6566,
      "step": 11361
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.306483737407896,
      "learning_rate": 1.5150130407259302e-05,
      "loss": 0.2505,
      "step": 11362
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5819974931231662,
      "learning_rate": 1.5149280156892663e-05,
      "loss": 0.4268,
      "step": 11363
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3014282064078046,
      "learning_rate": 1.5148429855865705e-05,
      "loss": 0.2507,
      "step": 11364
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9598102480514668,
      "learning_rate": 1.5147579504186805e-05,
      "loss": 0.5122,
      "step": 11365
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8296196088330094,
      "learning_rate": 1.5146729101864324e-05,
      "loss": 0.5037,
      "step": 11366
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.19768584056180674,
      "learning_rate": 1.5145878648906625e-05,
      "loss": 0.1736,
      "step": 11367
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37283524718614075,
      "learning_rate": 1.5145028145322078e-05,
      "loss": 0.2804,
      "step": 11368
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3989487127447089,
      "learning_rate": 1.5144177591119058e-05,
      "loss": 0.2027,
      "step": 11369
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5819819144300713,
      "learning_rate": 1.514332698630592e-05,
      "loss": 0.7654,
      "step": 11370
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8421614348877794,
      "learning_rate": 1.5142476330891041e-05,
      "loss": 0.3921,
      "step": 11371
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5263359334246548,
      "learning_rate": 1.5141625624882786e-05,
      "loss": 0.3593,
      "step": 11372
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3358326839399195,
      "learning_rate": 1.514077486828953e-05,
      "loss": 0.2256,
      "step": 11373
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36012259892078297,
      "learning_rate": 1.5139924061119635e-05,
      "loss": 0.3358,
      "step": 11374
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7571482794785888,
      "learning_rate": 1.5139073203381477e-05,
      "loss": 0.5603,
      "step": 11375
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29998614192680556,
      "learning_rate": 1.513822229508343e-05,
      "loss": 0.1937,
      "step": 11376
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2582918996429975,
      "learning_rate": 1.5137371336233858e-05,
      "loss": 0.071,
      "step": 11377
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41725807471025783,
      "learning_rate": 1.5136520326841133e-05,
      "loss": 0.3457,
      "step": 11378
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.28603135492076076,
      "learning_rate": 1.5135669266913638e-05,
      "loss": 0.2455,
      "step": 11379
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0613471199616513,
      "learning_rate": 1.5134818156459732e-05,
      "loss": 0.4091,
      "step": 11380
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8335776422375557,
      "learning_rate": 1.5133966995487797e-05,
      "loss": 0.4776,
      "step": 11381
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2681918712232077,
      "learning_rate": 1.5133115784006207e-05,
      "loss": 0.1899,
      "step": 11382
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3801057424824754,
      "learning_rate": 1.5132264522023335e-05,
      "loss": 0.9679,
      "step": 11383
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0193487071567535,
      "learning_rate": 1.5131413209547558e-05,
      "loss": 0.3806,
      "step": 11384
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35051096774653373,
      "learning_rate": 1.5130561846587246e-05,
      "loss": 0.3481,
      "step": 11385
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23701509347026986,
      "learning_rate": 1.5129710433150781e-05,
      "loss": 0.1439,
      "step": 11386
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3271419350360037,
      "learning_rate": 1.5128858969246534e-05,
      "loss": 0.2426,
      "step": 11387
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1396862177321059,
      "learning_rate": 1.5128007454882885e-05,
      "loss": 0.0708,
      "step": 11388
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8282663416689952,
      "learning_rate": 1.5127155890068213e-05,
      "loss": 0.5159,
      "step": 11389
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6128771926126216,
      "learning_rate": 1.5126304274810894e-05,
      "loss": 0.3097,
      "step": 11390
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3308301395559579,
      "learning_rate": 1.5125452609119307e-05,
      "loss": 0.2771,
      "step": 11391
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.31135508371710363,
      "learning_rate": 1.5124600893001831e-05,
      "loss": 0.2628,
      "step": 11392
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9703301008625349,
      "learning_rate": 1.5123749126466845e-05,
      "loss": 0.5749,
      "step": 11393
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6617885065507849,
      "learning_rate": 1.5122897309522726e-05,
      "loss": 0.2412,
      "step": 11394
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.239032692452651,
      "learning_rate": 1.5122045442177861e-05,
      "loss": 0.1319,
      "step": 11395
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5211262949278387,
      "learning_rate": 1.5121193524440627e-05,
      "loss": 0.3794,
      "step": 11396
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34318081324614474,
      "learning_rate": 1.5120341556319407e-05,
      "loss": 0.3245,
      "step": 11397
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3691745921191043,
      "learning_rate": 1.511948953782258e-05,
      "loss": 0.3782,
      "step": 11398
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4802862149375176,
      "learning_rate": 1.5118637468958532e-05,
      "loss": 0.2702,
      "step": 11399
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.978851468840713,
      "learning_rate": 1.5117785349735643e-05,
      "loss": 0.4812,
      "step": 11400
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3088220073595609,
      "learning_rate": 1.5116933180162296e-05,
      "loss": 0.2579,
      "step": 11401
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1391046221676677,
      "learning_rate": 1.511608096024688e-05,
      "loss": 0.6673,
      "step": 11402
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21434777946035843,
      "learning_rate": 1.5115228689997776e-05,
      "loss": 0.2039,
      "step": 11403
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8938675587287349,
      "learning_rate": 1.5114376369423367e-05,
      "loss": 0.4579,
      "step": 11404
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2319658655294548,
      "learning_rate": 1.5113523998532042e-05,
      "loss": 0.1716,
      "step": 11405
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.509491733966934,
      "learning_rate": 1.5112671577332182e-05,
      "loss": 0.3856,
      "step": 11406
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8964097841966485,
      "learning_rate": 1.5111819105832181e-05,
      "loss": 0.5658,
      "step": 11407
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3886291375109402,
      "learning_rate": 1.5110966584040424e-05,
      "loss": 0.2591,
      "step": 11408
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.40193551690918,
      "learning_rate": 1.5110114011965289e-05,
      "loss": 0.2981,
      "step": 11409
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3227567592904972,
      "learning_rate": 1.5109261389615175e-05,
      "loss": 0.2637,
      "step": 11410
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5639471211563456,
      "learning_rate": 1.5108408716998465e-05,
      "loss": 0.793,
      "step": 11411
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8433938546519811,
      "learning_rate": 1.510755599412355e-05,
      "loss": 0.3635,
      "step": 11412
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3562522629166875,
      "learning_rate": 1.5106703220998818e-05,
      "loss": 0.2475,
      "step": 11413
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3005110718215195,
      "learning_rate": 1.5105850397632658e-05,
      "loss": 0.2355,
      "step": 11414
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4113320669585691,
      "learning_rate": 1.5104997524033462e-05,
      "loss": 0.3631,
      "step": 11415
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.31817631865939,
      "learning_rate": 1.5104144600209622e-05,
      "loss": 0.2176,
      "step": 11416
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4388829677513604,
      "learning_rate": 1.510329162616953e-05,
      "loss": 0.8567,
      "step": 11417
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24904833835862897,
      "learning_rate": 1.5102438601921572e-05,
      "loss": 0.1899,
      "step": 11418
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9294398477212766,
      "learning_rate": 1.5101585527474147e-05,
      "loss": 0.4861,
      "step": 11419
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1033747316869609,
      "learning_rate": 1.5100732402835641e-05,
      "loss": 0.6629,
      "step": 11420
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3120692624816601,
      "learning_rate": 1.5099879228014456e-05,
      "loss": 0.2453,
      "step": 11421
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.46713977151338343,
      "learning_rate": 1.509902600301898e-05,
      "loss": 0.3529,
      "step": 11422
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37790106533806433,
      "learning_rate": 1.5098172727857608e-05,
      "loss": 0.2275,
      "step": 11423
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5791139950930456,
      "learning_rate": 1.5097319402538731e-05,
      "loss": 0.4533,
      "step": 11424
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.16113298133116843,
      "learning_rate": 1.5096466027070753e-05,
      "loss": 0.0699,
      "step": 11425
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36210193764150894,
      "learning_rate": 1.5095612601462063e-05,
      "loss": 0.3279,
      "step": 11426
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.19485732181463356,
      "learning_rate": 1.5094759125721061e-05,
      "loss": 0.0746,
      "step": 11427
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3851084755456622,
      "learning_rate": 1.5093905599856145e-05,
      "loss": 0.3794,
      "step": 11428
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9139983734235367,
      "learning_rate": 1.5093052023875706e-05,
      "loss": 0.4554,
      "step": 11429
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9487993575542777,
      "learning_rate": 1.5092198397788145e-05,
      "loss": 0.5557,
      "step": 11430
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8053820911954508,
      "learning_rate": 1.5091344721601862e-05,
      "loss": 0.3598,
      "step": 11431
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2796272965642652,
      "learning_rate": 1.5090490995325252e-05,
      "loss": 0.1945,
      "step": 11432
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3266073434269846,
      "learning_rate": 1.508963721896672e-05,
      "loss": 0.3168,
      "step": 11433
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2139586017334762,
      "learning_rate": 1.508878339253466e-05,
      "loss": 0.095,
      "step": 11434
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.1751376475316422,
      "learning_rate": 1.5087929516037477e-05,
      "loss": 0.7062,
      "step": 11435
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.30527442273180244,
      "learning_rate": 1.5087075589483567e-05,
      "loss": 0.195,
      "step": 11436
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5101562782737298,
      "learning_rate": 1.5086221612881332e-05,
      "loss": 0.4463,
      "step": 11437
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4098714217105897,
      "learning_rate": 1.508536758623918e-05,
      "loss": 0.2695,
      "step": 11438
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4515382076074891,
      "learning_rate": 1.5084513509565504e-05,
      "loss": 0.3112,
      "step": 11439
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6038745187330925,
      "learning_rate": 1.5083659382868713e-05,
      "loss": 0.3292,
      "step": 11440
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.32858344899591757,
      "learning_rate": 1.5082805206157209e-05,
      "loss": 0.2643,
      "step": 11441
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7007270565702548,
      "learning_rate": 1.5081950979439395e-05,
      "loss": 0.3348,
      "step": 11442
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.43909435493975923,
      "learning_rate": 1.5081096702723672e-05,
      "loss": 0.2302,
      "step": 11443
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.22850714480121648,
      "learning_rate": 1.508024237601845e-05,
      "loss": 0.2011,
      "step": 11444
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.336820830649209,
      "learning_rate": 1.5079387999332132e-05,
      "loss": 0.1986,
      "step": 11445
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5146462789964349,
      "learning_rate": 1.5078533572673123e-05,
      "loss": 0.3538,
      "step": 11446
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8531527613961399,
      "learning_rate": 1.5077679096049828e-05,
      "loss": 0.3462,
      "step": 11447
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7098800201825805,
      "learning_rate": 1.507682456947066e-05,
      "loss": 0.4704,
      "step": 11448
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37627152880349224,
      "learning_rate": 1.5075969992944018e-05,
      "loss": 0.2339,
      "step": 11449
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5085943368145482,
      "learning_rate": 1.5075115366478312e-05,
      "loss": 0.3603,
      "step": 11450
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.26527449263381564,
      "learning_rate": 1.5074260690081955e-05,
      "loss": 0.2501,
      "step": 11451
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42818342262256975,
      "learning_rate": 1.5073405963763348e-05,
      "loss": 0.2543,
      "step": 11452
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24308735858948263,
      "learning_rate": 1.5072551187530904e-05,
      "loss": 0.1581,
      "step": 11453
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9532537461840871,
      "learning_rate": 1.5071696361393033e-05,
      "loss": 0.5043,
      "step": 11454
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2928829411448005,
      "learning_rate": 1.5070841485358146e-05,
      "loss": 0.2255,
      "step": 11455
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8169206941655854,
      "learning_rate": 1.5069986559434648e-05,
      "loss": 0.487,
      "step": 11456
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3133597072494697,
      "learning_rate": 1.506913158363096e-05,
      "loss": 0.2791,
      "step": 11457
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.0328229219246183,
      "learning_rate": 1.5068276557955482e-05,
      "loss": 0.533,
      "step": 11458
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.32359646316859114,
      "learning_rate": 1.5067421482416636e-05,
      "loss": 0.1825,
      "step": 11459
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3998511049148779,
      "learning_rate": 1.5066566357022825e-05,
      "loss": 0.2623,
      "step": 11460
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4282347213833791,
      "learning_rate": 1.5065711181782471e-05,
      "loss": 0.3062,
      "step": 11461
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29533905628379187,
      "learning_rate": 1.506485595670398e-05,
      "loss": 0.2512,
      "step": 11462
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8493283979736683,
      "learning_rate": 1.5064000681795774e-05,
      "loss": 0.5429,
      "step": 11463
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3063518121423555,
      "learning_rate": 1.506314535706626e-05,
      "loss": 0.2205,
      "step": 11464
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5180591611508157,
      "learning_rate": 1.506228998252386e-05,
      "loss": 0.31,
      "step": 11465
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8904654792187405,
      "learning_rate": 1.5061434558176981e-05,
      "loss": 0.3853,
      "step": 11466
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6624065490518005,
      "learning_rate": 1.5060579084034044e-05,
      "loss": 0.7892,
      "step": 11467
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2659057219431571,
      "learning_rate": 1.5059723560103466e-05,
      "loss": 0.1944,
      "step": 11468
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.31424439823660266,
      "learning_rate": 1.505886798639366e-05,
      "loss": 0.3269,
      "step": 11469
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8335650438410891,
      "learning_rate": 1.505801236291305e-05,
      "loss": 0.2952,
      "step": 11470
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4390331614818794,
      "learning_rate": 1.505715668967005e-05,
      "loss": 0.3092,
      "step": 11471
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2727626317277736,
      "learning_rate": 1.5056300966673077e-05,
      "loss": 0.1659,
      "step": 11472
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3332182834645049,
      "learning_rate": 1.505544519393055e-05,
      "loss": 0.2348,
      "step": 11473
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8963581169896816,
      "learning_rate": 1.5054589371450891e-05,
      "loss": 0.4843,
      "step": 11474
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23755062437157284,
      "learning_rate": 1.505373349924252e-05,
      "loss": 0.2291,
      "step": 11475
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.6495466452902434,
      "learning_rate": 1.5052877577313853e-05,
      "loss": 0.7619,
      "step": 11476
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.22070289368822177,
      "learning_rate": 1.5052021605673318e-05,
      "loss": 0.0712,
      "step": 11477
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3786963515225539,
      "learning_rate": 1.5051165584329328e-05,
      "loss": 0.3071,
      "step": 11478
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9371896764582549,
      "learning_rate": 1.505030951329031e-05,
      "loss": 0.3995,
      "step": 11479
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3179259278003667,
      "learning_rate": 1.5049453392564686e-05,
      "loss": 0.315,
      "step": 11480
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4295791927876008,
      "learning_rate": 1.5048597222160877e-05,
      "loss": 0.1555,
      "step": 11481
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3320304174743044,
      "learning_rate": 1.5047741002087306e-05,
      "loss": 0.2887,
      "step": 11482
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4841427669765246,
      "learning_rate": 1.5046884732352402e-05,
      "loss": 0.3889,
      "step": 11483
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33040146772436474,
      "learning_rate": 1.5046028412964582e-05,
      "loss": 0.2336,
      "step": 11484
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.021133747584996,
      "learning_rate": 1.5045172043932277e-05,
      "loss": 0.3471,
      "step": 11485
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.26079601072235564,
      "learning_rate": 1.5044315625263908e-05,
      "loss": 0.2001,
      "step": 11486
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3494432066203465,
      "learning_rate": 1.5043459156967903e-05,
      "loss": 0.3156,
      "step": 11487
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9978119834200605,
      "learning_rate": 1.5042602639052683e-05,
      "loss": 0.0621,
      "step": 11488
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4529644857622062,
      "learning_rate": 1.5041746071526682e-05,
      "loss": 0.7818,
      "step": 11489
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.500047583445658,
      "learning_rate": 1.5040889454398326e-05,
      "loss": 0.3243,
      "step": 11490
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34406021667828385,
      "learning_rate": 1.5040032787676042e-05,
      "loss": 0.2798,
      "step": 11491
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3040037051129396,
      "learning_rate": 1.5039176071368253e-05,
      "loss": 0.288,
      "step": 11492
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.7864624340615614,
      "learning_rate": 1.5038319305483393e-05,
      "loss": 0.821,
      "step": 11493
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2062809373675834,
      "learning_rate": 1.503746249002989e-05,
      "loss": 0.1126,
      "step": 11494
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6291043914083211,
      "learning_rate": 1.5036605625016172e-05,
      "loss": 0.3105,
      "step": 11495
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3390637314876186,
      "learning_rate": 1.5035748710450672e-05,
      "loss": 0.2352,
      "step": 11496
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.8529747208175962,
      "learning_rate": 1.5034891746341822e-05,
      "loss": 0.7721,
      "step": 11497
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29910058617326574,
      "learning_rate": 1.5034034732698046e-05,
      "loss": 0.2653,
      "step": 11498
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.41183041188469427,
      "learning_rate": 1.5033177669527782e-05,
      "loss": 0.2508,
      "step": 11499
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.380828826783419,
      "learning_rate": 1.503232055683946e-05,
      "loss": 0.3093,
      "step": 11500
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6443040263310971,
      "learning_rate": 1.503146339464151e-05,
      "loss": 0.3652,
      "step": 11501
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42972630463725553,
      "learning_rate": 1.5030606182942369e-05,
      "loss": 0.2795,
      "step": 11502
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.20265771980371405,
      "learning_rate": 1.502974892175047e-05,
      "loss": 0.1978,
      "step": 11503
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6579698727322982,
      "learning_rate": 1.5028891611074245e-05,
      "loss": 0.3167,
      "step": 11504
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35341248516519486,
      "learning_rate": 1.5028034250922132e-05,
      "loss": 0.2345,
      "step": 11505
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3302027822286666,
      "learning_rate": 1.5027176841302563e-05,
      "loss": 0.718,
      "step": 11506
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5602833007102396,
      "learning_rate": 1.502631938222397e-05,
      "loss": 0.3802,
      "step": 11507
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8046033723745106,
      "learning_rate": 1.5025461873694795e-05,
      "loss": 0.5101,
      "step": 11508
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2878828816704314,
      "learning_rate": 1.5024604315723475e-05,
      "loss": 0.1658,
      "step": 11509
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37697408568368995,
      "learning_rate": 1.5023746708318445e-05,
      "loss": 0.3401,
      "step": 11510
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.38079396165021756,
      "learning_rate": 1.5022889051488139e-05,
      "loss": 0.2716,
      "step": 11511
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2265513993722839,
      "learning_rate": 1.5022031345240998e-05,
      "loss": 0.0723,
      "step": 11512
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7049044399554153,
      "learning_rate": 1.502117358958546e-05,
      "loss": 0.3739,
      "step": 11513
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3254040771638397,
      "learning_rate": 1.5020315784529964e-05,
      "loss": 0.2438,
      "step": 11514
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7898188356000366,
      "learning_rate": 1.5019457930082949e-05,
      "loss": 0.5165,
      "step": 11515
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2895914934400248,
      "learning_rate": 1.5018600026252857e-05,
      "loss": 0.2353,
      "step": 11516
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6315646739617493,
      "learning_rate": 1.5017742073048127e-05,
      "loss": 0.4529,
      "step": 11517
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2917356063483027,
      "learning_rate": 1.5016884070477198e-05,
      "loss": 0.2033,
      "step": 11518
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2890738449861456,
      "learning_rate": 1.5016026018548515e-05,
      "loss": 0.8062,
      "step": 11519
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8796895705312354,
      "learning_rate": 1.5015167917270513e-05,
      "loss": 0.3731,
      "step": 11520
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24773476352002927,
      "learning_rate": 1.501430976665164e-05,
      "loss": 0.2112,
      "step": 11521
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3072904091704389,
      "learning_rate": 1.5013451566700341e-05,
      "loss": 0.184,
      "step": 11522
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.367155040753291,
      "learning_rate": 1.5012593317425057e-05,
      "loss": 0.2845,
      "step": 11523
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2123634392001374,
      "learning_rate": 1.5011735018834226e-05,
      "loss": 0.0872,
      "step": 11524
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7198756731524767,
      "learning_rate": 1.5010876670936297e-05,
      "loss": 0.3915,
      "step": 11525
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8372728260675538,
      "learning_rate": 1.5010018273739716e-05,
      "loss": 0.5266,
      "step": 11526
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2900542096159155,
      "learning_rate": 1.5009159827252925e-05,
      "loss": 0.2032,
      "step": 11527
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4701870151532496,
      "learning_rate": 1.5008301331484375e-05,
      "loss": 0.3616,
      "step": 11528
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4103479330943538,
      "learning_rate": 1.5007442786442508e-05,
      "loss": 0.2709,
      "step": 11529
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3209773353588101,
      "learning_rate": 1.5006584192135769e-05,
      "loss": 0.1723,
      "step": 11530
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.48649028348476997,
      "learning_rate": 1.5005725548572607e-05,
      "loss": 0.1453,
      "step": 11531
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39369328574315215,
      "learning_rate": 1.500486685576147e-05,
      "loss": 0.3323,
      "step": 11532
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6941817699412552,
      "learning_rate": 1.5004008113710806e-05,
      "loss": 0.3719,
      "step": 11533
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33126648414777027,
      "learning_rate": 1.5003149322429064e-05,
      "loss": 0.3219,
      "step": 11534
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9351572200786983,
      "learning_rate": 1.5002290481924696e-05,
      "loss": 0.3512,
      "step": 11535
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6307308310840825,
      "learning_rate": 1.5001431592206146e-05,
      "loss": 0.3211,
      "step": 11536
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3367904275855828,
      "learning_rate": 1.5000572653281867e-05,
      "loss": 0.2484,
      "step": 11537
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.309512530293813,
      "learning_rate": 1.4999713665160305e-05,
      "loss": 0.3979,
      "step": 11538
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4783821634963157,
      "learning_rate": 1.499885462784992e-05,
      "loss": 0.293,
      "step": 11539
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.21398710208280403,
      "learning_rate": 1.4997995541359154e-05,
      "loss": 0.1852,
      "step": 11540
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33593113583050743,
      "learning_rate": 1.4997136405696467e-05,
      "loss": 0.2691,
      "step": 11541
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7286008304860833,
      "learning_rate": 1.4996277220870305e-05,
      "loss": 0.3972,
      "step": 11542
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9829860160936144,
      "learning_rate": 1.4995417986889124e-05,
      "loss": 0.5081,
      "step": 11543
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8798016084755524,
      "learning_rate": 1.4994558703761377e-05,
      "loss": 0.4488,
      "step": 11544
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4135529513328475,
      "learning_rate": 1.499369937149552e-05,
      "loss": 0.3111,
      "step": 11545
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.25793426843411943,
      "learning_rate": 1.4992839990100004e-05,
      "loss": 0.2582,
      "step": 11546
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.4069189686909913,
      "learning_rate": 1.4991980559583286e-05,
      "loss": 0.7882,
      "step": 11547
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18355084899616447,
      "learning_rate": 1.4991121079953824e-05,
      "loss": 0.0752,
      "step": 11548
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.39738241596337,
      "learning_rate": 1.4990261551220064e-05,
      "loss": 0.2611,
      "step": 11549
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3198766162613606,
      "learning_rate": 1.4989401973390471e-05,
      "loss": 0.2452,
      "step": 11550
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8376372138604345,
      "learning_rate": 1.4988542346473498e-05,
      "loss": 0.3874,
      "step": 11551
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33054019268315427,
      "learning_rate": 1.4987682670477608e-05,
      "loss": 0.3315,
      "step": 11552
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.7720273030286774,
      "learning_rate": 1.4986822945411252e-05,
      "loss": 0.4455,
      "step": 11553
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.35047792796615984,
      "learning_rate": 1.4985963171282893e-05,
      "loss": 0.226,
      "step": 11554
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.42778836441061446,
      "learning_rate": 1.4985103348100988e-05,
      "loss": 0.259,
      "step": 11555
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.518467985069718,
      "learning_rate": 1.4984243475873995e-05,
      "loss": 0.865,
      "step": 11556
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3015698644050514,
      "learning_rate": 1.4983383554610372e-05,
      "loss": 0.2458,
      "step": 11557
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.36687051945729493,
      "learning_rate": 1.4982523584318584e-05,
      "loss": 0.2128,
      "step": 11558
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.34997036199718806,
      "learning_rate": 1.498166356500709e-05,
      "loss": 0.2383,
      "step": 11559
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.6412206997682512,
      "learning_rate": 1.498080349668435e-05,
      "loss": 0.488,
      "step": 11560
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3469293454397584,
      "learning_rate": 1.4979943379358828e-05,
      "loss": 0.197,
      "step": 11561
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.433028235143435,
      "learning_rate": 1.4979083213038983e-05,
      "loss": 0.7626,
      "step": 11562
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.288736659792795,
      "learning_rate": 1.4978222997733278e-05,
      "loss": 0.1873,
      "step": 11563
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2799121580512702,
      "learning_rate": 1.4977362733450179e-05,
      "loss": 0.2463,
      "step": 11564
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.101875636983134,
      "learning_rate": 1.4976502420198146e-05,
      "loss": 0.6532,
      "step": 11565
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.8368970800878964,
      "learning_rate": 1.4975642057985647e-05,
      "loss": 0.3327,
      "step": 11566
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.524618648978852,
      "learning_rate": 1.497478164682114e-05,
      "loss": 0.4358,
      "step": 11567
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.28406007474909845,
      "learning_rate": 1.49739211867131e-05,
      "loss": 0.1956,
      "step": 11568
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29356211821102235,
      "learning_rate": 1.4973060677669985e-05,
      "loss": 0.2918,
      "step": 11569
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.23715956925428447,
      "learning_rate": 1.497220011970026e-05,
      "loss": 0.156,
      "step": 11570
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.5220998322897883,
      "learning_rate": 1.4971339512812395e-05,
      "loss": 0.7609,
      "step": 11571
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.32025809686571804,
      "learning_rate": 1.4970478857014856e-05,
      "loss": 0.1582,
      "step": 11572
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3921524000923708,
      "learning_rate": 1.4969618152316111e-05,
      "loss": 0.3312,
      "step": 11573
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.2051724863838,
      "learning_rate": 1.4968757398724628e-05,
      "loss": 0.1156,
      "step": 11574
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.4193825071345789,
      "learning_rate": 1.4967896596248874e-05,
      "loss": 0.3814,
      "step": 11575
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3369150720884294,
      "learning_rate": 1.496703574489732e-05,
      "loss": 0.2885,
      "step": 11576
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.29823707749283757,
      "learning_rate": 1.496617484467843e-05,
      "loss": 0.2417,
      "step": 11577
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3966901654362138,
      "learning_rate": 1.4965313895600684e-05,
      "loss": 0.5789,
      "step": 11578
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.18465101223168323,
      "learning_rate": 1.4964452897672543e-05,
      "loss": 0.0701,
      "step": 11579
    },
    {
      "epoch": 0.35,
      "grad_norm": 1.3884149763907805,
      "learning_rate": 1.4963591850902479e-05,
      "loss": 0.9539,
      "step": 11580
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.2526392484212357,
      "learning_rate": 1.496273075529897e-05,
      "loss": 0.1998,
      "step": 11581
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5272296808174055,
      "learning_rate": 1.4961869610870477e-05,
      "loss": 0.351,
      "step": 11582
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.3650397730794177,
      "learning_rate": 1.496100841762548e-05,
      "loss": 0.2634,
      "step": 11583
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.5432382110428777,
      "learning_rate": 1.496014717557245e-05,
      "loss": 0.4801,
      "step": 11584
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.51696354660535,
      "learning_rate": 1.495928588471986e-05,
      "loss": 0.305,
      "step": 11585
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.9776501988306263,
      "learning_rate": 1.4958424545076187e-05,
      "loss": 0.505,
      "step": 11586
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.27132296617415264,
      "learning_rate": 1.4957563156649897e-05,
      "loss": 0.237,
      "step": 11587
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.31279819649356255,
      "learning_rate": 1.4956701719449473e-05,
      "loss": 0.2559,
      "step": 11588
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.24815413282160642,
      "learning_rate": 1.4955840233483385e-05,
      "loss": 0.1324,
      "step": 11589
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.33336640219848485,
      "learning_rate": 1.4954978698760107e-05,
      "loss": 0.1224,
      "step": 11590
    },
    {
      "epoch": 0.35,
      "grad_norm": 0.37519585164362484,
      "learning_rate": 1.4954117115288123e-05,
      "loss": 0.3336,
      "step": 11591
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6780578572085296,
      "learning_rate": 1.4953255483075903e-05,
      "loss": 0.3574,
      "step": 11592
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31668495228840293,
      "learning_rate": 1.4952393802131926e-05,
      "loss": 0.2946,
      "step": 11593
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5448109034433157,
      "learning_rate": 1.495153207246467e-05,
      "loss": 0.3602,
      "step": 11594
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4145005798526742,
      "learning_rate": 1.495067029408261e-05,
      "loss": 0.2772,
      "step": 11595
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4341460177166544,
      "learning_rate": 1.4949808466994227e-05,
      "loss": 0.2453,
      "step": 11596
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.28946637255924607,
      "learning_rate": 1.4948946591208002e-05,
      "loss": 0.1903,
      "step": 11597
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0193564832761401,
      "learning_rate": 1.494808466673241e-05,
      "loss": 0.6059,
      "step": 11598
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3101565514954219,
      "learning_rate": 1.494722269357593e-05,
      "loss": 0.289,
      "step": 11599
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3238230237034585,
      "learning_rate": 1.4946360671747048e-05,
      "loss": 0.2279,
      "step": 11600
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.785061606766095,
      "learning_rate": 1.4945498601254244e-05,
      "loss": 0.8919,
      "step": 11601
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8370599063926656,
      "learning_rate": 1.4944636482105996e-05,
      "loss": 0.3769,
      "step": 11602
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0517496003833133,
      "learning_rate": 1.4943774314310789e-05,
      "loss": 0.2945,
      "step": 11603
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.392309631325537,
      "learning_rate": 1.49429120978771e-05,
      "loss": 0.3188,
      "step": 11604
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3442953225490531,
      "learning_rate": 1.4942049832813422e-05,
      "loss": 0.2387,
      "step": 11605
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.45912518821286613,
      "learning_rate": 1.4941187519128224e-05,
      "loss": 0.3601,
      "step": 11606
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2597658204010715,
      "learning_rate": 1.4940325156830002e-05,
      "loss": 0.1459,
      "step": 11607
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.636785230186141,
      "learning_rate": 1.4939462745927231e-05,
      "loss": 0.4183,
      "step": 11608
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31741725088663625,
      "learning_rate": 1.4938600286428404e-05,
      "loss": 0.239,
      "step": 11609
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5309000112471803,
      "learning_rate": 1.4937737778342003e-05,
      "loss": 0.349,
      "step": 11610
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3764534502613277,
      "learning_rate": 1.493687522167651e-05,
      "loss": 0.2963,
      "step": 11611
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8587303314544171,
      "learning_rate": 1.4936012616440416e-05,
      "loss": 0.527,
      "step": 11612
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4356528863809996,
      "learning_rate": 1.4935149962642204e-05,
      "loss": 0.074,
      "step": 11613
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37676719399017916,
      "learning_rate": 1.4934287260290365e-05,
      "loss": 0.3141,
      "step": 11614
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2680722814823804,
      "learning_rate": 1.4933424509393382e-05,
      "loss": 0.1377,
      "step": 11615
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40515602731712275,
      "learning_rate": 1.4932561709959745e-05,
      "loss": 0.2824,
      "step": 11616
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4804061736862003,
      "learning_rate": 1.493169886199795e-05,
      "loss": 0.3349,
      "step": 11617
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2954072524578508,
      "learning_rate": 1.4930835965516472e-05,
      "loss": 0.2035,
      "step": 11618
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6305389957445398,
      "learning_rate": 1.4929973020523807e-05,
      "loss": 0.4456,
      "step": 11619
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3328686237680903,
      "learning_rate": 1.4929110027028448e-05,
      "loss": 0.165,
      "step": 11620
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4923020681023456,
      "learning_rate": 1.4928246985038881e-05,
      "loss": 0.8309,
      "step": 11621
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33064855640122487,
      "learning_rate": 1.4927383894563599e-05,
      "loss": 0.1715,
      "step": 11622
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3626173506125513,
      "learning_rate": 1.4926520755611094e-05,
      "loss": 0.3167,
      "step": 11623
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38286781982785895,
      "learning_rate": 1.4925657568189858e-05,
      "loss": 0.2773,
      "step": 11624
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.719898057817542,
      "learning_rate": 1.4924794332308376e-05,
      "loss": 0.9083,
      "step": 11625
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.599882364224353,
      "learning_rate": 1.4923931047975151e-05,
      "loss": 0.3199,
      "step": 11626
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3659016654277536,
      "learning_rate": 1.4923067715198673e-05,
      "loss": 0.3007,
      "step": 11627
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5733876677576958,
      "learning_rate": 1.4922204333987434e-05,
      "loss": 0.3518,
      "step": 11628
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.22861630423512871,
      "learning_rate": 1.492134090434993e-05,
      "loss": 0.198,
      "step": 11629
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5374117703344856,
      "learning_rate": 1.4920477426294655e-05,
      "loss": 0.3229,
      "step": 11630
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31075937001888687,
      "learning_rate": 1.49196138998301e-05,
      "loss": 0.1835,
      "step": 11631
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5821896386539547,
      "learning_rate": 1.4918750324964768e-05,
      "loss": 0.4206,
      "step": 11632
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9305599585712221,
      "learning_rate": 1.4917886701707152e-05,
      "loss": 0.3636,
      "step": 11633
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43969744075780154,
      "learning_rate": 1.4917023030065747e-05,
      "loss": 0.3704,
      "step": 11634
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4006124848514175,
      "learning_rate": 1.4916159310049052e-05,
      "loss": 0.2731,
      "step": 11635
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9939443548592813,
      "learning_rate": 1.4915295541665566e-05,
      "loss": 0.5267,
      "step": 11636
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3581495205478988,
      "learning_rate": 1.4914431724923784e-05,
      "loss": 0.2771,
      "step": 11637
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.447389839267391,
      "learning_rate": 1.4913567859832205e-05,
      "loss": 0.2512,
      "step": 11638
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.30519363150395346,
      "learning_rate": 1.4912703946399328e-05,
      "loss": 0.1126,
      "step": 11639
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8646525480160073,
      "learning_rate": 1.4911839984633653e-05,
      "loss": 0.4032,
      "step": 11640
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.30701974713882274,
      "learning_rate": 1.4910975974543683e-05,
      "loss": 0.2311,
      "step": 11641
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4993594805147724,
      "learning_rate": 1.4910111916137913e-05,
      "loss": 0.3581,
      "step": 11642
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6891103237280889,
      "learning_rate": 1.4909247809424847e-05,
      "loss": 0.379,
      "step": 11643
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6645895460609428,
      "learning_rate": 1.4908383654412986e-05,
      "loss": 0.3178,
      "step": 11644
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39652344563225633,
      "learning_rate": 1.4907519451110828e-05,
      "loss": 0.2695,
      "step": 11645
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3016320177914414,
      "learning_rate": 1.4906655199526886e-05,
      "loss": 0.2527,
      "step": 11646
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32887640612538177,
      "learning_rate": 1.490579089966965e-05,
      "loss": 0.2681,
      "step": 11647
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2699150770538582,
      "learning_rate": 1.490492655154763e-05,
      "loss": 0.1193,
      "step": 11648
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9450530015852939,
      "learning_rate": 1.490406215516933e-05,
      "loss": 0.5086,
      "step": 11649
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31348052386135433,
      "learning_rate": 1.490319771054325e-05,
      "loss": 0.2053,
      "step": 11650
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6345663552654455,
      "learning_rate": 1.4902333217677901e-05,
      "loss": 0.9609,
      "step": 11651
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6880621260957228,
      "learning_rate": 1.490146867658178e-05,
      "loss": 0.3268,
      "step": 11652
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5825437612572795,
      "learning_rate": 1.4900604087263401e-05,
      "loss": 0.4231,
      "step": 11653
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2670883054287181,
      "learning_rate": 1.4899739449731265e-05,
      "loss": 0.2033,
      "step": 11654
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.399353016765396,
      "learning_rate": 1.4898874763993885e-05,
      "loss": 0.9545,
      "step": 11655
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8784187566131183,
      "learning_rate": 1.489801003005976e-05,
      "loss": 0.3248,
      "step": 11656
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.22362389750021133,
      "learning_rate": 1.4897145247937399e-05,
      "loss": 0.1137,
      "step": 11657
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33001061670343584,
      "learning_rate": 1.489628041763531e-05,
      "loss": 0.243,
      "step": 11658
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3473907239270189,
      "learning_rate": 1.4895415539162006e-05,
      "loss": 0.2273,
      "step": 11659
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5187705816228841,
      "learning_rate": 1.4894550612525994e-05,
      "loss": 0.4021,
      "step": 11660
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8674717506545762,
      "learning_rate": 1.4893685637735782e-05,
      "loss": 0.3861,
      "step": 11661
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9070617271773197,
      "learning_rate": 1.4892820614799882e-05,
      "loss": 0.6032,
      "step": 11662
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3094385621626232,
      "learning_rate": 1.48919555437268e-05,
      "loss": 0.0756,
      "step": 11663
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3466880402904537,
      "learning_rate": 1.4891090424525047e-05,
      "loss": 0.3142,
      "step": 11664
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.30246495539617585,
      "learning_rate": 1.4890225257203139e-05,
      "loss": 0.2562,
      "step": 11665
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.28636309238253166,
      "learning_rate": 1.4889360041769589e-05,
      "loss": 0.1507,
      "step": 11666
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.537975730174959,
      "learning_rate": 1.4888494778232905e-05,
      "loss": 0.1748,
      "step": 11667
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34263029893457897,
      "learning_rate": 1.4887629466601601e-05,
      "loss": 0.3071,
      "step": 11668
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8580381624283575,
      "learning_rate": 1.4886764106884188e-05,
      "loss": 0.4098,
      "step": 11669
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3737691379088133,
      "learning_rate": 1.4885898699089184e-05,
      "loss": 0.2997,
      "step": 11670
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.577786315360487,
      "learning_rate": 1.4885033243225098e-05,
      "loss": 0.9813,
      "step": 11671
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3011534384461911,
      "learning_rate": 1.488416773930045e-05,
      "loss": 0.1951,
      "step": 11672
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4794018672133354,
      "learning_rate": 1.4883302187323751e-05,
      "loss": 0.3786,
      "step": 11673
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2401299102042593,
      "learning_rate": 1.4882436587303521e-05,
      "loss": 0.14,
      "step": 11674
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42059435557783237,
      "learning_rate": 1.4881570939248274e-05,
      "loss": 0.2515,
      "step": 11675
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24983507551096268,
      "learning_rate": 1.4880705243166522e-05,
      "loss": 0.2213,
      "step": 11676
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36472329360376154,
      "learning_rate": 1.4879839499066786e-05,
      "loss": 0.332,
      "step": 11677
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6555299097408842,
      "learning_rate": 1.4878973706957585e-05,
      "loss": 0.3657,
      "step": 11678
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6442819019876983,
      "learning_rate": 1.4878107866847435e-05,
      "loss": 0.4543,
      "step": 11679
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7931766755157343,
      "learning_rate": 1.4877241978744857e-05,
      "loss": 0.4887,
      "step": 11680
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5875189682676036,
      "learning_rate": 1.4876376042658365e-05,
      "loss": 0.3497,
      "step": 11681
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32090477255586664,
      "learning_rate": 1.4875510058596481e-05,
      "loss": 0.2354,
      "step": 11682
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3177292482459823,
      "learning_rate": 1.4874644026567724e-05,
      "loss": 0.2649,
      "step": 11683
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4590566387651514,
      "learning_rate": 1.4873777946580617e-05,
      "loss": 0.2349,
      "step": 11684
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.17174736201585455,
      "learning_rate": 1.4872911818643675e-05,
      "loss": 0.0702,
      "step": 11685
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42269896629369924,
      "learning_rate": 1.4872045642765427e-05,
      "loss": 0.3113,
      "step": 11686
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9586185265569178,
      "learning_rate": 1.4871179418954392e-05,
      "loss": 0.4008,
      "step": 11687
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35465209268183984,
      "learning_rate": 1.4870313147219091e-05,
      "loss": 0.3247,
      "step": 11688
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8942895945096128,
      "learning_rate": 1.4869446827568043e-05,
      "loss": 0.56,
      "step": 11689
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9799273774211603,
      "learning_rate": 1.4868580460009776e-05,
      "loss": 0.3831,
      "step": 11690
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3187724543547174,
      "learning_rate": 1.486771404455281e-05,
      "loss": 0.2197,
      "step": 11691
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1697412581973579,
      "learning_rate": 1.4866847581205677e-05,
      "loss": 0.7022,
      "step": 11692
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.19823905050475377,
      "learning_rate": 1.4865981069976894e-05,
      "loss": 0.153,
      "step": 11693
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4903679670400213,
      "learning_rate": 1.4865114510874987e-05,
      "loss": 0.3719,
      "step": 11694
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3533306926498405,
      "learning_rate": 1.4864247903908485e-05,
      "loss": 0.2265,
      "step": 11695
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4266444070937546,
      "learning_rate": 1.4863381249085909e-05,
      "loss": 0.3011,
      "step": 11696
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9359645817914214,
      "learning_rate": 1.486251454641579e-05,
      "loss": 0.6559,
      "step": 11697
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9018568175708851,
      "learning_rate": 1.486164779590665e-05,
      "loss": 0.5143,
      "step": 11698
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5868314228131137,
      "learning_rate": 1.4860780997567023e-05,
      "loss": 0.3087,
      "step": 11699
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25848424549847915,
      "learning_rate": 1.4859914151405433e-05,
      "loss": 0.2458,
      "step": 11700
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.48114702151696476,
      "learning_rate": 1.4859047257430404e-05,
      "loss": 0.3233,
      "step": 11701
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9472968553472844,
      "learning_rate": 1.4858180315650471e-05,
      "loss": 0.067,
      "step": 11702
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7676515598914251,
      "learning_rate": 1.4857313326074162e-05,
      "loss": 0.4774,
      "step": 11703
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2874154741953622,
      "learning_rate": 1.4856446288710008e-05,
      "loss": 0.1914,
      "step": 11704
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33866013701693315,
      "learning_rate": 1.4855579203566537e-05,
      "loss": 0.2207,
      "step": 11705
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42641935973439077,
      "learning_rate": 1.485471207065228e-05,
      "loss": 0.2443,
      "step": 11706
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5038851664597925,
      "learning_rate": 1.4853844889975766e-05,
      "loss": 0.4154,
      "step": 11707
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4108483005024316,
      "learning_rate": 1.4852977661545534e-05,
      "loss": 0.1718,
      "step": 11708
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32249420364093695,
      "learning_rate": 1.4852110385370105e-05,
      "loss": 0.2282,
      "step": 11709
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0645806465878078,
      "learning_rate": 1.4851243061458019e-05,
      "loss": 0.6454,
      "step": 11710
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37271796467617346,
      "learning_rate": 1.4850375689817809e-05,
      "loss": 0.27,
      "step": 11711
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.40838474813662,
      "learning_rate": 1.4849508270458009e-05,
      "loss": 0.3857,
      "step": 11712
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32231604975259287,
      "learning_rate": 1.4848640803387144e-05,
      "loss": 0.0733,
      "step": 11713
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36024338242265747,
      "learning_rate": 1.4847773288613761e-05,
      "loss": 0.3117,
      "step": 11714
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.23262586256620021,
      "learning_rate": 1.4846905726146388e-05,
      "loss": 0.108,
      "step": 11715
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2882706269711215,
      "learning_rate": 1.4846038115993561e-05,
      "loss": 0.8921,
      "step": 11716
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5206646603352667,
      "learning_rate": 1.4845170458163819e-05,
      "loss": 0.1731,
      "step": 11717
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3499618512453475,
      "learning_rate": 1.4844302752665694e-05,
      "loss": 0.2858,
      "step": 11718
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3424915268584925,
      "learning_rate": 1.4843434999507721e-05,
      "loss": 0.2631,
      "step": 11719
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7665861368873594,
      "learning_rate": 1.4842567198698446e-05,
      "loss": 0.4588,
      "step": 11720
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7640097902897225,
      "learning_rate": 1.4841699350246399e-05,
      "loss": 0.3587,
      "step": 11721
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3353066013626239,
      "learning_rate": 1.484083145416012e-05,
      "loss": 0.1908,
      "step": 11722
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3290026597038734,
      "learning_rate": 1.483996351044815e-05,
      "loss": 0.2479,
      "step": 11723
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24526872551330572,
      "learning_rate": 1.4839095519119027e-05,
      "loss": 0.2182,
      "step": 11724
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.4899041981980674,
      "learning_rate": 1.4838227480181289e-05,
      "loss": 0.8221,
      "step": 11725
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44316559916234927,
      "learning_rate": 1.4837359393643476e-05,
      "loss": 0.2156,
      "step": 11726
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3438233194766608,
      "learning_rate": 1.483649125951413e-05,
      "loss": 0.2836,
      "step": 11727
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7211123611486969,
      "learning_rate": 1.4835623077801793e-05,
      "loss": 0.3763,
      "step": 11728
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9164809232328388,
      "learning_rate": 1.4834754848515002e-05,
      "loss": 0.5592,
      "step": 11729
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.30072123226506986,
      "learning_rate": 1.4833886571662306e-05,
      "loss": 0.2734,
      "step": 11730
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4051963323235853,
      "learning_rate": 1.4833018247252242e-05,
      "loss": 0.2793,
      "step": 11731
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.44075683519784264,
      "learning_rate": 1.4832149875293353e-05,
      "loss": 0.2695,
      "step": 11732
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5094657162344114,
      "learning_rate": 1.4831281455794181e-05,
      "loss": 0.3239,
      "step": 11733
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2081164910760718,
      "learning_rate": 1.4830412988763278e-05,
      "loss": 0.644,
      "step": 11734
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24339647755042423,
      "learning_rate": 1.4829544474209181e-05,
      "loss": 0.1558,
      "step": 11735
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29942481415158384,
      "learning_rate": 1.4828675912140434e-05,
      "loss": 0.285,
      "step": 11736
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1177538783578544,
      "learning_rate": 1.4827807302565586e-05,
      "loss": 0.0651,
      "step": 11737
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7326379493216492,
      "learning_rate": 1.4826938645493183e-05,
      "loss": 0.528,
      "step": 11738
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7108458970380009,
      "learning_rate": 1.4826069940931766e-05,
      "loss": 0.3613,
      "step": 11739
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6483319403427156,
      "learning_rate": 1.4825201188889886e-05,
      "loss": 0.3853,
      "step": 11740
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36761757239302295,
      "learning_rate": 1.4824332389376091e-05,
      "loss": 0.2475,
      "step": 11741
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.27191876220948386,
      "learning_rate": 1.4823463542398923e-05,
      "loss": 0.251,
      "step": 11742
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2978203983493173,
      "learning_rate": 1.4822594647966936e-05,
      "loss": 0.1632,
      "step": 11743
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1331524156864312,
      "learning_rate": 1.4821725706088676e-05,
      "loss": 0.7383,
      "step": 11744
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31172825570990464,
      "learning_rate": 1.482085671677269e-05,
      "loss": 0.1642,
      "step": 11745
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4487792137634959,
      "learning_rate": 1.4819987680027532e-05,
      "loss": 0.3977,
      "step": 11746
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37152378664288865,
      "learning_rate": 1.481911859586175e-05,
      "loss": 0.3207,
      "step": 11747
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3559643149718739,
      "learning_rate": 1.4818249464283889e-05,
      "loss": 0.2564,
      "step": 11748
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.69332675444637,
      "learning_rate": 1.481738028530251e-05,
      "loss": 0.3361,
      "step": 11749
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.30801884686497866,
      "learning_rate": 1.4816511058926156e-05,
      "loss": 0.2371,
      "step": 11750
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4515019255921163,
      "learning_rate": 1.4815641785163379e-05,
      "loss": 0.3034,
      "step": 11751
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9149319636361547,
      "learning_rate": 1.4814772464022735e-05,
      "loss": 0.4993,
      "step": 11752
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3275977294388798,
      "learning_rate": 1.4813903095512774e-05,
      "loss": 0.3241,
      "step": 11753
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2924410409244194,
      "learning_rate": 1.4813033679642055e-05,
      "loss": 0.1832,
      "step": 11754
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6163003971574321,
      "learning_rate": 1.4812164216419122e-05,
      "loss": 0.4994,
      "step": 11755
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31004845739348674,
      "learning_rate": 1.4811294705852537e-05,
      "loss": 0.1763,
      "step": 11756
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.639378708718672,
      "learning_rate": 1.4810425147950853e-05,
      "loss": 0.7643,
      "step": 11757
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36204904620112277,
      "learning_rate": 1.480955554272262e-05,
      "loss": 0.1262,
      "step": 11758
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4032623953710092,
      "learning_rate": 1.4808685890176398e-05,
      "loss": 0.3043,
      "step": 11759
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.312973267981397,
      "learning_rate": 1.480781619032074e-05,
      "loss": 0.2437,
      "step": 11760
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2829316123148928,
      "learning_rate": 1.4806946443164207e-05,
      "loss": 0.1494,
      "step": 11761
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6305098110995578,
      "learning_rate": 1.4806076648715353e-05,
      "loss": 0.4924,
      "step": 11762
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3127290508603966,
      "learning_rate": 1.4805206806982737e-05,
      "loss": 0.2357,
      "step": 11763
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8848627176170895,
      "learning_rate": 1.4804336917974913e-05,
      "loss": 0.5787,
      "step": 11764
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2549794610373622,
      "learning_rate": 1.480346698170044e-05,
      "loss": 0.2136,
      "step": 11765
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4359942981101259,
      "learning_rate": 1.480259699816788e-05,
      "loss": 0.3603,
      "step": 11766
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34740440491238744,
      "learning_rate": 1.4801726967385792e-05,
      "loss": 0.0777,
      "step": 11767
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42561065060549164,
      "learning_rate": 1.4800856889362733e-05,
      "loss": 0.3264,
      "step": 11768
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2879752695560254,
      "learning_rate": 1.4799986764107264e-05,
      "loss": 0.349,
      "step": 11769
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.191314667700168,
      "learning_rate": 1.4799116591627947e-05,
      "loss": 0.598,
      "step": 11770
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2869189933623955,
      "learning_rate": 1.479824637193334e-05,
      "loss": 0.2667,
      "step": 11771
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6720266069564632,
      "learning_rate": 1.4797376105032007e-05,
      "loss": 0.3588,
      "step": 11772
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31606810310257366,
      "learning_rate": 1.4796505790932512e-05,
      "loss": 0.2677,
      "step": 11773
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24745849999345032,
      "learning_rate": 1.4795635429643409e-05,
      "loss": 0.0853,
      "step": 11774
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0443505500223207,
      "learning_rate": 1.4794765021173271e-05,
      "loss": 0.688,
      "step": 11775
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4781998954954591,
      "learning_rate": 1.479389456553066e-05,
      "loss": 0.2179,
      "step": 11776
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3360821433736048,
      "learning_rate": 1.4793024062724131e-05,
      "loss": 0.274,
      "step": 11777
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2946923763713013,
      "learning_rate": 1.4792153512762255e-05,
      "loss": 0.2518,
      "step": 11778
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5201420231645941,
      "learning_rate": 1.4791282915653596e-05,
      "loss": 0.9413,
      "step": 11779
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5362521146385428,
      "learning_rate": 1.4790412271406722e-05,
      "loss": 0.3608,
      "step": 11780
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37594106776412856,
      "learning_rate": 1.4789541580030193e-05,
      "loss": 0.2766,
      "step": 11781
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42212688981113405,
      "learning_rate": 1.478867084153258e-05,
      "loss": 0.2718,
      "step": 11782
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4462999646139098,
      "learning_rate": 1.4787800055922446e-05,
      "loss": 0.2484,
      "step": 11783
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.22812246713325968,
      "learning_rate": 1.4786929223208359e-05,
      "loss": 0.1953,
      "step": 11784
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.1470896437067937,
      "learning_rate": 1.4786058343398887e-05,
      "loss": 0.5147,
      "step": 11785
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29276967021959976,
      "learning_rate": 1.4785187416502602e-05,
      "loss": 0.1951,
      "step": 11786
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9864105352246731,
      "learning_rate": 1.4784316442528064e-05,
      "loss": 0.367,
      "step": 11787
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6538657924266829,
      "learning_rate": 1.4783445421483852e-05,
      "loss": 0.5516,
      "step": 11788
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29515433845309114,
      "learning_rate": 1.4782574353378527e-05,
      "loss": 0.2725,
      "step": 11789
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42832647149251246,
      "learning_rate": 1.478170323822066e-05,
      "loss": 0.2816,
      "step": 11790
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4270747648103654,
      "learning_rate": 1.4780832076018824e-05,
      "loss": 0.2296,
      "step": 11791
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5121906430450713,
      "learning_rate": 1.4779960866781588e-05,
      "loss": 0.3043,
      "step": 11792
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2786491376218257,
      "learning_rate": 1.4779089610517527e-05,
      "loss": 0.138,
      "step": 11793
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7286880308701496,
      "learning_rate": 1.477821830723521e-05,
      "loss": 0.3131,
      "step": 11794
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34511289594650396,
      "learning_rate": 1.4777346956943208e-05,
      "loss": 0.2282,
      "step": 11795
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3260490074712503,
      "learning_rate": 1.4776475559650094e-05,
      "loss": 0.3139,
      "step": 11796
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7197533471334496,
      "learning_rate": 1.4775604115364441e-05,
      "loss": 0.3875,
      "step": 11797
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8233236246690427,
      "learning_rate": 1.4774732624094824e-05,
      "loss": 0.5701,
      "step": 11798
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2602710815191993,
      "learning_rate": 1.4773861085849819e-05,
      "loss": 0.0728,
      "step": 11799
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3155560163818084,
      "learning_rate": 1.4772989500637996e-05,
      "loss": 0.2387,
      "step": 11800
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34435335377673315,
      "learning_rate": 1.4772117868467933e-05,
      "loss": 0.3463,
      "step": 11801
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2155157395707358,
      "learning_rate": 1.4771246189348202e-05,
      "loss": 0.0729,
      "step": 11802
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9515193095801865,
      "learning_rate": 1.477037446328738e-05,
      "loss": 0.5439,
      "step": 11803
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.275084377059148,
      "learning_rate": 1.4769502690294046e-05,
      "loss": 0.184,
      "step": 11804
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8670174622105437,
      "learning_rate": 1.4768630870376774e-05,
      "loss": 0.5084,
      "step": 11805
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.755974484982989,
      "learning_rate": 1.4767759003544144e-05,
      "loss": 0.3812,
      "step": 11806
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31915518191536685,
      "learning_rate": 1.4766887089804734e-05,
      "loss": 0.312,
      "step": 11807
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36048716183215274,
      "learning_rate": 1.4766015129167117e-05,
      "loss": 0.1608,
      "step": 11808
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39043031024606456,
      "learning_rate": 1.4765143121639877e-05,
      "loss": 0.3289,
      "step": 11809
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8050377893425816,
      "learning_rate": 1.4764271067231591e-05,
      "loss": 0.0441,
      "step": 11810
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2551322657812312,
      "learning_rate": 1.4763398965950835e-05,
      "loss": 0.163,
      "step": 11811
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3932648116124413,
      "learning_rate": 1.4762526817806198e-05,
      "loss": 0.2497,
      "step": 11812
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24587789606206373,
      "learning_rate": 1.4761654622806252e-05,
      "loss": 0.2195,
      "step": 11813
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.267467938270543,
      "learning_rate": 1.476078238095958e-05,
      "loss": 0.7713,
      "step": 11814
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0602366374883503,
      "learning_rate": 1.4759910092274768e-05,
      "loss": 0.4415,
      "step": 11815
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.5257625845322818,
      "learning_rate": 1.475903775676039e-05,
      "loss": 0.8167,
      "step": 11816
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29375807448740937,
      "learning_rate": 1.4758165374425036e-05,
      "loss": 0.1892,
      "step": 11817
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5078043881331216,
      "learning_rate": 1.4757292945277283e-05,
      "loss": 0.3673,
      "step": 11818
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.27682852019372073,
      "learning_rate": 1.475642046932572e-05,
      "loss": 0.2246,
      "step": 11819
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.28946552963054273,
      "learning_rate": 1.4755547946578924e-05,
      "loss": 0.1559,
      "step": 11820
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4142430998849262,
      "learning_rate": 1.4754675377045483e-05,
      "loss": 0.2816,
      "step": 11821
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9215031910896825,
      "learning_rate": 1.4753802760733981e-05,
      "loss": 0.5983,
      "step": 11822
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3554284926913991,
      "learning_rate": 1.4752930097653005e-05,
      "loss": 0.2776,
      "step": 11823
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9060068535814395,
      "learning_rate": 1.4752057387811136e-05,
      "loss": 0.5731,
      "step": 11824
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3343271011371646,
      "learning_rate": 1.4751184631216967e-05,
      "loss": 0.2524,
      "step": 11825
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9224741759855574,
      "learning_rate": 1.4750311827879078e-05,
      "loss": 0.4671,
      "step": 11826
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3135623862345081,
      "learning_rate": 1.4749438977806056e-05,
      "loss": 0.226,
      "step": 11827
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.795884167635427,
      "learning_rate": 1.4748566081006492e-05,
      "loss": 0.3509,
      "step": 11828
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2615156469008406,
      "learning_rate": 1.4747693137488974e-05,
      "loss": 0.1585,
      "step": 11829
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3179054463194658,
      "learning_rate": 1.474682014726209e-05,
      "loss": 0.2934,
      "step": 11830
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39759222072713035,
      "learning_rate": 1.4745947110334424e-05,
      "loss": 0.2616,
      "step": 11831
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.31368648442494224,
      "learning_rate": 1.474507402671457e-05,
      "loss": 0.2874,
      "step": 11832
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.177973736941705,
      "learning_rate": 1.4744200896411116e-05,
      "loss": 0.7686,
      "step": 11833
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9065123473473443,
      "learning_rate": 1.4743327719432652e-05,
      "loss": 0.5888,
      "step": 11834
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8427950650209105,
      "learning_rate": 1.474245449578777e-05,
      "loss": 0.4193,
      "step": 11835
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32375176484993884,
      "learning_rate": 1.474158122548506e-05,
      "loss": 0.2141,
      "step": 11836
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35102396352280796,
      "learning_rate": 1.4740707908533111e-05,
      "loss": 0.3123,
      "step": 11837
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.25754195508893674,
      "learning_rate": 1.4739834544940524e-05,
      "loss": 0.1145,
      "step": 11838
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.43840489368776625,
      "learning_rate": 1.4738961134715881e-05,
      "loss": 0.3413,
      "step": 11839
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2987535934921597,
      "learning_rate": 1.473808767786778e-05,
      "loss": 0.1964,
      "step": 11840
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8445055406050915,
      "learning_rate": 1.4737214174404812e-05,
      "loss": 0.3499,
      "step": 11841
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.557301391338185,
      "learning_rate": 1.4736340624335574e-05,
      "loss": 0.8993,
      "step": 11842
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3018895143452887,
      "learning_rate": 1.4735467027668658e-05,
      "loss": 0.2627,
      "step": 11843
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5807437027171201,
      "learning_rate": 1.473459338441266e-05,
      "loss": 0.3586,
      "step": 11844
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.36064800968047667,
      "learning_rate": 1.4733719694576174e-05,
      "loss": 0.2277,
      "step": 11845
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.139494039288703,
      "learning_rate": 1.4732845958167794e-05,
      "loss": 0.6399,
      "step": 11846
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6014448713479841,
      "learning_rate": 1.4731972175196118e-05,
      "loss": 0.3587,
      "step": 11847
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3341735475843745,
      "learning_rate": 1.4731098345669748e-05,
      "loss": 0.3049,
      "step": 11848
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35998745799935067,
      "learning_rate": 1.4730224469597269e-05,
      "loss": 0.0734,
      "step": 11849
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3434752042235612,
      "learning_rate": 1.472935054698729e-05,
      "loss": 0.3094,
      "step": 11850
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2508939147445901,
      "learning_rate": 1.4728476577848403e-05,
      "loss": 0.0948,
      "step": 11851
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.2673715295907513,
      "learning_rate": 1.4727602562189208e-05,
      "loss": 0.8608,
      "step": 11852
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7298611596811662,
      "learning_rate": 1.4726728500018301e-05,
      "loss": 0.0526,
      "step": 11853
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3192744317724861,
      "learning_rate": 1.4725854391344283e-05,
      "loss": 0.2003,
      "step": 11854
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.39009618686420733,
      "learning_rate": 1.4724980236175758e-05,
      "loss": 0.3159,
      "step": 11855
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7034013823911983,
      "learning_rate": 1.4724106034521319e-05,
      "loss": 0.349,
      "step": 11856
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6871224249514237,
      "learning_rate": 1.4723231786389576e-05,
      "loss": 0.5181,
      "step": 11857
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.33423745943403366,
      "learning_rate": 1.472235749178912e-05,
      "loss": 0.1917,
      "step": 11858
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2615265105773626,
      "learning_rate": 1.4721483150728559e-05,
      "loss": 0.1979,
      "step": 11859
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3990165266093638,
      "learning_rate": 1.472060876321649e-05,
      "loss": 0.255,
      "step": 11860
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4667598683268405,
      "learning_rate": 1.4719734329261524e-05,
      "loss": 0.377,
      "step": 11861
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4477042140399744,
      "learning_rate": 1.4718859848872256e-05,
      "loss": 0.2273,
      "step": 11862
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37522919086746326,
      "learning_rate": 1.4717985322057291e-05,
      "loss": 0.3056,
      "step": 11863
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9831376056204618,
      "learning_rate": 1.471711074882524e-05,
      "loss": 0.3395,
      "step": 11864
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.9145894520250466,
      "learning_rate": 1.4716236129184693e-05,
      "loss": 0.5244,
      "step": 11865
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34154340628283564,
      "learning_rate": 1.4715361463144264e-05,
      "loss": 0.285,
      "step": 11866
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3001068845957349,
      "learning_rate": 1.471448675071256e-05,
      "loss": 0.169,
      "step": 11867
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5178936092577212,
      "learning_rate": 1.4713611991898184e-05,
      "loss": 0.3805,
      "step": 11868
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.24194828620430128,
      "learning_rate": 1.4712737186709745e-05,
      "loss": 0.092,
      "step": 11869
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.354361499832257,
      "learning_rate": 1.4711862335155841e-05,
      "loss": 0.8136,
      "step": 11870
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.4043698063249044,
      "learning_rate": 1.471098743724509e-05,
      "loss": 0.1444,
      "step": 11871
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5334876752885842,
      "learning_rate": 1.4710112492986092e-05,
      "loss": 0.3864,
      "step": 11872
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2834353788864971,
      "learning_rate": 1.4709237502387456e-05,
      "loss": 0.2429,
      "step": 11873
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8458963375359866,
      "learning_rate": 1.4708362465457795e-05,
      "loss": 0.4847,
      "step": 11874
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8179215094571954,
      "learning_rate": 1.4707487382205713e-05,
      "loss": 0.3738,
      "step": 11875
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.983507697361781,
      "learning_rate": 1.4706612252639823e-05,
      "loss": 0.4906,
      "step": 11876
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2734147301810431,
      "learning_rate": 1.4705737076768734e-05,
      "loss": 0.2007,
      "step": 11877
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2824980601390627,
      "learning_rate": 1.4704861854601052e-05,
      "loss": 0.2639,
      "step": 11878
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.41545826588362333,
      "learning_rate": 1.4703986586145389e-05,
      "loss": 0.3033,
      "step": 11879
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3238793706830479,
      "learning_rate": 1.4703111271410363e-05,
      "loss": 0.0726,
      "step": 11880
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.37073332158688266,
      "learning_rate": 1.4702235910404578e-05,
      "loss": 0.2816,
      "step": 11881
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6071118204675544,
      "learning_rate": 1.4701360503136651e-05,
      "loss": 0.3381,
      "step": 11882
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8014352826121662,
      "learning_rate": 1.470048504961519e-05,
      "loss": 0.5314,
      "step": 11883
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2807475290599578,
      "learning_rate": 1.4699609549848812e-05,
      "loss": 0.2419,
      "step": 11884
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8504960937315584,
      "learning_rate": 1.469873400384613e-05,
      "loss": 0.4867,
      "step": 11885
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2823084858818336,
      "learning_rate": 1.4697858411615753e-05,
      "loss": 0.2212,
      "step": 11886
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5155403326074393,
      "learning_rate": 1.4696982773166302e-05,
      "loss": 0.3522,
      "step": 11887
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.0711572704538586,
      "learning_rate": 1.4696107088506387e-05,
      "loss": 0.4968,
      "step": 11888
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.403765078082753,
      "learning_rate": 1.4695231357644628e-05,
      "loss": 0.3267,
      "step": 11889
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.28566407642706,
      "learning_rate": 1.4694355580589638e-05,
      "loss": 0.2088,
      "step": 11890
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34198445717923087,
      "learning_rate": 1.469347975735003e-05,
      "loss": 0.321,
      "step": 11891
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6689058786522243,
      "learning_rate": 1.4692603887934424e-05,
      "loss": 0.3826,
      "step": 11892
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.51793177911465,
      "learning_rate": 1.4691727972351437e-05,
      "loss": 0.0257,
      "step": 11893
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.6134484174520731,
      "learning_rate": 1.4690852010609691e-05,
      "loss": 0.3813,
      "step": 11894
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.34159275567137815,
      "learning_rate": 1.4689976002717796e-05,
      "loss": 0.223,
      "step": 11895
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.42217489612911646,
      "learning_rate": 1.4689099948684374e-05,
      "loss": 0.3603,
      "step": 11896
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.38330631588714653,
      "learning_rate": 1.4688223848518046e-05,
      "loss": 0.2904,
      "step": 11897
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5461245148368201,
      "learning_rate": 1.4687347702227426e-05,
      "loss": 0.281,
      "step": 11898
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3795586691468884,
      "learning_rate": 1.4686471509821136e-05,
      "loss": 0.0722,
      "step": 11899
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35160951750697,
      "learning_rate": 1.4685595271307799e-05,
      "loss": 0.341,
      "step": 11900
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.29516042489056643,
      "learning_rate": 1.4684718986696035e-05,
      "loss": 0.1537,
      "step": 11901
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.32812596792616516,
      "learning_rate": 1.4683842655994464e-05,
      "loss": 0.2917,
      "step": 11902
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.35788491511571235,
      "learning_rate": 1.4682966279211707e-05,
      "loss": 0.0652,
      "step": 11903
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3321770097638906,
      "learning_rate": 1.4682089856356387e-05,
      "loss": 0.2655,
      "step": 11904
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.734437048308058,
      "learning_rate": 1.4681213387437125e-05,
      "loss": 0.4182,
      "step": 11905
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5536673369791055,
      "learning_rate": 1.4680336872462546e-05,
      "loss": 0.3882,
      "step": 11906
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.8404996174831314,
      "learning_rate": 1.4679460311441273e-05,
      "loss": 0.5117,
      "step": 11907
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.2894553982587265,
      "learning_rate": 1.4678583704381932e-05,
      "loss": 0.1993,
      "step": 11908
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3367670937664439,
      "learning_rate": 1.4677707051293145e-05,
      "loss": 0.3037,
      "step": 11909
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3099559022018644,
      "learning_rate": 1.4676830352183535e-05,
      "loss": 0.1148,
      "step": 11910
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.8397389026656878,
      "learning_rate": 1.4675953607061728e-05,
      "loss": 0.9108,
      "step": 11911
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3992214640865184,
      "learning_rate": 1.467507681593635e-05,
      "loss": 0.0641,
      "step": 11912
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3505087734851505,
      "learning_rate": 1.4674199978816032e-05,
      "loss": 0.2947,
      "step": 11913
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3056742355973,
      "learning_rate": 1.4673323095709395e-05,
      "loss": 0.2649,
      "step": 11914
    },
    {
      "epoch": 0.36,
      "grad_norm": 1.6363578405450683,
      "learning_rate": 1.4672446166625067e-05,
      "loss": 0.9234,
      "step": 11915
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.7293241496303122,
      "learning_rate": 1.4671569191571677e-05,
      "loss": 0.3987,
      "step": 11916
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.5445480684245518,
      "learning_rate": 1.4670692170557853e-05,
      "loss": 0.2835,
      "step": 11917
    },
    {
      "epoch": 0.36,
      "grad_norm": 0.3522538522060001,
      "learning_rate": 1.466981510359222e-05,
      "loss": 0.2284,
      "step": 11918
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23229306603405603,
      "learning_rate": 1.4668937990683412e-05,
      "loss": 0.0767,
      "step": 11919
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31829362537328265,
      "learning_rate": 1.4668060831840054e-05,
      "loss": 0.3273,
      "step": 11920
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3716549844972129,
      "learning_rate": 1.4667183627070777e-05,
      "loss": 0.1742,
      "step": 11921
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37706031080816,
      "learning_rate": 1.4666306376384212e-05,
      "loss": 0.3068,
      "step": 11922
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8807544143374573,
      "learning_rate": 1.466542907978899e-05,
      "loss": 0.4485,
      "step": 11923
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7529351846190567,
      "learning_rate": 1.4664551737293742e-05,
      "loss": 0.5409,
      "step": 11924
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3181071180818755,
      "learning_rate": 1.4663674348907098e-05,
      "loss": 0.2769,
      "step": 11925
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.613526182933122,
      "learning_rate": 1.4662796914637693e-05,
      "loss": 0.3464,
      "step": 11926
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3044600566372195,
      "learning_rate": 1.466191943449416e-05,
      "loss": 0.2306,
      "step": 11927
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4870593498495671,
      "learning_rate": 1.4661041908485123e-05,
      "loss": 0.305,
      "step": 11928
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37762016164378265,
      "learning_rate": 1.4660164336619227e-05,
      "loss": 0.1871,
      "step": 11929
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.3604655399673713,
      "learning_rate": 1.46592867189051e-05,
      "loss": 0.9451,
      "step": 11930
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30811608760262377,
      "learning_rate": 1.4658409055351375e-05,
      "loss": 0.1928,
      "step": 11931
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.27270995717143737,
      "learning_rate": 1.4657531345966692e-05,
      "loss": 0.2454,
      "step": 11932
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8388217750857603,
      "learning_rate": 1.4656653590759681e-05,
      "loss": 0.4605,
      "step": 11933
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8049301916180155,
      "learning_rate": 1.465577578973898e-05,
      "loss": 0.3825,
      "step": 11934
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5878943903460806,
      "learning_rate": 1.4654897942913224e-05,
      "loss": 0.3337,
      "step": 11935
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3241452775299569,
      "learning_rate": 1.4654020050291052e-05,
      "loss": 0.2384,
      "step": 11936
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30123082611641083,
      "learning_rate": 1.4653142111881098e-05,
      "loss": 0.2171,
      "step": 11937
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39467857026884867,
      "learning_rate": 1.4652264127691999e-05,
      "loss": 0.2228,
      "step": 11938
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0260719333178883,
      "learning_rate": 1.4651386097732401e-05,
      "loss": 0.6182,
      "step": 11939
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2790570707556313,
      "learning_rate": 1.4650508022010932e-05,
      "loss": 0.1951,
      "step": 11940
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8631976646021173,
      "learning_rate": 1.4649629900536234e-05,
      "loss": 0.636,
      "step": 11941
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8065346690693422,
      "learning_rate": 1.464875173331695e-05,
      "loss": 0.3589,
      "step": 11942
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43247294596419816,
      "learning_rate": 1.4647873520361713e-05,
      "loss": 0.4112,
      "step": 11943
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.32465282090168346,
      "learning_rate": 1.4646995261679169e-05,
      "loss": 0.2085,
      "step": 11944
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.32972497766919495,
      "learning_rate": 1.4646116957277959e-05,
      "loss": 0.2228,
      "step": 11945
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2432557369144228,
      "learning_rate": 1.4645238607166719e-05,
      "loss": 0.6369,
      "step": 11946
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17249040897277768,
      "learning_rate": 1.4644360211354092e-05,
      "loss": 0.074,
      "step": 11947
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.072302178902192,
      "learning_rate": 1.4643481769848723e-05,
      "loss": 0.6371,
      "step": 11948
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2959120394922,
      "learning_rate": 1.4642603282659251e-05,
      "loss": 0.1635,
      "step": 11949
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3469047407148804,
      "learning_rate": 1.4641724749794322e-05,
      "loss": 0.2965,
      "step": 11950
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4121518930689261,
      "learning_rate": 1.4640846171262577e-05,
      "loss": 0.3091,
      "step": 11951
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3597212839384607,
      "learning_rate": 1.463996754707266e-05,
      "loss": 0.9108,
      "step": 11952
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22649526261524638,
      "learning_rate": 1.4639088877233216e-05,
      "loss": 0.0742,
      "step": 11953
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3763731159069657,
      "learning_rate": 1.4638210161752887e-05,
      "loss": 0.2787,
      "step": 11954
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20732409819404,
      "learning_rate": 1.463733140064032e-05,
      "loss": 0.1191,
      "step": 11955
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.32174532142056605,
      "learning_rate": 1.4636452593904164e-05,
      "loss": 0.3008,
      "step": 11956
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8421074367337762,
      "learning_rate": 1.463557374155306e-05,
      "loss": 0.345,
      "step": 11957
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28206902590451916,
      "learning_rate": 1.4634694843595658e-05,
      "loss": 0.2181,
      "step": 11958
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.43288278124176455,
      "learning_rate": 1.4633815900040604e-05,
      "loss": 0.3766,
      "step": 11959
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7215532090378626,
      "learning_rate": 1.463293691089654e-05,
      "loss": 0.4491,
      "step": 11960
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4939810593277541,
      "learning_rate": 1.4632057876172118e-05,
      "loss": 0.3589,
      "step": 11961
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3399682332773648,
      "learning_rate": 1.4631178795875994e-05,
      "loss": 0.2347,
      "step": 11962
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3441258472625264,
      "learning_rate": 1.46302996700168e-05,
      "loss": 0.3071,
      "step": 11963
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.1984361899789571,
      "learning_rate": 1.4629420498603199e-05,
      "loss": 0.0714,
      "step": 11964
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4424738805171979,
      "learning_rate": 1.4628541281643838e-05,
      "loss": 0.26,
      "step": 11965
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7342835909545934,
      "learning_rate": 1.4627662019147361e-05,
      "loss": 0.3408,
      "step": 11966
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40173946970556007,
      "learning_rate": 1.4626782711122422e-05,
      "loss": 0.3098,
      "step": 11967
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28128547209168675,
      "learning_rate": 1.4625903357577672e-05,
      "loss": 0.269,
      "step": 11968
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8479055472383178,
      "learning_rate": 1.4625023958521766e-05,
      "loss": 0.5259,
      "step": 11969
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8064192176320641,
      "learning_rate": 1.4624144513963355e-05,
      "loss": 0.6172,
      "step": 11970
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44861964723434344,
      "learning_rate": 1.4623265023911081e-05,
      "loss": 0.1836,
      "step": 11971
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36738095757399275,
      "learning_rate": 1.4622385488373612e-05,
      "loss": 0.3023,
      "step": 11972
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8446733566804967,
      "learning_rate": 1.4621505907359588e-05,
      "loss": 0.2469,
      "step": 11973
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25864854526569453,
      "learning_rate": 1.4620626280877668e-05,
      "loss": 0.2405,
      "step": 11974
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.551198363721185,
      "learning_rate": 1.461974660893651e-05,
      "loss": 0.2782,
      "step": 11975
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4314293762525294,
      "learning_rate": 1.4618866891544761e-05,
      "loss": 0.3103,
      "step": 11976
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49408649342330535,
      "learning_rate": 1.4617987128711084e-05,
      "loss": 0.2573,
      "step": 11977
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1780168938148765,
      "learning_rate": 1.4617107320444128e-05,
      "loss": 0.7489,
      "step": 11978
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3232978782161678,
      "learning_rate": 1.4616227466752548e-05,
      "loss": 0.2505,
      "step": 11979
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8231487854903445,
      "learning_rate": 1.4615347567645007e-05,
      "loss": 0.4851,
      "step": 11980
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31754656335173975,
      "learning_rate": 1.4614467623130155e-05,
      "loss": 0.2112,
      "step": 11981
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3464127684877754,
      "learning_rate": 1.4613587633216655e-05,
      "loss": 0.76,
      "step": 11982
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3514000377334936,
      "learning_rate": 1.4612707597913164e-05,
      "loss": 0.2233,
      "step": 11983
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6180865466036201,
      "learning_rate": 1.4611827517228331e-05,
      "loss": 0.3633,
      "step": 11984
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3460976833798903,
      "learning_rate": 1.4610947391170826e-05,
      "loss": 0.2818,
      "step": 11985
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21588252791444135,
      "learning_rate": 1.4610067219749304e-05,
      "loss": 0.1962,
      "step": 11986
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5958206395165537,
      "learning_rate": 1.460918700297242e-05,
      "loss": 0.8163,
      "step": 11987
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8860841521276819,
      "learning_rate": 1.4608306740848841e-05,
      "loss": 0.505,
      "step": 11988
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9349888252928616,
      "learning_rate": 1.4607426433387224e-05,
      "loss": 0.6451,
      "step": 11989
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.264670566144349,
      "learning_rate": 1.4606546080596227e-05,
      "loss": 0.1984,
      "step": 11990
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4251176512448173,
      "learning_rate": 1.4605665682484518e-05,
      "loss": 0.3358,
      "step": 11991
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39507943176364246,
      "learning_rate": 1.460478523906075e-05,
      "loss": 0.2447,
      "step": 11992
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5933214303553792,
      "learning_rate": 1.4603904750333591e-05,
      "loss": 0.4639,
      "step": 11993
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22926282615840937,
      "learning_rate": 1.4603024216311701e-05,
      "loss": 0.1435,
      "step": 11994
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2665383377965085,
      "learning_rate": 1.460214363700375e-05,
      "loss": 0.2159,
      "step": 11995
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9707034151193894,
      "learning_rate": 1.460126301241839e-05,
      "loss": 0.3997,
      "step": 11996
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3207743808071934,
      "learning_rate": 1.4600382342564292e-05,
      "loss": 0.2681,
      "step": 11997
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8457750624952198,
      "learning_rate": 1.4599501627450119e-05,
      "loss": 0.4696,
      "step": 11998
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29434757610235096,
      "learning_rate": 1.4598620867084537e-05,
      "loss": 0.2045,
      "step": 11999
    },
    {
      "epoch": 0.37,
      "grad_norm": 2.1124712906860754,
      "learning_rate": 1.4597740061476207e-05,
      "loss": 0.991,
      "step": 12000
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5711915873496111,
      "learning_rate": 1.4596859210633797e-05,
      "loss": 0.363,
      "step": 12001
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.507019721248073,
      "learning_rate": 1.4595978314565977e-05,
      "loss": 0.3744,
      "step": 12002
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2647827223662588,
      "learning_rate": 1.4595097373281408e-05,
      "loss": 0.2145,
      "step": 12003
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5167112727707155,
      "learning_rate": 1.459421638678876e-05,
      "loss": 0.3689,
      "step": 12004
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25604640176065385,
      "learning_rate": 1.4593335355096698e-05,
      "loss": 0.1124,
      "step": 12005
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.29711930595441,
      "learning_rate": 1.4592454278213893e-05,
      "loss": 0.8199,
      "step": 12006
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9048090747885765,
      "learning_rate": 1.4591573156149008e-05,
      "loss": 0.3212,
      "step": 12007
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6924644658683873,
      "learning_rate": 1.459069198891072e-05,
      "loss": 0.3965,
      "step": 12008
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29286318999074507,
      "learning_rate": 1.4589810776507691e-05,
      "loss": 0.2534,
      "step": 12009
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9446865214267239,
      "learning_rate": 1.4588929518948596e-05,
      "loss": 0.2958,
      "step": 12010
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2283080135752724,
      "learning_rate": 1.4588048216242098e-05,
      "loss": 0.775,
      "step": 12011
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.32680654370200407,
      "learning_rate": 1.4587166868396875e-05,
      "loss": 0.0732,
      "step": 12012
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2893514054705375,
      "learning_rate": 1.4586285475421596e-05,
      "loss": 0.2436,
      "step": 12013
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25058205508913817,
      "learning_rate": 1.458540403732493e-05,
      "loss": 0.1558,
      "step": 12014
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3323060132618757,
      "learning_rate": 1.458452255411555e-05,
      "loss": 0.3134,
      "step": 12015
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9937506495508668,
      "learning_rate": 1.4583641025802129e-05,
      "loss": 0.3494,
      "step": 12016
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40624404933297537,
      "learning_rate": 1.458275945239334e-05,
      "loss": 0.3064,
      "step": 12017
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6297644379012203,
      "learning_rate": 1.4581877833897855e-05,
      "loss": 0.2929,
      "step": 12018
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9410767085868392,
      "learning_rate": 1.4580996170324347e-05,
      "loss": 0.5578,
      "step": 12019
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5123829531178065,
      "learning_rate": 1.4580114461681494e-05,
      "loss": 0.2952,
      "step": 12020
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42594043672581844,
      "learning_rate": 1.4579232707977966e-05,
      "loss": 0.2934,
      "step": 12021
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23193042143863266,
      "learning_rate": 1.4578350909222439e-05,
      "loss": 0.1689,
      "step": 12022
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4874902235980257,
      "learning_rate": 1.457746906542359e-05,
      "loss": 0.2597,
      "step": 12023
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0000216507111983,
      "learning_rate": 1.4576587176590096e-05,
      "loss": 0.6445,
      "step": 12024
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7694892612047404,
      "learning_rate": 1.4575705242730629e-05,
      "loss": 0.3791,
      "step": 12025
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5947031752133075,
      "learning_rate": 1.4574823263853868e-05,
      "loss": 0.299,
      "step": 12026
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30701296141736584,
      "learning_rate": 1.4573941239968491e-05,
      "loss": 0.2376,
      "step": 12027
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3922121587203728,
      "learning_rate": 1.457305917108318e-05,
      "loss": 0.3612,
      "step": 12028
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8388005082175364,
      "learning_rate": 1.4572177057206602e-05,
      "loss": 0.3278,
      "step": 12029
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6242572274748623,
      "learning_rate": 1.4571294898347446e-05,
      "loss": 0.3031,
      "step": 12030
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.33141743931674855,
      "learning_rate": 1.4570412694514382e-05,
      "loss": 0.2239,
      "step": 12031
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5084241744627995,
      "learning_rate": 1.4569530445716096e-05,
      "loss": 0.3177,
      "step": 12032
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3030573152209726,
      "learning_rate": 1.4568648151961266e-05,
      "loss": 0.2584,
      "step": 12033
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38209134772300996,
      "learning_rate": 1.4567765813258575e-05,
      "loss": 0.3189,
      "step": 12034
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3557592700133339,
      "learning_rate": 1.4566883429616696e-05,
      "loss": 0.0763,
      "step": 12035
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3298701503458816,
      "learning_rate": 1.4566001001044317e-05,
      "loss": 0.3095,
      "step": 12036
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7058160916605581,
      "learning_rate": 1.4565118527550119e-05,
      "loss": 0.3482,
      "step": 12037
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4293725710021166,
      "learning_rate": 1.4564236009142781e-05,
      "loss": 0.2797,
      "step": 12038
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4709815328937467,
      "learning_rate": 1.4563353445830987e-05,
      "loss": 0.3791,
      "step": 12039
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28903834292455294,
      "learning_rate": 1.4562470837623424e-05,
      "loss": 0.2059,
      "step": 12040
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.092765690611939,
      "learning_rate": 1.456158818452877e-05,
      "loss": 0.7607,
      "step": 12041
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2407710575452408,
      "learning_rate": 1.4560705486555708e-05,
      "loss": 0.1286,
      "step": 12042
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6839571123762571,
      "learning_rate": 1.4559822743712926e-05,
      "loss": 0.4506,
      "step": 12043
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31674957620643773,
      "learning_rate": 1.4558939956009108e-05,
      "loss": 0.2053,
      "step": 12044
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30907510086934054,
      "learning_rate": 1.4558057123452937e-05,
      "loss": 0.2972,
      "step": 12045
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.41802835119512544,
      "learning_rate": 1.4557174246053102e-05,
      "loss": 0.1706,
      "step": 12046
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7176745361777688,
      "learning_rate": 1.4556291323818289e-05,
      "loss": 0.8944,
      "step": 12047
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4198749529045528,
      "learning_rate": 1.4555408356757179e-05,
      "loss": 0.0863,
      "step": 12048
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3479529533138503,
      "learning_rate": 1.4554525344878463e-05,
      "loss": 0.3006,
      "step": 12049
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.029463466385127,
      "learning_rate": 1.4553642288190833e-05,
      "loss": 0.5013,
      "step": 12050
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31502648726731863,
      "learning_rate": 1.4552759186702967e-05,
      "loss": 0.2628,
      "step": 12051
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7328030301624429,
      "learning_rate": 1.4551876040423558e-05,
      "loss": 0.4269,
      "step": 12052
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3240560701975987,
      "learning_rate": 1.4550992849361298e-05,
      "loss": 0.2211,
      "step": 12053
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.587969429039802,
      "learning_rate": 1.455010961352487e-05,
      "loss": 0.3983,
      "step": 12054
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2485533297195684,
      "learning_rate": 1.4549226332922968e-05,
      "loss": 0.1184,
      "step": 12055
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4184364711188768,
      "learning_rate": 1.4548343007564279e-05,
      "loss": 0.3098,
      "step": 12056
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3644113373185899,
      "learning_rate": 1.4547459637457494e-05,
      "loss": 0.2466,
      "step": 12057
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8663770622092954,
      "learning_rate": 1.4546576222611306e-05,
      "loss": 0.4702,
      "step": 12058
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.34388213707663245,
      "learning_rate": 1.4545692763034406e-05,
      "loss": 0.264,
      "step": 12059
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8839578554434355,
      "learning_rate": 1.4544809258735487e-05,
      "loss": 0.4772,
      "step": 12060
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.440946748715764,
      "learning_rate": 1.4543925709723235e-05,
      "loss": 0.3033,
      "step": 12061
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8627120301375781,
      "learning_rate": 1.4543042116006347e-05,
      "loss": 0.4886,
      "step": 12062
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2741633434344071,
      "learning_rate": 1.4542158477593518e-05,
      "loss": 0.2385,
      "step": 12063
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.21191855644420024,
      "learning_rate": 1.4541274794493438e-05,
      "loss": 0.0937,
      "step": 12064
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4249738218808796,
      "learning_rate": 1.4540391066714803e-05,
      "loss": 0.6015,
      "step": 12065
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0101982685573094,
      "learning_rate": 1.4539507294266308e-05,
      "loss": 0.3522,
      "step": 12066
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.38729091354666695,
      "learning_rate": 1.4538623477156644e-05,
      "loss": 0.2665,
      "step": 12067
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3351132470926256,
      "learning_rate": 1.4537739615394508e-05,
      "loss": 0.2604,
      "step": 12068
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.565470885060516,
      "learning_rate": 1.45368557089886e-05,
      "loss": 0.365,
      "step": 12069
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3722508330695646,
      "learning_rate": 1.4535971757947609e-05,
      "loss": 0.4112,
      "step": 12070
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8829779121564438,
      "learning_rate": 1.453508776228024e-05,
      "loss": 0.4905,
      "step": 12071
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25546979102985895,
      "learning_rate": 1.453420372199518e-05,
      "loss": 0.18,
      "step": 12072
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44244364167373434,
      "learning_rate": 1.4533319637101136e-05,
      "loss": 0.2799,
      "step": 12073
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.47881088632698643,
      "learning_rate": 1.4532435507606798e-05,
      "loss": 0.2917,
      "step": 12074
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49782298259657964,
      "learning_rate": 1.453155133352087e-05,
      "loss": 0.3975,
      "step": 12075
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2993266264970152,
      "learning_rate": 1.4530667114852051e-05,
      "loss": 0.2002,
      "step": 12076
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2179980587985277,
      "learning_rate": 1.4529782851609036e-05,
      "loss": 0.5162,
      "step": 12077
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1065192360268477,
      "learning_rate": 1.452889854380053e-05,
      "loss": 0.5067,
      "step": 12078
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8551964387028126,
      "learning_rate": 1.4528014191435227e-05,
      "loss": 0.4377,
      "step": 12079
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4023972176998344,
      "learning_rate": 1.4527129794521829e-05,
      "loss": 0.3496,
      "step": 12080
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.32250050280180426,
      "learning_rate": 1.452624535306904e-05,
      "loss": 0.2413,
      "step": 12081
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30351826491117273,
      "learning_rate": 1.4525360867085558e-05,
      "loss": 0.2553,
      "step": 12082
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2821059231569739,
      "learning_rate": 1.4524476336580089e-05,
      "loss": 0.1213,
      "step": 12083
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9776274460190904,
      "learning_rate": 1.4523591761561332e-05,
      "loss": 0.5756,
      "step": 12084
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29135671521922013,
      "learning_rate": 1.4522707142037993e-05,
      "loss": 0.0735,
      "step": 12085
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46869406956805837,
      "learning_rate": 1.4521822478018772e-05,
      "loss": 0.328,
      "step": 12086
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4652159354865569,
      "learning_rate": 1.4520937769512373e-05,
      "loss": 0.284,
      "step": 12087
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5805417477461472,
      "learning_rate": 1.4520053016527498e-05,
      "loss": 0.8817,
      "step": 12088
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9894583914423091,
      "learning_rate": 1.4519168219072858e-05,
      "loss": 0.2808,
      "step": 12089
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2593484431199836,
      "learning_rate": 1.4518283377157154e-05,
      "loss": 0.1873,
      "step": 12090
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2628195599399107,
      "learning_rate": 1.451739849078909e-05,
      "loss": 0.1488,
      "step": 12091
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3255920405945554,
      "learning_rate": 1.4516513559977374e-05,
      "loss": 0.2475,
      "step": 12092
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2266214774669522,
      "learning_rate": 1.451562858473071e-05,
      "loss": 0.6978,
      "step": 12093
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2874034934114295,
      "learning_rate": 1.4514743565057808e-05,
      "loss": 0.1768,
      "step": 12094
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4629862004461614,
      "learning_rate": 1.451385850096737e-05,
      "loss": 0.366,
      "step": 12095
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6813691809238401,
      "learning_rate": 1.4512973392468111e-05,
      "loss": 0.3932,
      "step": 12096
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.366446412201887,
      "learning_rate": 1.4512088239568734e-05,
      "loss": 0.8699,
      "step": 12097
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3075617628173928,
      "learning_rate": 1.4511203042277946e-05,
      "loss": 0.2429,
      "step": 12098
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3564000752193575,
      "learning_rate": 1.4510317800604461e-05,
      "loss": 0.3003,
      "step": 12099
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2669076054006391,
      "learning_rate": 1.4509432514556982e-05,
      "loss": 0.1179,
      "step": 12100
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46467253932711466,
      "learning_rate": 1.4508547184144224e-05,
      "loss": 0.2641,
      "step": 12101
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.842634861302153,
      "learning_rate": 1.4507661809374896e-05,
      "loss": 0.3752,
      "step": 12102
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31877417315166945,
      "learning_rate": 1.4506776390257709e-05,
      "loss": 0.2176,
      "step": 12103
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3809745315719004,
      "learning_rate": 1.450589092680137e-05,
      "loss": 0.3199,
      "step": 12104
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3540792971152121,
      "learning_rate": 1.4505005419014593e-05,
      "loss": 0.2712,
      "step": 12105
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3640175627254056,
      "learning_rate": 1.4504119866906093e-05,
      "loss": 0.8839,
      "step": 12106
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3795770361234952,
      "learning_rate": 1.4503234270484576e-05,
      "loss": 0.1816,
      "step": 12107
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5889516271803689,
      "learning_rate": 1.450234862975876e-05,
      "loss": 0.3678,
      "step": 12108
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28717816919945244,
      "learning_rate": 1.4501462944737359e-05,
      "loss": 0.1852,
      "step": 12109
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2431986990257214,
      "learning_rate": 1.4500577215429083e-05,
      "loss": 0.2488,
      "step": 12110
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5620130979379941,
      "learning_rate": 1.4499691441842648e-05,
      "loss": 0.3704,
      "step": 12111
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9729980091898107,
      "learning_rate": 1.449880562398677e-05,
      "loss": 0.5578,
      "step": 12112
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3189499295619074,
      "learning_rate": 1.4497919761870157e-05,
      "loss": 0.2421,
      "step": 12113
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.470342747911913,
      "learning_rate": 1.449703385550153e-05,
      "loss": 0.8166,
      "step": 12114
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3624504948681153,
      "learning_rate": 1.4496147904889606e-05,
      "loss": 0.2694,
      "step": 12115
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3634005459628244,
      "learning_rate": 1.4495261910043101e-05,
      "loss": 0.2288,
      "step": 12116
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3920640473065246,
      "learning_rate": 1.449437587097073e-05,
      "loss": 0.3202,
      "step": 12117
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28642523675356524,
      "learning_rate": 1.4493489787681209e-05,
      "loss": 0.1732,
      "step": 12118
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.580092683534525,
      "learning_rate": 1.4492603660183255e-05,
      "loss": 0.2839,
      "step": 12119
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7299655226065855,
      "learning_rate": 1.449171748848559e-05,
      "loss": 0.337,
      "step": 12120
    },
    {
      "epoch": 0.37,
      "grad_norm": 3.252841849558108,
      "learning_rate": 1.449083127259693e-05,
      "loss": 0.3536,
      "step": 12121
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.34896419144020074,
      "learning_rate": 1.4489945012525996e-05,
      "loss": 0.2378,
      "step": 12122
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.48331662785866303,
      "learning_rate": 1.4489058708281505e-05,
      "loss": 0.3762,
      "step": 12123
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9042704969200914,
      "learning_rate": 1.4488172359872174e-05,
      "loss": 0.4869,
      "step": 12124
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9013320378796896,
      "learning_rate": 1.448728596730673e-05,
      "loss": 0.548,
      "step": 12125
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29813795555539907,
      "learning_rate": 1.448639953059389e-05,
      "loss": 0.2011,
      "step": 12126
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4603694075245554,
      "learning_rate": 1.4485513049742372e-05,
      "loss": 0.269,
      "step": 12127
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3205201167753154,
      "learning_rate": 1.4484626524760904e-05,
      "loss": 0.2415,
      "step": 12128
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7897096999166621,
      "learning_rate": 1.4483739955658204e-05,
      "loss": 0.3749,
      "step": 12129
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3072110858879564,
      "learning_rate": 1.4482853342442993e-05,
      "loss": 0.1767,
      "step": 12130
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36804428442827253,
      "learning_rate": 1.4481966685124e-05,
      "loss": 0.235,
      "step": 12131
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5186308629507776,
      "learning_rate": 1.448107998370994e-05,
      "loss": 0.8784,
      "step": 12132
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3757836813021153,
      "learning_rate": 1.448019323820954e-05,
      "loss": 0.2641,
      "step": 12133
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.453274859353632,
      "learning_rate": 1.447930644863153e-05,
      "loss": 0.3831,
      "step": 12134
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.24566122672562074,
      "learning_rate": 1.4478419614984626e-05,
      "loss": 0.0744,
      "step": 12135
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39275788379141463,
      "learning_rate": 1.4477532737277556e-05,
      "loss": 0.3382,
      "step": 12136
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.093466040282311,
      "learning_rate": 1.4476645815519043e-05,
      "loss": 0.3731,
      "step": 12137
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42383239077037355,
      "learning_rate": 1.4475758849717821e-05,
      "loss": 0.2914,
      "step": 12138
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.17947494716380813,
      "learning_rate": 1.4474871839882605e-05,
      "loss": 0.0728,
      "step": 12139
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36564026355390766,
      "learning_rate": 1.4473984786022133e-05,
      "loss": 0.3281,
      "step": 12140
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3240438248324874,
      "learning_rate": 1.4473097688145123e-05,
      "loss": 0.2711,
      "step": 12141
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.832675406144498,
      "learning_rate": 1.4472210546260309e-05,
      "loss": 0.4474,
      "step": 12142
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2206480695477877,
      "learning_rate": 1.4471323360376413e-05,
      "loss": 0.7857,
      "step": 12143
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2883592313830199,
      "learning_rate": 1.4470436130502168e-05,
      "loss": 0.1917,
      "step": 12144
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5346650859204898,
      "learning_rate": 1.4469548856646301e-05,
      "loss": 0.3516,
      "step": 12145
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30585323075657633,
      "learning_rate": 1.4468661538817542e-05,
      "loss": 0.2734,
      "step": 12146
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8234414172046728,
      "learning_rate": 1.4467774177024619e-05,
      "loss": 0.5528,
      "step": 12147
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.228575531505799,
      "learning_rate": 1.4466886771276266e-05,
      "loss": 0.0712,
      "step": 12148
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3386081247095178,
      "learning_rate": 1.446599932158121e-05,
      "loss": 0.2977,
      "step": 12149
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9563545597883418,
      "learning_rate": 1.4465111827948182e-05,
      "loss": 0.63,
      "step": 12150
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.47814280892703,
      "learning_rate": 1.4464224290385915e-05,
      "loss": 0.6783,
      "step": 12151
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29310832291535494,
      "learning_rate": 1.4463336708903142e-05,
      "loss": 0.2475,
      "step": 12152
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40349211971907617,
      "learning_rate": 1.4462449083508592e-05,
      "loss": 0.3201,
      "step": 12153
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.46862927358680473,
      "learning_rate": 1.4461561414211002e-05,
      "loss": 0.3051,
      "step": 12154
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6396599805021889,
      "learning_rate": 1.4460673701019102e-05,
      "loss": 0.36,
      "step": 12155
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1144900362864003,
      "learning_rate": 1.4459785943941622e-05,
      "loss": 0.6389,
      "step": 12156
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.23014903619321883,
      "learning_rate": 1.4458898142987306e-05,
      "loss": 0.1556,
      "step": 12157
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3743366592910914,
      "learning_rate": 1.445801029816488e-05,
      "loss": 0.3023,
      "step": 12158
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40336692146108233,
      "learning_rate": 1.4457122409483082e-05,
      "loss": 0.2491,
      "step": 12159
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3939986316865671,
      "learning_rate": 1.445623447695065e-05,
      "loss": 0.2676,
      "step": 12160
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6978027299729588,
      "learning_rate": 1.4455346500576315e-05,
      "loss": 0.3617,
      "step": 12161
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6858010597181478,
      "learning_rate": 1.4454458480368814e-05,
      "loss": 0.4709,
      "step": 12162
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.36926499703004656,
      "learning_rate": 1.4453570416336886e-05,
      "loss": 0.2531,
      "step": 12163
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3459205525156701,
      "learning_rate": 1.4452682308489268e-05,
      "loss": 0.3281,
      "step": 12164
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.04763665535072,
      "learning_rate": 1.4451794156834692e-05,
      "loss": 0.2554,
      "step": 12165
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40491934856335104,
      "learning_rate": 1.4450905961381905e-05,
      "loss": 0.2422,
      "step": 12166
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.28441932868536374,
      "learning_rate": 1.445001772213964e-05,
      "loss": 0.1648,
      "step": 12167
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4408980115301862,
      "learning_rate": 1.4449129439116633e-05,
      "loss": 0.2944,
      "step": 12168
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.27298265388365606,
      "learning_rate": 1.4448241112321628e-05,
      "loss": 0.2505,
      "step": 12169
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2510743577291776,
      "learning_rate": 1.4447352741763366e-05,
      "loss": 0.3383,
      "step": 12170
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4922279565664179,
      "learning_rate": 1.4446464327450582e-05,
      "loss": 0.3004,
      "step": 12171
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3198262287275139,
      "learning_rate": 1.444557586939202e-05,
      "loss": 0.2691,
      "step": 12172
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8215734481505015,
      "learning_rate": 1.4444687367596417e-05,
      "loss": 0.4802,
      "step": 12173
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.44648632671021893,
      "learning_rate": 1.4443798822072521e-05,
      "loss": 0.0251,
      "step": 12174
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.34623568603745586,
      "learning_rate": 1.4442910232829067e-05,
      "loss": 0.3239,
      "step": 12175
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30931770874407644,
      "learning_rate": 1.44420215998748e-05,
      "loss": 0.1988,
      "step": 12176
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.5064788678285077,
      "learning_rate": 1.4441132923218466e-05,
      "loss": 0.8758,
      "step": 12177
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.240036679182883,
      "learning_rate": 1.4440244202868802e-05,
      "loss": 0.1043,
      "step": 12178
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.40630113732682466,
      "learning_rate": 1.4439355438834558e-05,
      "loss": 0.3006,
      "step": 12179
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.437314981445368,
      "learning_rate": 1.4438466631124472e-05,
      "loss": 0.249,
      "step": 12180
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.27945437369485526,
      "learning_rate": 1.4437577779747292e-05,
      "loss": 0.2341,
      "step": 12181
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5350314269773827,
      "learning_rate": 1.4436688884711759e-05,
      "loss": 0.3828,
      "step": 12182
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.2632685218989548,
      "learning_rate": 1.4435799946026623e-05,
      "loss": 0.3464,
      "step": 12183
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8613815403080647,
      "learning_rate": 1.4434910963700626e-05,
      "loss": 0.4551,
      "step": 12184
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3050258811560412,
      "learning_rate": 1.4434021937742517e-05,
      "loss": 0.2105,
      "step": 12185
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.6085440899463601,
      "learning_rate": 1.4433132868161042e-05,
      "loss": 0.8375,
      "step": 12186
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3089821442538929,
      "learning_rate": 1.4432243754964948e-05,
      "loss": 0.2529,
      "step": 12187
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7337166234698566,
      "learning_rate": 1.443135459816298e-05,
      "loss": 0.4883,
      "step": 12188
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4012546556345495,
      "learning_rate": 1.4430465397763885e-05,
      "loss": 0.2066,
      "step": 12189
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3750528481845103,
      "learning_rate": 1.4429576153776417e-05,
      "loss": 0.3343,
      "step": 12190
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25145887671487926,
      "learning_rate": 1.442868686620932e-05,
      "loss": 0.1167,
      "step": 12191
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.416392827914895,
      "learning_rate": 1.4427797535071344e-05,
      "loss": 0.7166,
      "step": 12192
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29495086811941423,
      "learning_rate": 1.442690816037124e-05,
      "loss": 0.2637,
      "step": 12193
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3209278617479359,
      "learning_rate": 1.4426018742117754e-05,
      "loss": 0.1711,
      "step": 12194
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5684716171558487,
      "learning_rate": 1.4425129280319641e-05,
      "loss": 0.3959,
      "step": 12195
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7641155752752227,
      "learning_rate": 1.4424239774985651e-05,
      "loss": 0.3901,
      "step": 12196
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.0190468491773235,
      "learning_rate": 1.4423350226124534e-05,
      "loss": 0.566,
      "step": 12197
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.20581921264710462,
      "learning_rate": 1.4422460633745044e-05,
      "loss": 0.1312,
      "step": 12198
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.34646160591707137,
      "learning_rate": 1.4421570997855927e-05,
      "loss": 0.2751,
      "step": 12199
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3353846668742128,
      "learning_rate": 1.4420681318465942e-05,
      "loss": 0.2521,
      "step": 12200
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3116023135493975,
      "learning_rate": 1.4419791595583838e-05,
      "loss": 0.6891,
      "step": 12201
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8674380694929857,
      "learning_rate": 1.441890182921837e-05,
      "loss": 0.4272,
      "step": 12202
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3826856390675236,
      "learning_rate": 1.4418012019378294e-05,
      "loss": 0.2525,
      "step": 12203
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.49671834064260906,
      "learning_rate": 1.4417122166072361e-05,
      "loss": 0.3183,
      "step": 12204
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9941536839150453,
      "learning_rate": 1.4416232269309326e-05,
      "loss": 0.5478,
      "step": 12205
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30927555237537196,
      "learning_rate": 1.4415342329097947e-05,
      "loss": 0.2945,
      "step": 12206
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.8239884214417842,
      "learning_rate": 1.4414452345446974e-05,
      "loss": 0.4924,
      "step": 12207
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.22390232021773088,
      "learning_rate": 1.4413562318365168e-05,
      "loss": 0.1652,
      "step": 12208
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.27720053843965126,
      "learning_rate": 1.4412672247861285e-05,
      "loss": 0.1201,
      "step": 12209
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.9111807143732114,
      "learning_rate": 1.4411782133944078e-05,
      "loss": 0.821,
      "step": 12210
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30882254418975297,
      "learning_rate": 1.441089197662231e-05,
      "loss": 0.2656,
      "step": 12211
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3830982104214318,
      "learning_rate": 1.4410001775904731e-05,
      "loss": 0.2734,
      "step": 12212
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42278868247659485,
      "learning_rate": 1.440911153180011e-05,
      "loss": 0.2384,
      "step": 12213
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7151531457877869,
      "learning_rate": 1.4408221244317192e-05,
      "loss": 0.4663,
      "step": 12214
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7789080051534133,
      "learning_rate": 1.4407330913464744e-05,
      "loss": 0.3953,
      "step": 12215
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.35490172114470625,
      "learning_rate": 1.4406440539251528e-05,
      "loss": 0.1874,
      "step": 12216
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3267202967892213,
      "learning_rate": 1.4405550121686298e-05,
      "loss": 0.2102,
      "step": 12217
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.29353689273934847,
      "learning_rate": 1.4404659660777816e-05,
      "loss": 0.2518,
      "step": 12218
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7244550570349755,
      "learning_rate": 1.4403769156534844e-05,
      "loss": 0.3492,
      "step": 12219
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.4758066340965157,
      "learning_rate": 1.4402878608966137e-05,
      "loss": 0.8671,
      "step": 12220
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31879871111112523,
      "learning_rate": 1.4401988018080466e-05,
      "loss": 0.1228,
      "step": 12221
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37286030605485976,
      "learning_rate": 1.4401097383886589e-05,
      "loss": 0.2936,
      "step": 12222
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.32892278425690086,
      "learning_rate": 1.440020670639327e-05,
      "loss": 0.3043,
      "step": 12223
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.6800245096672762,
      "learning_rate": 1.4399315985609264e-05,
      "loss": 0.4137,
      "step": 12224
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7101019832339066,
      "learning_rate": 1.4398425221543342e-05,
      "loss": 0.8265,
      "step": 12225
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.31288406034940697,
      "learning_rate": 1.4397534414204267e-05,
      "loss": 0.1878,
      "step": 12226
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.33988537749204234,
      "learning_rate": 1.4396643563600798e-05,
      "loss": 0.2199,
      "step": 12227
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1099832541282437,
      "learning_rate": 1.4395752669741705e-05,
      "loss": 0.0932,
      "step": 12228
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.37037295215492233,
      "learning_rate": 1.4394861732635752e-05,
      "loss": 0.3311,
      "step": 12229
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.42696139872302824,
      "learning_rate": 1.4393970752291702e-05,
      "loss": 0.2396,
      "step": 12230
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.39122323610573734,
      "learning_rate": 1.439307972871832e-05,
      "loss": 0.3219,
      "step": 12231
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5637949524460265,
      "learning_rate": 1.4392188661924378e-05,
      "loss": 0.3481,
      "step": 12232
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.1763852568416424,
      "learning_rate": 1.4391297551918634e-05,
      "loss": 0.7663,
      "step": 12233
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.3572103951429533,
      "learning_rate": 1.4390406398709864e-05,
      "loss": 0.2053,
      "step": 12234
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.25560442738628586,
      "learning_rate": 1.4389515202306829e-05,
      "loss": 0.2185,
      "step": 12235
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.3551798899471046,
      "learning_rate": 1.4388623962718303e-05,
      "loss": 0.7333,
      "step": 12236
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2617852523064891,
      "learning_rate": 1.4387732679953046e-05,
      "loss": 0.1181,
      "step": 12237
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.7505751871522728,
      "learning_rate": 1.4386841354019829e-05,
      "loss": 0.4997,
      "step": 12238
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.30940154332192554,
      "learning_rate": 1.438594998492743e-05,
      "loss": 0.2066,
      "step": 12239
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.5585361273242112,
      "learning_rate": 1.4385058572684607e-05,
      "loss": 0.3449,
      "step": 12240
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.2794555898795038,
      "learning_rate": 1.4384167117300138e-05,
      "loss": 0.2403,
      "step": 12241
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.160068134595765,
      "learning_rate": 1.4383275618782787e-05,
      "loss": 0.7059,
      "step": 12242
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.4299509419543013,
      "learning_rate": 1.4382384077141331e-05,
      "loss": 0.0739,
      "step": 12243
    },
    {
      "epoch": 0.37,
      "grad_norm": 0.9114679554407511,
      "learning_rate": 1.4381492492384537e-05,
      "loss": 0.3772,
      "step": 12244
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2235550125021566,
      "learning_rate": 1.438060086452118e-05,
      "loss": 0.1592,
      "step": 12245
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3123965410656109,
      "learning_rate": 1.4379709193560027e-05,
      "loss": 0.564,
      "step": 12246
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3319361846992417,
      "learning_rate": 1.4378817479509854e-05,
      "loss": 0.2444,
      "step": 12247
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6208080888523707,
      "learning_rate": 1.4377925722379439e-05,
      "loss": 0.2092,
      "step": 12248
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43018604359646134,
      "learning_rate": 1.4377033922177546e-05,
      "loss": 0.3152,
      "step": 12249
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7892927700927722,
      "learning_rate": 1.4376142078912953e-05,
      "loss": 0.435,
      "step": 12250
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.290825265480358,
      "learning_rate": 1.4375250192594434e-05,
      "loss": 0.7001,
      "step": 12251
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3104419485537175,
      "learning_rate": 1.4374358263230765e-05,
      "loss": 0.2494,
      "step": 12252
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4232199041152395,
      "learning_rate": 1.437346629083072e-05,
      "loss": 0.2759,
      "step": 12253
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2707887469401094,
      "learning_rate": 1.4372574275403075e-05,
      "loss": 0.1593,
      "step": 12254
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5223872874704831,
      "learning_rate": 1.4371682216956606e-05,
      "loss": 0.2622,
      "step": 12255
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.594302178534821,
      "learning_rate": 1.4370790115500089e-05,
      "loss": 0.3519,
      "step": 12256
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6884577519381838,
      "learning_rate": 1.4369897971042297e-05,
      "loss": 0.4381,
      "step": 12257
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30981956518961706,
      "learning_rate": 1.4369005783592014e-05,
      "loss": 0.2169,
      "step": 12258
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5052176439081429,
      "learning_rate": 1.4368113553158013e-05,
      "loss": 0.3805,
      "step": 12259
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4148848918635713,
      "learning_rate": 1.4367221279749073e-05,
      "loss": 0.274,
      "step": 12260
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.575390339714265,
      "learning_rate": 1.4366328963373976e-05,
      "loss": 0.1793,
      "step": 12261
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35421877902557924,
      "learning_rate": 1.4365436604041499e-05,
      "loss": 0.3001,
      "step": 12262
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.284222681727259,
      "learning_rate": 1.4364544201760415e-05,
      "loss": 0.0743,
      "step": 12263
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5081156612134122,
      "learning_rate": 1.4363651756539512e-05,
      "loss": 0.3012,
      "step": 12264
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3698777937038861,
      "learning_rate": 1.4362759268387567e-05,
      "loss": 0.2671,
      "step": 12265
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6068372044790932,
      "learning_rate": 1.4361866737313362e-05,
      "loss": 0.3733,
      "step": 12266
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30194813369528933,
      "learning_rate": 1.4360974163325676e-05,
      "loss": 0.2183,
      "step": 12267
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.143587772433085,
      "learning_rate": 1.4360081546433292e-05,
      "loss": 0.697,
      "step": 12268
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.896630302260747,
      "learning_rate": 1.4359188886644989e-05,
      "loss": 0.4409,
      "step": 12269
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.33846484840890895,
      "learning_rate": 1.435829618396955e-05,
      "loss": 0.2719,
      "step": 12270
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3613769852182497,
      "learning_rate": 1.4357403438415762e-05,
      "loss": 0.0725,
      "step": 12271
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3879453563773257,
      "learning_rate": 1.4356510649992405e-05,
      "loss": 0.3169,
      "step": 12272
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38052744096599367,
      "learning_rate": 1.4355617818708262e-05,
      "loss": 0.2312,
      "step": 12273
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38783605721343783,
      "learning_rate": 1.4354724944572119e-05,
      "loss": 0.2296,
      "step": 12274
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.86639041344971,
      "learning_rate": 1.435383202759276e-05,
      "loss": 0.8869,
      "step": 12275
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2923856532479673,
      "learning_rate": 1.4352939067778966e-05,
      "loss": 0.1933,
      "step": 12276
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3389367564836866,
      "learning_rate": 1.4352046065139524e-05,
      "loss": 0.3268,
      "step": 12277
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0185566130302461,
      "learning_rate": 1.4351153019683223e-05,
      "loss": 0.5134,
      "step": 12278
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9278890441052403,
      "learning_rate": 1.4350259931418849e-05,
      "loss": 0.8522,
      "step": 12279
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.318991697719373,
      "learning_rate": 1.4349366800355183e-05,
      "loss": 0.1751,
      "step": 12280
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4622899176882167,
      "learning_rate": 1.4348473626501019e-05,
      "loss": 0.3738,
      "step": 12281
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4068168000157794,
      "learning_rate": 1.4347580409865137e-05,
      "loss": 0.3013,
      "step": 12282
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4479140647917669,
      "learning_rate": 1.4346687150456328e-05,
      "loss": 0.3767,
      "step": 12283
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1913209207839169,
      "learning_rate": 1.4345793848283383e-05,
      "loss": 0.0928,
      "step": 12284
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35968971340236844,
      "learning_rate": 1.4344900503355086e-05,
      "loss": 0.2859,
      "step": 12285
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9825949422807008,
      "learning_rate": 1.434400711568023e-05,
      "loss": 0.5165,
      "step": 12286
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8083253415466765,
      "learning_rate": 1.4343113685267602e-05,
      "loss": 0.4437,
      "step": 12287
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32112175542589383,
      "learning_rate": 1.4342220212125994e-05,
      "loss": 0.3067,
      "step": 12288
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3187368887392385,
      "learning_rate": 1.4341326696264193e-05,
      "loss": 0.1969,
      "step": 12289
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5250078869290855,
      "learning_rate": 1.434043313769099e-05,
      "loss": 0.4075,
      "step": 12290
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9036618944228383,
      "learning_rate": 1.4339539536415182e-05,
      "loss": 0.367,
      "step": 12291
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3908012899682595,
      "learning_rate": 1.4338645892445553e-05,
      "loss": 0.2394,
      "step": 12292
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20343910347642258,
      "learning_rate": 1.4337752205790899e-05,
      "loss": 0.0924,
      "step": 12293
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5376948244985219,
      "learning_rate": 1.4336858476460012e-05,
      "loss": 0.3906,
      "step": 12294
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2671448790948922,
      "learning_rate": 1.433596470446168e-05,
      "loss": 0.2468,
      "step": 12295
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3730748122555414,
      "learning_rate": 1.4335070889804703e-05,
      "loss": 0.7727,
      "step": 12296
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8000458353837678,
      "learning_rate": 1.4334177032497875e-05,
      "loss": 0.3619,
      "step": 12297
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0939913708879359,
      "learning_rate": 1.4333283132549984e-05,
      "loss": 0.4845,
      "step": 12298
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.33828182510346877,
      "learning_rate": 1.433238918996983e-05,
      "loss": 0.2427,
      "step": 12299
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2979662330033811,
      "learning_rate": 1.4331495204766203e-05,
      "loss": 0.272,
      "step": 12300
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0982698947262117,
      "learning_rate": 1.4330601176947904e-05,
      "loss": 0.6442,
      "step": 12301
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2633458239201825,
      "learning_rate": 1.4329707106523724e-05,
      "loss": 0.1113,
      "step": 12302
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.416571840572812,
      "learning_rate": 1.432881299350246e-05,
      "loss": 0.2908,
      "step": 12303
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4050214255957136,
      "learning_rate": 1.4327918837892911e-05,
      "loss": 0.2907,
      "step": 12304
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40248135120156203,
      "learning_rate": 1.4327024639703875e-05,
      "loss": 0.2162,
      "step": 12305
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3088652805088555,
      "learning_rate": 1.4326130398944143e-05,
      "loss": 0.2681,
      "step": 12306
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7806095668723304,
      "learning_rate": 1.4325236115622518e-05,
      "loss": 0.3961,
      "step": 12307
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3039517841040105,
      "learning_rate": 1.4324341789747796e-05,
      "loss": 0.2374,
      "step": 12308
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8819820897158963,
      "learning_rate": 1.4323447421328775e-05,
      "loss": 0.5441,
      "step": 12309
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8526702817154896,
      "learning_rate": 1.4322553010374258e-05,
      "loss": 0.0886,
      "step": 12310
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3257446832724355,
      "learning_rate": 1.4321658556893043e-05,
      "loss": 0.2791,
      "step": 12311
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.29697574409834854,
      "learning_rate": 1.4320764060893929e-05,
      "loss": 0.2003,
      "step": 12312
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3818825338007789,
      "learning_rate": 1.4319869522385712e-05,
      "loss": 0.2302,
      "step": 12313
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45756278145197027,
      "learning_rate": 1.4318974941377202e-05,
      "loss": 0.3019,
      "step": 12314
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0528595530814644,
      "learning_rate": 1.4318080317877192e-05,
      "loss": 0.3955,
      "step": 12315
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5547625495254126,
      "learning_rate": 1.4317185651894488e-05,
      "loss": 0.3237,
      "step": 12316
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.36260543670459455,
      "learning_rate": 1.4316290943437893e-05,
      "loss": 0.2349,
      "step": 12317
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46189860822907064,
      "learning_rate": 1.4315396192516207e-05,
      "loss": 0.3754,
      "step": 12318
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4362585515023963,
      "learning_rate": 1.4314501399138233e-05,
      "loss": 0.2723,
      "step": 12319
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1952500691895582,
      "learning_rate": 1.4313606563312774e-05,
      "loss": 0.7228,
      "step": 12320
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3667336798531295,
      "learning_rate": 1.4312711685048633e-05,
      "loss": 0.0751,
      "step": 12321
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3869946233598142,
      "learning_rate": 1.4311816764354616e-05,
      "loss": 0.3075,
      "step": 12322
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2973702485747252,
      "learning_rate": 1.431092180123953e-05,
      "loss": 0.1553,
      "step": 12323
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3657602171277828,
      "learning_rate": 1.4310026795712174e-05,
      "loss": 0.3259,
      "step": 12324
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6782456154352209,
      "learning_rate": 1.4309131747781357e-05,
      "loss": 0.2617,
      "step": 12325
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3004879196133945,
      "learning_rate": 1.4308236657455882e-05,
      "loss": 0.2555,
      "step": 12326
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3267617687770743,
      "learning_rate": 1.430734152474456e-05,
      "loss": 0.757,
      "step": 12327
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7723768377959713,
      "learning_rate": 1.4306446349656193e-05,
      "loss": 0.3173,
      "step": 12328
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.48773416973574524,
      "learning_rate": 1.4305551132199588e-05,
      "loss": 0.3463,
      "step": 12329
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2863955391533219,
      "learning_rate": 1.430465587238356e-05,
      "loss": 0.2115,
      "step": 12330
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5493478023540824,
      "learning_rate": 1.4303760570216909e-05,
      "loss": 0.3946,
      "step": 12331
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6433438023087971,
      "learning_rate": 1.4302865225708442e-05,
      "loss": 0.3743,
      "step": 12332
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8447678933243183,
      "learning_rate": 1.4301969838866975e-05,
      "loss": 0.562,
      "step": 12333
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.23127999172044836,
      "learning_rate": 1.430107440970131e-05,
      "loss": 0.1254,
      "step": 12334
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31218404335337396,
      "learning_rate": 1.4300178938220261e-05,
      "loss": 0.2617,
      "step": 12335
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3004409648747883,
      "learning_rate": 1.4299283424432637e-05,
      "loss": 0.2308,
      "step": 12336
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.053275158318042,
      "learning_rate": 1.429838786834725e-05,
      "loss": 0.6874,
      "step": 12337
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0744517279842065,
      "learning_rate": 1.4297492269972905e-05,
      "loss": 0.4512,
      "step": 12338
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31849150456182024,
      "learning_rate": 1.4296596629318417e-05,
      "loss": 0.192,
      "step": 12339
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.645018032803306,
      "learning_rate": 1.4295700946392601e-05,
      "loss": 0.3604,
      "step": 12340
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9311072206925584,
      "learning_rate": 1.4294805221204265e-05,
      "loss": 0.3734,
      "step": 12341
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.34644666301307087,
      "learning_rate": 1.429390945376222e-05,
      "loss": 0.3327,
      "step": 12342
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21627192017023766,
      "learning_rate": 1.4293013644075284e-05,
      "loss": 0.0729,
      "step": 12343
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3548735508440291,
      "learning_rate": 1.4292117792152263e-05,
      "loss": 0.3015,
      "step": 12344
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37852743243818715,
      "learning_rate": 1.4291221898001976e-05,
      "loss": 0.1179,
      "step": 12345
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2292368614898075,
      "learning_rate": 1.4290325961633238e-05,
      "loss": 0.6965,
      "step": 12346
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32294328116305504,
      "learning_rate": 1.4289429983054858e-05,
      "loss": 0.2669,
      "step": 12347
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9132000128558583,
      "learning_rate": 1.4288533962275657e-05,
      "loss": 0.4483,
      "step": 12348
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2951433422169853,
      "learning_rate": 1.4287637899304448e-05,
      "loss": 0.2192,
      "step": 12349
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8663680788216633,
      "learning_rate": 1.4286741794150048e-05,
      "loss": 0.5491,
      "step": 12350
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.693213542247376,
      "learning_rate": 1.4285845646821268e-05,
      "loss": 0.3583,
      "step": 12351
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.1826254822475498,
      "learning_rate": 1.4284949457326929e-05,
      "loss": 0.0758,
      "step": 12352
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3646200279842673,
      "learning_rate": 1.428405322567585e-05,
      "loss": 0.2806,
      "step": 12353
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2689335426067597,
      "learning_rate": 1.4283156951876844e-05,
      "loss": 0.2505,
      "step": 12354
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9832496395697155,
      "learning_rate": 1.4282260635938731e-05,
      "loss": 0.5615,
      "step": 12355
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9007669308472902,
      "learning_rate": 1.4281364277870329e-05,
      "loss": 0.4941,
      "step": 12356
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5764391432783124,
      "learning_rate": 1.4280467877680456e-05,
      "loss": 0.3018,
      "step": 12357
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3725664586471827,
      "learning_rate": 1.427957143537793e-05,
      "loss": 0.2838,
      "step": 12358
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4806686376356404,
      "learning_rate": 1.4278674950971574e-05,
      "loss": 0.3835,
      "step": 12359
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35979971114712445,
      "learning_rate": 1.4277778424470206e-05,
      "loss": 0.246,
      "step": 12360
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.44873383500377617,
      "learning_rate": 1.4276881855882648e-05,
      "loss": 0.2505,
      "step": 12361
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2667084909618694,
      "learning_rate": 1.4275985245217715e-05,
      "loss": 0.1958,
      "step": 12362
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4348595362310943,
      "learning_rate": 1.4275088592484234e-05,
      "loss": 0.243,
      "step": 12363
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1605767867198593,
      "learning_rate": 1.4274191897691023e-05,
      "loss": 0.2967,
      "step": 12364
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32745473021137184,
      "learning_rate": 1.4273295160846907e-05,
      "loss": 0.2712,
      "step": 12365
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5603492759223798,
      "learning_rate": 1.4272398381960709e-05,
      "loss": 0.3116,
      "step": 12366
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4114928728722867,
      "learning_rate": 1.4271501561041248e-05,
      "loss": 0.2476,
      "step": 12367
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.714751069852233,
      "learning_rate": 1.427060469809735e-05,
      "loss": 0.5019,
      "step": 12368
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9724767924588175,
      "learning_rate": 1.4269707793137837e-05,
      "loss": 0.4379,
      "step": 12369
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30277656080633125,
      "learning_rate": 1.4268810846171535e-05,
      "loss": 0.2555,
      "step": 12370
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30504901886256375,
      "learning_rate": 1.4267913857207265e-05,
      "loss": 0.1702,
      "step": 12371
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3184829056458791,
      "learning_rate": 1.4267016826253856e-05,
      "loss": 0.2873,
      "step": 12372
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4007333843234362,
      "learning_rate": 1.4266119753320132e-05,
      "loss": 0.1409,
      "step": 12373
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0222841229445554,
      "learning_rate": 1.4265222638414918e-05,
      "loss": 0.5696,
      "step": 12374
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6276468351347492,
      "learning_rate": 1.4264325481547036e-05,
      "loss": 0.25,
      "step": 12375
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37305170407705723,
      "learning_rate": 1.4263428282725321e-05,
      "loss": 0.296,
      "step": 12376
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40233579676536974,
      "learning_rate": 1.4262531041958596e-05,
      "loss": 0.3193,
      "step": 12377
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42205217057010924,
      "learning_rate": 1.4261633759255685e-05,
      "loss": 0.2659,
      "step": 12378
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9516174099363344,
      "learning_rate": 1.4260736434625422e-05,
      "loss": 0.8004,
      "step": 12379
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.29594083348458594,
      "learning_rate": 1.4259839068076628e-05,
      "loss": 0.1591,
      "step": 12380
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24092001596968599,
      "learning_rate": 1.4258941659618141e-05,
      "loss": 0.1651,
      "step": 12381
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7509316541652916,
      "learning_rate": 1.4258044209258783e-05,
      "loss": 0.3746,
      "step": 12382
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4734885650259584,
      "learning_rate": 1.4257146717007385e-05,
      "loss": 0.383,
      "step": 12383
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35745375106491484,
      "learning_rate": 1.4256249182872772e-05,
      "loss": 0.2775,
      "step": 12384
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3357286202168292,
      "learning_rate": 1.4255351606863785e-05,
      "loss": 0.3081,
      "step": 12385
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6016490807971397,
      "learning_rate": 1.4254453988989246e-05,
      "loss": 0.4211,
      "step": 12386
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3509997453080858,
      "learning_rate": 1.425355632925799e-05,
      "loss": 0.6846,
      "step": 12387
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25671371664146453,
      "learning_rate": 1.4252658627678844e-05,
      "loss": 0.0624,
      "step": 12388
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38267254050029814,
      "learning_rate": 1.4251760884260649e-05,
      "loss": 0.2804,
      "step": 12389
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22937750697464637,
      "learning_rate": 1.4250863099012224e-05,
      "loss": 0.2031,
      "step": 12390
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.30833439588897,
      "learning_rate": 1.4249965271942415e-05,
      "loss": 0.6529,
      "step": 12391
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7579610900923054,
      "learning_rate": 1.4249067403060045e-05,
      "loss": 0.3626,
      "step": 12392
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3730509675231246,
      "learning_rate": 1.4248169492373958e-05,
      "loss": 0.2242,
      "step": 12393
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3559981114824219,
      "learning_rate": 1.4247271539892975e-05,
      "loss": 0.2933,
      "step": 12394
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4110417090592623,
      "learning_rate": 1.4246373545625942e-05,
      "loss": 0.2745,
      "step": 12395
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4887042003888776,
      "learning_rate": 1.4245475509581685e-05,
      "loss": 0.3809,
      "step": 12396
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8610165810706766,
      "learning_rate": 1.4244577431769044e-05,
      "loss": 0.3542,
      "step": 12397
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6449905361204427,
      "learning_rate": 1.4243679312196854e-05,
      "loss": 0.3561,
      "step": 12398
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26092351109496204,
      "learning_rate": 1.424278115087395e-05,
      "loss": 0.1873,
      "step": 12399
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4783501194458161,
      "learning_rate": 1.424188294780917e-05,
      "loss": 0.2881,
      "step": 12400
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3292373835529354,
      "learning_rate": 1.4240984703011348e-05,
      "loss": 0.2991,
      "step": 12401
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.736052738299682,
      "learning_rate": 1.4240086416489327e-05,
      "loss": 0.4,
      "step": 12402
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31773511132893545,
      "learning_rate": 1.4239188088251936e-05,
      "loss": 0.2032,
      "step": 12403
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.262419403304114,
      "learning_rate": 1.4238289718308021e-05,
      "loss": 0.8635,
      "step": 12404
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8440075723546162,
      "learning_rate": 1.4237391306666416e-05,
      "loss": 0.3884,
      "step": 12405
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.104851691055563,
      "learning_rate": 1.4236492853335961e-05,
      "loss": 0.2855,
      "step": 12406
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.37006863215707547,
      "learning_rate": 1.4235594358325496e-05,
      "loss": 0.3283,
      "step": 12407
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.20876867915773462,
      "learning_rate": 1.423469582164386e-05,
      "loss": 0.2002,
      "step": 12408
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.456283322558983,
      "learning_rate": 1.4233797243299891e-05,
      "loss": 0.3111,
      "step": 12409
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0755744449477878,
      "learning_rate": 1.4232898623302433e-05,
      "loss": 0.3657,
      "step": 12410
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3091226873672381,
      "learning_rate": 1.4231999961660328e-05,
      "loss": 0.7936,
      "step": 12411
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2875101081002666,
      "learning_rate": 1.4231101258382411e-05,
      "loss": 0.1999,
      "step": 12412
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3512842005153179,
      "learning_rate": 1.4230202513477529e-05,
      "loss": 0.3365,
      "step": 12413
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8388172274342746,
      "learning_rate": 1.4229303726954522e-05,
      "loss": 0.34,
      "step": 12414
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.0196481851188004,
      "learning_rate": 1.4228404898822238e-05,
      "loss": 0.7966,
      "step": 12415
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.547721297629924,
      "learning_rate": 1.4227506029089509e-05,
      "loss": 0.1998,
      "step": 12416
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4154341683152046,
      "learning_rate": 1.4226607117765186e-05,
      "loss": 0.3523,
      "step": 12417
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8994415074092329,
      "learning_rate": 1.4225708164858114e-05,
      "loss": 0.3836,
      "step": 12418
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24754739727230224,
      "learning_rate": 1.4224809170377134e-05,
      "loss": 0.2212,
      "step": 12419
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4006887440499954,
      "learning_rate": 1.4223910134331089e-05,
      "loss": 0.2328,
      "step": 12420
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3087205814319202,
      "learning_rate": 1.4223011056728828e-05,
      "loss": 0.2169,
      "step": 12421
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4480220710267453,
      "learning_rate": 1.4222111937579195e-05,
      "loss": 0.6836,
      "step": 12422
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8904843284031554,
      "learning_rate": 1.4221212776891033e-05,
      "loss": 0.3993,
      "step": 12423
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.42772807043387145,
      "learning_rate": 1.4220313574673195e-05,
      "loss": 0.3339,
      "step": 12424
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30302582188803506,
      "learning_rate": 1.421941433093452e-05,
      "loss": 0.2011,
      "step": 12425
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39115670268222047,
      "learning_rate": 1.4218515045683859e-05,
      "loss": 0.3066,
      "step": 12426
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6788810980897042,
      "learning_rate": 1.4217615718930058e-05,
      "loss": 0.3869,
      "step": 12427
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.43143159575569456,
      "learning_rate": 1.4216716350681967e-05,
      "loss": 0.3128,
      "step": 12428
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7395016928027123,
      "learning_rate": 1.4215816940948432e-05,
      "loss": 0.5697,
      "step": 12429
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5858235068400052,
      "learning_rate": 1.4214917489738304e-05,
      "loss": 0.2932,
      "step": 12430
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26837667099337065,
      "learning_rate": 1.4214017997060432e-05,
      "loss": 0.2426,
      "step": 12431
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.28063668894807975,
      "learning_rate": 1.4213118462923665e-05,
      "loss": 0.2061,
      "step": 12432
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.9232969440736867,
      "learning_rate": 1.421221888733685e-05,
      "loss": 0.8193,
      "step": 12433
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4659631866814912,
      "learning_rate": 1.4211319270308841e-05,
      "loss": 0.2089,
      "step": 12434
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.35648670743613425,
      "learning_rate": 1.4210419611848485e-05,
      "loss": 0.3269,
      "step": 12435
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3615031993939202,
      "learning_rate": 1.4209519911964635e-05,
      "loss": 0.2855,
      "step": 12436
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4830525215430182,
      "learning_rate": 1.4208620170666146e-05,
      "loss": 0.318,
      "step": 12437
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2863397452092742,
      "learning_rate": 1.4207720387961867e-05,
      "loss": 0.1629,
      "step": 12438
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3903479328634745,
      "learning_rate": 1.4206820563860648e-05,
      "loss": 0.2753,
      "step": 12439
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24539103437855805,
      "learning_rate": 1.4205920698371345e-05,
      "loss": 0.1674,
      "step": 12440
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.270253162080058,
      "learning_rate": 1.420502079150281e-05,
      "loss": 0.6195,
      "step": 12441
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3706095509161744,
      "learning_rate": 1.4204120843263896e-05,
      "loss": 0.2866,
      "step": 12442
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3742963340616885,
      "learning_rate": 1.420322085366346e-05,
      "loss": 0.3344,
      "step": 12443
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31021128853856644,
      "learning_rate": 1.4202320822710353e-05,
      "loss": 0.2338,
      "step": 12444
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7546381708042382,
      "learning_rate": 1.4201420750413433e-05,
      "loss": 0.3445,
      "step": 12445
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2655025116843552,
      "learning_rate": 1.420052063678155e-05,
      "loss": 0.667,
      "step": 12446
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9373103679337563,
      "learning_rate": 1.4199620481823564e-05,
      "loss": 0.6483,
      "step": 12447
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3839562601398635,
      "learning_rate": 1.4198720285548328e-05,
      "loss": 0.258,
      "step": 12448
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3102855060870533,
      "learning_rate": 1.41978200479647e-05,
      "loss": 0.231,
      "step": 12449
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21491970441665062,
      "learning_rate": 1.4196919769081543e-05,
      "loss": 0.1813,
      "step": 12450
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6412515297216724,
      "learning_rate": 1.4196019448907705e-05,
      "loss": 0.3784,
      "step": 12451
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5970078761991383,
      "learning_rate": 1.4195119087452045e-05,
      "loss": 0.3685,
      "step": 12452
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32937128215848677,
      "learning_rate": 1.4194218684723425e-05,
      "loss": 0.2239,
      "step": 12453
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7782680129705766,
      "learning_rate": 1.4193318240730702e-05,
      "loss": 0.4744,
      "step": 12454
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.27331110558990424,
      "learning_rate": 1.4192417755482733e-05,
      "loss": 0.2473,
      "step": 12455
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.3690523377848005,
      "learning_rate": 1.419151722898838e-05,
      "loss": 0.9143,
      "step": 12456
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3581748851044237,
      "learning_rate": 1.41906166612565e-05,
      "loss": 0.0789,
      "step": 12457
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.236489444819799,
      "learning_rate": 1.4189716052295954e-05,
      "loss": 0.1883,
      "step": 12458
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5105850248662873,
      "learning_rate": 1.4188815402115603e-05,
      "loss": 0.29,
      "step": 12459
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3319454529844433,
      "learning_rate": 1.4187914710724311e-05,
      "loss": 0.2702,
      "step": 12460
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9019654315973443,
      "learning_rate": 1.4187013978130932e-05,
      "loss": 0.5283,
      "step": 12461
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2729019287154985,
      "learning_rate": 1.4186113204344334e-05,
      "loss": 0.1987,
      "step": 12462
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.251519216466951,
      "learning_rate": 1.4185212389373377e-05,
      "loss": 0.761,
      "step": 12463
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9181030025063849,
      "learning_rate": 1.4184311533226923e-05,
      "loss": 0.2515,
      "step": 12464
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.2758315366375181,
      "learning_rate": 1.4183410635913832e-05,
      "loss": 0.7291,
      "step": 12465
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3116788092186273,
      "learning_rate": 1.4182509697442975e-05,
      "loss": 0.1776,
      "step": 12466
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2683002383758521,
      "learning_rate": 1.418160871782321e-05,
      "loss": 0.2793,
      "step": 12467
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25753473845699343,
      "learning_rate": 1.4180707697063403e-05,
      "loss": 0.1099,
      "step": 12468
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8205610851322445,
      "learning_rate": 1.4179806635172417e-05,
      "loss": 0.4792,
      "step": 12469
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.28437287492621816,
      "learning_rate": 1.417890553215912e-05,
      "loss": 0.1697,
      "step": 12470
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.29051246503807937,
      "learning_rate": 1.4178004388032374e-05,
      "loss": 0.2135,
      "step": 12471
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.8075599395879844,
      "learning_rate": 1.4177103202801043e-05,
      "loss": 0.6875,
      "step": 12472
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31043678566679755,
      "learning_rate": 1.4176201976474002e-05,
      "loss": 0.2473,
      "step": 12473
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.663835718558331,
      "learning_rate": 1.4175300709060108e-05,
      "loss": 0.7864,
      "step": 12474
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3025616609728854,
      "learning_rate": 1.4174399400568233e-05,
      "loss": 0.1798,
      "step": 12475
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5617720095199766,
      "learning_rate": 1.4173498051007244e-05,
      "loss": 0.4065,
      "step": 12476
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5785396516308297,
      "learning_rate": 1.4172596660386006e-05,
      "loss": 0.3865,
      "step": 12477
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3110848319595867,
      "learning_rate": 1.4171695228713388e-05,
      "loss": 0.3246,
      "step": 12478
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.24027233464292158,
      "learning_rate": 1.4170793755998262e-05,
      "loss": 0.1307,
      "step": 12479
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9774537770519213,
      "learning_rate": 1.4169892242249494e-05,
      "loss": 0.4989,
      "step": 12480
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.39058852634613983,
      "learning_rate": 1.4168990687475954e-05,
      "loss": 0.2549,
      "step": 12481
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.748160034095203,
      "learning_rate": 1.4168089091686513e-05,
      "loss": 0.9082,
      "step": 12482
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.47749951751970937,
      "learning_rate": 1.4167187454890039e-05,
      "loss": 0.279,
      "step": 12483
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.28424483048435384,
      "learning_rate": 1.4166285777095403e-05,
      "loss": 0.0782,
      "step": 12484
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2883415678813166,
      "learning_rate": 1.4165384058311477e-05,
      "loss": 0.2893,
      "step": 12485
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.606545730121895,
      "learning_rate": 1.4164482298547135e-05,
      "loss": 0.3653,
      "step": 12486
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.081743611527246,
      "learning_rate": 1.416358049781124e-05,
      "loss": 0.5813,
      "step": 12487
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.18703544215041012,
      "learning_rate": 1.4162678656112675e-05,
      "loss": 0.0921,
      "step": 12488
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41559203262339606,
      "learning_rate": 1.4161776773460306e-05,
      "loss": 0.2568,
      "step": 12489
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.33026915839842375,
      "learning_rate": 1.4160874849863009e-05,
      "loss": 0.256,
      "step": 12490
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4909972512763217,
      "learning_rate": 1.4159972885329653e-05,
      "loss": 0.3725,
      "step": 12491
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9245533527590627,
      "learning_rate": 1.4159070879869117e-05,
      "loss": 0.5584,
      "step": 12492
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6284590543452788,
      "learning_rate": 1.4158168833490273e-05,
      "loss": 0.392,
      "step": 12493
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3520941621009164,
      "learning_rate": 1.4157266746201996e-05,
      "loss": 0.2451,
      "step": 12494
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.6933588403496862,
      "learning_rate": 1.415636461801316e-05,
      "loss": 0.4771,
      "step": 12495
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3226139232648299,
      "learning_rate": 1.4155462448932644e-05,
      "loss": 0.2538,
      "step": 12496
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2402834862853972,
      "learning_rate": 1.4154560238969316e-05,
      "loss": 0.1048,
      "step": 12497
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8781397368284686,
      "learning_rate": 1.4153657988132062e-05,
      "loss": 0.2984,
      "step": 12498
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32456852229915856,
      "learning_rate": 1.4152755696429752e-05,
      "loss": 0.1195,
      "step": 12499
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.05873221687811,
      "learning_rate": 1.4151853363871263e-05,
      "loss": 0.6675,
      "step": 12500
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40937610901753135,
      "learning_rate": 1.415095099046548e-05,
      "loss": 0.2956,
      "step": 12501
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3820127813157294,
      "learning_rate": 1.4150048576221269e-05,
      "loss": 0.3234,
      "step": 12502
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32210555239967514,
      "learning_rate": 1.414914612114752e-05,
      "loss": 0.2168,
      "step": 12503
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7438321037421834,
      "learning_rate": 1.4148243625253102e-05,
      "loss": 0.452,
      "step": 12504
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7373221977769567,
      "learning_rate": 1.41473410885469e-05,
      "loss": 0.341,
      "step": 12505
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2816992489443058,
      "learning_rate": 1.4146438511037794e-05,
      "loss": 0.1809,
      "step": 12506
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25378680259236186,
      "learning_rate": 1.4145535892734658e-05,
      "loss": 0.0761,
      "step": 12507
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.38329505723487545,
      "learning_rate": 1.4144633233646378e-05,
      "loss": 0.3256,
      "step": 12508
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.32724714959779866,
      "learning_rate": 1.4143730533781832e-05,
      "loss": 0.2672,
      "step": 12509
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.7463208424028073,
      "learning_rate": 1.41428277931499e-05,
      "loss": 0.3932,
      "step": 12510
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8574490477426037,
      "learning_rate": 1.4141925011759466e-05,
      "loss": 0.5433,
      "step": 12511
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25373022643538,
      "learning_rate": 1.4141022189619414e-05,
      "loss": 0.1865,
      "step": 12512
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8575523302266912,
      "learning_rate": 1.414011932673862e-05,
      "loss": 0.5721,
      "step": 12513
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.28820548219436193,
      "learning_rate": 1.4139216423125968e-05,
      "loss": 0.245,
      "step": 12514
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.351921200580916,
      "learning_rate": 1.4138313478790344e-05,
      "loss": 0.7372,
      "step": 12515
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21680629826277123,
      "learning_rate": 1.4137410493740633e-05,
      "loss": 0.1321,
      "step": 12516
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3491928515903416,
      "learning_rate": 1.4136507467985712e-05,
      "loss": 0.2395,
      "step": 12517
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9263980357961585,
      "learning_rate": 1.4135604401534471e-05,
      "loss": 0.3352,
      "step": 12518
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.751451083467672,
      "learning_rate": 1.4134701294395796e-05,
      "loss": 0.5323,
      "step": 12519
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2924502791703869,
      "learning_rate": 1.4133798146578568e-05,
      "loss": 0.256,
      "step": 12520
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3376537598319004,
      "learning_rate": 1.4132894958091672e-05,
      "loss": 0.2808,
      "step": 12521
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.612587409766465,
      "learning_rate": 1.4131991728943996e-05,
      "loss": 0.4036,
      "step": 12522
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8059269923946463,
      "learning_rate": 1.4131088459144426e-05,
      "loss": 0.3833,
      "step": 12523
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2840994992637121,
      "learning_rate": 1.4130185148701848e-05,
      "loss": 0.1769,
      "step": 12524
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.31393623751614486,
      "learning_rate": 1.412928179762515e-05,
      "loss": 0.1741,
      "step": 12525
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5866571691079278,
      "learning_rate": 1.412837840592322e-05,
      "loss": 0.371,
      "step": 12526
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3058295315089623,
      "learning_rate": 1.4127474973604942e-05,
      "loss": 0.2709,
      "step": 12527
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8001747121698425,
      "learning_rate": 1.4126571500679209e-05,
      "loss": 0.507,
      "step": 12528
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5065618925394164,
      "learning_rate": 1.4125667987154908e-05,
      "loss": 0.2626,
      "step": 12529
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5702553257414128,
      "learning_rate": 1.4124764433040928e-05,
      "loss": 0.4019,
      "step": 12530
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4255567532684735,
      "learning_rate": 1.4123860838346154e-05,
      "loss": 0.2891,
      "step": 12531
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30860468370006794,
      "learning_rate": 1.4122957203079489e-05,
      "loss": 0.2796,
      "step": 12532
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3279487438733504,
      "learning_rate": 1.4122053527249806e-05,
      "loss": 0.14,
      "step": 12533
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9184521542707665,
      "learning_rate": 1.4121149810866008e-05,
      "loss": 0.4438,
      "step": 12534
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.22120832924945197,
      "learning_rate": 1.4120246053936981e-05,
      "loss": 0.1685,
      "step": 12535
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.033637866554493,
      "learning_rate": 1.4119342256471617e-05,
      "loss": 0.3632,
      "step": 12536
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3402111284653301,
      "learning_rate": 1.411843841847881e-05,
      "loss": 0.3246,
      "step": 12537
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.723779668362303,
      "learning_rate": 1.411753453996745e-05,
      "loss": 0.3791,
      "step": 12538
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3858154357190395,
      "learning_rate": 1.4116630620946432e-05,
      "loss": 0.295,
      "step": 12539
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.4369572127494845,
      "learning_rate": 1.4115726661424644e-05,
      "loss": 0.2894,
      "step": 12540
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.69328085243058,
      "learning_rate": 1.4114822661410985e-05,
      "loss": 0.8941,
      "step": 12541
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.001751367957498,
      "learning_rate": 1.4113918620914346e-05,
      "loss": 0.2581,
      "step": 12542
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40065260724752594,
      "learning_rate": 1.411301453994362e-05,
      "loss": 0.1925,
      "step": 12543
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25675717595419184,
      "learning_rate": 1.411211041850771e-05,
      "loss": 0.2476,
      "step": 12544
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.481587586834505,
      "learning_rate": 1.4111206256615502e-05,
      "loss": 0.3185,
      "step": 12545
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9756825696328005,
      "learning_rate": 1.411030205427589e-05,
      "loss": 0.3746,
      "step": 12546
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.1646514389858629,
      "learning_rate": 1.4109397811497776e-05,
      "loss": 0.7743,
      "step": 12547
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.26648025356860194,
      "learning_rate": 1.4108493528290054e-05,
      "loss": 0.1788,
      "step": 12548
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.9003009652777191,
      "learning_rate": 1.4107589204661622e-05,
      "loss": 0.4268,
      "step": 12549
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.34184094889854877,
      "learning_rate": 1.4106684840621375e-05,
      "loss": 0.331,
      "step": 12550
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.8761322532137719,
      "learning_rate": 1.4105780436178212e-05,
      "loss": 0.3625,
      "step": 12551
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30124623062747774,
      "learning_rate": 1.4104875991341032e-05,
      "loss": 0.1896,
      "step": 12552
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3562873969957644,
      "learning_rate": 1.4103971506118728e-05,
      "loss": 0.238,
      "step": 12553
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3885166162723708,
      "learning_rate": 1.41030669805202e-05,
      "loss": 0.3126,
      "step": 12554
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.40864323506733996,
      "learning_rate": 1.4102162414554355e-05,
      "loss": 0.31,
      "step": 12555
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.46934233640314676,
      "learning_rate": 1.410125780823008e-05,
      "loss": 0.3883,
      "step": 12556
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.2571820565978282,
      "learning_rate": 1.4100353161556286e-05,
      "loss": 0.0767,
      "step": 12557
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.41072087729500056,
      "learning_rate": 1.4099448474541868e-05,
      "loss": 0.3317,
      "step": 12558
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.4310477544997098,
      "learning_rate": 1.4098543747195725e-05,
      "loss": 0.4069,
      "step": 12559
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.0156155961813413,
      "learning_rate": 1.409763897952676e-05,
      "loss": 0.5561,
      "step": 12560
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3167988014713052,
      "learning_rate": 1.4096734171543877e-05,
      "loss": 0.2003,
      "step": 12561
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.25435285225746873,
      "learning_rate": 1.4095829323255971e-05,
      "loss": 0.2317,
      "step": 12562
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.669507548671577,
      "learning_rate": 1.4094924434671952e-05,
      "loss": 0.4667,
      "step": 12563
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.21846530143232434,
      "learning_rate": 1.409401950580072e-05,
      "loss": 0.1551,
      "step": 12564
    },
    {
      "epoch": 0.38,
      "grad_norm": 2.2629544988460286,
      "learning_rate": 1.4093114536651174e-05,
      "loss": 0.7978,
      "step": 12565
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3026076778115347,
      "learning_rate": 1.409220952723222e-05,
      "loss": 0.1871,
      "step": 12566
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.5633179114734211,
      "learning_rate": 1.4091304477552764e-05,
      "loss": 0.4286,
      "step": 12567
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.30395401413444295,
      "learning_rate": 1.4090399387621706e-05,
      "loss": 0.249,
      "step": 12568
    },
    {
      "epoch": 0.38,
      "grad_norm": 1.631723748128166,
      "learning_rate": 1.4089494257447956e-05,
      "loss": 0.7541,
      "step": 12569
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.45051743416114454,
      "learning_rate": 1.4088589087040414e-05,
      "loss": 0.3045,
      "step": 12570
    },
    {
      "epoch": 0.38,
      "grad_norm": 0.3479991697347697,
      "learning_rate": 1.408768387640799e-05,
      "loss": 0.2746,
      "step": 12571
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3329365000827616,
      "learning_rate": 1.4086778625559583e-05,
      "loss": 0.2059,
      "step": 12572
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4629281043712717,
      "learning_rate": 1.4085873334504105e-05,
      "loss": 0.275,
      "step": 12573
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2804645857880334,
      "learning_rate": 1.408496800325046e-05,
      "loss": 0.2553,
      "step": 12574
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3641585281557096,
      "learning_rate": 1.4084062631807558e-05,
      "loss": 0.2657,
      "step": 12575
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.41080679325969294,
      "learning_rate": 1.4083157220184302e-05,
      "loss": 0.2677,
      "step": 12576
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.173185380701296,
      "learning_rate": 1.4082251768389602e-05,
      "loss": 0.2875,
      "step": 12577
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7303432567193461,
      "learning_rate": 1.4081346276432366e-05,
      "loss": 0.4789,
      "step": 12578
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25375478491950937,
      "learning_rate": 1.4080440744321503e-05,
      "loss": 0.1939,
      "step": 12579
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3674626170454454,
      "learning_rate": 1.407953517206592e-05,
      "loss": 0.3166,
      "step": 12580
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4863906008207408,
      "learning_rate": 1.4078629559674528e-05,
      "loss": 0.3664,
      "step": 12581
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42774080242237755,
      "learning_rate": 1.4077723907156237e-05,
      "loss": 0.2587,
      "step": 12582
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8100897459424966,
      "learning_rate": 1.4076818214519957e-05,
      "loss": 0.6322,
      "step": 12583
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6081893196914891,
      "learning_rate": 1.4075912481774596e-05,
      "loss": 0.3149,
      "step": 12584
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2584729086759571,
      "learning_rate": 1.4075006708929068e-05,
      "loss": 0.1839,
      "step": 12585
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34776739066039375,
      "learning_rate": 1.4074100895992281e-05,
      "loss": 0.3063,
      "step": 12586
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6612090409429079,
      "learning_rate": 1.4073195042973153e-05,
      "loss": 0.3387,
      "step": 12587
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9487248014967222,
      "learning_rate": 1.4072289149880587e-05,
      "loss": 0.5628,
      "step": 12588
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2978757883525979,
      "learning_rate": 1.4071383216723504e-05,
      "loss": 0.2114,
      "step": 12589
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.45154157568456105,
      "learning_rate": 1.4070477243510814e-05,
      "loss": 0.2331,
      "step": 12590
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31929030285058646,
      "learning_rate": 1.4069571230251428e-05,
      "loss": 0.2923,
      "step": 12591
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9821276306034332,
      "learning_rate": 1.406866517695426e-05,
      "loss": 0.5119,
      "step": 12592
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6086636744451007,
      "learning_rate": 1.4067759083628226e-05,
      "loss": 0.3267,
      "step": 12593
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26489992837957743,
      "learning_rate": 1.406685295028224e-05,
      "loss": 0.1901,
      "step": 12594
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39611973711209203,
      "learning_rate": 1.4065946776925219e-05,
      "loss": 0.2841,
      "step": 12595
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.108561398917,
      "learning_rate": 1.406504056356607e-05,
      "loss": 0.3729,
      "step": 12596
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3159669301736082,
      "learning_rate": 1.4064134310213719e-05,
      "loss": 0.327,
      "step": 12597
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27289693534986276,
      "learning_rate": 1.4063228016877075e-05,
      "loss": 0.1973,
      "step": 12598
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9977400515635073,
      "learning_rate": 1.4062321683565057e-05,
      "loss": 0.5666,
      "step": 12599
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5735032841818826,
      "learning_rate": 1.4061415310286583e-05,
      "loss": 0.0291,
      "step": 12600
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3666230991682475,
      "learning_rate": 1.4060508897050569e-05,
      "loss": 0.9254,
      "step": 12601
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3023499001298213,
      "learning_rate": 1.4059602443865932e-05,
      "loss": 0.095,
      "step": 12602
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27481631191677214,
      "learning_rate": 1.405869595074159e-05,
      "loss": 0.2396,
      "step": 12603
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.30930939190493495,
      "learning_rate": 1.405778941768646e-05,
      "loss": 0.2647,
      "step": 12604
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8702733520709169,
      "learning_rate": 1.4056882844709465e-05,
      "loss": 0.4084,
      "step": 12605
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7403723921522254,
      "learning_rate": 1.405597623181952e-05,
      "loss": 0.4283,
      "step": 12606
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31198317699840317,
      "learning_rate": 1.4055069579025547e-05,
      "loss": 0.2194,
      "step": 12607
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7936947999406998,
      "learning_rate": 1.4054162886336466e-05,
      "loss": 0.7493,
      "step": 12608
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3056526472100493,
      "learning_rate": 1.4053256153761194e-05,
      "loss": 0.2603,
      "step": 12609
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3147412077143925,
      "learning_rate": 1.405234938130866e-05,
      "loss": 0.8589,
      "step": 12610
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.294170625899525,
      "learning_rate": 1.4051442568987773e-05,
      "loss": 0.1225,
      "step": 12611
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.35672384260232465,
      "learning_rate": 1.405053571680746e-05,
      "loss": 0.3067,
      "step": 12612
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3557147054477429,
      "learning_rate": 1.4049628824776652e-05,
      "loss": 0.2443,
      "step": 12613
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40997089502094936,
      "learning_rate": 1.4048721892904258e-05,
      "loss": 0.3325,
      "step": 12614
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27770187754182923,
      "learning_rate": 1.4047814921199202e-05,
      "loss": 0.2389,
      "step": 12615
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29337816931453825,
      "learning_rate": 1.4046907909670417e-05,
      "loss": 0.1638,
      "step": 12616
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4692584260439993,
      "learning_rate": 1.4046000858326818e-05,
      "loss": 0.3624,
      "step": 12617
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8781374008006112,
      "learning_rate": 1.4045093767177332e-05,
      "loss": 0.4457,
      "step": 12618
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8771612406797211,
      "learning_rate": 1.404418663623088e-05,
      "loss": 0.8314,
      "step": 12619
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3144254252658809,
      "learning_rate": 1.4043279465496393e-05,
      "loss": 0.2391,
      "step": 12620
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3329973531063294,
      "learning_rate": 1.404237225498279e-05,
      "loss": 0.2894,
      "step": 12621
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33953833957100377,
      "learning_rate": 1.4041465004699e-05,
      "loss": 0.3038,
      "step": 12622
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7938544417456894,
      "learning_rate": 1.4040557714653949e-05,
      "loss": 0.5679,
      "step": 12623
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.21364659869572,
      "learning_rate": 1.4039650384856555e-05,
      "loss": 0.0721,
      "step": 12624
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39919566372237736,
      "learning_rate": 1.4038743015315756e-05,
      "loss": 0.3097,
      "step": 12625
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.47775098963448853,
      "learning_rate": 1.4037835606040475e-05,
      "loss": 0.2721,
      "step": 12626
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5175497829655067,
      "learning_rate": 1.4036928157039638e-05,
      "loss": 0.3829,
      "step": 12627
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3973951030449082,
      "learning_rate": 1.4036020668322172e-05,
      "loss": 0.3016,
      "step": 12628
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36165716309054896,
      "learning_rate": 1.4035113139897008e-05,
      "loss": 0.1721,
      "step": 12629
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3445392238017658,
      "learning_rate": 1.4034205571773071e-05,
      "loss": 0.3281,
      "step": 12630
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6260995638134093,
      "learning_rate": 1.4033297963959296e-05,
      "loss": 0.3642,
      "step": 12631
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42152747688948666,
      "learning_rate": 1.4032390316464603e-05,
      "loss": 0.318,
      "step": 12632
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26876973509580016,
      "learning_rate": 1.4031482629297934e-05,
      "loss": 0.2167,
      "step": 12633
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38424235287594577,
      "learning_rate": 1.403057490246821e-05,
      "loss": 0.252,
      "step": 12634
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46992632749615854,
      "learning_rate": 1.4029667135984362e-05,
      "loss": 0.2466,
      "step": 12635
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.6624617410359717,
      "learning_rate": 1.4028759329855324e-05,
      "loss": 0.8316,
      "step": 12636
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7692631869445858,
      "learning_rate": 1.4027851484090025e-05,
      "loss": 0.4441,
      "step": 12637
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5145074421467605,
      "learning_rate": 1.4026943598697397e-05,
      "loss": 0.3574,
      "step": 12638
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3087624083614504,
      "learning_rate": 1.4026035673686376e-05,
      "loss": 0.2244,
      "step": 12639
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29802422509561494,
      "learning_rate": 1.4025127709065891e-05,
      "loss": 0.3156,
      "step": 12640
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7362469649431251,
      "learning_rate": 1.4024219704844873e-05,
      "loss": 0.3137,
      "step": 12641
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.14717199984414828,
      "learning_rate": 1.4023311661032257e-05,
      "loss": 0.0732,
      "step": 12642
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6687437008602579,
      "learning_rate": 1.4022403577636978e-05,
      "loss": 0.4232,
      "step": 12643
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36801658245654073,
      "learning_rate": 1.4021495454667971e-05,
      "loss": 0.2483,
      "step": 12644
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.32879393038443194,
      "learning_rate": 1.4020587292134168e-05,
      "loss": 0.3286,
      "step": 12645
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8506259564011945,
      "learning_rate": 1.4019679090044503e-05,
      "loss": 0.4969,
      "step": 12646
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6774508375778246,
      "learning_rate": 1.4018770848407912e-05,
      "loss": 0.464,
      "step": 12647
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2782730406395041,
      "learning_rate": 1.4017862567233329e-05,
      "loss": 0.1992,
      "step": 12648
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7761575240547254,
      "learning_rate": 1.4016954246529697e-05,
      "loss": 0.5811,
      "step": 12649
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5021878170220289,
      "learning_rate": 1.4016045886305944e-05,
      "loss": 0.025,
      "step": 12650
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26083731160077306,
      "learning_rate": 1.401513748657101e-05,
      "loss": 0.249,
      "step": 12651
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24439394589053143,
      "learning_rate": 1.4014229047333834e-05,
      "loss": 0.1019,
      "step": 12652
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3614534651495321,
      "learning_rate": 1.4013320568603353e-05,
      "loss": 0.3184,
      "step": 12653
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2352531896247336,
      "learning_rate": 1.40124120503885e-05,
      "loss": 0.3509,
      "step": 12654
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7490740904784069,
      "learning_rate": 1.4011503492698218e-05,
      "loss": 0.3972,
      "step": 12655
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37256420829819126,
      "learning_rate": 1.4010594895541448e-05,
      "loss": 0.3211,
      "step": 12656
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24306707833837368,
      "learning_rate": 1.4009686258927122e-05,
      "loss": 0.2175,
      "step": 12657
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8640810934680332,
      "learning_rate": 1.4008777582864183e-05,
      "loss": 0.523,
      "step": 12658
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8715091936054772,
      "learning_rate": 1.4007868867361576e-05,
      "loss": 0.3991,
      "step": 12659
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3150179503690552,
      "learning_rate": 1.4006960112428232e-05,
      "loss": 0.1802,
      "step": 12660
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3261196694538416,
      "learning_rate": 1.4006051318073095e-05,
      "loss": 0.1684,
      "step": 12661
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5228434960200636,
      "learning_rate": 1.400514248430511e-05,
      "loss": 0.3752,
      "step": 12662
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31905202865918614,
      "learning_rate": 1.4004233611133213e-05,
      "loss": 0.2743,
      "step": 12663
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6507983685474767,
      "learning_rate": 1.400332469856635e-05,
      "loss": 0.5322,
      "step": 12664
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6501886244559739,
      "learning_rate": 1.4002415746613462e-05,
      "loss": 0.3927,
      "step": 12665
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6603833636365276,
      "learning_rate": 1.400150675528349e-05,
      "loss": 0.3837,
      "step": 12666
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3303545664581942,
      "learning_rate": 1.4000597724585374e-05,
      "loss": 0.2442,
      "step": 12667
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2939724062117265,
      "learning_rate": 1.3999688654528065e-05,
      "loss": 0.2491,
      "step": 12668
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4336085019576066,
      "learning_rate": 1.3998779545120504e-05,
      "loss": 0.2748,
      "step": 12669
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.19568800401696335,
      "learning_rate": 1.399787039637163e-05,
      "loss": 0.0719,
      "step": 12670
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3577668448102279,
      "learning_rate": 1.3996961208290396e-05,
      "loss": 0.3216,
      "step": 12671
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.660505907127595,
      "learning_rate": 1.3996051980885743e-05,
      "loss": 0.3979,
      "step": 12672
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.693474983179257,
      "learning_rate": 1.3995142714166611e-05,
      "loss": 0.4854,
      "step": 12673
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2880537335150977,
      "learning_rate": 1.3994233408141952e-05,
      "loss": 0.2336,
      "step": 12674
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37248597264929867,
      "learning_rate": 1.3993324062820713e-05,
      "loss": 0.2998,
      "step": 12675
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3920008585511565,
      "learning_rate": 1.3992414678211837e-05,
      "loss": 0.2735,
      "step": 12676
    },
    {
      "epoch": 0.39,
      "grad_norm": 2.1293223510106754,
      "learning_rate": 1.3991505254324272e-05,
      "loss": 0.73,
      "step": 12677
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26968553772510895,
      "learning_rate": 1.3990595791166963e-05,
      "loss": 0.1359,
      "step": 12678
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38053058292627523,
      "learning_rate": 1.398968628874886e-05,
      "loss": 0.2455,
      "step": 12679
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3007328558612118,
      "learning_rate": 1.3988776747078909e-05,
      "loss": 0.2156,
      "step": 12680
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29634373577848167,
      "learning_rate": 1.398786716616606e-05,
      "loss": 0.2372,
      "step": 12681
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7274448910190046,
      "learning_rate": 1.3986957546019267e-05,
      "loss": 0.5124,
      "step": 12682
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.516342738443668,
      "learning_rate": 1.3986047886647468e-05,
      "loss": 0.3153,
      "step": 12683
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33371606718818214,
      "learning_rate": 1.3985138188059622e-05,
      "loss": 0.2828,
      "step": 12684
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.857992151764453,
      "learning_rate": 1.3984228450264674e-05,
      "loss": 0.3809,
      "step": 12685
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.48597292352941224,
      "learning_rate": 1.3983318673271575e-05,
      "loss": 0.3751,
      "step": 12686
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39564982649102215,
      "learning_rate": 1.3982408857089275e-05,
      "loss": 0.2673,
      "step": 12687
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33171338669628975,
      "learning_rate": 1.3981499001726728e-05,
      "loss": 0.1961,
      "step": 12688
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31674363497847596,
      "learning_rate": 1.3980589107192883e-05,
      "loss": 0.2214,
      "step": 12689
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39102907369623197,
      "learning_rate": 1.3979679173496695e-05,
      "loss": 0.3062,
      "step": 12690
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6594884430725279,
      "learning_rate": 1.3978769200647114e-05,
      "loss": 0.3904,
      "step": 12691
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3428176316046669,
      "learning_rate": 1.3977859188653087e-05,
      "loss": 0.31,
      "step": 12692
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2406048868075709,
      "learning_rate": 1.3976949137523574e-05,
      "loss": 0.0732,
      "step": 12693
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34762909837407396,
      "learning_rate": 1.3976039047267528e-05,
      "loss": 0.2534,
      "step": 12694
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7736717829478612,
      "learning_rate": 1.3975128917893904e-05,
      "loss": 0.833,
      "step": 12695
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2546561577067368,
      "learning_rate": 1.3974218749411651e-05,
      "loss": 0.1128,
      "step": 12696
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7426101872806613,
      "learning_rate": 1.3973308541829726e-05,
      "loss": 0.4838,
      "step": 12697
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2529184099495613,
      "learning_rate": 1.3972398295157089e-05,
      "loss": 0.1818,
      "step": 12698
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3222979763244616,
      "learning_rate": 1.3971488009402685e-05,
      "loss": 0.3182,
      "step": 12699
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29402954625158184,
      "learning_rate": 1.3970577684575476e-05,
      "loss": 0.1772,
      "step": 12700
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.536145530800699,
      "learning_rate": 1.396966732068442e-05,
      "loss": 0.9053,
      "step": 12701
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29255052596790176,
      "learning_rate": 1.3968756917738466e-05,
      "loss": 0.1789,
      "step": 12702
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4784188371143648,
      "learning_rate": 1.3967846475746578e-05,
      "loss": 0.3726,
      "step": 12703
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38343768245686993,
      "learning_rate": 1.3966935994717715e-05,
      "loss": 0.2592,
      "step": 12704
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.47093546761531363,
      "learning_rate": 1.3966025474660824e-05,
      "loss": 0.3534,
      "step": 12705
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4312719378744396,
      "learning_rate": 1.3965114915584872e-05,
      "loss": 0.2394,
      "step": 12706
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.30758653907081807,
      "learning_rate": 1.3964204317498813e-05,
      "loss": 0.2409,
      "step": 12707
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9233982120858618,
      "learning_rate": 1.3963293680411609e-05,
      "loss": 0.5355,
      "step": 12708
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.313321973745709,
      "learning_rate": 1.3962383004332218e-05,
      "loss": 0.2041,
      "step": 12709
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33688087112520837,
      "learning_rate": 1.39614722892696e-05,
      "loss": 0.3005,
      "step": 12710
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3167390909327862,
      "learning_rate": 1.3960561535232713e-05,
      "loss": 0.1911,
      "step": 12711
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5175639197310453,
      "learning_rate": 1.3959650742230519e-05,
      "loss": 0.3649,
      "step": 12712
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3034046141425028,
      "learning_rate": 1.3958739910271977e-05,
      "loss": 0.1401,
      "step": 12713
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6947651458044127,
      "learning_rate": 1.395782903936605e-05,
      "loss": 0.4584,
      "step": 12714
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.49069868141948136,
      "learning_rate": 1.39569181295217e-05,
      "loss": 0.248,
      "step": 12715
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5718401684234032,
      "learning_rate": 1.3956007180747885e-05,
      "loss": 0.3903,
      "step": 12716
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2727496628190509,
      "learning_rate": 1.395509619305357e-05,
      "loss": 0.2647,
      "step": 12717
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4179798266160845,
      "learning_rate": 1.3954185166447719e-05,
      "loss": 0.2605,
      "step": 12718
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26122478547475564,
      "learning_rate": 1.3953274100939291e-05,
      "loss": 0.1505,
      "step": 12719
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34848193804973737,
      "learning_rate": 1.3952362996537252e-05,
      "loss": 0.0755,
      "step": 12720
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37238996815788067,
      "learning_rate": 1.3951451853250567e-05,
      "loss": 0.3099,
      "step": 12721
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3060657263944517,
      "learning_rate": 1.3950540671088198e-05,
      "loss": 0.2477,
      "step": 12722
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8708563943702597,
      "learning_rate": 1.3949629450059109e-05,
      "loss": 0.5073,
      "step": 12723
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7934524102325791,
      "learning_rate": 1.3948718190172266e-05,
      "loss": 0.3499,
      "step": 12724
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3972261916147034,
      "learning_rate": 1.3947806891436636e-05,
      "loss": 0.2577,
      "step": 12725
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.42057377394529016,
      "learning_rate": 1.3946895553861176e-05,
      "loss": 0.2473,
      "step": 12726
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4788287334808907,
      "learning_rate": 1.3945984177454867e-05,
      "loss": 0.8261,
      "step": 12727
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.32119647339112506,
      "learning_rate": 1.3945072762226665e-05,
      "loss": 0.2529,
      "step": 12728
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0306973640389019,
      "learning_rate": 1.3944161308185538e-05,
      "loss": 0.2146,
      "step": 12729
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24574625712751258,
      "learning_rate": 1.3943249815340455e-05,
      "loss": 0.1758,
      "step": 12730
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3070126272299178,
      "learning_rate": 1.3942338283700383e-05,
      "loss": 0.6898,
      "step": 12731
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.2475700905139635,
      "learning_rate": 1.394142671327429e-05,
      "loss": 0.4263,
      "step": 12732
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4159683407662639,
      "learning_rate": 1.3940515104071141e-05,
      "loss": 0.2993,
      "step": 12733
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3440633518201734,
      "learning_rate": 1.3939603456099913e-05,
      "loss": 0.2812,
      "step": 12734
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3993989979846555,
      "learning_rate": 1.3938691769369565e-05,
      "loss": 0.2595,
      "step": 12735
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0247635109422573,
      "learning_rate": 1.3937780043889072e-05,
      "loss": 0.6022,
      "step": 12736
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0691676357813866,
      "learning_rate": 1.3936868279667405e-05,
      "loss": 0.5724,
      "step": 12737
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3660202327650326,
      "learning_rate": 1.3935956476713532e-05,
      "loss": 0.2004,
      "step": 12738
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.31045710133416404,
      "learning_rate": 1.3935044635036423e-05,
      "loss": 0.2219,
      "step": 12739
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2855852510056848,
      "learning_rate": 1.3934132754645053e-05,
      "loss": 0.2299,
      "step": 12740
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.411750419343164,
      "learning_rate": 1.393322083554839e-05,
      "loss": 0.3154,
      "step": 12741
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8226113063679783,
      "learning_rate": 1.3932308877755402e-05,
      "loss": 0.5019,
      "step": 12742
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.18311474209829068,
      "learning_rate": 1.393139688127507e-05,
      "loss": 0.0703,
      "step": 12743
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37014339455096107,
      "learning_rate": 1.393048484611636e-05,
      "loss": 0.2947,
      "step": 12744
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38166383079668037,
      "learning_rate": 1.3929572772288247e-05,
      "loss": 0.2826,
      "step": 12745
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4285480867114235,
      "learning_rate": 1.3928660659799704e-05,
      "loss": 0.2805,
      "step": 12746
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9367044894619193,
      "learning_rate": 1.3927748508659707e-05,
      "loss": 0.463,
      "step": 12747
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2188873720647492,
      "learning_rate": 1.3926836318877224e-05,
      "loss": 0.1667,
      "step": 12748
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3964400167416074,
      "learning_rate": 1.3925924090461234e-05,
      "loss": 0.2827,
      "step": 12749
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5137973153429809,
      "learning_rate": 1.3925011823420713e-05,
      "loss": 0.37,
      "step": 12750
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.47190839561534026,
      "learning_rate": 1.392409951776463e-05,
      "loss": 0.3702,
      "step": 12751
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26310298975822327,
      "learning_rate": 1.3923187173501969e-05,
      "loss": 0.1985,
      "step": 12752
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5220704998315766,
      "learning_rate": 1.3922274790641701e-05,
      "loss": 0.376,
      "step": 12753
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.121813084852527,
      "learning_rate": 1.3921362369192803e-05,
      "loss": 0.4914,
      "step": 12754
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5009011756172193,
      "learning_rate": 1.3920449909164251e-05,
      "loss": 0.8563,
      "step": 12755
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.23196497919877593,
      "learning_rate": 1.391953741056502e-05,
      "loss": 0.073,
      "step": 12756
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36359702879717326,
      "learning_rate": 1.3918624873404096e-05,
      "loss": 0.2955,
      "step": 12757
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3533173707704136,
      "learning_rate": 1.3917712297690446e-05,
      "loss": 0.2615,
      "step": 12758
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8213854590387938,
      "learning_rate": 1.3916799683433058e-05,
      "loss": 0.3809,
      "step": 12759
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3917810763890002,
      "learning_rate": 1.3915887030640905e-05,
      "loss": 0.2673,
      "step": 12760
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3209826655780607,
      "learning_rate": 1.3914974339322963e-05,
      "loss": 0.1942,
      "step": 12761
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5554032396649586,
      "learning_rate": 1.3914061609488217e-05,
      "loss": 0.3657,
      "step": 12762
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.927855338479161,
      "learning_rate": 1.3913148841145647e-05,
      "loss": 0.4375,
      "step": 12763
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.344059964938819,
      "learning_rate": 1.3912236034304227e-05,
      "loss": 0.3548,
      "step": 12764
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2955009439044337,
      "learning_rate": 1.3911323188972947e-05,
      "loss": 0.1174,
      "step": 12765
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5264129258414993,
      "learning_rate": 1.3910410305160779e-05,
      "loss": 0.3544,
      "step": 12766
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6171626393673973,
      "learning_rate": 1.3909497382876708e-05,
      "loss": 0.3259,
      "step": 12767
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3157126140549859,
      "learning_rate": 1.3908584422129717e-05,
      "loss": 0.2029,
      "step": 12768
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27272942114289933,
      "learning_rate": 1.3907671422928783e-05,
      "loss": 0.2503,
      "step": 12769
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9505099610134111,
      "learning_rate": 1.3906758385282894e-05,
      "loss": 0.4804,
      "step": 12770
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.28033100769562425,
      "learning_rate": 1.3905845309201032e-05,
      "loss": 0.1985,
      "step": 12771
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4601308596189968,
      "learning_rate": 1.3904932194692179e-05,
      "loss": 0.8901,
      "step": 12772
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.060773656060985,
      "learning_rate": 1.3904019041765316e-05,
      "loss": 0.4303,
      "step": 12773
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5596077058986071,
      "learning_rate": 1.3903105850429429e-05,
      "loss": 0.354,
      "step": 12774
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3774350059046737,
      "learning_rate": 1.39021926206935e-05,
      "loss": 0.252,
      "step": 12775
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27302208613536644,
      "learning_rate": 1.3901279352566518e-05,
      "loss": 0.2534,
      "step": 12776
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4315552862474967,
      "learning_rate": 1.3900366046057468e-05,
      "loss": 0.2743,
      "step": 12777
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27733074358887844,
      "learning_rate": 1.3899452701175334e-05,
      "loss": 0.0708,
      "step": 12778
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6376910503416421,
      "learning_rate": 1.3898539317929101e-05,
      "loss": 0.3413,
      "step": 12779
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3321144656763716,
      "learning_rate": 1.3897625896327751e-05,
      "loss": 0.2353,
      "step": 12780
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46725652317586897,
      "learning_rate": 1.3896712436380278e-05,
      "loss": 0.4002,
      "step": 12781
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3926968339074268,
      "learning_rate": 1.3895798938095663e-05,
      "loss": 0.2652,
      "step": 12782
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6482547367939685,
      "learning_rate": 1.38948854014829e-05,
      "loss": 0.4817,
      "step": 12783
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2856978766924302,
      "learning_rate": 1.3893971826550974e-05,
      "loss": 0.1979,
      "step": 12784
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6320079351557271,
      "learning_rate": 1.3893058213308867e-05,
      "loss": 0.478,
      "step": 12785
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1056272203614037,
      "learning_rate": 1.3892144561765577e-05,
      "loss": 0.2837,
      "step": 12786
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.23116380840119571,
      "learning_rate": 1.3891230871930085e-05,
      "loss": 0.164,
      "step": 12787
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7078235892424082,
      "learning_rate": 1.3890317143811383e-05,
      "loss": 0.3318,
      "step": 12788
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3290899530430569,
      "learning_rate": 1.3889403377418465e-05,
      "loss": 0.2427,
      "step": 12789
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4489768486704513,
      "learning_rate": 1.388848957276031e-05,
      "loss": 0.806,
      "step": 12790
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6710581312491479,
      "learning_rate": 1.3887575729845921e-05,
      "loss": 0.396,
      "step": 12791
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4939431373193638,
      "learning_rate": 1.3886661848684282e-05,
      "loss": 0.3756,
      "step": 12792
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.28192090155919947,
      "learning_rate": 1.3885747929284383e-05,
      "loss": 0.0725,
      "step": 12793
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2929774095916629,
      "learning_rate": 1.3884833971655216e-05,
      "loss": 0.2737,
      "step": 12794
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.862244062698515,
      "learning_rate": 1.3883919975805775e-05,
      "loss": 0.3917,
      "step": 12795
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3162625585116966,
      "learning_rate": 1.3883005941745054e-05,
      "loss": 0.2012,
      "step": 12796
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4829624981083225,
      "learning_rate": 1.3882091869482044e-05,
      "loss": 0.0715,
      "step": 12797
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33598555665436164,
      "learning_rate": 1.3881177759025732e-05,
      "loss": 0.2918,
      "step": 12798
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3692333962561546,
      "learning_rate": 1.3880263610385119e-05,
      "loss": 0.2801,
      "step": 12799
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.47906656083140453,
      "learning_rate": 1.3879349423569193e-05,
      "loss": 0.3141,
      "step": 12800
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9825429656710593,
      "learning_rate": 1.3878435198586952e-05,
      "loss": 0.5271,
      "step": 12801
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3257763562168308,
      "learning_rate": 1.3877520935447393e-05,
      "loss": 0.18,
      "step": 12802
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5642145447528115,
      "learning_rate": 1.3876606634159502e-05,
      "loss": 0.3839,
      "step": 12803
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33468593216651343,
      "learning_rate": 1.3875692294732283e-05,
      "loss": 0.071,
      "step": 12804
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.43654345155531277,
      "learning_rate": 1.387477791717473e-05,
      "loss": 0.3314,
      "step": 12805
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2329812551352205,
      "learning_rate": 1.387386350149583e-05,
      "loss": 0.1644,
      "step": 12806
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3511145764391154,
      "learning_rate": 1.3872949047704589e-05,
      "loss": 0.309,
      "step": 12807
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8974663457704254,
      "learning_rate": 1.3872034555810001e-05,
      "loss": 0.4142,
      "step": 12808
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.333012095920428,
      "learning_rate": 1.3871120025821066e-05,
      "loss": 0.5628,
      "step": 12809
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37079953735972326,
      "learning_rate": 1.3870205457746778e-05,
      "loss": 0.3022,
      "step": 12810
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38556926324017576,
      "learning_rate": 1.3869290851596131e-05,
      "loss": 0.2741,
      "step": 12811
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3071275295285254,
      "learning_rate": 1.3868376207378131e-05,
      "loss": 0.261,
      "step": 12812
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8962775332323477,
      "learning_rate": 1.3867461525101771e-05,
      "loss": 0.3364,
      "step": 12813
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3040649411362182,
      "learning_rate": 1.3866546804776052e-05,
      "loss": 0.1787,
      "step": 12814
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.39092620068470224,
      "learning_rate": 1.3865632046409974e-05,
      "loss": 0.1297,
      "step": 12815
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3651134439376178,
      "learning_rate": 1.3864717250012536e-05,
      "loss": 0.288,
      "step": 12816
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.46598013071738315,
      "learning_rate": 1.3863802415592738e-05,
      "loss": 0.3022,
      "step": 12817
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4900816277187086,
      "learning_rate": 1.3862887543159583e-05,
      "loss": 0.3837,
      "step": 12818
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8288102435766995,
      "learning_rate": 1.3861972632722065e-05,
      "loss": 0.4438,
      "step": 12819
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6177707719459419,
      "learning_rate": 1.386105768428919e-05,
      "loss": 0.3357,
      "step": 12820
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.30488722647576394,
      "learning_rate": 1.3860142697869959e-05,
      "loss": 0.2195,
      "step": 12821
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.8702258935528477,
      "learning_rate": 1.3859227673473377e-05,
      "loss": 0.8792,
      "step": 12822
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.24377483531800012,
      "learning_rate": 1.3858312611108442e-05,
      "loss": 0.205,
      "step": 12823
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5050905867042481,
      "learning_rate": 1.3857397510784156e-05,
      "loss": 0.3105,
      "step": 12824
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3082704849519015,
      "learning_rate": 1.3856482372509526e-05,
      "loss": 0.1724,
      "step": 12825
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5964354800861468,
      "learning_rate": 1.3855567196293551e-05,
      "loss": 0.3344,
      "step": 12826
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8795449751307235,
      "learning_rate": 1.385465198214524e-05,
      "loss": 0.5753,
      "step": 12827
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38268589429139105,
      "learning_rate": 1.3853736730073595e-05,
      "loss": 0.2419,
      "step": 12828
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.38499619099142435,
      "learning_rate": 1.3852821440087621e-05,
      "loss": 0.2847,
      "step": 12829
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3082068595916842,
      "learning_rate": 1.3851906112196316e-05,
      "loss": 0.2436,
      "step": 12830
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.004599654922777,
      "learning_rate": 1.3850990746408697e-05,
      "loss": 0.6023,
      "step": 12831
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9990630207161149,
      "learning_rate": 1.3850075342733762e-05,
      "loss": 0.3091,
      "step": 12832
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27242678298649775,
      "learning_rate": 1.3849159901180515e-05,
      "loss": 0.2083,
      "step": 12833
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27676525040453087,
      "learning_rate": 1.3848244421757973e-05,
      "loss": 0.1824,
      "step": 12834
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3277842719212574,
      "learning_rate": 1.3847328904475134e-05,
      "loss": 0.3289,
      "step": 12835
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7234985534641801,
      "learning_rate": 1.3846413349341006e-05,
      "loss": 0.3924,
      "step": 12836
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4425917284566987,
      "learning_rate": 1.3845497756364596e-05,
      "loss": 0.8911,
      "step": 12837
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.346949700961023,
      "learning_rate": 1.384458212555492e-05,
      "loss": 0.1242,
      "step": 12838
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3440727104763788,
      "learning_rate": 1.3843666456920973e-05,
      "loss": 0.2231,
      "step": 12839
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3596698707891413,
      "learning_rate": 1.3842750750471775e-05,
      "loss": 0.6933,
      "step": 12840
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3045086033190321,
      "learning_rate": 1.3841835006216328e-05,
      "loss": 0.2567,
      "step": 12841
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34225637598016695,
      "learning_rate": 1.3840919224163648e-05,
      "loss": 0.2512,
      "step": 12842
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2983752341262081,
      "learning_rate": 1.3840003404322737e-05,
      "loss": 0.2149,
      "step": 12843
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.7458351911899987,
      "learning_rate": 1.383908754670261e-05,
      "loss": 0.5154,
      "step": 12844
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.30232606333265855,
      "learning_rate": 1.3838171651312278e-05,
      "loss": 0.1966,
      "step": 12845
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.44036499865884277,
      "learning_rate": 1.3837255718160748e-05,
      "loss": 0.3776,
      "step": 12846
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.28826555539939913,
      "learning_rate": 1.3836339747257035e-05,
      "loss": 0.1789,
      "step": 12847
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3735055818473103,
      "learning_rate": 1.383542373861015e-05,
      "loss": 0.309,
      "step": 12848
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.0569570067274954,
      "learning_rate": 1.3834507692229103e-05,
      "loss": 0.2623,
      "step": 12849
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7211635538767643,
      "learning_rate": 1.3833591608122908e-05,
      "loss": 0.5074,
      "step": 12850
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5268238182063855,
      "learning_rate": 1.3832675486300577e-05,
      "loss": 0.3461,
      "step": 12851
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3222245021096179,
      "learning_rate": 1.3831759326771121e-05,
      "loss": 0.1663,
      "step": 12852
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.28999797235543817,
      "learning_rate": 1.3830843129543558e-05,
      "loss": 0.2904,
      "step": 12853
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.26970530778702434,
      "learning_rate": 1.3829926894626904e-05,
      "loss": 0.1291,
      "step": 12854
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4756672079525934,
      "learning_rate": 1.3829010622030166e-05,
      "loss": 0.8056,
      "step": 12855
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2536376799790955,
      "learning_rate": 1.382809431176236e-05,
      "loss": 0.0969,
      "step": 12856
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.37711902225302485,
      "learning_rate": 1.3827177963832503e-05,
      "loss": 0.2954,
      "step": 12857
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.40312647580799804,
      "learning_rate": 1.382626157824961e-05,
      "loss": 0.2728,
      "step": 12858
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.513012107344687,
      "learning_rate": 1.3825345155022695e-05,
      "loss": 0.412,
      "step": 12859
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.5892301509638508,
      "learning_rate": 1.382442869416078e-05,
      "loss": 0.3564,
      "step": 12860
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3922874663861703,
      "learning_rate": 1.3823512195672876e-05,
      "loss": 0.2763,
      "step": 12861
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.25830614379558825,
      "learning_rate": 1.3822595659567998e-05,
      "loss": 0.1881,
      "step": 12862
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.5947986011784556,
      "learning_rate": 1.3821679085855166e-05,
      "loss": 0.8666,
      "step": 12863
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34336072208039126,
      "learning_rate": 1.3820762474543403e-05,
      "loss": 0.2701,
      "step": 12864
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.21342207326525375,
      "learning_rate": 1.3819845825641715e-05,
      "loss": 0.1505,
      "step": 12865
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3826516743598013,
      "learning_rate": 1.3818929139159131e-05,
      "loss": 0.3206,
      "step": 12866
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9738208758604735,
      "learning_rate": 1.3818012415104667e-05,
      "loss": 0.2704,
      "step": 12867
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6986726870853324,
      "learning_rate": 1.3817095653487335e-05,
      "loss": 0.4674,
      "step": 12868
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4139242566405393,
      "learning_rate": 1.3816178854316165e-05,
      "loss": 0.3,
      "step": 12869
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.4088603191876395,
      "learning_rate": 1.3815262017600168e-05,
      "loss": 0.2947,
      "step": 12870
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3224896413412776,
      "learning_rate": 1.381434514334837e-05,
      "loss": 0.2356,
      "step": 12871
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.1874910610574514,
      "learning_rate": 1.3813428231569787e-05,
      "loss": 0.7363,
      "step": 12872
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2870897493183159,
      "learning_rate": 1.3812511282273447e-05,
      "loss": 0.1617,
      "step": 12873
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29182361498494863,
      "learning_rate": 1.3811594295468365e-05,
      "loss": 0.1721,
      "step": 12874
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3269069113190123,
      "learning_rate": 1.3810677271163561e-05,
      "loss": 0.2239,
      "step": 12875
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.7987589674487611,
      "learning_rate": 1.3809760209368062e-05,
      "loss": 0.5096,
      "step": 12876
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2906059785444794,
      "learning_rate": 1.3808843110090888e-05,
      "loss": 0.268,
      "step": 12877
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9636318056577805,
      "learning_rate": 1.3807925973341063e-05,
      "loss": 0.3401,
      "step": 12878
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.6390950108744453,
      "learning_rate": 1.380700879912761e-05,
      "loss": 0.29,
      "step": 12879
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.34458360083770634,
      "learning_rate": 1.3806091587459552e-05,
      "loss": 0.2383,
      "step": 12880
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.3563148773510743,
      "learning_rate": 1.380517433834591e-05,
      "loss": 0.8587,
      "step": 12881
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.36102701573582385,
      "learning_rate": 1.380425705179571e-05,
      "loss": 0.2406,
      "step": 12882
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3026051931972807,
      "learning_rate": 1.3803339727817982e-05,
      "loss": 0.2398,
      "step": 12883
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2551331560466552,
      "learning_rate": 1.3802422366421744e-05,
      "loss": 0.1944,
      "step": 12884
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.476651942454953,
      "learning_rate": 1.3801504967616023e-05,
      "loss": 0.2831,
      "step": 12885
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.112351650309694,
      "learning_rate": 1.3800587531409846e-05,
      "loss": 0.3796,
      "step": 12886
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8663279024809972,
      "learning_rate": 1.3799670057812237e-05,
      "loss": 0.5571,
      "step": 12887
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.32015691093352805,
      "learning_rate": 1.3798752546832225e-05,
      "loss": 0.17,
      "step": 12888
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.27950690674651457,
      "learning_rate": 1.3797834998478833e-05,
      "loss": 0.2891,
      "step": 12889
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9448644904121978,
      "learning_rate": 1.3796917412761095e-05,
      "loss": 0.3757,
      "step": 12890
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8149188575471071,
      "learning_rate": 1.3795999789688033e-05,
      "loss": 0.4929,
      "step": 12891
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.2612656179628127,
      "learning_rate": 1.3795082129268675e-05,
      "loss": 0.1477,
      "step": 12892
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.3186930755072452,
      "learning_rate": 1.3794164431512052e-05,
      "loss": 0.2448,
      "step": 12893
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.8966732943136881,
      "learning_rate": 1.3793246696427187e-05,
      "loss": 0.5781,
      "step": 12894
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.33084967365344775,
      "learning_rate": 1.3792328924023116e-05,
      "loss": 0.2831,
      "step": 12895
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.9363396877341825,
      "learning_rate": 1.3791411114308861e-05,
      "loss": 0.5072,
      "step": 12896
    },
    {
      "epoch": 0.39,
      "grad_norm": 0.29659237316421777,
      "learning_rate": 1.3790493267293461e-05,
      "loss": 0.1504,
      "step": 12897
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5091610837057222,
      "learning_rate": 1.378957538298594e-05,
      "loss": 0.378,
      "step": 12898
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8817994596496054,
      "learning_rate": 1.3788657461395328e-05,
      "loss": 0.5078,
      "step": 12899
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33824757471951405,
      "learning_rate": 1.378773950253066e-05,
      "loss": 0.298,
      "step": 12900
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4893637831707822,
      "learning_rate": 1.378682150640096e-05,
      "loss": 0.2541,
      "step": 12901
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8814056195181568,
      "learning_rate": 1.3785903473015266e-05,
      "loss": 0.4689,
      "step": 12902
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30964750325432294,
      "learning_rate": 1.3784985402382609e-05,
      "loss": 0.2726,
      "step": 12903
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21025869741882935,
      "learning_rate": 1.378406729451202e-05,
      "loss": 0.1439,
      "step": 12904
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0938836959438167,
      "learning_rate": 1.3783149149412532e-05,
      "loss": 0.6384,
      "step": 12905
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3327512858878001,
      "learning_rate": 1.3782230967093178e-05,
      "loss": 0.2004,
      "step": 12906
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3074625922478978,
      "learning_rate": 1.3781312747562987e-05,
      "loss": 0.2824,
      "step": 12907
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8493291902877816,
      "learning_rate": 1.3780394490831e-05,
      "loss": 0.5675,
      "step": 12908
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5882791414136217,
      "learning_rate": 1.3779476196906248e-05,
      "loss": 0.8753,
      "step": 12909
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9193881357689718,
      "learning_rate": 1.3778557865797766e-05,
      "loss": 0.3526,
      "step": 12910
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4263396641160316,
      "learning_rate": 1.377763949751459e-05,
      "loss": 0.3011,
      "step": 12911
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2508291454720314,
      "learning_rate": 1.377672109206575e-05,
      "loss": 0.2365,
      "step": 12912
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37857016932699467,
      "learning_rate": 1.3775802649460288e-05,
      "loss": 0.2592,
      "step": 12913
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30284180561243706,
      "learning_rate": 1.3774884169707237e-05,
      "loss": 0.1459,
      "step": 12914
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8111047683661793,
      "learning_rate": 1.3773965652815631e-05,
      "loss": 0.4129,
      "step": 12915
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30542960229047506,
      "learning_rate": 1.3773047098794511e-05,
      "loss": 0.2211,
      "step": 12916
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9198843104518928,
      "learning_rate": 1.3772128507652913e-05,
      "loss": 0.5169,
      "step": 12917
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3621404451295274,
      "learning_rate": 1.3771209879399871e-05,
      "loss": 0.3306,
      "step": 12918
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.45904506900229336,
      "learning_rate": 1.377029121404443e-05,
      "loss": 0.3019,
      "step": 12919
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3397526819590702,
      "learning_rate": 1.3769372511595617e-05,
      "loss": 0.2978,
      "step": 12920
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.55029581006555,
      "learning_rate": 1.376845377206248e-05,
      "loss": 0.348,
      "step": 12921
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42941623765092635,
      "learning_rate": 1.3767534995454054e-05,
      "loss": 0.2479,
      "step": 12922
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2515438535817782,
      "learning_rate": 1.3766616181779383e-05,
      "loss": 0.1825,
      "step": 12923
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3519628559621133,
      "learning_rate": 1.3765697331047499e-05,
      "loss": 0.2933,
      "step": 12924
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.32210208125543754,
      "learning_rate": 1.3764778443267446e-05,
      "loss": 0.2233,
      "step": 12925
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1285611659170454,
      "learning_rate": 1.3763859518448265e-05,
      "loss": 0.6491,
      "step": 12926
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5859959254664203,
      "learning_rate": 1.3762940556598993e-05,
      "loss": 0.3849,
      "step": 12927
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6947530269038164,
      "learning_rate": 1.3762021557728675e-05,
      "loss": 0.4353,
      "step": 12928
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.49590616769192397,
      "learning_rate": 1.3761102521846351e-05,
      "loss": 0.0699,
      "step": 12929
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27662550901047495,
      "learning_rate": 1.3760183448961067e-05,
      "loss": 0.2471,
      "step": 12930
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.46327074737921903,
      "learning_rate": 1.3759264339081855e-05,
      "loss": 0.3939,
      "step": 12931
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.21931064302617706,
      "learning_rate": 1.3758345192217766e-05,
      "loss": 0.0723,
      "step": 12932
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.448851008335838,
      "learning_rate": 1.375742600837784e-05,
      "loss": 0.7448,
      "step": 12933
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2528932883836116,
      "learning_rate": 1.375650678757112e-05,
      "loss": 0.1943,
      "step": 12934
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.264802676858032,
      "learning_rate": 1.375558752980665e-05,
      "loss": 0.8093,
      "step": 12935
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30968095308182286,
      "learning_rate": 1.3754668235093478e-05,
      "loss": 0.2898,
      "step": 12936
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8470911259909993,
      "learning_rate": 1.375374890344064e-05,
      "loss": 0.4616,
      "step": 12937
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3444734282389695,
      "learning_rate": 1.3752829534857183e-05,
      "loss": 0.2226,
      "step": 12938
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5428093986523485,
      "learning_rate": 1.3751910129352162e-05,
      "loss": 0.4103,
      "step": 12939
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.31127438810294505,
      "learning_rate": 1.3750990686934608e-05,
      "loss": 0.1358,
      "step": 12940
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5090327732977931,
      "learning_rate": 1.3750071207613576e-05,
      "loss": 0.3038,
      "step": 12941
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.25912164081545497,
      "learning_rate": 1.3749151691398109e-05,
      "loss": 0.2341,
      "step": 12942
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28037332475410265,
      "learning_rate": 1.3748232138297255e-05,
      "loss": 0.2163,
      "step": 12943
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.595006349499616,
      "learning_rate": 1.374731254832006e-05,
      "loss": 0.8581,
      "step": 12944
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7160198932803911,
      "learning_rate": 1.374639292147557e-05,
      "loss": 0.3917,
      "step": 12945
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8515019456779638,
      "learning_rate": 1.3745473257772833e-05,
      "loss": 0.4972,
      "step": 12946
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3006088316998129,
      "learning_rate": 1.3744553557220899e-05,
      "loss": 0.1674,
      "step": 12947
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5399212606159648,
      "learning_rate": 1.3743633819828814e-05,
      "loss": 0.3861,
      "step": 12948
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3056193345362273,
      "learning_rate": 1.374271404560563e-05,
      "loss": 0.274,
      "step": 12949
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2745435173026415,
      "learning_rate": 1.374179423456039e-05,
      "loss": 0.1636,
      "step": 12950
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.34753018704522476,
      "learning_rate": 1.3740874386702148e-05,
      "loss": 0.1248,
      "step": 12951
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5528104018091644,
      "learning_rate": 1.3739954502039956e-05,
      "loss": 0.3816,
      "step": 12952
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.698355868572841,
      "learning_rate": 1.3739034580582857e-05,
      "loss": 0.324,
      "step": 12953
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.318050006995604,
      "learning_rate": 1.3738114622339909e-05,
      "loss": 0.3074,
      "step": 12954
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7948798088680922,
      "learning_rate": 1.3737194627320158e-05,
      "loss": 0.4252,
      "step": 12955
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8658198975031643,
      "learning_rate": 1.3736274595532658e-05,
      "loss": 0.4266,
      "step": 12956
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2900205083722776,
      "learning_rate": 1.3735354526986456e-05,
      "loss": 0.1961,
      "step": 12957
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2949362566800032,
      "learning_rate": 1.3734434421690608e-05,
      "loss": 0.1074,
      "step": 12958
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.244857462138782,
      "learning_rate": 1.3733514279654165e-05,
      "loss": 0.2528,
      "step": 12959
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8269220959394004,
      "learning_rate": 1.3732594100886181e-05,
      "loss": 0.3781,
      "step": 12960
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3911040933044033,
      "learning_rate": 1.3731673885395709e-05,
      "loss": 0.2633,
      "step": 12961
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.410890946538967,
      "learning_rate": 1.37307536331918e-05,
      "loss": 0.3177,
      "step": 12962
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6721694899197016,
      "learning_rate": 1.372983334428351e-05,
      "loss": 0.4757,
      "step": 12963
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9254986401394427,
      "learning_rate": 1.3728913018679891e-05,
      "loss": 0.3838,
      "step": 12964
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4491284445370586,
      "learning_rate": 1.372799265639e-05,
      "loss": 0.3309,
      "step": 12965
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.25757753141673057,
      "learning_rate": 1.3727072257422889e-05,
      "loss": 0.2201,
      "step": 12966
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3409006025888335,
      "learning_rate": 1.3726151821787617e-05,
      "loss": 0.7376,
      "step": 12967
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22043540086558255,
      "learning_rate": 1.3725231349493238e-05,
      "loss": 0.0678,
      "step": 12968
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38907681703213554,
      "learning_rate": 1.3724310840548804e-05,
      "loss": 0.315,
      "step": 12969
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28379794124069047,
      "learning_rate": 1.3723390294963375e-05,
      "loss": 0.1974,
      "step": 12970
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5860626758714716,
      "learning_rate": 1.3722469712746009e-05,
      "loss": 0.3746,
      "step": 12971
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3268631520580847,
      "learning_rate": 1.372154909390576e-05,
      "loss": 0.3194,
      "step": 12972
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8838551506378408,
      "learning_rate": 1.3720628438451684e-05,
      "loss": 0.5594,
      "step": 12973
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6060474483830485,
      "learning_rate": 1.3719707746392845e-05,
      "loss": 0.3461,
      "step": 12974
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3214600398506971,
      "learning_rate": 1.3718787017738297e-05,
      "loss": 0.2216,
      "step": 12975
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0822592795814543,
      "learning_rate": 1.3717866252497098e-05,
      "loss": 0.6585,
      "step": 12976
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.34716465474787256,
      "learning_rate": 1.3716945450678305e-05,
      "loss": 0.2741,
      "step": 12977
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4183817058086702,
      "learning_rate": 1.3716024612290984e-05,
      "loss": 0.3336,
      "step": 12978
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3446758382036182,
      "learning_rate": 1.3715103737344186e-05,
      "loss": 0.0753,
      "step": 12979
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3671599748954523,
      "learning_rate": 1.3714182825846979e-05,
      "loss": 0.337,
      "step": 12980
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7786247485119997,
      "learning_rate": 1.3713261877808419e-05,
      "loss": 0.4305,
      "step": 12981
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4128566084820051,
      "learning_rate": 1.3712340893237563e-05,
      "loss": 0.2294,
      "step": 12982
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38268311078161144,
      "learning_rate": 1.3711419872143477e-05,
      "loss": 0.2366,
      "step": 12983
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2603286586361415,
      "learning_rate": 1.3710498814535222e-05,
      "loss": 0.229,
      "step": 12984
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.250024444490284,
      "learning_rate": 1.3709577720421858e-05,
      "loss": 0.5947,
      "step": 12985
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.36130442345404096,
      "learning_rate": 1.370865658981245e-05,
      "loss": 0.1836,
      "step": 12986
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9960009254023107,
      "learning_rate": 1.3707735422716055e-05,
      "loss": 0.6154,
      "step": 12987
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4459276958178927,
      "learning_rate": 1.3706814219141742e-05,
      "loss": 0.2245,
      "step": 12988
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4272572509356239,
      "learning_rate": 1.3705892979098569e-05,
      "loss": 0.3281,
      "step": 12989
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29920142014466905,
      "learning_rate": 1.3704971702595602e-05,
      "loss": 0.2587,
      "step": 12990
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5860604526249935,
      "learning_rate": 1.3704050389641905e-05,
      "loss": 0.8376,
      "step": 12991
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.23501301708399402,
      "learning_rate": 1.3703129040246539e-05,
      "loss": 0.1039,
      "step": 12992
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3259013773246806,
      "learning_rate": 1.3702207654418573e-05,
      "loss": 0.2875,
      "step": 12993
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9068183662427634,
      "learning_rate": 1.3701286232167071e-05,
      "loss": 0.3215,
      "step": 12994
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4376317485310132,
      "learning_rate": 1.3700364773501096e-05,
      "loss": 0.3531,
      "step": 12995
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.47394091318102743,
      "learning_rate": 1.3699443278429714e-05,
      "loss": 0.3184,
      "step": 12996
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2781118293311213,
      "learning_rate": 1.3698521746961993e-05,
      "loss": 0.2019,
      "step": 12997
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7568736090100634,
      "learning_rate": 1.3697600179106996e-05,
      "loss": 0.5536,
      "step": 12998
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.910900830867477,
      "learning_rate": 1.3696678574873794e-05,
      "loss": 0.5853,
      "step": 12999
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2520649335728272,
      "learning_rate": 1.3695756934271449e-05,
      "loss": 0.1749,
      "step": 13000
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35271469897849295,
      "learning_rate": 1.3694835257309037e-05,
      "loss": 0.2256,
      "step": 13001
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4133604315950884,
      "learning_rate": 1.3693913543995614e-05,
      "loss": 0.3514,
      "step": 13002
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8180595891302985,
      "learning_rate": 1.3692991794340254e-05,
      "loss": 0.0722,
      "step": 13003
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7414989205869859,
      "learning_rate": 1.369207000835203e-05,
      "loss": 0.478,
      "step": 13004
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6526037762992339,
      "learning_rate": 1.3691148186040004e-05,
      "loss": 0.3461,
      "step": 13005
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8889665951646002,
      "learning_rate": 1.3690226327413248e-05,
      "loss": 0.4153,
      "step": 13006
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2857660366358686,
      "learning_rate": 1.3689304432480832e-05,
      "loss": 0.2155,
      "step": 13007
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3260193860307822,
      "learning_rate": 1.3688382501251823e-05,
      "loss": 0.3132,
      "step": 13008
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1861313677941756,
      "learning_rate": 1.3687460533735292e-05,
      "loss": 0.1068,
      "step": 13009
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7899283452109761,
      "learning_rate": 1.3686538529940315e-05,
      "loss": 0.3243,
      "step": 13010
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38890571457409934,
      "learning_rate": 1.3685616489875957e-05,
      "loss": 0.2669,
      "step": 13011
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.47020414865970556,
      "learning_rate": 1.3684694413551292e-05,
      "loss": 0.3002,
      "step": 13012
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3359420308956548,
      "learning_rate": 1.3683772300975392e-05,
      "loss": 0.331,
      "step": 13013
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9203474638636698,
      "learning_rate": 1.3682850152157325e-05,
      "loss": 0.3816,
      "step": 13014
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6517815137239098,
      "learning_rate": 1.3681927967106169e-05,
      "loss": 0.3222,
      "step": 13015
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3262755294574674,
      "learning_rate": 1.3681005745830991e-05,
      "loss": 0.2228,
      "step": 13016
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2708615769841933,
      "learning_rate": 1.3680083488340869e-05,
      "loss": 0.8554,
      "step": 13017
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20710170065334033,
      "learning_rate": 1.3679161194644877e-05,
      "loss": 0.0987,
      "step": 13018
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2592462429518386,
      "learning_rate": 1.3678238864752083e-05,
      "loss": 0.2608,
      "step": 13019
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2491185495022784,
      "learning_rate": 1.3677316498671567e-05,
      "loss": 0.1817,
      "step": 13020
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8012366938535918,
      "learning_rate": 1.3676394096412399e-05,
      "loss": 0.5444,
      "step": 13021
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.690445404096864,
      "learning_rate": 1.3675471657983657e-05,
      "loss": 0.38,
      "step": 13022
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7060645466944516,
      "learning_rate": 1.3674549183394414e-05,
      "loss": 0.383,
      "step": 13023
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.568785559355267,
      "learning_rate": 1.3673626672653751e-05,
      "loss": 0.273,
      "step": 13024
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2525019769852689,
      "learning_rate": 1.367270412577074e-05,
      "loss": 0.2231,
      "step": 13025
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4721831174822573,
      "learning_rate": 1.3671781542754455e-05,
      "loss": 0.3493,
      "step": 13026
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8723444405350524,
      "learning_rate": 1.3670858923613977e-05,
      "loss": 0.5769,
      "step": 13027
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2512203316530263,
      "learning_rate": 1.3669936268358377e-05,
      "loss": 0.1576,
      "step": 13028
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3007892178815952,
      "learning_rate": 1.366901357699674e-05,
      "loss": 0.2096,
      "step": 13029
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7019098068288335,
      "learning_rate": 1.3668090849538138e-05,
      "loss": 0.4986,
      "step": 13030
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29787288640592885,
      "learning_rate": 1.3667168085991653e-05,
      "loss": 0.2864,
      "step": 13031
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9432421152973135,
      "learning_rate": 1.3666245286366362e-05,
      "loss": 0.5718,
      "step": 13032
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42663291609476006,
      "learning_rate": 1.3665322450671343e-05,
      "loss": 0.085,
      "step": 13033
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3501209401280223,
      "learning_rate": 1.3664399578915674e-05,
      "loss": 0.2856,
      "step": 13034
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0587590805917888,
      "learning_rate": 1.3663476671108437e-05,
      "loss": 0.5775,
      "step": 13035
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.303334380298772,
      "learning_rate": 1.3662553727258712e-05,
      "loss": 0.2513,
      "step": 13036
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.44231865546175336,
      "learning_rate": 1.3661630747375579e-05,
      "loss": 0.2591,
      "step": 13037
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2005021894743636,
      "learning_rate": 1.3660707731468118e-05,
      "loss": 0.1342,
      "step": 13038
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4161853624325368,
      "learning_rate": 1.3659784679545405e-05,
      "loss": 0.3812,
      "step": 13039
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6143491211713007,
      "learning_rate": 1.365886159161653e-05,
      "loss": 0.3491,
      "step": 13040
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.6783059067050592,
      "learning_rate": 1.365793846769057e-05,
      "loss": 0.8592,
      "step": 13041
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3352684720486224,
      "learning_rate": 1.3657015307776607e-05,
      "loss": 0.2242,
      "step": 13042
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3465363768126497,
      "learning_rate": 1.3656092111883728e-05,
      "loss": 0.285,
      "step": 13043
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3940829323719547,
      "learning_rate": 1.3655168880021008e-05,
      "loss": 0.3013,
      "step": 13044
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.5496126667925159,
      "learning_rate": 1.3654245612197534e-05,
      "loss": 0.7536,
      "step": 13045
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7274426222378068,
      "learning_rate": 1.3653322308422393e-05,
      "loss": 0.3588,
      "step": 13046
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39546575919367066,
      "learning_rate": 1.365239896870466e-05,
      "loss": 0.2664,
      "step": 13047
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3010453453777875,
      "learning_rate": 1.3651475593053424e-05,
      "loss": 0.2263,
      "step": 13048
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2799375830487329,
      "learning_rate": 1.3650552181477774e-05,
      "loss": 0.2214,
      "step": 13049
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5082142385347171,
      "learning_rate": 1.364962873398679e-05,
      "loss": 0.2524,
      "step": 13050
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.228612071997846,
      "learning_rate": 1.3648705250589553e-05,
      "loss": 0.0664,
      "step": 13051
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3793664484344957,
      "learning_rate": 1.3647781731295158e-05,
      "loss": 0.3121,
      "step": 13052
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9475833432139668,
      "learning_rate": 1.3646858176112683e-05,
      "loss": 0.4948,
      "step": 13053
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7515617020916221,
      "learning_rate": 1.3645934585051219e-05,
      "loss": 0.5987,
      "step": 13054
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3026799445641537,
      "learning_rate": 1.3645010958119851e-05,
      "loss": 0.2752,
      "step": 13055
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35647472313551165,
      "learning_rate": 1.3644087295327666e-05,
      "loss": 0.2486,
      "step": 13056
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4517424213276807,
      "learning_rate": 1.3643163596683748e-05,
      "loss": 0.2628,
      "step": 13057
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.26978233865055334,
      "learning_rate": 1.3642239862197191e-05,
      "loss": 0.1998,
      "step": 13058
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8048514431695322,
      "learning_rate": 1.3641316091877079e-05,
      "loss": 0.301,
      "step": 13059
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.40546747887255263,
      "learning_rate": 1.36403922857325e-05,
      "loss": 0.2942,
      "step": 13060
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2990110699629535,
      "learning_rate": 1.3639468443772543e-05,
      "loss": 0.21,
      "step": 13061
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3996575646599228,
      "learning_rate": 1.3638544566006303e-05,
      "loss": 0.3055,
      "step": 13062
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7271132280155171,
      "learning_rate": 1.3637620652442861e-05,
      "loss": 0.5084,
      "step": 13063
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6395752745348872,
      "learning_rate": 1.363669670309131e-05,
      "loss": 0.3487,
      "step": 13064
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5374675686735578,
      "learning_rate": 1.363577271796074e-05,
      "loss": 0.3482,
      "step": 13065
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.34807428861462925,
      "learning_rate": 1.3634848697060243e-05,
      "loss": 0.2223,
      "step": 13066
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28043571169895004,
      "learning_rate": 1.3633924640398906e-05,
      "loss": 0.2631,
      "step": 13067
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20847488248106788,
      "learning_rate": 1.3633000547985824e-05,
      "loss": 0.0742,
      "step": 13068
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0762689795653706,
      "learning_rate": 1.3632076419830088e-05,
      "loss": 0.644,
      "step": 13069
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29776118897605525,
      "learning_rate": 1.363115225594079e-05,
      "loss": 0.1919,
      "step": 13070
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.3945821101713476,
      "learning_rate": 1.3630228056327017e-05,
      "loss": 0.8777,
      "step": 13071
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7052392197571222,
      "learning_rate": 1.362930382099787e-05,
      "loss": 0.3744,
      "step": 13072
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33911814442394583,
      "learning_rate": 1.3628379549962435e-05,
      "loss": 0.3221,
      "step": 13073
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37793145143233325,
      "learning_rate": 1.362745524322981e-05,
      "loss": 0.2187,
      "step": 13074
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3221827689828231,
      "learning_rate": 1.3626530900809088e-05,
      "loss": 0.2343,
      "step": 13075
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4883197421740871,
      "learning_rate": 1.362560652270936e-05,
      "loss": 0.2768,
      "step": 13076
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.34237009054788137,
      "learning_rate": 1.362468210893972e-05,
      "loss": 0.1375,
      "step": 13077
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2863248548004024,
      "learning_rate": 1.3623757659509267e-05,
      "loss": 0.2743,
      "step": 13078
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2833436678273509,
      "learning_rate": 1.3622833174427094e-05,
      "loss": 0.2166,
      "step": 13079
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.671393031667869,
      "learning_rate": 1.3621908653702294e-05,
      "loss": 0.9198,
      "step": 13080
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.007130305430538,
      "learning_rate": 1.3620984097343967e-05,
      "loss": 0.3754,
      "step": 13081
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1799250466536442,
      "learning_rate": 1.362005950536121e-05,
      "loss": 0.6037,
      "step": 13082
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2071758716707846,
      "learning_rate": 1.3619134877763112e-05,
      "loss": 0.0725,
      "step": 13083
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.42166829148244617,
      "learning_rate": 1.3618210214558774e-05,
      "loss": 0.3185,
      "step": 13084
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3252620089673102,
      "learning_rate": 1.3617285515757294e-05,
      "loss": 0.2595,
      "step": 13085
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2746800919111858,
      "learning_rate": 1.361636078136777e-05,
      "loss": 0.1559,
      "step": 13086
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7432334359251517,
      "learning_rate": 1.3615436011399299e-05,
      "loss": 0.467,
      "step": 13087
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3079756775718892,
      "learning_rate": 1.3614511205860977e-05,
      "loss": 0.1694,
      "step": 13088
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4636076487747108,
      "learning_rate": 1.3613586364761906e-05,
      "loss": 0.3736,
      "step": 13089
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33346450675780226,
      "learning_rate": 1.3612661488111183e-05,
      "loss": 0.2911,
      "step": 13090
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8188959689406721,
      "learning_rate": 1.3611736575917905e-05,
      "loss": 0.6264,
      "step": 13091
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.37024646050661414,
      "learning_rate": 1.361081162819118e-05,
      "loss": 0.1292,
      "step": 13092
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.38278163044084,
      "learning_rate": 1.3609886644940095e-05,
      "loss": 0.3206,
      "step": 13093
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2822216973850822,
      "learning_rate": 1.3608961626173763e-05,
      "loss": 0.1538,
      "step": 13094
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48765157307727564,
      "learning_rate": 1.3608036571901277e-05,
      "loss": 0.2763,
      "step": 13095
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29956900086737426,
      "learning_rate": 1.360711148213174e-05,
      "loss": 0.2416,
      "step": 13096
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3680772235199231,
      "learning_rate": 1.3606186356874251e-05,
      "loss": 0.2611,
      "step": 13097
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43178240237037613,
      "learning_rate": 1.3605261196137918e-05,
      "loss": 0.2999,
      "step": 13098
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8559646985177232,
      "learning_rate": 1.3604335999931837e-05,
      "loss": 0.5493,
      "step": 13099
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5835282955504458,
      "learning_rate": 1.3603410768265112e-05,
      "loss": 0.3893,
      "step": 13100
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29435872752348746,
      "learning_rate": 1.3602485501146846e-05,
      "loss": 0.0729,
      "step": 13101
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.279746124620076,
      "learning_rate": 1.3601560198586145e-05,
      "loss": 0.2811,
      "step": 13102
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.40888565035943986,
      "learning_rate": 1.3600634860592108e-05,
      "loss": 0.2691,
      "step": 13103
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30349855423749755,
      "learning_rate": 1.359970948717384e-05,
      "loss": 0.2208,
      "step": 13104
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6495695171289325,
      "learning_rate": 1.3598784078340446e-05,
      "loss": 0.3707,
      "step": 13105
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4040366556772874,
      "learning_rate": 1.359785863410103e-05,
      "loss": 0.3039,
      "step": 13106
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4162638156382272,
      "learning_rate": 1.3596933154464698e-05,
      "loss": 0.2711,
      "step": 13107
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.412917334710488,
      "learning_rate": 1.3596007639440553e-05,
      "loss": 0.363,
      "step": 13108
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.39379511129281464,
      "learning_rate": 1.3595082089037702e-05,
      "loss": 0.2772,
      "step": 13109
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5881613981268222,
      "learning_rate": 1.359415650326525e-05,
      "loss": 0.2951,
      "step": 13110
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33848561389609233,
      "learning_rate": 1.3593230882132305e-05,
      "loss": 0.2219,
      "step": 13111
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.239643450164017,
      "learning_rate": 1.359230522564797e-05,
      "loss": 0.6645,
      "step": 13112
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.17783940972640946,
      "learning_rate": 1.3591379533821355e-05,
      "loss": 0.1051,
      "step": 13113
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3010869083027784,
      "learning_rate": 1.3590453806661566e-05,
      "loss": 0.26,
      "step": 13114
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6517768973756479,
      "learning_rate": 1.3589528044177714e-05,
      "loss": 0.3613,
      "step": 13115
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3105185002113907,
      "learning_rate": 1.3588602246378899e-05,
      "loss": 0.2725,
      "step": 13116
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8631185287817421,
      "learning_rate": 1.3587676413274236e-05,
      "loss": 0.6283,
      "step": 13117
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0693873281625164,
      "learning_rate": 1.3586750544872836e-05,
      "loss": 0.3911,
      "step": 13118
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4959629155301506,
      "learning_rate": 1.35858246411838e-05,
      "loss": 0.3885,
      "step": 13119
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.26048704853308985,
      "learning_rate": 1.3584898702216243e-05,
      "loss": 0.194,
      "step": 13120
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4818896245299882,
      "learning_rate": 1.3583972727979274e-05,
      "loss": 0.3762,
      "step": 13121
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2665380280634399,
      "learning_rate": 1.3583046718481999e-05,
      "loss": 0.0743,
      "step": 13122
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33918855739786014,
      "learning_rate": 1.358212067373353e-05,
      "loss": 0.2954,
      "step": 13123
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2997093691981131,
      "learning_rate": 1.3581194593742984e-05,
      "loss": 0.2074,
      "step": 13124
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.43439576911599453,
      "learning_rate": 1.3580268478519463e-05,
      "loss": 0.3651,
      "step": 13125
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35285872303467786,
      "learning_rate": 1.3579342328072085e-05,
      "loss": 0.2296,
      "step": 13126
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3622856068044529,
      "learning_rate": 1.3578416142409958e-05,
      "loss": 0.2856,
      "step": 13127
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8104788280558474,
      "learning_rate": 1.3577489921542197e-05,
      "loss": 0.4937,
      "step": 13128
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27431124140539304,
      "learning_rate": 1.3576563665477913e-05,
      "loss": 0.2135,
      "step": 13129
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.7685945409598824,
      "learning_rate": 1.3575637374226217e-05,
      "loss": 0.8848,
      "step": 13130
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5587551660769764,
      "learning_rate": 1.3574711047796227e-05,
      "loss": 0.327,
      "step": 13131
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2908882492218071,
      "learning_rate": 1.3573784686197054e-05,
      "loss": 0.2833,
      "step": 13132
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2999982942182956,
      "learning_rate": 1.3572858289437807e-05,
      "loss": 0.2145,
      "step": 13133
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5603465247548933,
      "learning_rate": 1.3571931857527607e-05,
      "loss": 0.3983,
      "step": 13134
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22773286000512588,
      "learning_rate": 1.3571005390475565e-05,
      "loss": 0.1466,
      "step": 13135
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.48974917787867234,
      "learning_rate": 1.3570078888290796e-05,
      "loss": 0.3109,
      "step": 13136
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.302009221683481,
      "learning_rate": 1.3569152350982419e-05,
      "loss": 0.2164,
      "step": 13137
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5225498282016926,
      "learning_rate": 1.3568225778559548e-05,
      "loss": 0.355,
      "step": 13138
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3170779592673483,
      "learning_rate": 1.3567299171031294e-05,
      "loss": 0.2428,
      "step": 13139
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7908599036532421,
      "learning_rate": 1.3566372528406777e-05,
      "loss": 0.4613,
      "step": 13140
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6688922512176043,
      "learning_rate": 1.3565445850695117e-05,
      "loss": 0.348,
      "step": 13141
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2967729418282591,
      "learning_rate": 1.3564519137905423e-05,
      "loss": 0.0686,
      "step": 13142
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3783309515615563,
      "learning_rate": 1.3563592390046819e-05,
      "loss": 0.3222,
      "step": 13143
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29377982372155503,
      "learning_rate": 1.3562665607128423e-05,
      "loss": 0.2627,
      "step": 13144
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.31511466695425633,
      "learning_rate": 1.3561738789159348e-05,
      "loss": 0.1913,
      "step": 13145
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7563273188886401,
      "learning_rate": 1.3560811936148715e-05,
      "loss": 0.0259,
      "step": 13146
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.379886768119158,
      "learning_rate": 1.3559885048105643e-05,
      "loss": 0.2775,
      "step": 13147
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5175327920411075,
      "learning_rate": 1.355895812503925e-05,
      "loss": 0.2726,
      "step": 13148
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9846953439016219,
      "learning_rate": 1.3558031166958653e-05,
      "loss": 0.4572,
      "step": 13149
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28630814510061403,
      "learning_rate": 1.355710417387298e-05,
      "loss": 0.2506,
      "step": 13150
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7674711009755734,
      "learning_rate": 1.3556177145791344e-05,
      "loss": 0.4434,
      "step": 13151
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30802845009336316,
      "learning_rate": 1.3555250082722866e-05,
      "loss": 0.2095,
      "step": 13152
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4649604499811828,
      "learning_rate": 1.3554322984676667e-05,
      "loss": 0.9137,
      "step": 13153
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22179879952169948,
      "learning_rate": 1.3553395851661873e-05,
      "loss": 0.0904,
      "step": 13154
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.32922413204960205,
      "learning_rate": 1.3552468683687597e-05,
      "loss": 0.2403,
      "step": 13155
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33269211389825964,
      "learning_rate": 1.3551541480762968e-05,
      "loss": 0.2985,
      "step": 13156
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3915649500318913,
      "learning_rate": 1.3550614242897105e-05,
      "loss": 0.2427,
      "step": 13157
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9118748306558189,
      "learning_rate": 1.354968697009913e-05,
      "loss": 0.4949,
      "step": 13158
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7096013464392856,
      "learning_rate": 1.3548759662378163e-05,
      "loss": 0.3521,
      "step": 13159
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6108793813428551,
      "learning_rate": 1.3547832319743335e-05,
      "loss": 0.3526,
      "step": 13160
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.31645755104892626,
      "learning_rate": 1.3546904942203765e-05,
      "loss": 0.2151,
      "step": 13161
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33926929127167416,
      "learning_rate": 1.3545977529768575e-05,
      "loss": 0.3153,
      "step": 13162
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2663939573322461,
      "learning_rate": 1.3545050082446894e-05,
      "loss": 0.1605,
      "step": 13163
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4935353244777786,
      "learning_rate": 1.3544122600247845e-05,
      "loss": 0.2763,
      "step": 13164
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.40235673804910804,
      "learning_rate": 1.3543195083180547e-05,
      "loss": 0.0785,
      "step": 13165
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4215327642244214,
      "learning_rate": 1.3542267531254132e-05,
      "loss": 0.3223,
      "step": 13166
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5820286210271959,
      "learning_rate": 1.3541339944477725e-05,
      "loss": 0.3317,
      "step": 13167
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.35893151717772787,
      "learning_rate": 1.3540412322860448e-05,
      "loss": 0.235,
      "step": 13168
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8650600711211595,
      "learning_rate": 1.353948466641143e-05,
      "loss": 0.3747,
      "step": 13169
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.279087979873793,
      "learning_rate": 1.35385569751398e-05,
      "loss": 0.2186,
      "step": 13170
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2757985844499486,
      "learning_rate": 1.353762924905468e-05,
      "loss": 0.8527,
      "step": 13171
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.20638027332746048,
      "learning_rate": 1.3536701488165197e-05,
      "loss": 0.0687,
      "step": 13172
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.475940672221943,
      "learning_rate": 1.3535773692480484e-05,
      "loss": 0.3958,
      "step": 13173
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2301330115899577,
      "learning_rate": 1.3534845862009665e-05,
      "loss": 0.1867,
      "step": 13174
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.41346697662443727,
      "learning_rate": 1.353391799676187e-05,
      "loss": 0.3729,
      "step": 13175
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5511944651100481,
      "learning_rate": 1.3532990096746228e-05,
      "loss": 0.3694,
      "step": 13176
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1082399269711298,
      "learning_rate": 1.3532062161971866e-05,
      "loss": 0.6865,
      "step": 13177
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4044180345264008,
      "learning_rate": 1.3531134192447913e-05,
      "loss": 0.2636,
      "step": 13178
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4382102572098456,
      "learning_rate": 1.3530206188183498e-05,
      "loss": 0.2752,
      "step": 13179
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.29444431185910763,
      "learning_rate": 1.352927814918776e-05,
      "loss": 0.2439,
      "step": 13180
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.892615125890066,
      "learning_rate": 1.3528350075469816e-05,
      "loss": 0.3542,
      "step": 13181
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.40576750695753083,
      "learning_rate": 1.3527421967038806e-05,
      "loss": 0.2966,
      "step": 13182
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.32215300575959843,
      "learning_rate": 1.3526493823903856e-05,
      "loss": 0.2033,
      "step": 13183
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5153894788341246,
      "learning_rate": 1.3525565646074102e-05,
      "loss": 0.3204,
      "step": 13184
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.26034748226079596,
      "learning_rate": 1.352463743355867e-05,
      "loss": 0.1949,
      "step": 13185
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.46135008582178266,
      "learning_rate": 1.3523709186366698e-05,
      "loss": 0.3426,
      "step": 13186
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3820373106593515,
      "learning_rate": 1.3522780904507313e-05,
      "loss": 0.1141,
      "step": 13187
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5289760428279601,
      "learning_rate": 1.3521852587989652e-05,
      "loss": 0.3775,
      "step": 13188
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4149936127447829,
      "learning_rate": 1.3520924236822845e-05,
      "loss": 0.2783,
      "step": 13189
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.044517335187084,
      "learning_rate": 1.351999585101603e-05,
      "loss": 0.5503,
      "step": 13190
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.30137744232551417,
      "learning_rate": 1.3519067430578333e-05,
      "loss": 0.2801,
      "step": 13191
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0094091384043795,
      "learning_rate": 1.3518138975518895e-05,
      "loss": 0.5719,
      "step": 13192
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.33818998577782317,
      "learning_rate": 1.3517210485846847e-05,
      "loss": 0.2417,
      "step": 13193
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.1917302637558445,
      "learning_rate": 1.3516281961571327e-05,
      "loss": 0.1258,
      "step": 13194
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.1670658362800068,
      "learning_rate": 1.3515353402701468e-05,
      "loss": 0.7126,
      "step": 13195
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.181756482384515,
      "learning_rate": 1.3514424809246407e-05,
      "loss": 0.5042,
      "step": 13196
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3953152060690158,
      "learning_rate": 1.3513496181215273e-05,
      "loss": 0.2806,
      "step": 13197
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2668717504740436,
      "learning_rate": 1.351256751861721e-05,
      "loss": 0.2481,
      "step": 13198
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7731584941608196,
      "learning_rate": 1.3511638821461355e-05,
      "loss": 0.5966,
      "step": 13199
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4521394958369913,
      "learning_rate": 1.3510710089756837e-05,
      "loss": 0.3233,
      "step": 13200
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7228683901460159,
      "learning_rate": 1.3509781323512803e-05,
      "loss": 0.3496,
      "step": 13201
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.241684767882314,
      "learning_rate": 1.350885252273838e-05,
      "loss": 0.1831,
      "step": 13202
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.24828564747506768,
      "learning_rate": 1.3507923687442715e-05,
      "loss": 0.2423,
      "step": 13203
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.7952392347226034,
      "learning_rate": 1.3506994817634942e-05,
      "loss": 0.3155,
      "step": 13204
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.2888673875885261,
      "learning_rate": 1.3506065913324197e-05,
      "loss": 0.8031,
      "step": 13205
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2634638006172671,
      "learning_rate": 1.3505136974519626e-05,
      "loss": 0.2064,
      "step": 13206
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.8651632734201848,
      "learning_rate": 1.3504208001230363e-05,
      "loss": 0.488,
      "step": 13207
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6896557182336892,
      "learning_rate": 1.3503278993465547e-05,
      "loss": 0.4869,
      "step": 13208
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28268118679580634,
      "learning_rate": 1.3502349951234322e-05,
      "loss": 0.2935,
      "step": 13209
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.5985183082566297,
      "learning_rate": 1.3501420874545823e-05,
      "loss": 0.3649,
      "step": 13210
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.31617604213765377,
      "learning_rate": 1.3500491763409195e-05,
      "loss": 0.2237,
      "step": 13211
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.40500500268166184,
      "learning_rate": 1.349956261783358e-05,
      "loss": 0.2443,
      "step": 13212
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.22036279677290443,
      "learning_rate": 1.349863343782811e-05,
      "loss": 0.1035,
      "step": 13213
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.4525524791606178,
      "learning_rate": 1.3497704223401937e-05,
      "loss": 0.3618,
      "step": 13214
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.27393841363512816,
      "learning_rate": 1.34967749745642e-05,
      "loss": 0.0764,
      "step": 13215
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.28280627831524563,
      "learning_rate": 1.349584569132404e-05,
      "loss": 0.2777,
      "step": 13216
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.9134681236493664,
      "learning_rate": 1.3494916373690596e-05,
      "loss": 0.3847,
      "step": 13217
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.6767587350037843,
      "learning_rate": 1.3493987021673016e-05,
      "loss": 0.4323,
      "step": 13218
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.3256853098023978,
      "learning_rate": 1.3493057635280443e-05,
      "loss": 0.0665,
      "step": 13219
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2917737413642441,
      "learning_rate": 1.3492128214522023e-05,
      "loss": 0.2128,
      "step": 13220
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.25969461379439013,
      "learning_rate": 1.3491198759406891e-05,
      "loss": 0.262,
      "step": 13221
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.379953489734698,
      "learning_rate": 1.34902692699442e-05,
      "loss": 0.1868,
      "step": 13222
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.4017469943274043,
      "learning_rate": 1.3489339746143092e-05,
      "loss": 0.853,
      "step": 13223
    },
    {
      "epoch": 0.4,
      "grad_norm": 0.2909382087126853,
      "learning_rate": 1.3488410188012708e-05,
      "loss": 0.1924,
      "step": 13224
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47104577305650613,
      "learning_rate": 1.3487480595562199e-05,
      "loss": 0.3963,
      "step": 13225
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5448310800977659,
      "learning_rate": 1.3486550968800708e-05,
      "loss": 0.3556,
      "step": 13226
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8443722297739119,
      "learning_rate": 1.3485621307737382e-05,
      "loss": 0.3914,
      "step": 13227
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.549778155777633,
      "learning_rate": 1.3484691612381365e-05,
      "loss": 0.2489,
      "step": 13228
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33249623984412163,
      "learning_rate": 1.3483761882741808e-05,
      "loss": 0.2804,
      "step": 13229
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.17198453209020037,
      "learning_rate": 1.3482832118827853e-05,
      "loss": 0.0695,
      "step": 13230
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6732331244253689,
      "learning_rate": 1.348190232064865e-05,
      "loss": 0.8954,
      "step": 13231
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3416379416638508,
      "learning_rate": 1.3480972488213348e-05,
      "loss": 0.272,
      "step": 13232
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2426980794472363,
      "learning_rate": 1.3480042621531094e-05,
      "loss": 0.1833,
      "step": 13233
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4035333840703987,
      "learning_rate": 1.3479112720611032e-05,
      "loss": 0.3657,
      "step": 13234
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5233495981576781,
      "learning_rate": 1.3478182785462317e-05,
      "loss": 0.3614,
      "step": 13235
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4675285665997102,
      "learning_rate": 1.3477252816094092e-05,
      "loss": 0.8521,
      "step": 13236
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39961750678641994,
      "learning_rate": 1.347632281251551e-05,
      "loss": 0.1775,
      "step": 13237
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3683087780981555,
      "learning_rate": 1.3475392774735725e-05,
      "loss": 0.2921,
      "step": 13238
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38378302937885894,
      "learning_rate": 1.347446270276388e-05,
      "loss": 0.2313,
      "step": 13239
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.25588493554224095,
      "learning_rate": 1.3473532596609125e-05,
      "loss": 0.1817,
      "step": 13240
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6006493129422353,
      "learning_rate": 1.3472602456280615e-05,
      "loss": 0.4298,
      "step": 13241
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6652391956518828,
      "learning_rate": 1.34716722817875e-05,
      "loss": 0.3935,
      "step": 13242
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3260236272346569,
      "learning_rate": 1.347074207313893e-05,
      "loss": 0.2565,
      "step": 13243
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5783776477999015,
      "learning_rate": 1.3469811830344056e-05,
      "loss": 0.4706,
      "step": 13244
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31094080298415655,
      "learning_rate": 1.3468881553412036e-05,
      "loss": 0.2507,
      "step": 13245
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9413119904160342,
      "learning_rate": 1.3467951242352015e-05,
      "loss": 0.2953,
      "step": 13246
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4087437938984924,
      "learning_rate": 1.346702089717315e-05,
      "loss": 0.2782,
      "step": 13247
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.255874334700937,
      "learning_rate": 1.3466090517884593e-05,
      "loss": 0.1624,
      "step": 13248
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40653202782009923,
      "learning_rate": 1.3465160104495493e-05,
      "loss": 0.2584,
      "step": 13249
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3681590439520714,
      "learning_rate": 1.346422965701501e-05,
      "loss": 0.2917,
      "step": 13250
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37379405830033563,
      "learning_rate": 1.3463299175452298e-05,
      "loss": 0.2948,
      "step": 13251
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.45923861809331934,
      "learning_rate": 1.3462368659816507e-05,
      "loss": 0.2561,
      "step": 13252
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8114900144621601,
      "learning_rate": 1.3461438110116794e-05,
      "loss": 0.5358,
      "step": 13253
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.866807616525078,
      "learning_rate": 1.3460507526362311e-05,
      "loss": 0.2653,
      "step": 13254
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0532428744224025,
      "learning_rate": 1.3459576908562222e-05,
      "loss": 0.5916,
      "step": 13255
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2974012070360838,
      "learning_rate": 1.3458646256725674e-05,
      "loss": 0.1978,
      "step": 13256
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37276434102642747,
      "learning_rate": 1.3457715570861827e-05,
      "loss": 0.3368,
      "step": 13257
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.27180449386563893,
      "learning_rate": 1.3456784850979835e-05,
      "loss": 0.1194,
      "step": 13258
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6476539976584872,
      "learning_rate": 1.3455854097088858e-05,
      "loss": 0.3609,
      "step": 13259
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5671483638329141,
      "learning_rate": 1.345492330919805e-05,
      "loss": 0.3318,
      "step": 13260
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31302981379186023,
      "learning_rate": 1.3453992487316569e-05,
      "loss": 0.2167,
      "step": 13261
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.496898224127483,
      "learning_rate": 1.3453061631453574e-05,
      "loss": 0.8151,
      "step": 13262
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23372039280192206,
      "learning_rate": 1.3452130741618223e-05,
      "loss": 0.2004,
      "step": 13263
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9915687044991576,
      "learning_rate": 1.3451199817819674e-05,
      "loss": 0.4473,
      "step": 13264
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2935101853053489,
      "learning_rate": 1.3450268860067087e-05,
      "loss": 0.1956,
      "step": 13265
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6745961138872758,
      "learning_rate": 1.3449337868369617e-05,
      "loss": 0.824,
      "step": 13266
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4096751191306196,
      "learning_rate": 1.3448406842736426e-05,
      "loss": 0.1345,
      "step": 13267
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3661403331076219,
      "learning_rate": 1.3447475783176676e-05,
      "loss": 0.3277,
      "step": 13268
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4452854811588187,
      "learning_rate": 1.3446544689699521e-05,
      "loss": 0.2063,
      "step": 13269
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4121471136562524,
      "learning_rate": 1.3445613562314131e-05,
      "loss": 0.3031,
      "step": 13270
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6658842241359473,
      "learning_rate": 1.3444682401029659e-05,
      "loss": 0.4291,
      "step": 13271
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9060490788955516,
      "learning_rate": 1.3443751205855265e-05,
      "loss": 0.3899,
      "step": 13272
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3198572033762181,
      "learning_rate": 1.3442819976800114e-05,
      "loss": 0.7373,
      "step": 13273
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2800042910348198,
      "learning_rate": 1.344188871387337e-05,
      "loss": 0.158,
      "step": 13274
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2912392390329239,
      "learning_rate": 1.344095741708419e-05,
      "loss": 0.2769,
      "step": 13275
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8627710727565175,
      "learning_rate": 1.3440026086441741e-05,
      "loss": 0.3491,
      "step": 13276
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9559482650247569,
      "learning_rate": 1.3439094721955182e-05,
      "loss": 0.6074,
      "step": 13277
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.26973745858795733,
      "learning_rate": 1.343816332363368e-05,
      "loss": 0.1357,
      "step": 13278
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37672905002946533,
      "learning_rate": 1.343723189148639e-05,
      "loss": 0.3197,
      "step": 13279
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3598028488031943,
      "learning_rate": 1.3436300425522485e-05,
      "loss": 0.27,
      "step": 13280
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4681650307711601,
      "learning_rate": 1.3435368925751128e-05,
      "loss": 0.3092,
      "step": 13281
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2540705452724633,
      "learning_rate": 1.3434437392181478e-05,
      "loss": 0.1116,
      "step": 13282
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42142229033111434,
      "learning_rate": 1.3433505824822706e-05,
      "loss": 0.284,
      "step": 13283
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.597596511060184,
      "learning_rate": 1.3432574223683972e-05,
      "loss": 0.3012,
      "step": 13284
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2699590028495609,
      "learning_rate": 1.3431642588774443e-05,
      "loss": 0.4021,
      "step": 13285
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3427688502132667,
      "learning_rate": 1.3430710920103285e-05,
      "loss": 0.3502,
      "step": 13286
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31823028136946957,
      "learning_rate": 1.3429779217679664e-05,
      "loss": 0.0676,
      "step": 13287
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3791761769205611,
      "learning_rate": 1.3428847481512746e-05,
      "loss": 0.312,
      "step": 13288
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0609231050458723,
      "learning_rate": 1.3427915711611702e-05,
      "loss": 0.504,
      "step": 13289
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9683970483275224,
      "learning_rate": 1.3426983907985691e-05,
      "loss": 0.1805,
      "step": 13290
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4502380315726039,
      "learning_rate": 1.3426052070643887e-05,
      "loss": 0.2783,
      "step": 13291
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42519790542223374,
      "learning_rate": 1.3425120199595455e-05,
      "loss": 0.3096,
      "step": 13292
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3358080864249705,
      "learning_rate": 1.342418829484956e-05,
      "loss": 0.2485,
      "step": 13293
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.01653876743226,
      "learning_rate": 1.342325635641538e-05,
      "loss": 0.5255,
      "step": 13294
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1096434527203045,
      "learning_rate": 1.3422324384302073e-05,
      "loss": 0.4072,
      "step": 13295
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6268704843836529,
      "learning_rate": 1.3421392378518814e-05,
      "loss": 0.3294,
      "step": 13296
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3594339567427323,
      "learning_rate": 1.3420460339074772e-05,
      "loss": 0.2328,
      "step": 13297
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4938635793642175,
      "learning_rate": 1.3419528265979114e-05,
      "loss": 0.2911,
      "step": 13298
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2579486049611084,
      "learning_rate": 1.3418596159241011e-05,
      "loss": 0.205,
      "step": 13299
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8369379228823183,
      "learning_rate": 1.3417664018869636e-05,
      "loss": 0.0353,
      "step": 13300
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5895813282018062,
      "learning_rate": 1.3416731844874158e-05,
      "loss": 0.2706,
      "step": 13301
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4915133214461988,
      "learning_rate": 1.3415799637263749e-05,
      "loss": 0.2822,
      "step": 13302
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9379243504520915,
      "learning_rate": 1.3414867396047575e-05,
      "loss": 0.518,
      "step": 13303
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3447878509937173,
      "learning_rate": 1.3413935121234816e-05,
      "loss": 0.2531,
      "step": 13304
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9886179973713316,
      "learning_rate": 1.3413002812834636e-05,
      "loss": 0.5719,
      "step": 13305
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2875985648684578,
      "learning_rate": 1.3412070470856216e-05,
      "loss": 0.1924,
      "step": 13306
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2988857565084664,
      "learning_rate": 1.3411138095308722e-05,
      "loss": 0.8545,
      "step": 13307
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3955657571130113,
      "learning_rate": 1.341020568620133e-05,
      "loss": 0.1153,
      "step": 13308
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.419075004624025,
      "learning_rate": 1.3409273243543207e-05,
      "loss": 0.3071,
      "step": 13309
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.30072346932243005,
      "learning_rate": 1.340834076734354e-05,
      "loss": 0.2096,
      "step": 13310
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3189166092553283,
      "learning_rate": 1.340740825761149e-05,
      "loss": 0.2812,
      "step": 13311
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8307064199936524,
      "learning_rate": 1.3406475714356235e-05,
      "loss": 0.5835,
      "step": 13312
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2260976578491165,
      "learning_rate": 1.3405543137586956e-05,
      "loss": 0.0779,
      "step": 13313
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8809857393803598,
      "learning_rate": 1.340461052731282e-05,
      "loss": 0.4632,
      "step": 13314
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3401464434832565,
      "learning_rate": 1.3403677883543007e-05,
      "loss": 0.2222,
      "step": 13315
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.8733286521472812,
      "learning_rate": 1.3402745206286686e-05,
      "loss": 0.7532,
      "step": 13316
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2600770543141138,
      "learning_rate": 1.3401812495553047e-05,
      "loss": 0.2019,
      "step": 13317
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44507505444303025,
      "learning_rate": 1.340087975135125e-05,
      "loss": 0.2835,
      "step": 13318
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31979143424845297,
      "learning_rate": 1.3399946973690478e-05,
      "loss": 0.2307,
      "step": 13319
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5728649237724711,
      "learning_rate": 1.3399014162579915e-05,
      "loss": 0.3222,
      "step": 13320
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6098628265137152,
      "learning_rate": 1.3398081318028731e-05,
      "loss": 0.3474,
      "step": 13321
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3766612228662442,
      "learning_rate": 1.3397148440046102e-05,
      "loss": 0.3166,
      "step": 13322
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8590812507327371,
      "learning_rate": 1.339621552864121e-05,
      "loss": 0.4871,
      "step": 13323
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6787683629574707,
      "learning_rate": 1.3395282583823233e-05,
      "loss": 0.3119,
      "step": 13324
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3617604326768864,
      "learning_rate": 1.3394349605601346e-05,
      "loss": 0.2358,
      "step": 13325
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3805674140897076,
      "learning_rate": 1.3393416593984734e-05,
      "loss": 0.1903,
      "step": 13326
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4801691800683539,
      "learning_rate": 1.3392483548982573e-05,
      "loss": 0.3726,
      "step": 13327
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2971241416825036,
      "learning_rate": 1.339155047060404e-05,
      "loss": 0.1956,
      "step": 13328
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4220282010564474,
      "learning_rate": 1.3390617358858317e-05,
      "loss": 0.3186,
      "step": 13329
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3427107290376139,
      "learning_rate": 1.338968421375459e-05,
      "loss": 0.1682,
      "step": 13330
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1891265458691849,
      "learning_rate": 1.338875103530203e-05,
      "loss": 0.5939,
      "step": 13331
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8035338066013978,
      "learning_rate": 1.3387817823509821e-05,
      "loss": 0.572,
      "step": 13332
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3996128786441799,
      "learning_rate": 1.338688457838715e-05,
      "loss": 0.2635,
      "step": 13333
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.27210619399942526,
      "learning_rate": 1.3385951299943192e-05,
      "loss": 0.2457,
      "step": 13334
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2996172843498155,
      "learning_rate": 1.338501798818713e-05,
      "loss": 0.7024,
      "step": 13335
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6700275771028089,
      "learning_rate": 1.3384084643128148e-05,
      "loss": 0.3484,
      "step": 13336
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5956188504584484,
      "learning_rate": 1.3383151264775427e-05,
      "loss": 0.4259,
      "step": 13337
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.21549390399796567,
      "learning_rate": 1.3382217853138151e-05,
      "loss": 0.1668,
      "step": 13338
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3091156120601266,
      "learning_rate": 1.3381284408225504e-05,
      "loss": 0.1167,
      "step": 13339
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3607568279969847,
      "learning_rate": 1.3380350930046668e-05,
      "loss": 0.3343,
      "step": 13340
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47215164112435193,
      "learning_rate": 1.3379417418610827e-05,
      "loss": 0.2297,
      "step": 13341
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3845375013331468,
      "learning_rate": 1.3378483873927164e-05,
      "loss": 0.3002,
      "step": 13342
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0140109600314582,
      "learning_rate": 1.3377550296004866e-05,
      "loss": 0.5015,
      "step": 13343
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9235132858674402,
      "learning_rate": 1.3376616684853118e-05,
      "loss": 0.4863,
      "step": 13344
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29541788310101064,
      "learning_rate": 1.3375683040481102e-05,
      "loss": 0.2605,
      "step": 13345
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4890990634941575,
      "learning_rate": 1.3374749362898007e-05,
      "loss": 0.3832,
      "step": 13346
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3856929709524751,
      "learning_rate": 1.3373815652113017e-05,
      "loss": 0.2252,
      "step": 13347
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3111898844344171,
      "learning_rate": 1.337288190813532e-05,
      "loss": 0.1773,
      "step": 13348
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2132495817624234,
      "learning_rate": 1.33719481309741e-05,
      "loss": 0.3574,
      "step": 13349
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2856655088685462,
      "learning_rate": 1.3371014320638542e-05,
      "loss": 0.7712,
      "step": 13350
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3183796849915359,
      "learning_rate": 1.337008047713784e-05,
      "loss": 0.0713,
      "step": 13351
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23613244906387532,
      "learning_rate": 1.3369146600481175e-05,
      "loss": 0.2356,
      "step": 13352
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.756898233174155,
      "learning_rate": 1.336821269067774e-05,
      "loss": 0.4598,
      "step": 13353
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.800097070951396,
      "learning_rate": 1.3367278747736716e-05,
      "loss": 0.3664,
      "step": 13354
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0932959052771156,
      "learning_rate": 1.3366344771667296e-05,
      "loss": 0.5495,
      "step": 13355
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2863008856268974,
      "learning_rate": 1.3365410762478673e-05,
      "loss": 0.1922,
      "step": 13356
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.28675651396641916,
      "learning_rate": 1.3364476720180026e-05,
      "loss": 0.1811,
      "step": 13357
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2918981248485986,
      "learning_rate": 1.3363542644780555e-05,
      "loss": 0.2355,
      "step": 13358
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.6971135447219503,
      "learning_rate": 1.3362608536289442e-05,
      "loss": 0.8974,
      "step": 13359
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3020614272940728,
      "learning_rate": 1.3361674394715878e-05,
      "loss": 0.1854,
      "step": 13360
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.47096284412275535,
      "learning_rate": 1.3360740220069057e-05,
      "loss": 0.3899,
      "step": 13361
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8070200664633489,
      "learning_rate": 1.335980601235817e-05,
      "loss": 0.367,
      "step": 13362
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8815343133711682,
      "learning_rate": 1.3358871771592402e-05,
      "loss": 0.6027,
      "step": 13363
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.28489833871147485,
      "learning_rate": 1.335793749778095e-05,
      "loss": 0.2438,
      "step": 13364
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3134149022197837,
      "learning_rate": 1.3357003190933004e-05,
      "loss": 0.2216,
      "step": 13365
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5884957654631179,
      "learning_rate": 1.3356068851057758e-05,
      "loss": 0.3141,
      "step": 13366
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.32474684377293184,
      "learning_rate": 1.33551344781644e-05,
      "loss": 0.1454,
      "step": 13367
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.3279390541320564,
      "learning_rate": 1.3354200072262124e-05,
      "loss": 0.8181,
      "step": 13368
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24346127308647983,
      "learning_rate": 1.3353265633360129e-05,
      "loss": 0.2039,
      "step": 13369
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5289117520172751,
      "learning_rate": 1.3352331161467598e-05,
      "loss": 0.3801,
      "step": 13370
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4456254922924829,
      "learning_rate": 1.3351396656593733e-05,
      "loss": 0.3304,
      "step": 13371
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9070826637376108,
      "learning_rate": 1.3350462118747724e-05,
      "loss": 0.5618,
      "step": 13372
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33030786282216534,
      "learning_rate": 1.3349527547938768e-05,
      "loss": 0.1329,
      "step": 13373
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5761191181226533,
      "learning_rate": 1.3348592944176053e-05,
      "loss": 0.4655,
      "step": 13374
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.21482405359293816,
      "learning_rate": 1.3347658307468783e-05,
      "loss": 0.1404,
      "step": 13375
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3298868328495002,
      "learning_rate": 1.3346723637826147e-05,
      "loss": 0.3132,
      "step": 13376
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.669008756788804,
      "learning_rate": 1.3345788935257345e-05,
      "loss": 0.4231,
      "step": 13377
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.264074241013074,
      "learning_rate": 1.3344854199771571e-05,
      "loss": 0.0759,
      "step": 13378
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3552199056597677,
      "learning_rate": 1.3343919431378017e-05,
      "loss": 0.3493,
      "step": 13379
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5529035672471877,
      "learning_rate": 1.3342984630085885e-05,
      "loss": 0.3598,
      "step": 13380
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3241435708244063,
      "learning_rate": 1.3342049795904368e-05,
      "loss": 0.3324,
      "step": 13381
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9180901583675007,
      "learning_rate": 1.3341114928842669e-05,
      "loss": 0.2707,
      "step": 13382
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4327426784168595,
      "learning_rate": 1.3340180028909982e-05,
      "loss": 0.2673,
      "step": 13383
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.26687148711457137,
      "learning_rate": 1.3339245096115502e-05,
      "loss": 0.1722,
      "step": 13384
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42381613772371013,
      "learning_rate": 1.3338310130468432e-05,
      "loss": 0.2794,
      "step": 13385
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6623746684276164,
      "learning_rate": 1.3337375131977969e-05,
      "loss": 0.3588,
      "step": 13386
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31186313214895384,
      "learning_rate": 1.3336440100653309e-05,
      "loss": 0.2791,
      "step": 13387
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3122901091399947,
      "learning_rate": 1.3335505036503656e-05,
      "loss": 0.2214,
      "step": 13388
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8643533918407331,
      "learning_rate": 1.3334569939538204e-05,
      "loss": 0.5739,
      "step": 13389
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9531995861118439,
      "learning_rate": 1.3333634809766157e-05,
      "loss": 0.3961,
      "step": 13390
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5294521608202383,
      "learning_rate": 1.3332699647196719e-05,
      "loss": 0.2188,
      "step": 13391
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.35167319518171924,
      "learning_rate": 1.3331764451839078e-05,
      "loss": 0.2759,
      "step": 13392
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.18949301962199744,
      "learning_rate": 1.3330829223702444e-05,
      "loss": 0.0654,
      "step": 13393
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23646980814757507,
      "learning_rate": 1.3329893962796017e-05,
      "loss": 0.2234,
      "step": 13394
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7354217604624749,
      "learning_rate": 1.3328958669128997e-05,
      "loss": 0.4022,
      "step": 13395
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5441650462638591,
      "learning_rate": 1.3328023342710586e-05,
      "loss": 0.3406,
      "step": 13396
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3421563238710373,
      "learning_rate": 1.3327087983549987e-05,
      "loss": 0.2344,
      "step": 13397
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8865920289591991,
      "learning_rate": 1.3326152591656403e-05,
      "loss": 0.6737,
      "step": 13398
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29566168038903934,
      "learning_rate": 1.3325217167039033e-05,
      "loss": 0.229,
      "step": 13399
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0273480375955653,
      "learning_rate": 1.332428170970708e-05,
      "loss": 0.6471,
      "step": 13400
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3227049397583967,
      "learning_rate": 1.3323346219669756e-05,
      "loss": 0.0742,
      "step": 13401
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3869910848330682,
      "learning_rate": 1.3322410696936255e-05,
      "loss": 0.3045,
      "step": 13402
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.35793193022528047,
      "learning_rate": 1.3321475141515783e-05,
      "loss": 0.2028,
      "step": 13403
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.497051985833483,
      "learning_rate": 1.3320539553417549e-05,
      "loss": 0.3108,
      "step": 13404
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9148523697572066,
      "learning_rate": 1.3319603932650753e-05,
      "loss": 0.5243,
      "step": 13405
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23639460607316515,
      "learning_rate": 1.3318668279224598e-05,
      "loss": 0.2243,
      "step": 13406
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2746417566334052,
      "learning_rate": 1.3317732593148293e-05,
      "loss": 0.9169,
      "step": 13407
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.262871493558061,
      "learning_rate": 1.3316796874431048e-05,
      "loss": 0.1141,
      "step": 13408
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2352532205125712,
      "learning_rate": 1.3315861123082062e-05,
      "loss": 0.6327,
      "step": 13409
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2895688917327039,
      "learning_rate": 1.3314925339110542e-05,
      "loss": 0.1649,
      "step": 13410
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3710683521090903,
      "learning_rate": 1.3313989522525697e-05,
      "loss": 0.3329,
      "step": 13411
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.25831404606135844,
      "learning_rate": 1.3313053673336732e-05,
      "loss": 0.1931,
      "step": 13412
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6986911439466672,
      "learning_rate": 1.3312117791552852e-05,
      "loss": 0.4934,
      "step": 13413
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.617828142120011,
      "learning_rate": 1.3311181877183272e-05,
      "loss": 0.2923,
      "step": 13414
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3728675932306088,
      "learning_rate": 1.3310245930237192e-05,
      "loss": 0.2943,
      "step": 13415
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24374794277352654,
      "learning_rate": 1.3309309950723825e-05,
      "loss": 0.1265,
      "step": 13416
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2939111481502502,
      "learning_rate": 1.3308373938652377e-05,
      "loss": 0.2487,
      "step": 13417
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2592302084841345,
      "learning_rate": 1.3307437894032058e-05,
      "loss": 0.6187,
      "step": 13418
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2904338378552266,
      "learning_rate": 1.3306501816872076e-05,
      "loss": 0.2075,
      "step": 13419
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4672211705202645,
      "learning_rate": 1.330556570718164e-05,
      "loss": 0.6137,
      "step": 13420
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7209964967358063,
      "learning_rate": 1.3304629564969963e-05,
      "loss": 0.3738,
      "step": 13421
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6964963397008933,
      "learning_rate": 1.3303693390246255e-05,
      "loss": 0.4978,
      "step": 13422
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.26148754363667354,
      "learning_rate": 1.3302757183019722e-05,
      "loss": 0.2163,
      "step": 13423
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38182353808585234,
      "learning_rate": 1.3301820943299577e-05,
      "loss": 0.3038,
      "step": 13424
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9946050507325847,
      "learning_rate": 1.3300884671095032e-05,
      "loss": 0.5736,
      "step": 13425
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3025758859169573,
      "learning_rate": 1.3299948366415293e-05,
      "loss": 0.1743,
      "step": 13426
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.081587249036263,
      "learning_rate": 1.3299012029269583e-05,
      "loss": 0.0836,
      "step": 13427
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.008605559676933,
      "learning_rate": 1.3298075659667105e-05,
      "loss": 0.4414,
      "step": 13428
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2713991100031392,
      "learning_rate": 1.3297139257617072e-05,
      "loss": 0.2278,
      "step": 13429
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.564893150054211,
      "learning_rate": 1.3296202823128694e-05,
      "loss": 0.3196,
      "step": 13430
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.900985071594894,
      "learning_rate": 1.3295266356211193e-05,
      "loss": 0.5938,
      "step": 13431
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8632909299048118,
      "learning_rate": 1.3294329856873778e-05,
      "loss": 0.5123,
      "step": 13432
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37038958039338077,
      "learning_rate": 1.3293393325125661e-05,
      "loss": 0.272,
      "step": 13433
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.252572298526176,
      "learning_rate": 1.3292456760976056e-05,
      "loss": 0.1758,
      "step": 13434
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2338065890421223,
      "learning_rate": 1.329152016443418e-05,
      "loss": 0.2521,
      "step": 13435
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7966094779556372,
      "learning_rate": 1.3290583535509243e-05,
      "loss": 0.3139,
      "step": 13436
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.609966112455296,
      "learning_rate": 1.3289646874210465e-05,
      "loss": 0.3291,
      "step": 13437
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33124114381226016,
      "learning_rate": 1.3288710180547056e-05,
      "loss": 0.2756,
      "step": 13438
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6275814105172666,
      "learning_rate": 1.3287773454528235e-05,
      "loss": 0.4697,
      "step": 13439
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.967916261358992,
      "learning_rate": 1.3286836696163216e-05,
      "loss": 0.3666,
      "step": 13440
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34646867930989494,
      "learning_rate": 1.3285899905461219e-05,
      "loss": 0.3386,
      "step": 13441
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2604445458826645,
      "learning_rate": 1.3284963082431455e-05,
      "loss": 0.187,
      "step": 13442
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.160582650209478,
      "learning_rate": 1.3284026227083144e-05,
      "loss": 0.4719,
      "step": 13443
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2797682214963157,
      "learning_rate": 1.3283089339425504e-05,
      "loss": 0.1631,
      "step": 13444
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5996095621004647,
      "learning_rate": 1.3282152419467746e-05,
      "loss": 0.3186,
      "step": 13445
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5354923753015681,
      "learning_rate": 1.3281215467219095e-05,
      "loss": 0.3558,
      "step": 13446
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.268497699378797,
      "learning_rate": 1.328027848268877e-05,
      "loss": 0.2215,
      "step": 13447
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4090034770757578,
      "learning_rate": 1.3279341465885982e-05,
      "loss": 0.3506,
      "step": 13448
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.636464144694538,
      "learning_rate": 1.3278404416819954e-05,
      "loss": 0.3564,
      "step": 13449
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8113684540750274,
      "learning_rate": 1.3277467335499907e-05,
      "loss": 0.4332,
      "step": 13450
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3417650039501984,
      "learning_rate": 1.3276530221935056e-05,
      "loss": 0.1888,
      "step": 13451
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5251832900104857,
      "learning_rate": 1.3275593076134623e-05,
      "loss": 0.4183,
      "step": 13452
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2382707010152054,
      "learning_rate": 1.3274655898107826e-05,
      "loss": 0.2023,
      "step": 13453
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.432191780683665,
      "learning_rate": 1.327371868786389e-05,
      "loss": 0.2814,
      "step": 13454
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7363116940584317,
      "learning_rate": 1.327278144541203e-05,
      "loss": 0.2513,
      "step": 13455
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.28742585220590794,
      "learning_rate": 1.327184417076147e-05,
      "loss": 0.2248,
      "step": 13456
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7499365864937292,
      "learning_rate": 1.327090686392143e-05,
      "loss": 0.472,
      "step": 13457
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29593273741055115,
      "learning_rate": 1.3269969524901132e-05,
      "loss": 0.25,
      "step": 13458
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5766652973977322,
      "learning_rate": 1.32690321537098e-05,
      "loss": 0.8648,
      "step": 13459
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3052679957466771,
      "learning_rate": 1.3268094750356653e-05,
      "loss": 0.1859,
      "step": 13460
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.526790108007187,
      "learning_rate": 1.3267157314850913e-05,
      "loss": 0.3792,
      "step": 13461
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.24440121273089097,
      "learning_rate": 1.3266219847201804e-05,
      "loss": 0.1238,
      "step": 13462
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0180235559858544,
      "learning_rate": 1.3265282347418551e-05,
      "loss": 0.5569,
      "step": 13463
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34560676474667623,
      "learning_rate": 1.3264344815510375e-05,
      "loss": 0.2788,
      "step": 13464
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3379114383646877,
      "learning_rate": 1.32634072514865e-05,
      "loss": 0.2734,
      "step": 13465
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5304379029337145,
      "learning_rate": 1.3262469655356151e-05,
      "loss": 0.2634,
      "step": 13466
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4598623365802756,
      "learning_rate": 1.3261532027128554e-05,
      "loss": 0.7279,
      "step": 13467
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8317746225963709,
      "learning_rate": 1.3260594366812929e-05,
      "loss": 0.6263,
      "step": 13468
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.27775014025344724,
      "learning_rate": 1.3259656674418504e-05,
      "loss": 0.1959,
      "step": 13469
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43119314178096957,
      "learning_rate": 1.3258718949954506e-05,
      "loss": 0.3499,
      "step": 13470
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38929716523299823,
      "learning_rate": 1.3257781193430157e-05,
      "loss": 0.2389,
      "step": 13471
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4520138734958615,
      "learning_rate": 1.3256843404854687e-05,
      "loss": 0.2957,
      "step": 13472
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5528217667970142,
      "learning_rate": 1.3255905584237318e-05,
      "loss": 0.2545,
      "step": 13473
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2814317612304722,
      "learning_rate": 1.325496773158728e-05,
      "loss": 0.2417,
      "step": 13474
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8548345218080342,
      "learning_rate": 1.3254029846913797e-05,
      "loss": 0.0545,
      "step": 13475
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.4959007072335581,
      "learning_rate": 1.3253091930226098e-05,
      "loss": 0.3953,
      "step": 13476
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3919707027156653,
      "learning_rate": 1.3252153981533409e-05,
      "loss": 0.2769,
      "step": 13477
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6698921701677231,
      "learning_rate": 1.3251216000844963e-05,
      "loss": 0.3514,
      "step": 13478
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3411200073936611,
      "learning_rate": 1.325027798816998e-05,
      "loss": 0.2223,
      "step": 13479
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.5390098818956515,
      "learning_rate": 1.3249339943517695e-05,
      "loss": 0.4854,
      "step": 13480
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.875143116149916,
      "learning_rate": 1.3248401866897331e-05,
      "loss": 0.3739,
      "step": 13481
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3258392239134746,
      "learning_rate": 1.3247463758318122e-05,
      "loss": 0.2011,
      "step": 13482
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.45527505567353566,
      "learning_rate": 1.3246525617789299e-05,
      "loss": 0.2864,
      "step": 13483
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.1969635689110648,
      "learning_rate": 1.3245587445320084e-05,
      "loss": 0.1433,
      "step": 13484
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2212244442348272,
      "learning_rate": 1.3244649240919719e-05,
      "loss": 0.7576,
      "step": 13485
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9671807176497466,
      "learning_rate": 1.3243711004597422e-05,
      "loss": 0.511,
      "step": 13486
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.603566701722412,
      "learning_rate": 1.3242772736362427e-05,
      "loss": 0.3943,
      "step": 13487
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43576160153133453,
      "learning_rate": 1.324183443622397e-05,
      "loss": 0.2802,
      "step": 13488
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3713465068255128,
      "learning_rate": 1.324089610419128e-05,
      "loss": 0.3301,
      "step": 13489
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1118876439518832,
      "learning_rate": 1.3239957740273583e-05,
      "loss": 0.3966,
      "step": 13490
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.42836261795543024,
      "learning_rate": 1.3239019344480122e-05,
      "loss": 0.2246,
      "step": 13491
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.25128851659962553,
      "learning_rate": 1.323808091682012e-05,
      "loss": 0.2071,
      "step": 13492
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41065720685601753,
      "learning_rate": 1.3237142457302813e-05,
      "loss": 0.247,
      "step": 13493
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3690168324463555,
      "learning_rate": 1.323620396593743e-05,
      "loss": 0.2358,
      "step": 13494
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38888510371564494,
      "learning_rate": 1.323526544273321e-05,
      "loss": 0.2767,
      "step": 13495
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.8191524359283847,
      "learning_rate": 1.3234326887699384e-05,
      "loss": 0.4493,
      "step": 13496
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3685045947017403,
      "learning_rate": 1.3233388300845186e-05,
      "loss": 0.2654,
      "step": 13497
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.9058847602377548,
      "learning_rate": 1.323244968217985e-05,
      "loss": 0.5594,
      "step": 13498
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7280333287389795,
      "learning_rate": 1.3231511031712612e-05,
      "loss": 0.3752,
      "step": 13499
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31034173023302897,
      "learning_rate": 1.3230572349452704e-05,
      "loss": 0.2797,
      "step": 13500
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.30181124206240595,
      "learning_rate": 1.3229633635409358e-05,
      "loss": 0.1988,
      "step": 13501
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39266965251995917,
      "learning_rate": 1.322869488959182e-05,
      "loss": 0.2027,
      "step": 13502
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29041348567757197,
      "learning_rate": 1.3227756112009316e-05,
      "loss": 0.1298,
      "step": 13503
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.09965990549501,
      "learning_rate": 1.3226817302671086e-05,
      "loss": 0.7311,
      "step": 13504
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43807377187618834,
      "learning_rate": 1.3225878461586367e-05,
      "loss": 0.1949,
      "step": 13505
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37058818826587614,
      "learning_rate": 1.3224939588764392e-05,
      "loss": 0.2908,
      "step": 13506
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31366744337262614,
      "learning_rate": 1.32240006842144e-05,
      "loss": 0.2749,
      "step": 13507
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6948546688532186,
      "learning_rate": 1.322306174794563e-05,
      "loss": 0.3484,
      "step": 13508
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.41876850323689624,
      "learning_rate": 1.3222122779967316e-05,
      "loss": 0.2121,
      "step": 13509
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33113751177509054,
      "learning_rate": 1.3221183780288702e-05,
      "loss": 0.1815,
      "step": 13510
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.31007501365801177,
      "learning_rate": 1.3220244748919018e-05,
      "loss": 0.2356,
      "step": 13511
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2909156673405068,
      "learning_rate": 1.3219305685867507e-05,
      "loss": 0.2443,
      "step": 13512
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.4808643283900549,
      "learning_rate": 1.3218366591143409e-05,
      "loss": 0.7286,
      "step": 13513
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.43238318738120407,
      "learning_rate": 1.321742746475596e-05,
      "loss": 0.2292,
      "step": 13514
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3647852484296593,
      "learning_rate": 1.3216488306714403e-05,
      "loss": 0.3346,
      "step": 13515
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6180804776389379,
      "learning_rate": 1.3215549117027973e-05,
      "loss": 0.3374,
      "step": 13516
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.2965023033861047,
      "learning_rate": 1.3214609895705914e-05,
      "loss": 0.6529,
      "step": 13517
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.30428655692040885,
      "learning_rate": 1.3213670642757467e-05,
      "loss": 0.2665,
      "step": 13518
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38578841510402073,
      "learning_rate": 1.3212731358191868e-05,
      "loss": 0.2812,
      "step": 13519
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2569824803037135,
      "learning_rate": 1.3211792042018361e-05,
      "loss": 0.1402,
      "step": 13520
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.7198397203565616,
      "learning_rate": 1.3210852694246187e-05,
      "loss": 0.9406,
      "step": 13521
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0972106282803702,
      "learning_rate": 1.3209913314884589e-05,
      "loss": 0.4315,
      "step": 13522
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5553665871304991,
      "learning_rate": 1.320897390394281e-05,
      "loss": 0.2544,
      "step": 13523
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.29284396753241637,
      "learning_rate": 1.3208034461430086e-05,
      "loss": 0.2937,
      "step": 13524
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3668793593134167,
      "learning_rate": 1.3207094987355664e-05,
      "loss": 0.2828,
      "step": 13525
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.1911564164903963,
      "learning_rate": 1.3206155481728787e-05,
      "loss": 0.6116,
      "step": 13526
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0443488314711211,
      "learning_rate": 1.3205215944558694e-05,
      "loss": 0.2851,
      "step": 13527
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3926369854579921,
      "learning_rate": 1.3204276375854637e-05,
      "loss": 0.2941,
      "step": 13528
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2638267675283386,
      "learning_rate": 1.3203336775625855e-05,
      "loss": 0.151,
      "step": 13529
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.34369380516559506,
      "learning_rate": 1.3202397143881586e-05,
      "loss": 0.2618,
      "step": 13530
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.40502939924999376,
      "learning_rate": 1.3201457480631083e-05,
      "loss": 0.2937,
      "step": 13531
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.45398059195292095,
      "learning_rate": 1.3200517785883589e-05,
      "loss": 0.3043,
      "step": 13532
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2910302709924453,
      "learning_rate": 1.3199578059648346e-05,
      "loss": 0.2116,
      "step": 13533
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7574681057222252,
      "learning_rate": 1.3198638301934601e-05,
      "loss": 0.5374,
      "step": 13534
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.37724523155084777,
      "learning_rate": 1.3197698512751605e-05,
      "loss": 0.279,
      "step": 13535
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.39531260573839816,
      "learning_rate": 1.3196758692108596e-05,
      "loss": 0.2192,
      "step": 13536
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.7037412053137614,
      "learning_rate": 1.319581884001482e-05,
      "loss": 0.3501,
      "step": 13537
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.2495916458699024,
      "learning_rate": 1.319487895647953e-05,
      "loss": 0.1899,
      "step": 13538
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.376577879665349,
      "learning_rate": 1.319393904151197e-05,
      "loss": 0.2698,
      "step": 13539
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6510252344324361,
      "learning_rate": 1.3192999095121381e-05,
      "loss": 0.3485,
      "step": 13540
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.38673406292749785,
      "learning_rate": 1.3192059117317024e-05,
      "loss": 0.3442,
      "step": 13541
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.23718905842992347,
      "learning_rate": 1.3191119108108137e-05,
      "loss": 0.1772,
      "step": 13542
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.44224717543913267,
      "learning_rate": 1.3190179067503968e-05,
      "loss": 0.3647,
      "step": 13543
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0644910301366104,
      "learning_rate": 1.3189238995513767e-05,
      "loss": 0.3024,
      "step": 13544
    },
    {
      "epoch": 0.41,
      "grad_norm": 1.0188472606519319,
      "learning_rate": 1.3188298892146784e-05,
      "loss": 0.5187,
      "step": 13545
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.3118990427350687,
      "learning_rate": 1.3187358757412269e-05,
      "loss": 0.162,
      "step": 13546
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.33635523366513076,
      "learning_rate": 1.318641859131947e-05,
      "loss": 0.2329,
      "step": 13547
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.411370282425017,
      "learning_rate": 1.3185478393877638e-05,
      "loss": 0.2611,
      "step": 13548
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.5871951174695542,
      "learning_rate": 1.3184538165096016e-05,
      "loss": 0.3307,
      "step": 13549
    },
    {
      "epoch": 0.41,
      "grad_norm": 0.6578161062121481,
      "learning_rate": 1.3183597904983862e-05,
      "loss": 0.4625,
      "step": 13550
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2950387082758535,
      "learning_rate": 1.3182657613550427e-05,
      "loss": 0.2153,
      "step": 13551
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6675102584977715,
      "learning_rate": 1.3181717290804957e-05,
      "loss": 0.9209,
      "step": 13552
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2772767915624193,
      "learning_rate": 1.3180776936756703e-05,
      "loss": 0.185,
      "step": 13553
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.48952762756394125,
      "learning_rate": 1.3179836551414922e-05,
      "loss": 0.3732,
      "step": 13554
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.29322177481198386,
      "learning_rate": 1.3178896134788864e-05,
      "loss": 0.1691,
      "step": 13555
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4879041316465817,
      "learning_rate": 1.3177955686887777e-05,
      "loss": 0.3487,
      "step": 13556
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40737413431798014,
      "learning_rate": 1.3177015207720916e-05,
      "loss": 0.2225,
      "step": 13557
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.750015209717822,
      "learning_rate": 1.3176074697297536e-05,
      "loss": 0.5159,
      "step": 13558
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2782533642206479,
      "learning_rate": 1.3175134155626885e-05,
      "loss": 0.203,
      "step": 13559
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5049518054358141,
      "learning_rate": 1.3174193582718224e-05,
      "loss": 0.3352,
      "step": 13560
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2903922663992176,
      "learning_rate": 1.31732529785808e-05,
      "loss": 0.2577,
      "step": 13561
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19261976707886588,
      "learning_rate": 1.3172312343223868e-05,
      "loss": 0.0712,
      "step": 13562
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8667801527956793,
      "learning_rate": 1.3171371676656684e-05,
      "loss": 0.5824,
      "step": 13563
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3046373532895769,
      "learning_rate": 1.3170430978888502e-05,
      "loss": 0.0688,
      "step": 13564
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3891335262422191,
      "learning_rate": 1.3169490249928576e-05,
      "loss": 0.3214,
      "step": 13565
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5662448511687779,
      "learning_rate": 1.3168549489786161e-05,
      "loss": 0.2787,
      "step": 13566
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7939969151223366,
      "learning_rate": 1.3167608698470518e-05,
      "loss": 0.5499,
      "step": 13567
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0135095604131512,
      "learning_rate": 1.3166667875990897e-05,
      "loss": 0.5819,
      "step": 13568
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.504706874678029,
      "learning_rate": 1.3165727022356554e-05,
      "loss": 0.2588,
      "step": 13569
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2118851379661781,
      "learning_rate": 1.3164786137576744e-05,
      "loss": 0.1315,
      "step": 13570
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0730224979208778,
      "learning_rate": 1.3163845221660732e-05,
      "loss": 0.6695,
      "step": 13571
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31788977983812894,
      "learning_rate": 1.3162904274617766e-05,
      "loss": 0.2528,
      "step": 13572
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9327071264227418,
      "learning_rate": 1.316196329645711e-05,
      "loss": 0.4786,
      "step": 13573
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31155286208800925,
      "learning_rate": 1.3161022287188017e-05,
      "loss": 0.2304,
      "step": 13574
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7397935478097835,
      "learning_rate": 1.3160081246819747e-05,
      "loss": 0.3558,
      "step": 13575
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7852217858878745,
      "learning_rate": 1.3159140175361555e-05,
      "loss": 0.5567,
      "step": 13576
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26021108233675005,
      "learning_rate": 1.3158199072822707e-05,
      "loss": 0.2164,
      "step": 13577
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3282780681885589,
      "learning_rate": 1.3157257939212454e-05,
      "loss": 0.281,
      "step": 13578
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9993047643205843,
      "learning_rate": 1.3156316774540061e-05,
      "loss": 0.4964,
      "step": 13579
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.28442989291869863,
      "learning_rate": 1.3155375578814783e-05,
      "loss": 0.1774,
      "step": 13580
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9044482119243294,
      "learning_rate": 1.3154434352045884e-05,
      "loss": 0.3533,
      "step": 13581
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5850990041243748,
      "learning_rate": 1.315349309424262e-05,
      "loss": 0.333,
      "step": 13582
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3374897935228265,
      "learning_rate": 1.3152551805414254e-05,
      "loss": 0.2125,
      "step": 13583
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32252644208882264,
      "learning_rate": 1.3151610485570045e-05,
      "loss": 0.319,
      "step": 13584
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8302631468416994,
      "learning_rate": 1.3150669134719257e-05,
      "loss": 0.3556,
      "step": 13585
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2514476819021136,
      "learning_rate": 1.3149727752871147e-05,
      "loss": 0.8372,
      "step": 13586
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2720479446559292,
      "learning_rate": 1.3148786340034982e-05,
      "loss": 0.0724,
      "step": 13587
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24628618977055508,
      "learning_rate": 1.3147844896220018e-05,
      "loss": 0.1944,
      "step": 13588
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.30112400911516646,
      "learning_rate": 1.314690342143552e-05,
      "loss": 0.2539,
      "step": 13589
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.49003115262583163,
      "learning_rate": 1.3145961915690752e-05,
      "loss": 0.3034,
      "step": 13590
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8258761352754226,
      "learning_rate": 1.3145020378994975e-05,
      "loss": 0.448,
      "step": 13591
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6045309618707275,
      "learning_rate": 1.3144078811357454e-05,
      "loss": 0.2336,
      "step": 13592
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9817665585917295,
      "learning_rate": 1.314313721278745e-05,
      "loss": 0.567,
      "step": 13593
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9148254411245894,
      "learning_rate": 1.3142195583294229e-05,
      "loss": 0.4596,
      "step": 13594
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3152326627212477,
      "learning_rate": 1.3141253922887054e-05,
      "loss": 0.2971,
      "step": 13595
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3192871311513737,
      "learning_rate": 1.3140312231575186e-05,
      "loss": 0.1972,
      "step": 13596
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5374926384263745,
      "learning_rate": 1.3139370509367897e-05,
      "loss": 0.4113,
      "step": 13597
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18249671949307314,
      "learning_rate": 1.3138428756274445e-05,
      "loss": 0.0707,
      "step": 13598
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.80285112195513,
      "learning_rate": 1.3137486972304098e-05,
      "loss": 0.5219,
      "step": 13599
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6017365461244679,
      "learning_rate": 1.3136545157466127e-05,
      "loss": 0.3215,
      "step": 13600
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.25361019897923703,
      "learning_rate": 1.3135603311769787e-05,
      "loss": 0.2197,
      "step": 13601
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.45867452971497924,
      "learning_rate": 1.3134661435224353e-05,
      "loss": 0.3248,
      "step": 13602
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0488524749491503,
      "learning_rate": 1.3133719527839086e-05,
      "loss": 0.3715,
      "step": 13603
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2490007378757648,
      "learning_rate": 1.3132777589623256e-05,
      "loss": 0.78,
      "step": 13604
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.30419027800079845,
      "learning_rate": 1.313183562058613e-05,
      "loss": 0.1922,
      "step": 13605
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5547054218162591,
      "learning_rate": 1.3130893620736975e-05,
      "loss": 0.3737,
      "step": 13606
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.18179052167743592,
      "learning_rate": 1.3129951590085058e-05,
      "loss": 0.133,
      "step": 13607
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.45153092770847275,
      "learning_rate": 1.3129009528639644e-05,
      "loss": 0.3422,
      "step": 13608
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44164006404878525,
      "learning_rate": 1.3128067436410008e-05,
      "loss": 0.2538,
      "step": 13609
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5752069158251744,
      "learning_rate": 1.3127125313405416e-05,
      "loss": 0.3623,
      "step": 13610
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.436039993095369,
      "learning_rate": 1.3126183159635138e-05,
      "loss": 0.2179,
      "step": 13611
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2317889275196021,
      "learning_rate": 1.3125240975108436e-05,
      "loss": 0.7686,
      "step": 13612
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27765879405948485,
      "learning_rate": 1.3124298759834588e-05,
      "loss": 0.2536,
      "step": 13613
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2861254468303855,
      "learning_rate": 1.3123356513822861e-05,
      "loss": 0.0686,
      "step": 13614
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3513773731570389,
      "learning_rate": 1.3122414237082521e-05,
      "loss": 0.3004,
      "step": 13615
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4649276015539292,
      "learning_rate": 1.312147192962285e-05,
      "loss": 0.353,
      "step": 13616
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3902578362318703,
      "learning_rate": 1.3120529591453104e-05,
      "loss": 0.3225,
      "step": 13617
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.25942157386223363,
      "learning_rate": 1.3119587222582564e-05,
      "loss": 0.1801,
      "step": 13618
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3926609700371166,
      "learning_rate": 1.3118644823020499e-05,
      "loss": 0.2781,
      "step": 13619
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32392464551802685,
      "learning_rate": 1.311770239277618e-05,
      "loss": 0.2504,
      "step": 13620
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3882253963206104,
      "learning_rate": 1.3116759931858876e-05,
      "loss": 0.7886,
      "step": 13621
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8523985102051058,
      "learning_rate": 1.3115817440277862e-05,
      "loss": 0.4992,
      "step": 13622
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7220247193025118,
      "learning_rate": 1.3114874918042415e-05,
      "loss": 0.443,
      "step": 13623
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3481386610882419,
      "learning_rate": 1.3113932365161803e-05,
      "loss": 0.2227,
      "step": 13624
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3328096498417038,
      "learning_rate": 1.3112989781645298e-05,
      "loss": 0.3354,
      "step": 13625
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6383125388754832,
      "learning_rate": 1.3112047167502176e-05,
      "loss": 0.3388,
      "step": 13626
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2587062284349897,
      "learning_rate": 1.311110452274171e-05,
      "loss": 0.1197,
      "step": 13627
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3449971786779584,
      "learning_rate": 1.311016184737317e-05,
      "loss": 0.2968,
      "step": 13628
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9139322324071141,
      "learning_rate": 1.310921914140584e-05,
      "loss": 0.328,
      "step": 13629
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.459278125459873,
      "learning_rate": 1.310827640484899e-05,
      "loss": 0.9251,
      "step": 13630
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3167727124213534,
      "learning_rate": 1.310733363771189e-05,
      "loss": 0.256,
      "step": 13631
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6696679022436979,
      "learning_rate": 1.310639084000382e-05,
      "loss": 0.3376,
      "step": 13632
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36179915888683883,
      "learning_rate": 1.3105448011734054e-05,
      "loss": 0.2379,
      "step": 13633
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7555654034748776,
      "learning_rate": 1.3104505152911871e-05,
      "loss": 0.4947,
      "step": 13634
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6208841977187987,
      "learning_rate": 1.3103562263546541e-05,
      "loss": 0.3565,
      "step": 13635
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2670831094681937,
      "learning_rate": 1.3102619343647348e-05,
      "loss": 0.2636,
      "step": 13636
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3156673941138368,
      "learning_rate": 1.3101676393223563e-05,
      "loss": 0.0729,
      "step": 13637
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2700064552037671,
      "learning_rate": 1.3100733412284466e-05,
      "loss": 0.2306,
      "step": 13638
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.786698405565127,
      "learning_rate": 1.3099790400839334e-05,
      "loss": 0.4442,
      "step": 13639
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.754357958066652,
      "learning_rate": 1.3098847358897442e-05,
      "loss": 0.4002,
      "step": 13640
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.983845266526218,
      "learning_rate": 1.309790428646807e-05,
      "loss": 0.4876,
      "step": 13641
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2932225957670509,
      "learning_rate": 1.3096961183560502e-05,
      "loss": 0.2377,
      "step": 13642
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31788310869249653,
      "learning_rate": 1.3096018050184006e-05,
      "loss": 0.3113,
      "step": 13643
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6788275882065551,
      "learning_rate": 1.3095074886347865e-05,
      "loss": 0.288,
      "step": 13644
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.317627259816558,
      "learning_rate": 1.309413169206136e-05,
      "loss": 0.6665,
      "step": 13645
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20385205751693664,
      "learning_rate": 1.309318846733377e-05,
      "loss": 0.133,
      "step": 13646
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2946138470499604,
      "learning_rate": 1.3092245212174372e-05,
      "loss": 0.2402,
      "step": 13647
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4139231646503034,
      "learning_rate": 1.309130192659245e-05,
      "loss": 0.2995,
      "step": 13648
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.46739898009711883,
      "learning_rate": 1.3090358610597282e-05,
      "loss": 0.3618,
      "step": 13649
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5581630803384546,
      "learning_rate": 1.3089415264198148e-05,
      "loss": 0.2896,
      "step": 13650
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34887625931199706,
      "learning_rate": 1.3088471887404332e-05,
      "loss": 0.2932,
      "step": 13651
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5748610384093834,
      "learning_rate": 1.3087528480225114e-05,
      "loss": 0.3357,
      "step": 13652
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9828724577051282,
      "learning_rate": 1.3086585042669773e-05,
      "loss": 0.2565,
      "step": 13653
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36617592961501466,
      "learning_rate": 1.3085641574747592e-05,
      "loss": 0.2547,
      "step": 13654
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2748022899154252,
      "learning_rate": 1.3084698076467854e-05,
      "loss": 0.1858,
      "step": 13655
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3273517473861913,
      "learning_rate": 1.3083754547839844e-05,
      "loss": 0.2413,
      "step": 13656
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7757981434930757,
      "learning_rate": 1.3082810988872841e-05,
      "loss": 0.4659,
      "step": 13657
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8810565162713647,
      "learning_rate": 1.308186739957613e-05,
      "loss": 0.6019,
      "step": 13658
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4409760951533843,
      "learning_rate": 1.308092377995899e-05,
      "loss": 0.2471,
      "step": 13659
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3808651868100209,
      "learning_rate": 1.307998013003071e-05,
      "loss": 0.2898,
      "step": 13660
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27558695579388687,
      "learning_rate": 1.307903644980057e-05,
      "loss": 0.2454,
      "step": 13661
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.376818480418448,
      "learning_rate": 1.3078092739277862e-05,
      "loss": 0.6875,
      "step": 13662
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3053071225797253,
      "learning_rate": 1.307714899847186e-05,
      "loss": 0.1423,
      "step": 13663
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6853405162513014,
      "learning_rate": 1.3076205227391853e-05,
      "loss": 0.2628,
      "step": 13664
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24732801131899246,
      "learning_rate": 1.307526142604713e-05,
      "loss": 0.178,
      "step": 13665
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38935549943215797,
      "learning_rate": 1.307431759444697e-05,
      "loss": 0.2967,
      "step": 13666
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.421695439706174,
      "learning_rate": 1.3073373732600665e-05,
      "loss": 0.3607,
      "step": 13667
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6663654427808462,
      "learning_rate": 1.3072429840517495e-05,
      "loss": 0.3506,
      "step": 13668
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39376171951936956,
      "learning_rate": 1.3071485918206755e-05,
      "loss": 0.3097,
      "step": 13669
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33787093809912727,
      "learning_rate": 1.307054196567772e-05,
      "loss": 0.2062,
      "step": 13670
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8937433938794881,
      "learning_rate": 1.3069597982939683e-05,
      "loss": 0.5422,
      "step": 13671
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3897732811518311,
      "learning_rate": 1.3068653970001935e-05,
      "loss": 0.2167,
      "step": 13672
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3614362089352876,
      "learning_rate": 1.3067709926873754e-05,
      "loss": 0.2843,
      "step": 13673
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.21331455121253723,
      "learning_rate": 1.3066765853564438e-05,
      "loss": 0.1567,
      "step": 13674
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2626254970414381,
      "learning_rate": 1.3065821750083269e-05,
      "loss": 0.8545,
      "step": 13675
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8449559431333531,
      "learning_rate": 1.3064877616439538e-05,
      "loss": 0.3757,
      "step": 13676
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7165564312674851,
      "learning_rate": 1.306393345264253e-05,
      "loss": 0.4454,
      "step": 13677
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.28375931280641287,
      "learning_rate": 1.3062989258701538e-05,
      "loss": 0.1947,
      "step": 13678
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3170468369931884,
      "learning_rate": 1.3062045034625847e-05,
      "loss": 0.2449,
      "step": 13679
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6211391512798885,
      "learning_rate": 1.3061100780424756e-05,
      "loss": 0.7522,
      "step": 13680
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0139278559232032,
      "learning_rate": 1.3060156496107543e-05,
      "loss": 0.2951,
      "step": 13681
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33513251076240475,
      "learning_rate": 1.3059212181683505e-05,
      "loss": 0.1786,
      "step": 13682
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3113276170533054,
      "learning_rate": 1.3058267837161928e-05,
      "loss": 0.2112,
      "step": 13683
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2494097565399942,
      "learning_rate": 1.305732346255211e-05,
      "loss": 0.5767,
      "step": 13684
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35854663763971406,
      "learning_rate": 1.3056379057863335e-05,
      "loss": 0.2805,
      "step": 13685
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6937047457111827,
      "learning_rate": 1.3055434623104899e-05,
      "loss": 0.4293,
      "step": 13686
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2957559877913382,
      "learning_rate": 1.305449015828609e-05,
      "loss": 0.201,
      "step": 13687
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4400246773403261,
      "learning_rate": 1.3053545663416207e-05,
      "loss": 0.2538,
      "step": 13688
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1502641503363615,
      "learning_rate": 1.305260113850453e-05,
      "loss": 0.2751,
      "step": 13689
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3330202958120867,
      "learning_rate": 1.3051656583560361e-05,
      "loss": 0.2847,
      "step": 13690
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32505364861587766,
      "learning_rate": 1.3050711998592993e-05,
      "loss": 0.0672,
      "step": 13691
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31657131118080284,
      "learning_rate": 1.3049767383611713e-05,
      "loss": 0.2225,
      "step": 13692
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7787016931656089,
      "learning_rate": 1.3048822738625821e-05,
      "loss": 0.4873,
      "step": 13693
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7255615415251374,
      "learning_rate": 1.3047878063644606e-05,
      "loss": 0.3995,
      "step": 13694
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9549737650642824,
      "learning_rate": 1.3046933358677363e-05,
      "loss": 0.8289,
      "step": 13695
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2514554897877072,
      "learning_rate": 1.3045988623733387e-05,
      "loss": 0.1343,
      "step": 13696
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3145453614493877,
      "learning_rate": 1.3045043858821973e-05,
      "loss": 0.2867,
      "step": 13697
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3832033201593931,
      "learning_rate": 1.3044099063952417e-05,
      "loss": 0.1726,
      "step": 13698
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2388338388785414,
      "learning_rate": 1.3043154239134013e-05,
      "loss": 0.5257,
      "step": 13699
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.582534360025201,
      "learning_rate": 1.3042209384376055e-05,
      "loss": 0.2208,
      "step": 13700
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.378488332162393,
      "learning_rate": 1.304126449968784e-05,
      "loss": 0.2714,
      "step": 13701
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33918966424874286,
      "learning_rate": 1.3040319585078662e-05,
      "loss": 0.2677,
      "step": 13702
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9902764514523116,
      "learning_rate": 1.3039374640557822e-05,
      "loss": 0.5732,
      "step": 13703
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.761831812329778,
      "learning_rate": 1.3038429666134616e-05,
      "loss": 0.4413,
      "step": 13704
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3809146962226694,
      "learning_rate": 1.3037484661818336e-05,
      "loss": 0.2739,
      "step": 13705
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.17199328544660322,
      "learning_rate": 1.3036539627618286e-05,
      "loss": 0.1186,
      "step": 13706
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1781326955863862,
      "learning_rate": 1.3035594563543757e-05,
      "loss": 0.0554,
      "step": 13707
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3358556126606523,
      "learning_rate": 1.303464946960405e-05,
      "loss": 0.2973,
      "step": 13708
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3677936072751775,
      "learning_rate": 1.3033704345808463e-05,
      "loss": 0.134,
      "step": 13709
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.41759785138176897,
      "learning_rate": 1.3032759192166293e-05,
      "loss": 0.3018,
      "step": 13710
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9016199214921989,
      "learning_rate": 1.3031814008686842e-05,
      "loss": 0.3948,
      "step": 13711
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8943466458274641,
      "learning_rate": 1.303086879537941e-05,
      "loss": 0.6199,
      "step": 13712
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.385565364939778,
      "learning_rate": 1.302992355225329e-05,
      "loss": 0.3006,
      "step": 13713
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4057644181772708,
      "learning_rate": 1.3028978279317786e-05,
      "loss": 0.2805,
      "step": 13714
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27506466705120197,
      "learning_rate": 1.3028032976582197e-05,
      "loss": 0.2089,
      "step": 13715
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.35488852576901153,
      "learning_rate": 1.302708764405582e-05,
      "loss": 0.185,
      "step": 13716
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6146227531559822,
      "learning_rate": 1.3026142281747965e-05,
      "loss": 0.3539,
      "step": 13717
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.758563168805746,
      "learning_rate": 1.3025196889667921e-05,
      "loss": 0.4918,
      "step": 13718
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3047026899322604,
      "learning_rate": 1.3024251467825e-05,
      "loss": 0.204,
      "step": 13719
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3166165638579881,
      "learning_rate": 1.3023306016228495e-05,
      "loss": 0.2712,
      "step": 13720
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1544646390660356,
      "learning_rate": 1.302236053488771e-05,
      "loss": 0.7592,
      "step": 13721
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9186411470842174,
      "learning_rate": 1.3021415023811948e-05,
      "loss": 0.5085,
      "step": 13722
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.710227148514658,
      "learning_rate": 1.3020469483010508e-05,
      "loss": 0.4056,
      "step": 13723
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19727007219632317,
      "learning_rate": 1.30195239124927e-05,
      "loss": 0.156,
      "step": 13724
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1854173727122381,
      "learning_rate": 1.301857831226782e-05,
      "loss": 0.6246,
      "step": 13725
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.299736061495754,
      "learning_rate": 1.3017632682345173e-05,
      "loss": 0.2523,
      "step": 13726
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7104052242556893,
      "learning_rate": 1.3016687022734064e-05,
      "loss": 0.5297,
      "step": 13727
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26874626850633504,
      "learning_rate": 1.3015741333443795e-05,
      "loss": 0.1911,
      "step": 13728
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6221185466061538,
      "learning_rate": 1.3014795614483671e-05,
      "loss": 0.5007,
      "step": 13729
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2796678853175993,
      "learning_rate": 1.3013849865862996e-05,
      "loss": 0.5112,
      "step": 13730
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.344430604188818,
      "learning_rate": 1.3012904087591074e-05,
      "loss": 0.7291,
      "step": 13731
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26314788690949015,
      "learning_rate": 1.3011958279677208e-05,
      "loss": 0.2018,
      "step": 13732
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.24804295888818173,
      "learning_rate": 1.3011012442130708e-05,
      "loss": 0.197,
      "step": 13733
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.435059361812747,
      "learning_rate": 1.3010066574960875e-05,
      "loss": 0.2204,
      "step": 13734
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5918874543714322,
      "learning_rate": 1.3009120678177016e-05,
      "loss": 0.356,
      "step": 13735
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5324164117289886,
      "learning_rate": 1.3008174751788438e-05,
      "loss": 0.3833,
      "step": 13736
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3168684492342467,
      "learning_rate": 1.300722879580445e-05,
      "loss": 0.2161,
      "step": 13737
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.739540538694483,
      "learning_rate": 1.3006282810234355e-05,
      "loss": 0.6575,
      "step": 13738
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3078488310149096,
      "learning_rate": 1.3005336795087458e-05,
      "loss": 0.261,
      "step": 13739
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.4126960641682973,
      "learning_rate": 1.300439075037307e-05,
      "loss": 0.8227,
      "step": 13740
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26488178740505014,
      "learning_rate": 1.3003444676100496e-05,
      "loss": 0.0754,
      "step": 13741
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3681641835022297,
      "learning_rate": 1.3002498572279045e-05,
      "loss": 0.2984,
      "step": 13742
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20692864381767379,
      "learning_rate": 1.3001552438918028e-05,
      "loss": 0.1246,
      "step": 13743
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3485792575406554,
      "learning_rate": 1.300060627602675e-05,
      "loss": 0.3022,
      "step": 13744
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5985807043803382,
      "learning_rate": 1.2999660083614518e-05,
      "loss": 0.346,
      "step": 13745
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2856717139045986,
      "learning_rate": 1.2998713861690644e-05,
      "loss": 0.1609,
      "step": 13746
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5221881530016438,
      "learning_rate": 1.2997767610264436e-05,
      "loss": 0.4087,
      "step": 13747
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.859427639889581,
      "learning_rate": 1.2996821329345204e-05,
      "loss": 0.62,
      "step": 13748
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4927559963017672,
      "learning_rate": 1.2995875018942257e-05,
      "loss": 0.3743,
      "step": 13749
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.33611401194852847,
      "learning_rate": 1.299492867906491e-05,
      "loss": 0.2062,
      "step": 13750
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3690011157172776,
      "learning_rate": 1.2993982309722465e-05,
      "loss": 0.2862,
      "step": 13751
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2618083040679718,
      "learning_rate": 1.2993035910924234e-05,
      "loss": 0.1255,
      "step": 13752
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.716445987219733,
      "learning_rate": 1.2992089482679537e-05,
      "loss": 0.4272,
      "step": 13753
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6527430427401812,
      "learning_rate": 1.2991143024997675e-05,
      "loss": 0.3436,
      "step": 13754
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3981168178724804,
      "learning_rate": 1.2990196537887965e-05,
      "loss": 0.2733,
      "step": 13755
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27571971692097125,
      "learning_rate": 1.2989250021359719e-05,
      "loss": 0.2408,
      "step": 13756
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.569904162950048,
      "learning_rate": 1.2988303475422248e-05,
      "loss": 0.8696,
      "step": 13757
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.1148400494391737,
      "learning_rate": 1.2987356900084862e-05,
      "loss": 0.5669,
      "step": 13758
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3878525334802268,
      "learning_rate": 1.2986410295356877e-05,
      "loss": 0.1341,
      "step": 13759
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3634574422794113,
      "learning_rate": 1.2985463661247604e-05,
      "loss": 0.3096,
      "step": 13760
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8893394553522258,
      "learning_rate": 1.2984516997766354e-05,
      "loss": 0.3678,
      "step": 13761
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3282322310235549,
      "learning_rate": 1.298357030492245e-05,
      "loss": 0.3237,
      "step": 13762
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.19769110889252184,
      "learning_rate": 1.2982623582725198e-05,
      "loss": 0.1215,
      "step": 13763
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.34725971039115333,
      "learning_rate": 1.2981676831183911e-05,
      "loss": 0.2804,
      "step": 13764
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8733697413384577,
      "learning_rate": 1.2980730050307909e-05,
      "loss": 0.2616,
      "step": 13765
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5130736849797208,
      "learning_rate": 1.2979783240106503e-05,
      "loss": 0.8304,
      "step": 13766
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3213935510156325,
      "learning_rate": 1.2978836400589008e-05,
      "loss": 0.2647,
      "step": 13767
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.635018139780314,
      "learning_rate": 1.297788953176474e-05,
      "loss": 0.3212,
      "step": 13768
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3407786095612601,
      "learning_rate": 1.2976942633643018e-05,
      "loss": 0.2304,
      "step": 13769
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7286787057465312,
      "learning_rate": 1.2975995706233153e-05,
      "loss": 0.4778,
      "step": 13770
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.38158113857805287,
      "learning_rate": 1.2975048749544463e-05,
      "loss": 0.2232,
      "step": 13771
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2574259545980665,
      "learning_rate": 1.2974101763586266e-05,
      "loss": 0.0714,
      "step": 13772
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6960778433857323,
      "learning_rate": 1.2973154748367874e-05,
      "loss": 0.3839,
      "step": 13773
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26615439752305164,
      "learning_rate": 1.2972207703898608e-05,
      "loss": 0.2363,
      "step": 13774
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2940914012253684,
      "learning_rate": 1.2971260630187788e-05,
      "loss": 0.893,
      "step": 13775
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.5537970420867124,
      "learning_rate": 1.2970313527244724e-05,
      "loss": 0.3631,
      "step": 13776
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7723396086937885,
      "learning_rate": 1.296936639507874e-05,
      "loss": 0.5215,
      "step": 13777
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2599092170041057,
      "learning_rate": 1.2968419233699153e-05,
      "loss": 0.1884,
      "step": 13778
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6150338174145725,
      "learning_rate": 1.296747204311528e-05,
      "loss": 0.4914,
      "step": 13779
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.30728930211414707,
      "learning_rate": 1.2966524823336441e-05,
      "loss": 0.2561,
      "step": 13780
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4404901393307588,
      "learning_rate": 1.2965577574371954e-05,
      "loss": 0.2569,
      "step": 13781
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.23764863597549127,
      "learning_rate": 1.2964630296231142e-05,
      "loss": 0.1299,
      "step": 13782
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7766352916089211,
      "learning_rate": 1.2963682988923319e-05,
      "loss": 0.3801,
      "step": 13783
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0087147384861663,
      "learning_rate": 1.2962735652457805e-05,
      "loss": 0.4956,
      "step": 13784
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.30390740317127435,
      "learning_rate": 1.2961788286843925e-05,
      "loss": 0.2624,
      "step": 13785
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6120780952004762,
      "learning_rate": 1.2960840892090998e-05,
      "loss": 0.4064,
      "step": 13786
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2700049364110586,
      "learning_rate": 1.2959893468208345e-05,
      "loss": 0.2276,
      "step": 13787
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.005913974556191,
      "learning_rate": 1.2958946015205285e-05,
      "loss": 0.5042,
      "step": 13788
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6737807589598159,
      "learning_rate": 1.2957998533091142e-05,
      "loss": 0.3386,
      "step": 13789
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2923916996164707,
      "learning_rate": 1.2957051021875233e-05,
      "loss": 0.1705,
      "step": 13790
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2632437950097573,
      "learning_rate": 1.2956103481566883e-05,
      "loss": 0.1934,
      "step": 13791
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5442805530702017,
      "learning_rate": 1.2955155912175418e-05,
      "loss": 0.3425,
      "step": 13792
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.40308379000825495,
      "learning_rate": 1.2954208313710152e-05,
      "loss": 0.2859,
      "step": 13793
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.010110333430954,
      "learning_rate": 1.2953260686180417e-05,
      "loss": 0.6051,
      "step": 13794
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7497937956336888,
      "learning_rate": 1.2952313029595533e-05,
      "loss": 0.2558,
      "step": 13795
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.48634912295997484,
      "learning_rate": 1.2951365343964817e-05,
      "loss": 0.2899,
      "step": 13796
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.43897234268772606,
      "learning_rate": 1.2950417629297597e-05,
      "loss": 0.2756,
      "step": 13797
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31464691013916296,
      "learning_rate": 1.2949469885603201e-05,
      "loss": 0.2365,
      "step": 13798
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31715398049294313,
      "learning_rate": 1.2948522112890949e-05,
      "loss": 0.174,
      "step": 13799
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2925226201933212,
      "learning_rate": 1.2947574311170167e-05,
      "loss": 0.0741,
      "step": 13800
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.44810633740826633,
      "learning_rate": 1.2946626480450178e-05,
      "loss": 0.3295,
      "step": 13801
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.6338887148187495,
      "learning_rate": 1.294567862074031e-05,
      "loss": 0.6318,
      "step": 13802
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36075501776750735,
      "learning_rate": 1.2944730732049883e-05,
      "loss": 0.3299,
      "step": 13803
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.015491430458304,
      "learning_rate": 1.2943782814388227e-05,
      "loss": 0.3662,
      "step": 13804
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.39169782627798405,
      "learning_rate": 1.2942834867764669e-05,
      "loss": 0.2656,
      "step": 13805
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4002593949059707,
      "learning_rate": 1.2941886892188532e-05,
      "loss": 0.277,
      "step": 13806
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3421867469295417,
      "learning_rate": 1.2940938887669144e-05,
      "loss": 0.6127,
      "step": 13807
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1906544847152842,
      "learning_rate": 1.2939990854215831e-05,
      "loss": 0.0674,
      "step": 13808
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3138791943371694,
      "learning_rate": 1.2939042791837921e-05,
      "loss": 0.2949,
      "step": 13809
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.23463494482047922,
      "learning_rate": 1.293809470054474e-05,
      "loss": 0.1664,
      "step": 13810
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2791830789094,
      "learning_rate": 1.2937146580345616e-05,
      "loss": 0.5206,
      "step": 13811
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.212083277307365,
      "learning_rate": 1.293619843124988e-05,
      "loss": 0.4884,
      "step": 13812
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6772836563152748,
      "learning_rate": 1.2935250253266857e-05,
      "loss": 0.3334,
      "step": 13813
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31384941305385117,
      "learning_rate": 1.2934302046405875e-05,
      "loss": 0.2694,
      "step": 13814
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9703464748293157,
      "learning_rate": 1.2933353810676267e-05,
      "loss": 0.2665,
      "step": 13815
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.31127555432979886,
      "learning_rate": 1.2932405546087354e-05,
      "loss": 0.2879,
      "step": 13816
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3442884732845208,
      "learning_rate": 1.2931457252648473e-05,
      "loss": 0.1684,
      "step": 13817
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3045698016449447,
      "learning_rate": 1.2930508930368952e-05,
      "loss": 0.1908,
      "step": 13818
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36347077276998,
      "learning_rate": 1.2929560579258117e-05,
      "loss": 0.2261,
      "step": 13819
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9327396562663018,
      "learning_rate": 1.2928612199325306e-05,
      "loss": 0.6074,
      "step": 13820
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37266736204359774,
      "learning_rate": 1.2927663790579843e-05,
      "loss": 0.2812,
      "step": 13821
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9094792296889694,
      "learning_rate": 1.2926715353031058e-05,
      "loss": 0.5764,
      "step": 13822
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2735475894835584,
      "learning_rate": 1.2925766886688283e-05,
      "loss": 0.0731,
      "step": 13823
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.29531015329989496,
      "learning_rate": 1.292481839156085e-05,
      "loss": 0.2204,
      "step": 13824
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.327178315499281,
      "learning_rate": 1.2923869867658096e-05,
      "loss": 0.6341,
      "step": 13825
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3076701281603168,
      "learning_rate": 1.292292131498935e-05,
      "loss": 0.2716,
      "step": 13826
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0144473054759637,
      "learning_rate": 1.2921972733563937e-05,
      "loss": 0.535,
      "step": 13827
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.20859087827370154,
      "learning_rate": 1.2921024123391197e-05,
      "loss": 0.1623,
      "step": 13828
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.762098809044959,
      "learning_rate": 1.2920075484480459e-05,
      "loss": 0.5708,
      "step": 13829
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5282310859089365,
      "learning_rate": 1.2919126816841058e-05,
      "loss": 0.3715,
      "step": 13830
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.368947191118144,
      "learning_rate": 1.2918178120482328e-05,
      "loss": 0.8969,
      "step": 13831
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2763204511466149,
      "learning_rate": 1.2917229395413603e-05,
      "loss": 0.1389,
      "step": 13832
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3621749033706758,
      "learning_rate": 1.2916280641644212e-05,
      "loss": 0.3197,
      "step": 13833
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.37498610096338525,
      "learning_rate": 1.2915331859183496e-05,
      "loss": 0.2343,
      "step": 13834
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.327891517112945,
      "learning_rate": 1.2914383048040781e-05,
      "loss": 0.7902,
      "step": 13835
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4206701924483595,
      "learning_rate": 1.2913434208225408e-05,
      "loss": 0.1584,
      "step": 13836
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2870557315750563,
      "learning_rate": 1.2912485339746712e-05,
      "loss": 0.2445,
      "step": 13837
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.655202293059956,
      "learning_rate": 1.2911536442614026e-05,
      "loss": 0.472,
      "step": 13838
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27958930080818817,
      "learning_rate": 1.2910587516836688e-05,
      "loss": 0.2622,
      "step": 13839
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.3158151521327934,
      "learning_rate": 1.2909638562424032e-05,
      "loss": 0.7915,
      "step": 13840
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.2554640003161821,
      "learning_rate": 1.2908689579385393e-05,
      "loss": 0.1365,
      "step": 13841
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3421277764911629,
      "learning_rate": 1.2907740567730109e-05,
      "loss": 0.2561,
      "step": 13842
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.9802911432297033,
      "learning_rate": 1.2906791527467517e-05,
      "loss": 0.309,
      "step": 13843
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5104156429786816,
      "learning_rate": 1.2905842458606954e-05,
      "loss": 0.3416,
      "step": 13844
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3413989501251172,
      "learning_rate": 1.2904893361157757e-05,
      "loss": 0.2195,
      "step": 13845
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.526616542036814,
      "learning_rate": 1.2903944235129262e-05,
      "loss": 0.3168,
      "step": 13846
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.42311727605337046,
      "learning_rate": 1.2902995080530812e-05,
      "loss": 0.3046,
      "step": 13847
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8671717220302936,
      "learning_rate": 1.2902045897371735e-05,
      "loss": 0.5153,
      "step": 13848
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.16818519333957654,
      "learning_rate": 1.2901096685661378e-05,
      "loss": 0.1032,
      "step": 13849
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.285519554317375,
      "learning_rate": 1.290014744540908e-05,
      "loss": 0.0744,
      "step": 13850
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.32026345894400143,
      "learning_rate": 1.2899198176624175e-05,
      "loss": 0.2711,
      "step": 13851
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.4223013430493551,
      "learning_rate": 1.2898248879316001e-05,
      "loss": 0.2823,
      "step": 13852
    },
    {
      "epoch": 0.42,
      "grad_norm": 2.2434316520924398,
      "learning_rate": 1.2897299553493905e-05,
      "loss": 0.6803,
      "step": 13853
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.8664648051071023,
      "learning_rate": 1.2896350199167223e-05,
      "loss": 0.3507,
      "step": 13854
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3869465841243997,
      "learning_rate": 1.2895400816345294e-05,
      "loss": 0.2438,
      "step": 13855
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3908778817060554,
      "learning_rate": 1.2894451405037456e-05,
      "loss": 0.2936,
      "step": 13856
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.30827620432717046,
      "learning_rate": 1.2893501965253057e-05,
      "loss": 0.2964,
      "step": 13857
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0078537729240957,
      "learning_rate": 1.2892552497001431e-05,
      "loss": 0.4923,
      "step": 13858
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7414306190257248,
      "learning_rate": 1.2891603000291923e-05,
      "loss": 0.254,
      "step": 13859
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.1955530562988834,
      "learning_rate": 1.2890653475133874e-05,
      "loss": 0.1528,
      "step": 13860
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0597867126660785,
      "learning_rate": 1.2889703921536623e-05,
      "loss": 0.5523,
      "step": 13861
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6282550884195215,
      "learning_rate": 1.2888754339509513e-05,
      "loss": 0.3106,
      "step": 13862
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27957716426947854,
      "learning_rate": 1.288780472906189e-05,
      "loss": 0.2466,
      "step": 13863
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36394669806505264,
      "learning_rate": 1.2886855090203094e-05,
      "loss": 0.2894,
      "step": 13864
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.5329531748794845,
      "learning_rate": 1.2885905422942467e-05,
      "loss": 0.353,
      "step": 13865
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.2983757366747808,
      "learning_rate": 1.2884955727289353e-05,
      "loss": 0.7675,
      "step": 13866
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.26925274618425404,
      "learning_rate": 1.2884006003253093e-05,
      "loss": 0.1482,
      "step": 13867
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.36565944014768514,
      "learning_rate": 1.2883056250843034e-05,
      "loss": 0.2801,
      "step": 13868
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3184979112271253,
      "learning_rate": 1.288210647006852e-05,
      "loss": 0.2145,
      "step": 13869
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27027155784587154,
      "learning_rate": 1.2881156660938897e-05,
      "loss": 0.2141,
      "step": 13870
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.6450782075626598,
      "learning_rate": 1.2880206823463502e-05,
      "loss": 0.3404,
      "step": 13871
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.525143718892351,
      "learning_rate": 1.2879256957651684e-05,
      "loss": 0.389,
      "step": 13872
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.3692528029002002,
      "learning_rate": 1.2878307063512794e-05,
      "loss": 0.2029,
      "step": 13873
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.548707807543504,
      "learning_rate": 1.2877357141056167e-05,
      "loss": 0.3504,
      "step": 13874
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.27551345464515703,
      "learning_rate": 1.2876407190291156e-05,
      "loss": 0.259,
      "step": 13875
    },
    {
      "epoch": 0.42,
      "grad_norm": 0.7785464481465229,
      "learning_rate": 1.2875457211227104e-05,
      "loss": 0.4123,
      "step": 13876
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.0362213856311204,
      "learning_rate": 1.2874507203873359e-05,
      "loss": 0.4238,
      "step": 13877
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21010381411785561,
      "learning_rate": 1.2873557168239263e-05,
      "loss": 0.1628,
      "step": 13878
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4885738632694853,
      "learning_rate": 1.287260710433417e-05,
      "loss": 0.2626,
      "step": 13879
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3644094383670302,
      "learning_rate": 1.287165701216742e-05,
      "loss": 0.2845,
      "step": 13880
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9675149514856614,
      "learning_rate": 1.2870706891748364e-05,
      "loss": 0.4983,
      "step": 13881
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34119972649370056,
      "learning_rate": 1.2869756743086353e-05,
      "loss": 0.2086,
      "step": 13882
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4932748421466094,
      "learning_rate": 1.2868806566190728e-05,
      "loss": 0.3448,
      "step": 13883
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9702700067798539,
      "learning_rate": 1.2867856361070839e-05,
      "loss": 0.6108,
      "step": 13884
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2685482524422371,
      "learning_rate": 1.2866906127736035e-05,
      "loss": 0.7247,
      "step": 13885
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2763506407704699,
      "learning_rate": 1.2865955866195667e-05,
      "loss": 0.2394,
      "step": 13886
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3358583504162419,
      "learning_rate": 1.286500557645908e-05,
      "loss": 0.2842,
      "step": 13887
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24491335219114094,
      "learning_rate": 1.2864055258535626e-05,
      "loss": 0.1316,
      "step": 13888
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1074604052374981,
      "learning_rate": 1.2863104912434656e-05,
      "loss": 0.4755,
      "step": 13889
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.668076981968892,
      "learning_rate": 1.2862154538165518e-05,
      "loss": 0.3622,
      "step": 13890
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28408059672139574,
      "learning_rate": 1.2861204135737559e-05,
      "loss": 0.1983,
      "step": 13891
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.49613696324242507,
      "learning_rate": 1.2860253705160135e-05,
      "loss": 0.3689,
      "step": 13892
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4018512017795834,
      "learning_rate": 1.2859303246442592e-05,
      "loss": 0.2747,
      "step": 13893
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4253685247509986,
      "learning_rate": 1.2858352759594284e-05,
      "loss": 0.7528,
      "step": 13894
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41184414655288243,
      "learning_rate": 1.2857402244624562e-05,
      "loss": 0.1713,
      "step": 13895
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37578799835735915,
      "learning_rate": 1.2856451701542776e-05,
      "loss": 0.3045,
      "step": 13896
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4876338221503818,
      "learning_rate": 1.2855501130358276e-05,
      "loss": 0.2304,
      "step": 13897
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.49019046839166613,
      "learning_rate": 1.2854550531080416e-05,
      "loss": 0.343,
      "step": 13898
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23428798542566798,
      "learning_rate": 1.2853599903718553e-05,
      "loss": 0.1719,
      "step": 13899
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8682877823594951,
      "learning_rate": 1.285264924828203e-05,
      "loss": 0.3693,
      "step": 13900
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3451242642157284,
      "learning_rate": 1.2851698564780208e-05,
      "loss": 0.2437,
      "step": 13901
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5376930148299954,
      "learning_rate": 1.2850747853222436e-05,
      "loss": 0.7919,
      "step": 13902
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5133018601939935,
      "learning_rate": 1.2849797113618073e-05,
      "loss": 0.2708,
      "step": 13903
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5008362955004716,
      "learning_rate": 1.2848846345976463e-05,
      "loss": 0.2084,
      "step": 13904
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3646252656229395,
      "learning_rate": 1.2847895550306966e-05,
      "loss": 0.3057,
      "step": 13905
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7676591560746377,
      "learning_rate": 1.2846944726618938e-05,
      "loss": 0.3595,
      "step": 13906
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4458200284409561,
      "learning_rate": 1.2845993874921727e-05,
      "loss": 0.3056,
      "step": 13907
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3370309209616631,
      "learning_rate": 1.2845042995224695e-05,
      "loss": 0.1211,
      "step": 13908
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6457558160698427,
      "learning_rate": 1.2844092087537194e-05,
      "loss": 0.4592,
      "step": 13909
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.26658240677387074,
      "learning_rate": 1.2843141151868577e-05,
      "loss": 0.236,
      "step": 13910
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.47944600924125175,
      "learning_rate": 1.2842190188228203e-05,
      "loss": 0.4306,
      "step": 13911
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8813011658565721,
      "learning_rate": 1.2841239196625427e-05,
      "loss": 0.352,
      "step": 13912
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8314658414588232,
      "learning_rate": 1.2840288177069604e-05,
      "loss": 0.5343,
      "step": 13913
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.26563246097806836,
      "learning_rate": 1.2839337129570093e-05,
      "loss": 0.1934,
      "step": 13914
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7388626755235831,
      "learning_rate": 1.2838386054136246e-05,
      "loss": 0.4411,
      "step": 13915
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43437183907253907,
      "learning_rate": 1.2837434950777427e-05,
      "loss": 0.2564,
      "step": 13916
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.17978577109302663,
      "learning_rate": 1.2836483819502987e-05,
      "loss": 0.1273,
      "step": 13917
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6070161257509774,
      "learning_rate": 1.2835532660322285e-05,
      "loss": 0.3109,
      "step": 13918
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3517849456910577,
      "learning_rate": 1.2834581473244682e-05,
      "loss": 0.2354,
      "step": 13919
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4031843786266425,
      "learning_rate": 1.2833630258279535e-05,
      "loss": 0.787,
      "step": 13920
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8254727793653485,
      "learning_rate": 1.2832679015436197e-05,
      "loss": 0.3654,
      "step": 13921
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3263549773044947,
      "learning_rate": 1.2831727744724034e-05,
      "loss": 0.303,
      "step": 13922
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31064410158513805,
      "learning_rate": 1.28307764461524e-05,
      "loss": 0.2135,
      "step": 13923
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6522575049060124,
      "learning_rate": 1.2829825119730654e-05,
      "loss": 0.478,
      "step": 13924
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6746645119162289,
      "learning_rate": 1.2828873765468162e-05,
      "loss": 0.362,
      "step": 13925
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.27620447862996506,
      "learning_rate": 1.2827922383374274e-05,
      "loss": 0.167,
      "step": 13926
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2842225974953924,
      "learning_rate": 1.282697097345836e-05,
      "loss": 0.1194,
      "step": 13927
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38593306368123725,
      "learning_rate": 1.2826019535729773e-05,
      "loss": 0.3115,
      "step": 13928
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29259490638917957,
      "learning_rate": 1.2825068070197875e-05,
      "loss": 0.2496,
      "step": 13929
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7582342977381195,
      "learning_rate": 1.2824116576872028e-05,
      "loss": 0.3658,
      "step": 13930
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8454534392209638,
      "learning_rate": 1.2823165055761592e-05,
      "loss": 0.5047,
      "step": 13931
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3057998686064359,
      "learning_rate": 1.2822213506875931e-05,
      "loss": 0.1655,
      "step": 13932
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4657011439452116,
      "learning_rate": 1.2821261930224407e-05,
      "loss": 0.362,
      "step": 13933
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2844795977372801,
      "learning_rate": 1.2820310325816377e-05,
      "loss": 0.251,
      "step": 13934
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.2033397140681648,
      "learning_rate": 1.2819358693661208e-05,
      "loss": 0.8533,
      "step": 13935
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.16492618315000612,
      "learning_rate": 1.2818407033768256e-05,
      "loss": 0.0721,
      "step": 13936
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3591785054481736,
      "learning_rate": 1.2817455346146893e-05,
      "loss": 0.3111,
      "step": 13937
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.195597391331287,
      "learning_rate": 1.2816503630806474e-05,
      "loss": 0.5734,
      "step": 13938
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.939256777131362,
      "learning_rate": 1.2815551887756368e-05,
      "loss": 0.5669,
      "step": 13939
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31940098110913767,
      "learning_rate": 1.2814600117005936e-05,
      "loss": 0.2856,
      "step": 13940
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38183200440792986,
      "learning_rate": 1.281364831856454e-05,
      "loss": 0.297,
      "step": 13941
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40476407772097106,
      "learning_rate": 1.2812696492441548e-05,
      "loss": 0.259,
      "step": 13942
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8958913637729035,
      "learning_rate": 1.2811744638646322e-05,
      "loss": 0.324,
      "step": 13943
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.25634573289345575,
      "learning_rate": 1.2810792757188224e-05,
      "loss": 0.1813,
      "step": 13944
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2548579843225368,
      "learning_rate": 1.2809840848076624e-05,
      "loss": 0.1687,
      "step": 13945
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2914589241304161,
      "learning_rate": 1.2808888911320889e-05,
      "loss": 0.289,
      "step": 13946
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9773148451406707,
      "learning_rate": 1.2807936946930375e-05,
      "loss": 0.5005,
      "step": 13947
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.681517475609854,
      "learning_rate": 1.2806984954914459e-05,
      "loss": 0.5034,
      "step": 13948
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5405213595141382,
      "learning_rate": 1.2806032935282496e-05,
      "loss": 0.3359,
      "step": 13949
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3860914775591299,
      "learning_rate": 1.2805080888043859e-05,
      "loss": 0.2465,
      "step": 13950
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4372251333331069,
      "learning_rate": 1.2804128813207915e-05,
      "loss": 0.2269,
      "step": 13951
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3473938868670792,
      "learning_rate": 1.2803176710784029e-05,
      "loss": 0.2921,
      "step": 13952
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2011595794568553,
      "learning_rate": 1.2802224580781566e-05,
      "loss": 0.0711,
      "step": 13953
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6030257199491517,
      "learning_rate": 1.2801272423209896e-05,
      "loss": 0.3313,
      "step": 13954
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3458868797558704,
      "learning_rate": 1.2800320238078388e-05,
      "loss": 0.2309,
      "step": 13955
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7624553325418357,
      "learning_rate": 1.2799368025396407e-05,
      "loss": 0.4194,
      "step": 13956
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40345216432181913,
      "learning_rate": 1.279841578517332e-05,
      "loss": 0.3432,
      "step": 13957
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3789383693229438,
      "learning_rate": 1.27974635174185e-05,
      "loss": 0.2655,
      "step": 13958
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30735575574340046,
      "learning_rate": 1.2796511222141316e-05,
      "loss": 0.1985,
      "step": 13959
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3387083546032073,
      "learning_rate": 1.279555889935113e-05,
      "loss": 0.2521,
      "step": 13960
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9055367386627126,
      "learning_rate": 1.2794606549057318e-05,
      "loss": 0.5279,
      "step": 13961
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2355742512486255,
      "learning_rate": 1.2793654171269245e-05,
      "loss": 0.0705,
      "step": 13962
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48660476745260267,
      "learning_rate": 1.2792701765996285e-05,
      "loss": 0.4139,
      "step": 13963
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2804969979958005,
      "learning_rate": 1.2791749333247807e-05,
      "loss": 0.19,
      "step": 13964
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3897756081496353,
      "learning_rate": 1.279079687303318e-05,
      "loss": 0.334,
      "step": 13965
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.898898098796666,
      "learning_rate": 1.2789844385361774e-05,
      "loss": 0.3544,
      "step": 13966
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41400098091884846,
      "learning_rate": 1.278889187024296e-05,
      "loss": 0.3024,
      "step": 13967
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3029045089936445,
      "learning_rate": 1.2787939327686112e-05,
      "loss": 0.1735,
      "step": 13968
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3265816881049178,
      "learning_rate": 1.2786986757700599e-05,
      "loss": 0.2426,
      "step": 13969
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48492137369839,
      "learning_rate": 1.2786034160295795e-05,
      "loss": 0.3504,
      "step": 13970
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1938437143671998,
      "learning_rate": 1.2785081535481068e-05,
      "loss": 0.3174,
      "step": 13971
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8578355677031776,
      "learning_rate": 1.2784128883265791e-05,
      "loss": 0.4454,
      "step": 13972
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30756466365224067,
      "learning_rate": 1.278317620365934e-05,
      "loss": 0.2319,
      "step": 13973
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8865299078866226,
      "learning_rate": 1.2782223496671086e-05,
      "loss": 0.55,
      "step": 13974
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3770852476334245,
      "learning_rate": 1.27812707623104e-05,
      "loss": 0.3045,
      "step": 13975
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29272458542551794,
      "learning_rate": 1.2780318000586658e-05,
      "loss": 0.2305,
      "step": 13976
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2367374261318254,
      "learning_rate": 1.2779365211509234e-05,
      "loss": 0.1322,
      "step": 13977
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5108050048453683,
      "learning_rate": 1.27784123950875e-05,
      "loss": 0.3317,
      "step": 13978
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1962264112907814,
      "learning_rate": 1.277745955133083e-05,
      "loss": 0.3328,
      "step": 13979
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.054079453111561,
      "learning_rate": 1.27765066802486e-05,
      "loss": 0.6174,
      "step": 13980
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.26553219977481407,
      "learning_rate": 1.2775553781850182e-05,
      "loss": 0.2286,
      "step": 13981
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3312559212875812,
      "learning_rate": 1.2774600856144951e-05,
      "loss": 0.1802,
      "step": 13982
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6311272613204052,
      "learning_rate": 1.277364790314229e-05,
      "loss": 0.3869,
      "step": 13983
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7400273778445557,
      "learning_rate": 1.2772694922851565e-05,
      "loss": 0.3921,
      "step": 13984
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4862634481794784,
      "learning_rate": 1.2771741915282155e-05,
      "loss": 0.3094,
      "step": 13985
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.17802651924407537,
      "learning_rate": 1.2770788880443434e-05,
      "loss": 0.0752,
      "step": 13986
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4128809807283035,
      "learning_rate": 1.2769835818344782e-05,
      "loss": 0.3157,
      "step": 13987
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.27671170865218436,
      "learning_rate": 1.2768882728995574e-05,
      "loss": 0.224,
      "step": 13988
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.523432166209923,
      "learning_rate": 1.2767929612405186e-05,
      "loss": 0.6681,
      "step": 13989
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5771343076470583,
      "learning_rate": 1.2766976468582996e-05,
      "loss": 0.3569,
      "step": 13990
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3749097267705456,
      "learning_rate": 1.2766023297538382e-05,
      "loss": 0.2377,
      "step": 13991
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3988183402315742,
      "learning_rate": 1.2765070099280719e-05,
      "loss": 0.3011,
      "step": 13992
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4570972190751606,
      "learning_rate": 1.2764116873819386e-05,
      "loss": 0.3565,
      "step": 13993
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3713793609041038,
      "learning_rate": 1.2763163621163765e-05,
      "loss": 0.2619,
      "step": 13994
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1331425011085024,
      "learning_rate": 1.2762210341323227e-05,
      "loss": 0.0706,
      "step": 13995
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3769808323770567,
      "learning_rate": 1.276125703430716e-05,
      "loss": 0.2874,
      "step": 13996
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8717515483629523,
      "learning_rate": 1.2760303700124934e-05,
      "loss": 0.3202,
      "step": 13997
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9754893601832119,
      "learning_rate": 1.2759350338785932e-05,
      "loss": 0.568,
      "step": 13998
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2724726590854142,
      "learning_rate": 1.2758396950299532e-05,
      "loss": 0.2399,
      "step": 13999
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36436234991381267,
      "learning_rate": 1.2757443534675118e-05,
      "loss": 0.2594,
      "step": 14000
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9020186890201312,
      "learning_rate": 1.2756490091922067e-05,
      "loss": 0.3432,
      "step": 14001
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2723160428602909,
      "learning_rate": 1.2755536622049758e-05,
      "loss": 0.8097,
      "step": 14002
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.275733437678952,
      "learning_rate": 1.2754583125067574e-05,
      "loss": 0.105,
      "step": 14003
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3197196977963217,
      "learning_rate": 1.2753629600984896e-05,
      "loss": 0.1755,
      "step": 14004
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35891382232236463,
      "learning_rate": 1.27526760498111e-05,
      "loss": 0.2276,
      "step": 14005
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39543894326350765,
      "learning_rate": 1.2751722471555575e-05,
      "loss": 0.2999,
      "step": 14006
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7260297193433252,
      "learning_rate": 1.2750768866227699e-05,
      "loss": 0.3605,
      "step": 14007
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9223741232415089,
      "learning_rate": 1.274981523383685e-05,
      "loss": 0.3698,
      "step": 14008
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7017537779323509,
      "learning_rate": 1.2748861574392417e-05,
      "loss": 0.3739,
      "step": 14009
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3546440507211722,
      "learning_rate": 1.274790788790378e-05,
      "loss": 0.247,
      "step": 14010
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3180663213974068,
      "learning_rate": 1.2746954174380319e-05,
      "loss": 0.3122,
      "step": 14011
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7948673667270119,
      "learning_rate": 1.274600043383142e-05,
      "loss": 0.3354,
      "step": 14012
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3380501181809057,
      "learning_rate": 1.2745046666266465e-05,
      "loss": 0.1785,
      "step": 14013
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2567844203046375,
      "learning_rate": 1.2744092871694837e-05,
      "loss": 0.1709,
      "step": 14014
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1007918158544767,
      "learning_rate": 1.274313905012592e-05,
      "loss": 0.5439,
      "step": 14015
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.035647704417448,
      "learning_rate": 1.27421852015691e-05,
      "loss": 0.3769,
      "step": 14016
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33475523912396754,
      "learning_rate": 1.274123132603376e-05,
      "loss": 0.3152,
      "step": 14017
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32511258316644975,
      "learning_rate": 1.2740277423529284e-05,
      "loss": 0.2121,
      "step": 14018
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6467329287242353,
      "learning_rate": 1.2739323494065055e-05,
      "loss": 0.362,
      "step": 14019
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9383814907201876,
      "learning_rate": 1.2738369537650462e-05,
      "loss": 0.5075,
      "step": 14020
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.075583598040031,
      "learning_rate": 1.273741555429489e-05,
      "loss": 0.3297,
      "step": 14021
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.22879176260594314,
      "learning_rate": 1.273646154400772e-05,
      "loss": 0.1445,
      "step": 14022
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24205182087544458,
      "learning_rate": 1.2735507506798341e-05,
      "loss": 0.2206,
      "step": 14023
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4624962581148558,
      "learning_rate": 1.2734553442676138e-05,
      "loss": 0.3341,
      "step": 14024
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7039607322552796,
      "learning_rate": 1.27335993516505e-05,
      "loss": 0.3738,
      "step": 14025
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7315943859334021,
      "learning_rate": 1.2732645233730814e-05,
      "loss": 0.4924,
      "step": 14026
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29838051288650924,
      "learning_rate": 1.2731691088926463e-05,
      "loss": 0.185,
      "step": 14027
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.527060314634418,
      "learning_rate": 1.2730736917246837e-05,
      "loss": 0.3234,
      "step": 14028
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3128761713796243,
      "learning_rate": 1.2729782718701322e-05,
      "loss": 0.2643,
      "step": 14029
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9530407303080998,
      "learning_rate": 1.2728828493299307e-05,
      "loss": 0.5389,
      "step": 14030
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.38754221161979036,
      "learning_rate": 1.272787424105018e-05,
      "loss": 0.1698,
      "step": 14031
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5512707695545678,
      "learning_rate": 1.2726919961963326e-05,
      "loss": 0.3938,
      "step": 14032
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2462948716711295,
      "learning_rate": 1.2725965656048138e-05,
      "loss": 0.1609,
      "step": 14033
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5357242969696472,
      "learning_rate": 1.2725011323314005e-05,
      "loss": 0.3218,
      "step": 14034
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44628152774993046,
      "learning_rate": 1.272405696377031e-05,
      "loss": 0.3342,
      "step": 14035
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2455106376345936,
      "learning_rate": 1.2723102577426449e-05,
      "loss": 0.0731,
      "step": 14036
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3669509505647753,
      "learning_rate": 1.2722148164291807e-05,
      "loss": 0.2999,
      "step": 14037
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.330454845378476,
      "learning_rate": 1.2721193724375775e-05,
      "loss": 0.656,
      "step": 14038
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.134711898569294,
      "learning_rate": 1.2720239257687747e-05,
      "loss": 0.6576,
      "step": 14039
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.24430031276348943,
      "learning_rate": 1.2719284764237108e-05,
      "loss": 0.1608,
      "step": 14040
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.42032912845161546,
      "learning_rate": 1.2718330244033252e-05,
      "loss": 0.318,
      "step": 14041
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3194674695230428,
      "learning_rate": 1.2717375697085567e-05,
      "loss": 0.287,
      "step": 14042
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3812412301059096,
      "learning_rate": 1.2716421123403446e-05,
      "loss": 0.3204,
      "step": 14043
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.064053809607816,
      "learning_rate": 1.2715466522996281e-05,
      "loss": 0.2885,
      "step": 14044
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6791251412310185,
      "learning_rate": 1.2714511895873463e-05,
      "loss": 0.3572,
      "step": 14045
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3367626416644221,
      "learning_rate": 1.2713557242044383e-05,
      "loss": 0.255,
      "step": 14046
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4072447780578872,
      "learning_rate": 1.2712602561518435e-05,
      "loss": 0.2787,
      "step": 14047
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.48643341656129996,
      "learning_rate": 1.2711647854305009e-05,
      "loss": 0.4056,
      "step": 14048
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.47039088717136857,
      "learning_rate": 1.27106931204135e-05,
      "loss": 0.2662,
      "step": 14049
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32198097037179413,
      "learning_rate": 1.2709738359853301e-05,
      "loss": 0.2754,
      "step": 14050
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5289127577819284,
      "learning_rate": 1.2708783572633801e-05,
      "loss": 0.346,
      "step": 14051
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3577403053919837,
      "learning_rate": 1.27078287587644e-05,
      "loss": 0.267,
      "step": 14052
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.21912743872143783,
      "learning_rate": 1.270687391825449e-05,
      "loss": 0.206,
      "step": 14053
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6654356101339496,
      "learning_rate": 1.270591905111346e-05,
      "loss": 0.3543,
      "step": 14054
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3311297238011291,
      "learning_rate": 1.270496415735071e-05,
      "loss": 0.2268,
      "step": 14055
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.334550890377375,
      "learning_rate": 1.2704009236975631e-05,
      "loss": 0.8783,
      "step": 14056
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9914945402890646,
      "learning_rate": 1.2703054289997621e-05,
      "loss": 0.3757,
      "step": 14057
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33425161270071796,
      "learning_rate": 1.2702099316426072e-05,
      "loss": 0.3343,
      "step": 14058
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28074249104061233,
      "learning_rate": 1.2701144316270381e-05,
      "loss": 0.0712,
      "step": 14059
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31390572862552324,
      "learning_rate": 1.2700189289539948e-05,
      "loss": 0.2597,
      "step": 14060
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40957490852737083,
      "learning_rate": 1.269923423624416e-05,
      "loss": 0.2497,
      "step": 14061
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5045464426939522,
      "learning_rate": 1.2698279156392417e-05,
      "loss": 0.1679,
      "step": 14062
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0147364128229865,
      "learning_rate": 1.2697324049994118e-05,
      "loss": 0.4809,
      "step": 14063
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28442165023112276,
      "learning_rate": 1.2696368917058655e-05,
      "loss": 0.2229,
      "step": 14064
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37000777902827564,
      "learning_rate": 1.2695413757595429e-05,
      "loss": 0.3082,
      "step": 14065
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6496696868963966,
      "learning_rate": 1.2694458571613837e-05,
      "loss": 0.3815,
      "step": 14066
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9264853881349205,
      "learning_rate": 1.2693503359123272e-05,
      "loss": 0.5766,
      "step": 14067
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29716532329397866,
      "learning_rate": 1.2692548120133135e-05,
      "loss": 0.2097,
      "step": 14068
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5388237748547746,
      "learning_rate": 1.2691592854652826e-05,
      "loss": 0.3406,
      "step": 14069
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2992055153050712,
      "learning_rate": 1.2690637562691737e-05,
      "loss": 0.2269,
      "step": 14070
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5024341233701026,
      "learning_rate": 1.2689682244259274e-05,
      "loss": 0.2812,
      "step": 14071
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23832834904807232,
      "learning_rate": 1.2688726899364832e-05,
      "loss": 0.1187,
      "step": 14072
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32877886454342253,
      "learning_rate": 1.2687771528017806e-05,
      "loss": 0.2154,
      "step": 14073
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.5936965621483086,
      "learning_rate": 1.2686816130227605e-05,
      "loss": 0.8537,
      "step": 14074
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6800122683632591,
      "learning_rate": 1.2685860706003619e-05,
      "loss": 0.3432,
      "step": 14075
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32911432511832567,
      "learning_rate": 1.2684905255355251e-05,
      "loss": 0.3149,
      "step": 14076
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.27764776850429845,
      "learning_rate": 1.2683949778291903e-05,
      "loss": 0.1971,
      "step": 14077
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9374215414414551,
      "learning_rate": 1.2682994274822974e-05,
      "loss": 0.5225,
      "step": 14078
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8347289817596089,
      "learning_rate": 1.2682038744957868e-05,
      "loss": 0.3384,
      "step": 14079
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30885467655732857,
      "learning_rate": 1.2681083188705976e-05,
      "loss": 0.1675,
      "step": 14080
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28670216244689706,
      "learning_rate": 1.2680127606076708e-05,
      "loss": 0.1673,
      "step": 14081
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39677898529106426,
      "learning_rate": 1.2679171997079465e-05,
      "loss": 0.3272,
      "step": 14082
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2745890622355604,
      "learning_rate": 1.2678216361723644e-05,
      "loss": 0.2483,
      "step": 14083
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9106634175207269,
      "learning_rate": 1.267726070001865e-05,
      "loss": 0.6131,
      "step": 14084
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6662232116490031,
      "learning_rate": 1.2676305011973884e-05,
      "loss": 0.3481,
      "step": 14085
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9618507307393305,
      "learning_rate": 1.2675349297598749e-05,
      "loss": 0.4244,
      "step": 14086
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.32788328739010747,
      "learning_rate": 1.2674393556902644e-05,
      "loss": 0.2093,
      "step": 14087
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.39398780276765544,
      "learning_rate": 1.267343778989498e-05,
      "loss": 0.2513,
      "step": 14088
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.22987843900852362,
      "learning_rate": 1.2672481996585153e-05,
      "loss": 0.1879,
      "step": 14089
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9214094313340437,
      "learning_rate": 1.2671526176982567e-05,
      "loss": 0.4953,
      "step": 14090
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.41107623881641614,
      "learning_rate": 1.2670570331096632e-05,
      "loss": 0.2534,
      "step": 14091
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.47752657927419123,
      "learning_rate": 1.2669614458936745e-05,
      "loss": 0.2998,
      "step": 14092
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8349084714313689,
      "learning_rate": 1.2668658560512313e-05,
      "loss": 0.4366,
      "step": 14093
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30894504879407936,
      "learning_rate": 1.2667702635832737e-05,
      "loss": 0.2381,
      "step": 14094
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7943828724007884,
      "learning_rate": 1.2666746684907429e-05,
      "loss": 0.3397,
      "step": 14095
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3385693954322257,
      "learning_rate": 1.2665790707745786e-05,
      "loss": 0.2238,
      "step": 14096
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1027041246138267,
      "learning_rate": 1.2664834704357221e-05,
      "loss": 0.7196,
      "step": 14097
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.22995644912979515,
      "learning_rate": 1.2663878674751134e-05,
      "loss": 0.0704,
      "step": 14098
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.285305070841671,
      "learning_rate": 1.2662922618936931e-05,
      "loss": 0.8569,
      "step": 14099
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2904817410298773,
      "learning_rate": 1.2661966536924014e-05,
      "loss": 0.19,
      "step": 14100
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3516797931873704,
      "learning_rate": 1.2661010428721803e-05,
      "loss": 0.2644,
      "step": 14101
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8924834051216897,
      "learning_rate": 1.2660054294339692e-05,
      "loss": 0.5029,
      "step": 14102
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34838932848654375,
      "learning_rate": 1.265909813378709e-05,
      "loss": 0.1878,
      "step": 14103
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5732065925535869,
      "learning_rate": 1.2658141947073409e-05,
      "loss": 0.332,
      "step": 14104
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.348213336661014,
      "learning_rate": 1.2657185734208052e-05,
      "loss": 0.2347,
      "step": 14105
    },
    {
      "epoch": 0.43,
      "grad_norm": 3.543050295019816,
      "learning_rate": 1.2656229495200425e-05,
      "loss": 0.758,
      "step": 14106
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2554687719070672,
      "learning_rate": 1.2655273230059939e-05,
      "loss": 0.2087,
      "step": 14107
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.794870682316015,
      "learning_rate": 1.2654316938796003e-05,
      "loss": 0.4835,
      "step": 14108
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3637466627869658,
      "learning_rate": 1.2653360621418023e-05,
      "loss": 0.1969,
      "step": 14109
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5205179054386592,
      "learning_rate": 1.2652404277935408e-05,
      "loss": 0.4347,
      "step": 14110
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7171356202102602,
      "learning_rate": 1.2651447908357568e-05,
      "loss": 0.3931,
      "step": 14111
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23857367877865548,
      "learning_rate": 1.2650491512693907e-05,
      "loss": 0.2538,
      "step": 14112
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.47844892273798245,
      "learning_rate": 1.2649535090953841e-05,
      "loss": 0.1661,
      "step": 14113
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33410444974372455,
      "learning_rate": 1.2648578643146778e-05,
      "loss": 0.2154,
      "step": 14114
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0704626861750957,
      "learning_rate": 1.2647622169282126e-05,
      "loss": 0.5604,
      "step": 14115
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.849243740907869,
      "learning_rate": 1.2646665669369297e-05,
      "loss": 0.4769,
      "step": 14116
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7551368770801568,
      "learning_rate": 1.2645709143417699e-05,
      "loss": 0.4818,
      "step": 14117
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.26311557953725895,
      "learning_rate": 1.2644752591436744e-05,
      "loss": 0.194,
      "step": 14118
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34178756834019564,
      "learning_rate": 1.2643796013435843e-05,
      "loss": 0.3177,
      "step": 14119
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6199717656681071,
      "learning_rate": 1.2642839409424405e-05,
      "loss": 0.3916,
      "step": 14120
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4664460522559209,
      "learning_rate": 1.2641882779411846e-05,
      "loss": 0.2599,
      "step": 14121
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.23453433207628702,
      "learning_rate": 1.2640926123407575e-05,
      "loss": 0.1031,
      "step": 14122
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3149395931345155,
      "learning_rate": 1.2639969441421003e-05,
      "loss": 0.2607,
      "step": 14123
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2902798890733522,
      "learning_rate": 1.2639012733461545e-05,
      "loss": 0.2224,
      "step": 14124
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.6992888493833516,
      "learning_rate": 1.2638055999538606e-05,
      "loss": 0.8849,
      "step": 14125
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5537222719488956,
      "learning_rate": 1.2637099239661605e-05,
      "loss": 0.3737,
      "step": 14126
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28353088359336015,
      "learning_rate": 1.2636142453839959e-05,
      "loss": 0.1925,
      "step": 14127
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6555344558273647,
      "learning_rate": 1.2635185642083073e-05,
      "loss": 0.4514,
      "step": 14128
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7535344913026998,
      "learning_rate": 1.2634228804400363e-05,
      "loss": 0.4804,
      "step": 14129
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.33432368106234106,
      "learning_rate": 1.2633271940801243e-05,
      "loss": 0.2821,
      "step": 14130
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.18062704963231463,
      "learning_rate": 1.263231505129513e-05,
      "loss": 0.071,
      "step": 14131
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37957982880246627,
      "learning_rate": 1.2631358135891435e-05,
      "loss": 0.2871,
      "step": 14132
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2063986155146667,
      "learning_rate": 1.2630401194599568e-05,
      "loss": 0.2801,
      "step": 14133
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3261875982250533,
      "learning_rate": 1.2629444227428953e-05,
      "loss": 0.6432,
      "step": 14134
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31919049678810124,
      "learning_rate": 1.2628487234389001e-05,
      "loss": 0.2871,
      "step": 14135
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6612738749031452,
      "learning_rate": 1.2627530215489125e-05,
      "loss": 0.3503,
      "step": 14136
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3275852719362007,
      "learning_rate": 1.2626573170738743e-05,
      "loss": 0.2565,
      "step": 14137
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9203953721961728,
      "learning_rate": 1.2625616100147269e-05,
      "loss": 0.5951,
      "step": 14138
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.1918828978366171,
      "learning_rate": 1.2624659003724118e-05,
      "loss": 0.1091,
      "step": 14139
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5341025020557519,
      "learning_rate": 1.2623701881478711e-05,
      "loss": 0.172,
      "step": 14140
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3542019481812258,
      "learning_rate": 1.262274473342046e-05,
      "loss": 0.3207,
      "step": 14141
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3299082547485415,
      "learning_rate": 1.2621787559558785e-05,
      "loss": 0.2521,
      "step": 14142
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6955140808422623,
      "learning_rate": 1.26208303599031e-05,
      "loss": 0.5123,
      "step": 14143
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9767324515281601,
      "learning_rate": 1.2619873134462825e-05,
      "loss": 0.3385,
      "step": 14144
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5082209310325909,
      "learning_rate": 1.2618915883247373e-05,
      "loss": 0.3412,
      "step": 14145
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30601179934388184,
      "learning_rate": 1.2617958606266167e-05,
      "loss": 0.24,
      "step": 14146
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.2008926665293234,
      "learning_rate": 1.2617001303528621e-05,
      "loss": 0.6565,
      "step": 14147
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2870733969644732,
      "learning_rate": 1.2616043975044161e-05,
      "loss": 0.236,
      "step": 14148
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2939140909125695,
      "learning_rate": 1.2615086620822194e-05,
      "loss": 0.1633,
      "step": 14149
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2569704299476681,
      "learning_rate": 1.2614129240872146e-05,
      "loss": 0.1899,
      "step": 14150
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7975147954331665,
      "learning_rate": 1.2613171835203433e-05,
      "loss": 0.7664,
      "step": 14151
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6811532215005561,
      "learning_rate": 1.2612214403825476e-05,
      "loss": 0.3346,
      "step": 14152
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3266982166202743,
      "learning_rate": 1.2611256946747696e-05,
      "loss": 0.2846,
      "step": 14153
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.37372934594995555,
      "learning_rate": 1.261029946397951e-05,
      "loss": 0.2687,
      "step": 14154
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.43998521837775545,
      "learning_rate": 1.2609341955530338e-05,
      "loss": 0.2697,
      "step": 14155
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.7818336637839958,
      "learning_rate": 1.26083844214096e-05,
      "loss": 0.8605,
      "step": 14156
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2754858052443977,
      "learning_rate": 1.2607426861626722e-05,
      "loss": 0.1575,
      "step": 14157
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.35035552214606824,
      "learning_rate": 1.2606469276191117e-05,
      "loss": 0.1794,
      "step": 14158
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.276389050647485,
      "learning_rate": 1.2605511665112208e-05,
      "loss": 0.2159,
      "step": 14159
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.46420641466234147,
      "learning_rate": 1.2604554028399423e-05,
      "loss": 0.3425,
      "step": 14160
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4174910435199916,
      "learning_rate": 1.2603596366062178e-05,
      "loss": 0.286,
      "step": 14161
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7341750469742133,
      "learning_rate": 1.2602638678109893e-05,
      "loss": 0.4672,
      "step": 14162
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.27434010979272605,
      "learning_rate": 1.2601680964551994e-05,
      "loss": 0.1252,
      "step": 14163
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.36673822375530574,
      "learning_rate": 1.2600723225397899e-05,
      "loss": 0.2925,
      "step": 14164
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8615449745596113,
      "learning_rate": 1.2599765460657034e-05,
      "loss": 0.5671,
      "step": 14165
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2756775089748052,
      "learning_rate": 1.2598807670338821e-05,
      "loss": 0.2224,
      "step": 14166
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2901245044078623,
      "learning_rate": 1.2597849854452685e-05,
      "loss": 0.1829,
      "step": 14167
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.30728549809977784,
      "learning_rate": 1.2596892013008047e-05,
      "loss": 0.1665,
      "step": 14168
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44182541126355745,
      "learning_rate": 1.2595934146014328e-05,
      "loss": 0.3547,
      "step": 14169
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5764044836013547,
      "learning_rate": 1.2594976253480957e-05,
      "loss": 0.3495,
      "step": 14170
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4543249501166719,
      "learning_rate": 1.2594018335417355e-05,
      "loss": 0.384,
      "step": 14171
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.27455328597557943,
      "learning_rate": 1.2593060391832949e-05,
      "loss": 0.1609,
      "step": 14172
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3827556080273514,
      "learning_rate": 1.259210242273716e-05,
      "loss": 0.284,
      "step": 14173
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.7788527206762085,
      "learning_rate": 1.2591144428139414e-05,
      "loss": 0.4729,
      "step": 14174
    },
    {
      "epoch": 0.43,
      "grad_norm": 2.515131232480947,
      "learning_rate": 1.2590186408049137e-05,
      "loss": 0.8636,
      "step": 14175
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31832622272883826,
      "learning_rate": 1.2589228362475754e-05,
      "loss": 0.1286,
      "step": 14176
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3735216348598547,
      "learning_rate": 1.2588270291428688e-05,
      "loss": 0.2513,
      "step": 14177
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2873123817136271,
      "learning_rate": 1.2587312194917369e-05,
      "loss": 0.2629,
      "step": 14178
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.0722916995185534,
      "learning_rate": 1.2586354072951224e-05,
      "loss": 0.3492,
      "step": 14179
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.4670797486233556,
      "learning_rate": 1.2585395925539674e-05,
      "loss": 0.7252,
      "step": 14180
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.31344030598791184,
      "learning_rate": 1.258443775269215e-05,
      "loss": 0.1617,
      "step": 14181
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40503788825598736,
      "learning_rate": 1.2583479554418072e-05,
      "loss": 0.3131,
      "step": 14182
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.9165488216288747,
      "learning_rate": 1.2582521330726878e-05,
      "loss": 0.3633,
      "step": 14183
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3579415319871058,
      "learning_rate": 1.2581563081627986e-05,
      "loss": 0.3098,
      "step": 14184
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.273696167048866,
      "learning_rate": 1.2580604807130828e-05,
      "loss": 0.103,
      "step": 14185
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.34916501066563593,
      "learning_rate": 1.2579646507244832e-05,
      "loss": 0.2835,
      "step": 14186
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.8609422707174853,
      "learning_rate": 1.2578688181979422e-05,
      "loss": 0.3684,
      "step": 14187
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.4864720882966771,
      "learning_rate": 1.257772983134403e-05,
      "loss": 0.295,
      "step": 14188
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.28565400814428304,
      "learning_rate": 1.2576771455348084e-05,
      "loss": 0.2336,
      "step": 14189
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.549739180921609,
      "learning_rate": 1.2575813054001013e-05,
      "loss": 0.2957,
      "step": 14190
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3613523065725142,
      "learning_rate": 1.2574854627312245e-05,
      "loss": 0.2277,
      "step": 14191
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.015831799441523,
      "learning_rate": 1.2573896175291211e-05,
      "loss": 0.485,
      "step": 14192
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.1479564498767294,
      "learning_rate": 1.2572937697947341e-05,
      "loss": 0.6885,
      "step": 14193
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.5064822385644294,
      "learning_rate": 1.257197919529006e-05,
      "loss": 0.3088,
      "step": 14194
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6107411766193581,
      "learning_rate": 1.2571020667328802e-05,
      "loss": 0.261,
      "step": 14195
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.29653966781261937,
      "learning_rate": 1.2570062114072998e-05,
      "loss": 0.2501,
      "step": 14196
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.40495951952986836,
      "learning_rate": 1.2569103535532076e-05,
      "loss": 0.2588,
      "step": 14197
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.2517944205001372,
      "learning_rate": 1.2568144931715469e-05,
      "loss": 0.073,
      "step": 14198
    },
    {
      "epoch": 0.43,
      "grad_norm": 1.3975139327193427,
      "learning_rate": 1.2567186302632608e-05,
      "loss": 0.7867,
      "step": 14199
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.25927891367479333,
      "learning_rate": 1.2566227648292921e-05,
      "loss": 0.1863,
      "step": 14200
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.46372124172118523,
      "learning_rate": 1.2565268968705844e-05,
      "loss": 0.3696,
      "step": 14201
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.44525088900627996,
      "learning_rate": 1.2564310263880807e-05,
      "loss": 0.2319,
      "step": 14202
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.6792423226330648,
      "learning_rate": 1.256335153382724e-05,
      "loss": 0.4647,
      "step": 14203
    },
    {
      "epoch": 0.43,
      "grad_norm": 0.3532804139579427,
      "learning_rate": 1.2562392778554577e-05,
      "loss": 0.2139,
      "step": 14204
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3917140174928071,
      "learning_rate": 1.2561433998072252e-05,
      "loss": 0.2749,
      "step": 14205
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5013382886468604,
      "learning_rate": 1.2560475192389698e-05,
      "loss": 0.2493,
      "step": 14206
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24576307522760496,
      "learning_rate": 1.2559516361516345e-05,
      "loss": 0.2089,
      "step": 14207
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9271244471757818,
      "learning_rate": 1.2558557505461625e-05,
      "loss": 0.5293,
      "step": 14208
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28332771144454016,
      "learning_rate": 1.2557598624234979e-05,
      "loss": 0.2126,
      "step": 14209
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6349351239651317,
      "learning_rate": 1.2556639717845833e-05,
      "loss": 0.8863,
      "step": 14210
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7915224575205486,
      "learning_rate": 1.2555680786303625e-05,
      "loss": 0.3422,
      "step": 14211
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6740450935257251,
      "learning_rate": 1.2554721829617793e-05,
      "loss": 0.4013,
      "step": 14212
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.26032509015191796,
      "learning_rate": 1.2553762847797761e-05,
      "loss": 0.2049,
      "step": 14213
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5398025962688403,
      "learning_rate": 1.2552803840852969e-05,
      "loss": 0.3706,
      "step": 14214
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.398622380447021,
      "learning_rate": 1.255184480879286e-05,
      "loss": 0.2591,
      "step": 14215
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29644127150205385,
      "learning_rate": 1.2550885751626855e-05,
      "loss": 0.2025,
      "step": 14216
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3366876849105136,
      "learning_rate": 1.2549926669364401e-05,
      "loss": 0.1611,
      "step": 14217
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3064471888049268,
      "learning_rate": 1.2548967562014927e-05,
      "loss": 0.2188,
      "step": 14218
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5105163717472978,
      "learning_rate": 1.2548008429587873e-05,
      "loss": 0.3759,
      "step": 14219
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6170929265416346,
      "learning_rate": 1.254704927209267e-05,
      "loss": 0.3308,
      "step": 14220
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.013802709068326,
      "learning_rate": 1.2546090089538761e-05,
      "loss": 0.4606,
      "step": 14221
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23666158159132522,
      "learning_rate": 1.2545130881935582e-05,
      "loss": 0.0697,
      "step": 14222
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3603816333562272,
      "learning_rate": 1.2544171649292567e-05,
      "loss": 0.2944,
      "step": 14223
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.15078401709691072,
      "learning_rate": 1.254321239161915e-05,
      "loss": 0.0694,
      "step": 14224
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35578555022286834,
      "learning_rate": 1.2542253108924778e-05,
      "loss": 0.3465,
      "step": 14225
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8111758450509405,
      "learning_rate": 1.254129380121888e-05,
      "loss": 0.5688,
      "step": 14226
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35224258488095006,
      "learning_rate": 1.2540334468510898e-05,
      "loss": 0.2589,
      "step": 14227
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.44758771100058575,
      "learning_rate": 1.253937511081027e-05,
      "loss": 0.2967,
      "step": 14228
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8989754302816013,
      "learning_rate": 1.2538415728126434e-05,
      "loss": 0.4394,
      "step": 14229
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8676673309409271,
      "learning_rate": 1.253745632046883e-05,
      "loss": 0.3278,
      "step": 14230
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.26139102001536985,
      "learning_rate": 1.2536496887846894e-05,
      "loss": 0.2095,
      "step": 14231
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34824769689432267,
      "learning_rate": 1.2535537430270069e-05,
      "loss": 0.3026,
      "step": 14232
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2385613685249406,
      "learning_rate": 1.253457794774779e-05,
      "loss": 0.0968,
      "step": 14233
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42239432982934777,
      "learning_rate": 1.2533618440289499e-05,
      "loss": 0.2701,
      "step": 14234
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5472091114759526,
      "learning_rate": 1.2532658907904641e-05,
      "loss": 0.2825,
      "step": 14235
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4119806308072988,
      "learning_rate": 1.2531699350602646e-05,
      "loss": 0.2971,
      "step": 14236
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2546042961361509,
      "learning_rate": 1.2530739768392962e-05,
      "loss": 0.1702,
      "step": 14237
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5415577932238965,
      "learning_rate": 1.2529780161285028e-05,
      "loss": 0.3632,
      "step": 14238
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.215972728097449,
      "learning_rate": 1.2528820529288282e-05,
      "loss": 0.3974,
      "step": 14239
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6174343567129027,
      "learning_rate": 1.2527860872412168e-05,
      "loss": 0.3981,
      "step": 14240
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3368118626451948,
      "learning_rate": 1.252690119066613e-05,
      "loss": 0.2164,
      "step": 14241
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8546455418636799,
      "learning_rate": 1.2525941484059606e-05,
      "loss": 0.6226,
      "step": 14242
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3098722966526807,
      "learning_rate": 1.2524981752602034e-05,
      "loss": 0.28,
      "step": 14243
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4525559364988735,
      "learning_rate": 1.2524021996302865e-05,
      "loss": 0.2692,
      "step": 14244
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9249177121403911,
      "learning_rate": 1.2523062215171535e-05,
      "loss": 0.4516,
      "step": 14245
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.31952976455225346,
      "learning_rate": 1.2522102409217489e-05,
      "loss": 0.2661,
      "step": 14246
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.45411811135529034,
      "learning_rate": 1.252114257845017e-05,
      "loss": 0.2722,
      "step": 14247
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2699504704033374,
      "learning_rate": 1.2520182722879022e-05,
      "loss": 0.1815,
      "step": 14248
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4200359717476852,
      "learning_rate": 1.2519222842513488e-05,
      "loss": 0.3271,
      "step": 14249
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.30424542768038126,
      "learning_rate": 1.2518262937363007e-05,
      "loss": 0.203,
      "step": 14250
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3249501238100605,
      "learning_rate": 1.2517303007437027e-05,
      "loss": 0.7167,
      "step": 14251
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1385668824355912,
      "learning_rate": 1.2516343052744992e-05,
      "loss": 0.4321,
      "step": 14252
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7616366178933954,
      "learning_rate": 1.2515383073296348e-05,
      "loss": 0.5772,
      "step": 14253
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2902318192482707,
      "learning_rate": 1.2514423069100539e-05,
      "loss": 0.2026,
      "step": 14254
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34124059164365395,
      "learning_rate": 1.2513463040167007e-05,
      "loss": 0.3218,
      "step": 14255
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3563519796084318,
      "learning_rate": 1.2512502986505197e-05,
      "loss": 0.2791,
      "step": 14256
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24645139978714384,
      "learning_rate": 1.2511542908124555e-05,
      "loss": 0.0744,
      "step": 14257
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.739395114866222,
      "learning_rate": 1.251058280503453e-05,
      "loss": 0.4002,
      "step": 14258
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29094090715232007,
      "learning_rate": 1.2509622677244564e-05,
      "loss": 0.2068,
      "step": 14259
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7616702127170225,
      "learning_rate": 1.2508662524764104e-05,
      "loss": 0.8138,
      "step": 14260
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29791765189273994,
      "learning_rate": 1.2507702347602597e-05,
      "loss": 0.2504,
      "step": 14261
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8606532904359023,
      "learning_rate": 1.2506742145769493e-05,
      "loss": 0.5786,
      "step": 14262
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4572749829325925,
      "learning_rate": 1.2505781919274229e-05,
      "loss": 0.2142,
      "step": 14263
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5696838565808143,
      "learning_rate": 1.2504821668126262e-05,
      "loss": 0.335,
      "step": 14264
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5247698950441206,
      "learning_rate": 1.2503861392335033e-05,
      "loss": 0.3639,
      "step": 14265
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24224496969185355,
      "learning_rate": 1.2502901091909991e-05,
      "loss": 0.1998,
      "step": 14266
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2950305707495702,
      "learning_rate": 1.2501940766860587e-05,
      "loss": 0.1962,
      "step": 14267
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5248118971800415,
      "learning_rate": 1.2500980417196268e-05,
      "loss": 0.3574,
      "step": 14268
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.44534641466360175,
      "learning_rate": 1.2500020042926475e-05,
      "loss": 0.2538,
      "step": 14269
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3264631263125657,
      "learning_rate": 1.2499059644060665e-05,
      "loss": 0.8183,
      "step": 14270
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5550257905295557,
      "learning_rate": 1.2498099220608287e-05,
      "loss": 0.3635,
      "step": 14271
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4505804496309485,
      "learning_rate": 1.2497138772578783e-05,
      "loss": 0.0709,
      "step": 14272
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29988082677464467,
      "learning_rate": 1.2496178299981608e-05,
      "loss": 0.2927,
      "step": 14273
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42970730585901673,
      "learning_rate": 1.249521780282621e-05,
      "loss": 0.2926,
      "step": 14274
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24312528424354543,
      "learning_rate": 1.2494257281122035e-05,
      "loss": 0.1524,
      "step": 14275
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6906170363728918,
      "learning_rate": 1.2493296734878538e-05,
      "loss": 0.3184,
      "step": 14276
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37310079070722285,
      "learning_rate": 1.2492336164105167e-05,
      "loss": 0.2517,
      "step": 14277
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2991528624704888,
      "learning_rate": 1.2491375568811374e-05,
      "loss": 0.2218,
      "step": 14278
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.45980481291884767,
      "learning_rate": 1.2490414949006607e-05,
      "loss": 0.3759,
      "step": 14279
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5071176004372114,
      "learning_rate": 1.248945430470032e-05,
      "loss": 0.3224,
      "step": 14280
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5163066970679099,
      "learning_rate": 1.2488493635901962e-05,
      "loss": 0.3454,
      "step": 14281
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3301797394546768,
      "learning_rate": 1.2487532942620984e-05,
      "loss": 0.2396,
      "step": 14282
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.477657368947164,
      "learning_rate": 1.2486572224866836e-05,
      "loss": 0.8517,
      "step": 14283
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24674526306440728,
      "learning_rate": 1.2485611482648977e-05,
      "loss": 0.1635,
      "step": 14284
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2737946736540372,
      "learning_rate": 1.2484650715976851e-05,
      "loss": 0.1729,
      "step": 14285
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3551702606377832,
      "learning_rate": 1.2483689924859914e-05,
      "loss": 0.2872,
      "step": 14286
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2279392547276917,
      "learning_rate": 1.248272910930762e-05,
      "loss": 0.282,
      "step": 14287
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8566758232291924,
      "learning_rate": 1.248176826932942e-05,
      "loss": 0.5002,
      "step": 14288
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6708352374923539,
      "learning_rate": 1.2480807404934765e-05,
      "loss": 0.3687,
      "step": 14289
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37581777573077446,
      "learning_rate": 1.2479846516133114e-05,
      "loss": 0.2835,
      "step": 14290
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3376285057448004,
      "learning_rate": 1.2478885602933912e-05,
      "loss": 0.2292,
      "step": 14291
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4620533638898705,
      "learning_rate": 1.2477924665346623e-05,
      "loss": 0.3541,
      "step": 14292
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3111252203696426,
      "learning_rate": 1.2476963703380694e-05,
      "loss": 0.1689,
      "step": 14293
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3953512056812325,
      "learning_rate": 1.2476002717045578e-05,
      "loss": 0.2278,
      "step": 14294
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3194267226030133,
      "learning_rate": 1.2475041706350735e-05,
      "loss": 0.1889,
      "step": 14295
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4988119248315285,
      "learning_rate": 1.2474080671305617e-05,
      "loss": 0.3717,
      "step": 14296
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3853254540835073,
      "learning_rate": 1.2473119611919682e-05,
      "loss": 0.2721,
      "step": 14297
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9106553103730229,
      "learning_rate": 1.2472158528202377e-05,
      "loss": 0.3915,
      "step": 14298
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9136293165271938,
      "learning_rate": 1.2471197420163169e-05,
      "loss": 0.4277,
      "step": 14299
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29615039686316197,
      "learning_rate": 1.2470236287811504e-05,
      "loss": 0.2243,
      "step": 14300
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3658494592669896,
      "learning_rate": 1.2469275131156842e-05,
      "loss": 0.7939,
      "step": 14301
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2125307612488174,
      "learning_rate": 1.2468313950208637e-05,
      "loss": 0.1921,
      "step": 14302
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4932254948170817,
      "learning_rate": 1.246735274497635e-05,
      "loss": 0.2763,
      "step": 14303
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3044243214075851,
      "learning_rate": 1.2466391515469434e-05,
      "loss": 0.1817,
      "step": 14304
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.43620926455038295,
      "learning_rate": 1.246543026169735e-05,
      "loss": 0.3484,
      "step": 14305
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6474231737993774,
      "learning_rate": 1.246446898366955e-05,
      "loss": 0.3683,
      "step": 14306
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7559787699431388,
      "learning_rate": 1.2463507681395488e-05,
      "loss": 0.9477,
      "step": 14307
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2816199011590928,
      "learning_rate": 1.246254635488463e-05,
      "loss": 0.2407,
      "step": 14308
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33800748845477646,
      "learning_rate": 1.246158500414643e-05,
      "loss": 0.2856,
      "step": 14309
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8397224549602202,
      "learning_rate": 1.246062362919035e-05,
      "loss": 0.4857,
      "step": 14310
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0033339552299847,
      "learning_rate": 1.2459662230025842e-05,
      "loss": 0.291,
      "step": 14311
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2677734374548011,
      "learning_rate": 1.245870080666237e-05,
      "loss": 0.1658,
      "step": 14312
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3971193368818008,
      "learning_rate": 1.245773935910939e-05,
      "loss": 0.1817,
      "step": 14313
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7327789223009501,
      "learning_rate": 1.2456777887376359e-05,
      "loss": 0.4974,
      "step": 14314
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29152403236858965,
      "learning_rate": 1.2455816391472738e-05,
      "loss": 0.2596,
      "step": 14315
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7143183926985138,
      "learning_rate": 1.2454854871407993e-05,
      "loss": 0.4935,
      "step": 14316
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24626508654212045,
      "learning_rate": 1.2453893327191575e-05,
      "loss": 0.163,
      "step": 14317
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4059794497452731,
      "learning_rate": 1.2452931758832947e-05,
      "loss": 0.3246,
      "step": 14318
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9175300332391914,
      "learning_rate": 1.2451970166341569e-05,
      "loss": 0.4969,
      "step": 14319
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.32766495971139736,
      "learning_rate": 1.2451008549726902e-05,
      "loss": 0.2799,
      "step": 14320
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2265470826579849,
      "learning_rate": 1.2450046908998405e-05,
      "loss": 0.1418,
      "step": 14321
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6858851579927872,
      "learning_rate": 1.244908524416554e-05,
      "loss": 0.3266,
      "step": 14322
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3750827657322715,
      "learning_rate": 1.2448123555237773e-05,
      "loss": 0.2452,
      "step": 14323
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6066551925625892,
      "learning_rate": 1.2447161842224558e-05,
      "loss": 0.3561,
      "step": 14324
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4393084712875891,
      "learning_rate": 1.2446200105135357e-05,
      "loss": 0.3722,
      "step": 14325
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3097103526958872,
      "learning_rate": 1.2445238343979639e-05,
      "loss": 0.1999,
      "step": 14326
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33163548359215755,
      "learning_rate": 1.2444276558766859e-05,
      "loss": 0.2764,
      "step": 14327
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.836059246836097,
      "learning_rate": 1.2443314749506482e-05,
      "loss": 0.4832,
      "step": 14328
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.7317174810739315,
      "learning_rate": 1.2442352916207972e-05,
      "loss": 0.8726,
      "step": 14329
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7158103500807285,
      "learning_rate": 1.2441391058880787e-05,
      "loss": 0.3376,
      "step": 14330
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5222577456657407,
      "learning_rate": 1.2440429177534399e-05,
      "loss": 0.3718,
      "step": 14331
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.26777374821680944,
      "learning_rate": 1.243946727217826e-05,
      "loss": 0.2473,
      "step": 14332
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2382196086228101,
      "learning_rate": 1.2438505342821844e-05,
      "loss": 0.2005,
      "step": 14333
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.314009513657279,
      "learning_rate": 1.2437543389474607e-05,
      "loss": 0.0574,
      "step": 14334
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2567085163290892,
      "learning_rate": 1.2436581412146014e-05,
      "loss": 0.8729,
      "step": 14335
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2836211857939575,
      "learning_rate": 1.2435619410845537e-05,
      "loss": 0.1885,
      "step": 14336
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9787060666088682,
      "learning_rate": 1.2434657385582634e-05,
      "loss": 0.3487,
      "step": 14337
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4608681342962662,
      "learning_rate": 1.2433695336366767e-05,
      "loss": 0.3661,
      "step": 14338
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4966556753929437,
      "learning_rate": 1.2432733263207407e-05,
      "loss": 0.2896,
      "step": 14339
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5472030438276679,
      "learning_rate": 1.2431771166114016e-05,
      "loss": 0.3797,
      "step": 14340
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.35915537768474876,
      "learning_rate": 1.2430809045096057e-05,
      "loss": 0.2437,
      "step": 14341
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42690306258249966,
      "learning_rate": 1.2429846900163002e-05,
      "loss": 0.243,
      "step": 14342
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2861453369425586,
      "learning_rate": 1.2428884731324315e-05,
      "loss": 0.1928,
      "step": 14343
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42856034227328216,
      "learning_rate": 1.2427922538589459e-05,
      "loss": 0.2919,
      "step": 14344
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.31325742508307974,
      "learning_rate": 1.24269603219679e-05,
      "loss": 0.2012,
      "step": 14345
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5266241927488269,
      "learning_rate": 1.242599808146911e-05,
      "loss": 0.822,
      "step": 14346
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8821650857490374,
      "learning_rate": 1.2425035817102553e-05,
      "loss": 0.3577,
      "step": 14347
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7121606041240008,
      "learning_rate": 1.2424073528877694e-05,
      "loss": 0.4198,
      "step": 14348
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3331019689404692,
      "learning_rate": 1.2423111216804002e-05,
      "loss": 0.0774,
      "step": 14349
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29406070730556483,
      "learning_rate": 1.2422148880890945e-05,
      "loss": 0.2313,
      "step": 14350
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3217375206584011,
      "learning_rate": 1.242118652114799e-05,
      "loss": 0.2795,
      "step": 14351
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23580647903267393,
      "learning_rate": 1.2420224137584606e-05,
      "loss": 0.1177,
      "step": 14352
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.831488982152564,
      "learning_rate": 1.241926173021026e-05,
      "loss": 0.8658,
      "step": 14353
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3119774139624782,
      "learning_rate": 1.241829929903442e-05,
      "loss": 0.1912,
      "step": 14354
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5452729486616719,
      "learning_rate": 1.2417336844066559e-05,
      "loss": 0.3663,
      "step": 14355
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.47957887855609366,
      "learning_rate": 1.2416374365316141e-05,
      "loss": 0.2948,
      "step": 14356
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.169321890253188,
      "learning_rate": 1.2415411862792637e-05,
      "loss": 0.5254,
      "step": 14357
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.49193255028583915,
      "learning_rate": 1.2414449336505514e-05,
      "loss": 0.2301,
      "step": 14358
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37050168235392184,
      "learning_rate": 1.2413486786464248e-05,
      "loss": 0.3131,
      "step": 14359
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2428679412822692,
      "learning_rate": 1.24125242126783e-05,
      "loss": 0.1075,
      "step": 14360
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5587930263755014,
      "learning_rate": 1.2411561615157148e-05,
      "loss": 0.8259,
      "step": 14361
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.30036517327446616,
      "learning_rate": 1.2410598993910262e-05,
      "loss": 0.2548,
      "step": 14362
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2796623829445003,
      "learning_rate": 1.2409636348947106e-05,
      "loss": 0.19,
      "step": 14363
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.589246634093989,
      "learning_rate": 1.2408673680277154e-05,
      "loss": 0.9331,
      "step": 14364
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9186404908631955,
      "learning_rate": 1.2407710987909882e-05,
      "loss": 0.3778,
      "step": 14365
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8151271854400112,
      "learning_rate": 1.2406748271854752e-05,
      "loss": 0.4807,
      "step": 14366
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29383528964002026,
      "learning_rate": 1.2405785532121244e-05,
      "loss": 0.2016,
      "step": 14367
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29376775466157906,
      "learning_rate": 1.2404822768718824e-05,
      "loss": 0.2725,
      "step": 14368
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0252104407201148,
      "learning_rate": 1.2403859981656972e-05,
      "loss": 0.0301,
      "step": 14369
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3113085007521669,
      "learning_rate": 1.2402897170945148e-05,
      "loss": 0.1386,
      "step": 14370
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7585538983002342,
      "learning_rate": 1.2401934336592831e-05,
      "loss": 0.4955,
      "step": 14371
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4102605920168164,
      "learning_rate": 1.2400971478609498e-05,
      "loss": 0.2639,
      "step": 14372
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3718690408069273,
      "learning_rate": 1.2400008597004613e-05,
      "loss": 0.2514,
      "step": 14373
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34786395856245295,
      "learning_rate": 1.2399045691787659e-05,
      "loss": 0.3267,
      "step": 14374
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6147940634922257,
      "learning_rate": 1.2398082762968102e-05,
      "loss": 0.3678,
      "step": 14375
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6498325594002053,
      "learning_rate": 1.2397119810555416e-05,
      "loss": 0.1551,
      "step": 14376
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34585952665882835,
      "learning_rate": 1.2396156834559077e-05,
      "loss": 0.2998,
      "step": 14377
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2800053921873929,
      "learning_rate": 1.239519383498856e-05,
      "loss": 0.1055,
      "step": 14378
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.504898909872371,
      "learning_rate": 1.2394230811853336e-05,
      "loss": 0.4121,
      "step": 14379
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3972715828046138,
      "learning_rate": 1.2393267765162881e-05,
      "loss": 0.2591,
      "step": 14380
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6767016132826913,
      "learning_rate": 1.2392304694926673e-05,
      "loss": 0.352,
      "step": 14381
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.34806519083802573,
      "learning_rate": 1.2391341601154185e-05,
      "loss": 0.2799,
      "step": 14382
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38059214319466955,
      "learning_rate": 1.2390378483854888e-05,
      "loss": 0.2596,
      "step": 14383
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2757365468000985,
      "learning_rate": 1.2389415343038262e-05,
      "loss": 0.0844,
      "step": 14384
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.33678868360061104,
      "learning_rate": 1.2388452178713783e-05,
      "loss": 0.3116,
      "step": 14385
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2907266098252182,
      "learning_rate": 1.2387488990890924e-05,
      "loss": 0.1877,
      "step": 14386
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.286861984325302,
      "learning_rate": 1.2386525779579164e-05,
      "loss": 0.7034,
      "step": 14387
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.892981549505667,
      "learning_rate": 1.238556254478798e-05,
      "loss": 0.3454,
      "step": 14388
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5512625120535349,
      "learning_rate": 1.2384599286526842e-05,
      "loss": 0.4169,
      "step": 14389
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41473125130695704,
      "learning_rate": 1.2383636004805233e-05,
      "loss": 0.2592,
      "step": 14390
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28614878117543935,
      "learning_rate": 1.2382672699632629e-05,
      "loss": 0.2628,
      "step": 14391
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4705334884894815,
      "learning_rate": 1.2381709371018507e-05,
      "loss": 0.3598,
      "step": 14392
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19444282202927204,
      "learning_rate": 1.2380746018972348e-05,
      "loss": 0.1246,
      "step": 14393
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.0933929157640128,
      "learning_rate": 1.2379782643503622e-05,
      "loss": 0.4237,
      "step": 14394
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2903309676794221,
      "learning_rate": 1.2378819244621812e-05,
      "loss": 0.2124,
      "step": 14395
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6321055546240206,
      "learning_rate": 1.2377855822336395e-05,
      "loss": 0.9296,
      "step": 14396
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.37941318843037997,
      "learning_rate": 1.237689237665685e-05,
      "loss": 0.2839,
      "step": 14397
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42561937729649707,
      "learning_rate": 1.2375928907592658e-05,
      "loss": 0.3705,
      "step": 14398
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3128563223649229,
      "learning_rate": 1.2374965415153292e-05,
      "loss": 0.209,
      "step": 14399
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5668750269658792,
      "learning_rate": 1.237400189934824e-05,
      "loss": 0.3905,
      "step": 14400
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3438676133371997,
      "learning_rate": 1.2373038360186975e-05,
      "loss": 0.2208,
      "step": 14401
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2556722505806652,
      "learning_rate": 1.2372074797678975e-05,
      "loss": 0.1752,
      "step": 14402
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4690195551939234,
      "learning_rate": 1.2371111211833722e-05,
      "loss": 0.3686,
      "step": 14403
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28531165861937263,
      "learning_rate": 1.23701476026607e-05,
      "loss": 0.165,
      "step": 14404
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4953430844321192,
      "learning_rate": 1.2369183970169383e-05,
      "loss": 0.3354,
      "step": 14405
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6617951390939809,
      "learning_rate": 1.2368220314369257e-05,
      "loss": 0.4005,
      "step": 14406
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.606361785956806,
      "learning_rate": 1.23672566352698e-05,
      "loss": 0.4839,
      "step": 14407
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4684505895478287,
      "learning_rate": 1.2366292932880492e-05,
      "loss": 0.1891,
      "step": 14408
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4831844708113188,
      "learning_rate": 1.2365329207210816e-05,
      "loss": 0.3246,
      "step": 14409
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29718688628505363,
      "learning_rate": 1.2364365458270251e-05,
      "loss": 0.2523,
      "step": 14410
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.285047967732759,
      "learning_rate": 1.2363401686068283e-05,
      "loss": 0.1502,
      "step": 14411
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9206470841236237,
      "learning_rate": 1.2362437890614392e-05,
      "loss": 0.2444,
      "step": 14412
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3580280708730799,
      "learning_rate": 1.236147407191806e-05,
      "loss": 0.2778,
      "step": 14413
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4279551790785086,
      "learning_rate": 1.236051022998877e-05,
      "loss": 0.2379,
      "step": 14414
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42477318128376224,
      "learning_rate": 1.2359546364836e-05,
      "loss": 0.2716,
      "step": 14415
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4662397244329837,
      "learning_rate": 1.2358582476469236e-05,
      "loss": 0.3596,
      "step": 14416
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5784924175539399,
      "learning_rate": 1.2357618564897966e-05,
      "loss": 0.251,
      "step": 14417
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.399058855576923,
      "learning_rate": 1.2356654630131664e-05,
      "loss": 0.3444,
      "step": 14418
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.139859008076179,
      "learning_rate": 1.2355690672179823e-05,
      "loss": 0.3435,
      "step": 14419
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2693297728705254,
      "learning_rate": 1.235472669105192e-05,
      "loss": 0.1441,
      "step": 14420
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2898276037728378,
      "learning_rate": 1.235376268675744e-05,
      "loss": 0.234,
      "step": 14421
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3549372078601808,
      "learning_rate": 1.2352798659305866e-05,
      "loss": 0.2997,
      "step": 14422
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1155550866803365,
      "learning_rate": 1.2351834608706689e-05,
      "loss": 0.3015,
      "step": 14423
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9249915935443783,
      "learning_rate": 1.2350870534969388e-05,
      "loss": 0.5518,
      "step": 14424
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7740949625911666,
      "learning_rate": 1.234990643810345e-05,
      "loss": 0.3644,
      "step": 14425
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5628835997820905,
      "learning_rate": 1.2348942318118358e-05,
      "loss": 0.3085,
      "step": 14426
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3264142970084781,
      "learning_rate": 1.2347978175023601e-05,
      "loss": 0.2177,
      "step": 14427
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.30055903436104076,
      "learning_rate": 1.2347014008828657e-05,
      "loss": 0.252,
      "step": 14428
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28073989596693993,
      "learning_rate": 1.234604981954302e-05,
      "loss": 0.1667,
      "step": 14429
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.2008745708559503,
      "learning_rate": 1.2345085607176173e-05,
      "loss": 0.2908,
      "step": 14430
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6392147133032666,
      "learning_rate": 1.23441213717376e-05,
      "loss": 0.3172,
      "step": 14431
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.39312994655937333,
      "learning_rate": 1.2343157113236795e-05,
      "loss": 0.2663,
      "step": 14432
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3623462893278643,
      "learning_rate": 1.2342192831683235e-05,
      "loss": 0.2901,
      "step": 14433
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9877289045721296,
      "learning_rate": 1.2341228527086412e-05,
      "loss": 0.3924,
      "step": 14434
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.841998072826407,
      "learning_rate": 1.2340264199455811e-05,
      "loss": 0.5059,
      "step": 14435
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2946995010029069,
      "learning_rate": 1.2339299848800921e-05,
      "loss": 0.2274,
      "step": 14436
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5981451713426151,
      "learning_rate": 1.2338335475131233e-05,
      "loss": 0.6928,
      "step": 14437
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.19817016684745967,
      "learning_rate": 1.2337371078456229e-05,
      "loss": 0.0696,
      "step": 14438
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3212658439441694,
      "learning_rate": 1.2336406658785398e-05,
      "loss": 0.3053,
      "step": 14439
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28035461288632935,
      "learning_rate": 1.2335442216128233e-05,
      "loss": 0.1532,
      "step": 14440
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4654218811511652,
      "learning_rate": 1.2334477750494216e-05,
      "loss": 0.2915,
      "step": 14441
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9995972245769942,
      "learning_rate": 1.2333513261892838e-05,
      "loss": 0.6423,
      "step": 14442
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8357005279789748,
      "learning_rate": 1.2332548750333591e-05,
      "loss": 0.3936,
      "step": 14443
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4403110857196971,
      "learning_rate": 1.233158421582596e-05,
      "loss": 0.3525,
      "step": 14444
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.22877326963875805,
      "learning_rate": 1.2330619658379437e-05,
      "loss": 0.2131,
      "step": 14445
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5840234429669637,
      "learning_rate": 1.2329655078003513e-05,
      "loss": 0.8446,
      "step": 14446
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2505922958765548,
      "learning_rate": 1.2328690474707672e-05,
      "loss": 0.106,
      "step": 14447
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4440809229284989,
      "learning_rate": 1.2327725848501408e-05,
      "loss": 0.2609,
      "step": 14448
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.38723694264891156,
      "learning_rate": 1.2326761199394212e-05,
      "loss": 0.2226,
      "step": 14449
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6063524855201627,
      "learning_rate": 1.2325796527395577e-05,
      "loss": 0.4005,
      "step": 14450
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42643918798345376,
      "learning_rate": 1.232483183251499e-05,
      "loss": 0.2865,
      "step": 14451
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9040424742170932,
      "learning_rate": 1.2323867114761938e-05,
      "loss": 0.5272,
      "step": 14452
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.329643913054395,
      "learning_rate": 1.2322902374145922e-05,
      "loss": 0.1655,
      "step": 14453
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5850373594644914,
      "learning_rate": 1.2321937610676424e-05,
      "loss": 0.3791,
      "step": 14454
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42075480118850084,
      "learning_rate": 1.232097282436294e-05,
      "loss": 0.2762,
      "step": 14455
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.24979922842701852,
      "learning_rate": 1.2320008015214964e-05,
      "loss": 0.1725,
      "step": 14456
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.339028525985444,
      "learning_rate": 1.2319043183241986e-05,
      "loss": 0.2866,
      "step": 14457
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29346521832802025,
      "learning_rate": 1.2318078328453495e-05,
      "loss": 0.0698,
      "step": 14458
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.46998785416963884,
      "learning_rate": 1.2317113450858987e-05,
      "loss": 0.3097,
      "step": 14459
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5874108870554678,
      "learning_rate": 1.2316148550467958e-05,
      "loss": 0.3557,
      "step": 14460
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.3653414397881805,
      "learning_rate": 1.2315183627289892e-05,
      "loss": 0.8709,
      "step": 14461
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3684546454900704,
      "learning_rate": 1.2314218681334291e-05,
      "loss": 0.2522,
      "step": 14462
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4195848540418576,
      "learning_rate": 1.2313253712610647e-05,
      "loss": 0.2787,
      "step": 14463
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3227702628056426,
      "learning_rate": 1.231228872112845e-05,
      "loss": 0.2442,
      "step": 14464
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.4773971660602143,
      "learning_rate": 1.2311323706897196e-05,
      "loss": 0.6219,
      "step": 14465
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.20696041957424893,
      "learning_rate": 1.231035866992638e-05,
      "loss": 0.1257,
      "step": 14466
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5575819143584245,
      "learning_rate": 1.230939361022549e-05,
      "loss": 0.3564,
      "step": 14467
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3264906146729867,
      "learning_rate": 1.2308428527804028e-05,
      "loss": 0.284,
      "step": 14468
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2917965476873125,
      "learning_rate": 1.2307463422671491e-05,
      "loss": 0.2376,
      "step": 14469
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1739167373399761,
      "learning_rate": 1.2306498294837366e-05,
      "loss": 0.6917,
      "step": 14470
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.46545995942025437,
      "learning_rate": 1.2305533144311153e-05,
      "loss": 0.2791,
      "step": 14471
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3360906616722459,
      "learning_rate": 1.2304567971102344e-05,
      "loss": 0.2426,
      "step": 14472
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.1378076548550777,
      "learning_rate": 1.230360277522044e-05,
      "loss": 0.2649,
      "step": 14473
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.962124222708152,
      "learning_rate": 1.230263755667493e-05,
      "loss": 0.6364,
      "step": 14474
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2692735364442573,
      "learning_rate": 1.230167231547532e-05,
      "loss": 0.2809,
      "step": 14475
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.704107264835856,
      "learning_rate": 1.2300707051631098e-05,
      "loss": 0.3055,
      "step": 14476
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.320870627324379,
      "learning_rate": 1.229974176515176e-05,
      "loss": 0.2174,
      "step": 14477
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.25634755938263554,
      "learning_rate": 1.2298776456046805e-05,
      "loss": 0.1968,
      "step": 14478
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.008694955856295,
      "learning_rate": 1.2297811124325735e-05,
      "loss": 0.5667,
      "step": 14479
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.32821568750032815,
      "learning_rate": 1.2296845769998042e-05,
      "loss": 0.3053,
      "step": 14480
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3354983926229187,
      "learning_rate": 1.2295880393073222e-05,
      "loss": 0.0749,
      "step": 14481
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3620235967165793,
      "learning_rate": 1.229491499356078e-05,
      "loss": 0.2196,
      "step": 14482
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8562918815995864,
      "learning_rate": 1.2293949571470207e-05,
      "loss": 0.5507,
      "step": 14483
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5940901029901151,
      "learning_rate": 1.2292984126811004e-05,
      "loss": 0.3374,
      "step": 14484
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8634727401839496,
      "learning_rate": 1.2292018659592667e-05,
      "loss": 0.4542,
      "step": 14485
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.31357886243877764,
      "learning_rate": 1.2291053169824696e-05,
      "loss": 0.2248,
      "step": 14486
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.25425816539987206,
      "learning_rate": 1.229008765751659e-05,
      "loss": 0.2275,
      "step": 14487
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2359139259220525,
      "learning_rate": 1.228912212267785e-05,
      "loss": 0.1567,
      "step": 14488
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.6429982450753549,
      "learning_rate": 1.2288156565317975e-05,
      "loss": 0.8821,
      "step": 14489
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.29440423704953333,
      "learning_rate": 1.2287190985446458e-05,
      "loss": 0.1699,
      "step": 14490
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5834650229728938,
      "learning_rate": 1.2286225383072805e-05,
      "loss": 0.3934,
      "step": 14491
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4042556863934852,
      "learning_rate": 1.2285259758206517e-05,
      "loss": 0.2809,
      "step": 14492
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.42215782528822837,
      "learning_rate": 1.228429411085709e-05,
      "loss": 0.3462,
      "step": 14493
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6350442679700838,
      "learning_rate": 1.2283328441034024e-05,
      "loss": 0.2785,
      "step": 14494
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.27321727085899206,
      "learning_rate": 1.2282362748746826e-05,
      "loss": 0.1979,
      "step": 14495
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.23478857686913282,
      "learning_rate": 1.2281397034004987e-05,
      "loss": 0.1385,
      "step": 14496
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8542777125941148,
      "learning_rate": 1.2280431296818014e-05,
      "loss": 0.5801,
      "step": 14497
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3452845129705547,
      "learning_rate": 1.227946553719541e-05,
      "loss": 0.3161,
      "step": 14498
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2722357120412941,
      "learning_rate": 1.2278499755146673e-05,
      "loss": 0.1872,
      "step": 14499
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.8031965287042826,
      "learning_rate": 1.2277533950681307e-05,
      "loss": 0.8966,
      "step": 14500
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7720172945925533,
      "learning_rate": 1.227656812380881e-05,
      "loss": 0.3535,
      "step": 14501
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7638583986730603,
      "learning_rate": 1.227560227453869e-05,
      "loss": 0.4891,
      "step": 14502
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.40832334123041847,
      "learning_rate": 1.2274636402880442e-05,
      "loss": 0.1752,
      "step": 14503
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5385140944461316,
      "learning_rate": 1.2273670508843576e-05,
      "loss": 0.3152,
      "step": 14504
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.27965663919097733,
      "learning_rate": 1.2272704592437589e-05,
      "loss": 0.2541,
      "step": 14505
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.4179241081055684,
      "learning_rate": 1.2271738653671987e-05,
      "loss": 0.2872,
      "step": 14506
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.21690818572340007,
      "learning_rate": 1.2270772692556277e-05,
      "loss": 0.1176,
      "step": 14507
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.22064370079674697,
      "learning_rate": 1.2269806709099954e-05,
      "loss": 0.0696,
      "step": 14508
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.41824464201207023,
      "learning_rate": 1.2268840703312526e-05,
      "loss": 0.3636,
      "step": 14509
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3839030852510465,
      "learning_rate": 1.2267874675203495e-05,
      "loss": 0.2962,
      "step": 14510
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.9452344074582629,
      "learning_rate": 1.2266908624782367e-05,
      "loss": 0.5789,
      "step": 14511
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.7928259236831926,
      "learning_rate": 1.226594255205865e-05,
      "loss": 0.0863,
      "step": 14512
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3497429450900124,
      "learning_rate": 1.2264976457041839e-05,
      "loss": 0.2747,
      "step": 14513
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2437107120432191,
      "learning_rate": 1.2264010339741448e-05,
      "loss": 0.1649,
      "step": 14514
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.250389389082044,
      "learning_rate": 1.226304420016698e-05,
      "loss": 0.9208,
      "step": 14515
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.28842405346708283,
      "learning_rate": 1.2262078038327933e-05,
      "loss": 0.2492,
      "step": 14516
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.524172433918055,
      "learning_rate": 1.2261111854233822e-05,
      "loss": 0.1801,
      "step": 14517
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3451099392632842,
      "learning_rate": 1.2260145647894149e-05,
      "loss": 0.2481,
      "step": 14518
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.6584019438384746,
      "learning_rate": 1.2259179419318416e-05,
      "loss": 0.4554,
      "step": 14519
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8255969815684663,
      "learning_rate": 1.2258213168516137e-05,
      "loss": 0.2865,
      "step": 14520
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.297737254370743,
      "learning_rate": 1.2257246895496812e-05,
      "loss": 0.0591,
      "step": 14521
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3517907716506005,
      "learning_rate": 1.2256280600269948e-05,
      "loss": 0.2913,
      "step": 14522
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3217422709029122,
      "learning_rate": 1.2255314282845053e-05,
      "loss": 0.2582,
      "step": 14523
    },
    {
      "epoch": 0.44,
      "grad_norm": 1.5132632951460734,
      "learning_rate": 1.2254347943231632e-05,
      "loss": 0.9987,
      "step": 14524
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.5386928671039688,
      "learning_rate": 1.2253381581439199e-05,
      "loss": 0.2011,
      "step": 14525
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.48624208137794395,
      "learning_rate": 1.2252415197477257e-05,
      "loss": 0.3028,
      "step": 14526
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3373388101756372,
      "learning_rate": 1.2251448791355309e-05,
      "loss": 0.2174,
      "step": 14527
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.3962728550008328,
      "learning_rate": 1.225048236308287e-05,
      "loss": 0.3492,
      "step": 14528
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.2765969155282551,
      "learning_rate": 1.224951591266944e-05,
      "loss": 0.1893,
      "step": 14529
    },
    {
      "epoch": 0.44,
      "grad_norm": 0.8099722484884535,
      "learning_rate": 1.2248549440124537e-05,
      "loss": 0.4311,
      "step": 14530
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3749689153559041,
      "learning_rate": 1.2247582945457663e-05,
      "loss": 0.1902,
      "step": 14531
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5455491958011087,
      "learning_rate": 1.2246616428678329e-05,
      "loss": 0.3314,
      "step": 14532
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8199869159850578,
      "learning_rate": 1.2245649889796045e-05,
      "loss": 0.6084,
      "step": 14533
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2776792263267183,
      "learning_rate": 1.2244683328820318e-05,
      "loss": 0.2693,
      "step": 14534
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9361002141481599,
      "learning_rate": 1.2243716745760654e-05,
      "loss": 0.4148,
      "step": 14535
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3345677009589683,
      "learning_rate": 1.2242750140626569e-05,
      "loss": 0.2134,
      "step": 14536
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3897104173457065,
      "learning_rate": 1.224178351342757e-05,
      "loss": 0.2946,
      "step": 14537
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36793850341144335,
      "learning_rate": 1.2240816864173169e-05,
      "loss": 0.134,
      "step": 14538
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.98088027343569,
      "learning_rate": 1.2239850192872873e-05,
      "loss": 0.8792,
      "step": 14539
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.266425249000332,
      "learning_rate": 1.2238883499536192e-05,
      "loss": 0.2071,
      "step": 14540
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36262827918575435,
      "learning_rate": 1.223791678417264e-05,
      "loss": 0.301,
      "step": 14541
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9235600668321561,
      "learning_rate": 1.2236950046791726e-05,
      "loss": 0.6428,
      "step": 14542
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8013404686741644,
      "learning_rate": 1.223598328740296e-05,
      "loss": 0.476,
      "step": 14543
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4133967532563397,
      "learning_rate": 1.2235016506015857e-05,
      "loss": 0.289,
      "step": 14544
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3444611149793887,
      "learning_rate": 1.2234049702639928e-05,
      "loss": 0.3245,
      "step": 14545
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23267601230215318,
      "learning_rate": 1.2233082877284679e-05,
      "loss": 0.1794,
      "step": 14546
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2630357691879442,
      "learning_rate": 1.2232116029959628e-05,
      "loss": 0.0726,
      "step": 14547
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1779790544846873,
      "learning_rate": 1.2231149160674284e-05,
      "loss": 0.6262,
      "step": 14548
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28343418465647224,
      "learning_rate": 1.223018226943816e-05,
      "loss": 0.1879,
      "step": 14549
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0232631682294946,
      "learning_rate": 1.222921535626077e-05,
      "loss": 0.5347,
      "step": 14550
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6985454645887749,
      "learning_rate": 1.2228248421151624e-05,
      "loss": 0.3292,
      "step": 14551
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33163347873737437,
      "learning_rate": 1.222728146412024e-05,
      "loss": 0.3105,
      "step": 14552
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4324480929070522,
      "learning_rate": 1.2226314485176125e-05,
      "loss": 0.2389,
      "step": 14553
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3661845665550424,
      "learning_rate": 1.2225347484328797e-05,
      "loss": 0.2848,
      "step": 14554
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.006214784477762,
      "learning_rate": 1.2224380461587768e-05,
      "loss": 0.4412,
      "step": 14555
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29645621811277245,
      "learning_rate": 1.2223413416962552e-05,
      "loss": 0.1775,
      "step": 14556
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28274711095420646,
      "learning_rate": 1.2222446350462665e-05,
      "loss": 0.2369,
      "step": 14557
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6583037400603938,
      "learning_rate": 1.2221479262097619e-05,
      "loss": 0.2909,
      "step": 14558
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3185942032331688,
      "learning_rate": 1.2220512151876926e-05,
      "loss": 0.2064,
      "step": 14559
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.807176903433668,
      "learning_rate": 1.2219545019810109e-05,
      "loss": 0.3539,
      "step": 14560
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6695473352722979,
      "learning_rate": 1.2218577865906673e-05,
      "loss": 0.4569,
      "step": 14561
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3227881616968608,
      "learning_rate": 1.2217610690176139e-05,
      "loss": 0.1513,
      "step": 14562
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32787983558682754,
      "learning_rate": 1.221664349262802e-05,
      "loss": 0.2628,
      "step": 14563
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2987002282300534,
      "learning_rate": 1.221567627327184e-05,
      "loss": 0.2418,
      "step": 14564
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3383263950456291,
      "learning_rate": 1.22147090321171e-05,
      "loss": 0.1846,
      "step": 14565
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.166774699678442,
      "learning_rate": 1.2213741769173327e-05,
      "loss": 0.2647,
      "step": 14566
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6715082284771654,
      "learning_rate": 1.2212774484450033e-05,
      "loss": 0.3626,
      "step": 14567
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3182959542838629,
      "learning_rate": 1.2211807177956736e-05,
      "loss": 0.243,
      "step": 14568
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5896678402258064,
      "learning_rate": 1.2210839849702951e-05,
      "loss": 0.3669,
      "step": 14569
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43690463924949485,
      "learning_rate": 1.2209872499698198e-05,
      "loss": 0.2775,
      "step": 14570
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8385001498358072,
      "learning_rate": 1.2208905127951994e-05,
      "loss": 0.5052,
      "step": 14571
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2768580329181273,
      "learning_rate": 1.220793773447385e-05,
      "loss": 0.1938,
      "step": 14572
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26924132899207215,
      "learning_rate": 1.220697031927329e-05,
      "loss": 0.1418,
      "step": 14573
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.507076095324346,
      "learning_rate": 1.220600288235983e-05,
      "loss": 0.2549,
      "step": 14574
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3091729052063594,
      "learning_rate": 1.2205035423742989e-05,
      "loss": 0.2579,
      "step": 14575
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6423595478384581,
      "learning_rate": 1.2204067943432283e-05,
      "loss": 0.3533,
      "step": 14576
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3490673754203793,
      "learning_rate": 1.220310044143723e-05,
      "loss": 0.2579,
      "step": 14577
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0569513578329615,
      "learning_rate": 1.2202132917767351e-05,
      "loss": 0.5924,
      "step": 14578
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7410431789233789,
      "learning_rate": 1.2201165372432162e-05,
      "loss": 0.3969,
      "step": 14579
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7889023285476866,
      "learning_rate": 1.2200197805441187e-05,
      "loss": 0.4757,
      "step": 14580
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2812649160331026,
      "learning_rate": 1.219923021680394e-05,
      "loss": 0.21,
      "step": 14581
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43407746386447216,
      "learning_rate": 1.2198262606529942e-05,
      "loss": 0.3248,
      "step": 14582
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2609081179202756,
      "learning_rate": 1.2197294974628714e-05,
      "loss": 0.1856,
      "step": 14583
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3940538692662107,
      "learning_rate": 1.2196327321109778e-05,
      "loss": 0.2725,
      "step": 14584
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6410685649607331,
      "learning_rate": 1.2195359645982647e-05,
      "loss": 0.2192,
      "step": 14585
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4065674358650993,
      "learning_rate": 1.2194391949256846e-05,
      "loss": 0.2441,
      "step": 14586
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9060884535559349,
      "learning_rate": 1.2193424230941896e-05,
      "loss": 0.4763,
      "step": 14587
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2740592601071316,
      "learning_rate": 1.2192456491047316e-05,
      "loss": 0.2535,
      "step": 14588
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3330893779347315,
      "learning_rate": 1.2191488729582628e-05,
      "loss": 0.8578,
      "step": 14589
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2954333816336275,
      "learning_rate": 1.2190520946557354e-05,
      "loss": 0.1506,
      "step": 14590
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5356588347428023,
      "learning_rate": 1.2189553141981012e-05,
      "loss": 0.368,
      "step": 14591
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.16458316683166987,
      "learning_rate": 1.2188585315863126e-05,
      "loss": 0.0656,
      "step": 14592
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5035765301395949,
      "learning_rate": 1.218761746821322e-05,
      "loss": 0.371,
      "step": 14593
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28645817534881823,
      "learning_rate": 1.218664959904081e-05,
      "loss": 0.1857,
      "step": 14594
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4837465582847517,
      "learning_rate": 1.2185681708355426e-05,
      "loss": 0.3413,
      "step": 14595
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1624256059166194,
      "learning_rate": 1.2184713796166583e-05,
      "loss": 0.3917,
      "step": 14596
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6374210309230155,
      "learning_rate": 1.2183745862483807e-05,
      "loss": 0.8288,
      "step": 14597
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8076522843124188,
      "learning_rate": 1.2182777907316621e-05,
      "loss": 0.4932,
      "step": 14598
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28941447916877505,
      "learning_rate": 1.2181809930674546e-05,
      "loss": 0.1837,
      "step": 14599
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35549527130353176,
      "learning_rate": 1.218084193256711e-05,
      "loss": 0.3006,
      "step": 14600
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9229053201712768,
      "learning_rate": 1.2179873913003832e-05,
      "loss": 0.2731,
      "step": 14601
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38677476086521856,
      "learning_rate": 1.2178905871994237e-05,
      "loss": 0.2266,
      "step": 14602
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5473743500047534,
      "learning_rate": 1.217793780954785e-05,
      "loss": 0.2535,
      "step": 14603
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.410006895576875,
      "learning_rate": 1.2176969725674194e-05,
      "loss": 0.3328,
      "step": 14604
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9547585695566575,
      "learning_rate": 1.2176001620382793e-05,
      "loss": 0.2565,
      "step": 14605
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.333591979576122,
      "learning_rate": 1.2175033493683173e-05,
      "loss": 0.3095,
      "step": 14606
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5195809854031462,
      "learning_rate": 1.2174065345584855e-05,
      "loss": 0.2782,
      "step": 14607
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3255877192363579,
      "learning_rate": 1.2173097176097371e-05,
      "loss": 0.2328,
      "step": 14608
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0021962681521548,
      "learning_rate": 1.217212898523024e-05,
      "loss": 0.3315,
      "step": 14609
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4468757427336681,
      "learning_rate": 1.217116077299299e-05,
      "loss": 0.2538,
      "step": 14610
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.31684843928149503,
      "learning_rate": 1.2170192539395142e-05,
      "loss": 0.2701,
      "step": 14611
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36741095815199976,
      "learning_rate": 1.2169224284446228e-05,
      "loss": 0.1939,
      "step": 14612
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4216415459479409,
      "learning_rate": 1.2168256008155775e-05,
      "loss": 0.3237,
      "step": 14613
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32121508745287236,
      "learning_rate": 1.2167287710533304e-05,
      "loss": 0.1633,
      "step": 14614
    },
    {
      "epoch": 0.45,
      "grad_norm": 3.454702643629183,
      "learning_rate": 1.2166319391588342e-05,
      "loss": 0.8158,
      "step": 14615
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.91684239877943,
      "learning_rate": 1.2165351051330418e-05,
      "loss": 0.3292,
      "step": 14616
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3415786862413457,
      "learning_rate": 1.2164382689769059e-05,
      "loss": 0.283,
      "step": 14617
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26632507471731565,
      "learning_rate": 1.2163414306913787e-05,
      "loss": 0.24,
      "step": 14618
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1832308162537313,
      "learning_rate": 1.2162445902774138e-05,
      "loss": 0.5795,
      "step": 14619
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5448044733576155,
      "learning_rate": 1.2161477477359631e-05,
      "loss": 0.3425,
      "step": 14620
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3578769153960264,
      "learning_rate": 1.2160509030679799e-05,
      "loss": 0.2955,
      "step": 14621
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2890456847677631,
      "learning_rate": 1.2159540562744169e-05,
      "loss": 0.1978,
      "step": 14622
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3068531289891886,
      "learning_rate": 1.2158572073562266e-05,
      "loss": 0.237,
      "step": 14623
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3924995787081976,
      "learning_rate": 1.2157603563143623e-05,
      "loss": 0.2435,
      "step": 14624
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0835727127353427,
      "learning_rate": 1.2156635031497763e-05,
      "loss": 0.5638,
      "step": 14625
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3843092353334442,
      "learning_rate": 1.2155666478634222e-05,
      "loss": 0.2707,
      "step": 14626
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4249495919653171,
      "learning_rate": 1.2154697904562525e-05,
      "loss": 0.207,
      "step": 14627
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0084947685411785,
      "learning_rate": 1.2153729309292196e-05,
      "loss": 0.5352,
      "step": 14628
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29413529253145604,
      "learning_rate": 1.2152760692832775e-05,
      "loss": 0.2797,
      "step": 14629
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8074685920672993,
      "learning_rate": 1.2151792055193782e-05,
      "loss": 0.4671,
      "step": 14630
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.295100352153749,
      "learning_rate": 1.2150823396384752e-05,
      "loss": 0.2164,
      "step": 14631
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5384910068572281,
      "learning_rate": 1.2149854716415215e-05,
      "loss": 0.2353,
      "step": 14632
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.21639829379537776,
      "learning_rate": 1.21488860152947e-05,
      "loss": 0.1499,
      "step": 14633
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.42565857774420385,
      "learning_rate": 1.2147917293032735e-05,
      "loss": 0.3616,
      "step": 14634
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27096305130165815,
      "learning_rate": 1.2146948549638855e-05,
      "loss": 0.2039,
      "step": 14635
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5577142258660863,
      "learning_rate": 1.2145979785122589e-05,
      "loss": 0.3873,
      "step": 14636
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6686360099106283,
      "learning_rate": 1.2145010999493462e-05,
      "loss": 0.3566,
      "step": 14637
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5832244669600781,
      "learning_rate": 1.2144042192761017e-05,
      "loss": 0.4594,
      "step": 14638
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.20165413464749163,
      "learning_rate": 1.2143073364934779e-05,
      "loss": 0.0692,
      "step": 14639
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3070015339074694,
      "learning_rate": 1.214210451602428e-05,
      "loss": 0.2098,
      "step": 14640
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.22573507256406167,
      "learning_rate": 1.2141135646039048e-05,
      "loss": 0.1913,
      "step": 14641
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39513567251450876,
      "learning_rate": 1.2140166754988624e-05,
      "loss": 0.2933,
      "step": 14642
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1953477517960733,
      "learning_rate": 1.2139197842882532e-05,
      "loss": 0.7004,
      "step": 14643
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27792227064444025,
      "learning_rate": 1.2138228909730307e-05,
      "loss": 0.0792,
      "step": 14644
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40459208529369506,
      "learning_rate": 1.2137259955541487e-05,
      "loss": 0.3021,
      "step": 14645
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7824305111239086,
      "learning_rate": 1.2136290980325598e-05,
      "loss": 0.3727,
      "step": 14646
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3114763966532751,
      "learning_rate": 1.2135321984092175e-05,
      "loss": 0.3024,
      "step": 14647
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9376523189887618,
      "learning_rate": 1.2134352966850752e-05,
      "loss": 0.2515,
      "step": 14648
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3761644270229265,
      "learning_rate": 1.2133383928610863e-05,
      "loss": 0.2767,
      "step": 14649
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2618099442585738,
      "learning_rate": 1.2132414869382036e-05,
      "loss": 0.1718,
      "step": 14650
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4111789209691248,
      "learning_rate": 1.2131445789173814e-05,
      "loss": 0.8574,
      "step": 14651
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26369454619073024,
      "learning_rate": 1.2130476687995726e-05,
      "loss": 0.2018,
      "step": 14652
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3007936120127406,
      "learning_rate": 1.212950756585731e-05,
      "loss": 0.1886,
      "step": 14653
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36616915531677063,
      "learning_rate": 1.2128538422768096e-05,
      "loss": 0.339,
      "step": 14654
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7282961845113063,
      "learning_rate": 1.212756925873762e-05,
      "loss": 0.3814,
      "step": 14655
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7963178197860655,
      "learning_rate": 1.2126600073775418e-05,
      "loss": 0.5663,
      "step": 14656
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.46342680162879263,
      "learning_rate": 1.2125630867891025e-05,
      "loss": 0.1517,
      "step": 14657
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3329215401479283,
      "learning_rate": 1.2124661641093979e-05,
      "loss": 0.2812,
      "step": 14658
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26694405950271316,
      "learning_rate": 1.2123692393393809e-05,
      "loss": 0.1655,
      "step": 14659
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4373624202345396,
      "learning_rate": 1.2122723124800054e-05,
      "loss": 0.3342,
      "step": 14660
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6072662840213566,
      "learning_rate": 1.2121753835322252e-05,
      "loss": 0.2049,
      "step": 14661
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5829254459422787,
      "learning_rate": 1.2120784524969935e-05,
      "loss": 0.3862,
      "step": 14662
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3198864797781038,
      "learning_rate": 1.2119815193752644e-05,
      "loss": 0.2201,
      "step": 14663
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.034093583063153,
      "learning_rate": 1.2118845841679916e-05,
      "loss": 0.4638,
      "step": 14664
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2867963728291499,
      "learning_rate": 1.2117876468761283e-05,
      "loss": 0.2438,
      "step": 14665
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8255885311513852,
      "learning_rate": 1.2116907075006284e-05,
      "loss": 0.2971,
      "step": 14666
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6192921677324987,
      "learning_rate": 1.2115937660424457e-05,
      "loss": 0.3229,
      "step": 14667
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3012364055101946,
      "learning_rate": 1.2114968225025338e-05,
      "loss": 0.2371,
      "step": 14668
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3062934255535756,
      "learning_rate": 1.2113998768818466e-05,
      "loss": 0.8048,
      "step": 14669
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2707889114717557,
      "learning_rate": 1.211302929181338e-05,
      "loss": 0.2174,
      "step": 14670
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39882058669117826,
      "learning_rate": 1.2112059794019617e-05,
      "loss": 0.2881,
      "step": 14671
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36633137544079375,
      "learning_rate": 1.2111090275446715e-05,
      "loss": 0.2272,
      "step": 14672
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36499982708725176,
      "learning_rate": 1.2110120736104209e-05,
      "loss": 0.2737,
      "step": 14673
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8035384032330448,
      "learning_rate": 1.2109151176001646e-05,
      "loss": 0.3109,
      "step": 14674
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6830143627105412,
      "learning_rate": 1.2108181595148554e-05,
      "loss": 0.716,
      "step": 14675
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.31033031358201024,
      "learning_rate": 1.210721199355448e-05,
      "loss": 0.1676,
      "step": 14676
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27517616211247936,
      "learning_rate": 1.2106242371228963e-05,
      "loss": 0.2962,
      "step": 14677
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9609579436356555,
      "learning_rate": 1.210527272818154e-05,
      "loss": 0.5628,
      "step": 14678
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.563788854723257,
      "learning_rate": 1.210430306442175e-05,
      "loss": 0.3534,
      "step": 14679
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5666530949701609,
      "learning_rate": 1.2103333379959133e-05,
      "loss": 0.4078,
      "step": 14680
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29916385976864424,
      "learning_rate": 1.2102363674803233e-05,
      "loss": 0.2072,
      "step": 14681
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.47900448944159707,
      "learning_rate": 1.2101393948963583e-05,
      "loss": 0.2847,
      "step": 14682
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2450092451004654,
      "learning_rate": 1.210042420244973e-05,
      "loss": 0.2018,
      "step": 14683
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0707833137931864,
      "learning_rate": 1.2099454435271215e-05,
      "loss": 0.5867,
      "step": 14684
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2955037476542611,
      "learning_rate": 1.2098484647437576e-05,
      "loss": 0.1607,
      "step": 14685
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5042833212777499,
      "learning_rate": 1.209751483895835e-05,
      "loss": 0.3253,
      "step": 14686
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6754750869160276,
      "learning_rate": 1.2096545009843089e-05,
      "loss": 0.4411,
      "step": 14687
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39188334117918355,
      "learning_rate": 1.2095575160101322e-05,
      "loss": 0.3555,
      "step": 14688
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28347872598229146,
      "learning_rate": 1.20946052897426e-05,
      "loss": 0.2177,
      "step": 14689
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5290803025214958,
      "learning_rate": 1.2093635398776464e-05,
      "loss": 0.372,
      "step": 14690
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2777261924284394,
      "learning_rate": 1.2092665487212455e-05,
      "loss": 0.1767,
      "step": 14691
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2541219692203249,
      "learning_rate": 1.209169555506011e-05,
      "loss": 0.1087,
      "step": 14692
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4594942118216228,
      "learning_rate": 1.2090725602328977e-05,
      "loss": 0.3905,
      "step": 14693
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2578826705490231,
      "learning_rate": 1.2089755629028599e-05,
      "loss": 0.0723,
      "step": 14694
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28608187853973965,
      "learning_rate": 1.2088785635168515e-05,
      "loss": 0.2743,
      "step": 14695
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1195873846433986,
      "learning_rate": 1.2087815620758274e-05,
      "loss": 0.4215,
      "step": 14696
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.750487257712142,
      "learning_rate": 1.2086845585807416e-05,
      "loss": 0.4311,
      "step": 14697
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9301172386895494,
      "learning_rate": 1.2085875530325482e-05,
      "loss": 0.322,
      "step": 14698
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38473855602995166,
      "learning_rate": 1.2084905454322015e-05,
      "loss": 0.25,
      "step": 14699
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24222858895763344,
      "learning_rate": 1.2083935357806568e-05,
      "loss": 0.1872,
      "step": 14700
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2907660055694781,
      "learning_rate": 1.2082965240788677e-05,
      "loss": 0.2417,
      "step": 14701
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9030195778315397,
      "learning_rate": 1.2081995103277889e-05,
      "loss": 0.2959,
      "step": 14702
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7480349659943633,
      "learning_rate": 1.2081024945283749e-05,
      "loss": 0.4258,
      "step": 14703
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3391263347543874,
      "learning_rate": 1.20800547668158e-05,
      "loss": 0.2477,
      "step": 14704
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0811566528959684,
      "learning_rate": 1.2079084567883586e-05,
      "loss": 0.3679,
      "step": 14705
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43353208027728424,
      "learning_rate": 1.2078114348496655e-05,
      "loss": 0.3648,
      "step": 14706
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33162816841126125,
      "learning_rate": 1.2077144108664554e-05,
      "loss": 0.2337,
      "step": 14707
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3291386320405113,
      "learning_rate": 1.2076173848396822e-05,
      "loss": 0.2794,
      "step": 14708
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.049851236358136,
      "learning_rate": 1.2075203567703012e-05,
      "loss": 0.2737,
      "step": 14709
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28560436483199964,
      "learning_rate": 1.2074233266592664e-05,
      "loss": 0.1697,
      "step": 14710
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4216522594333825,
      "learning_rate": 1.2073262945075326e-05,
      "loss": 0.2518,
      "step": 14711
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.39241573125160945,
      "learning_rate": 1.2072292603160545e-05,
      "loss": 0.2993,
      "step": 14712
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.32047840392324783,
      "learning_rate": 1.207132224085787e-05,
      "loss": 0.2225,
      "step": 14713
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6915717811552473,
      "learning_rate": 1.2070351858176843e-05,
      "loss": 0.5187,
      "step": 14714
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7450021588775944,
      "learning_rate": 1.2069381455127016e-05,
      "loss": 0.3855,
      "step": 14715
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.1877023484315177,
      "learning_rate": 1.2068411031717931e-05,
      "loss": 0.7118,
      "step": 14716
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2555869225087978,
      "learning_rate": 1.2067440587959138e-05,
      "loss": 0.0744,
      "step": 14717
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2734849443777979,
      "learning_rate": 1.2066470123860183e-05,
      "loss": 0.2502,
      "step": 14718
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.247909232008035,
      "learning_rate": 1.2065499639430615e-05,
      "loss": 0.1886,
      "step": 14719
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0109784723546018,
      "learning_rate": 1.2064529134679984e-05,
      "loss": 0.364,
      "step": 14720
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8256656255322466,
      "learning_rate": 1.2063558609617834e-05,
      "loss": 0.4351,
      "step": 14721
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3127424081619479,
      "learning_rate": 1.2062588064253718e-05,
      "loss": 0.2273,
      "step": 14722
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8800379650780277,
      "learning_rate": 1.2061617498597182e-05,
      "loss": 0.4847,
      "step": 14723
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2870255887412648,
      "learning_rate": 1.2060646912657772e-05,
      "loss": 0.2526,
      "step": 14724
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.5495932476533154,
      "learning_rate": 1.2059676306445041e-05,
      "loss": 0.8224,
      "step": 14725
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29699087954412723,
      "learning_rate": 1.2058705679968535e-05,
      "loss": 0.1781,
      "step": 14726
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5182569516393952,
      "learning_rate": 1.2057735033237807e-05,
      "loss": 0.3652,
      "step": 14727
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23209176378165342,
      "learning_rate": 1.2056764366262406e-05,
      "loss": 0.0913,
      "step": 14728
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2176079174030405,
      "learning_rate": 1.2055793679051877e-05,
      "loss": 0.6598,
      "step": 14729
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27588473942334274,
      "learning_rate": 1.2054822971615777e-05,
      "loss": 0.2454,
      "step": 14730
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28703152890597394,
      "learning_rate": 1.2053852243963648e-05,
      "loss": 0.2411,
      "step": 14731
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6516419586136054,
      "learning_rate": 1.2052881496105047e-05,
      "loss": 0.5021,
      "step": 14732
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8049870309690167,
      "learning_rate": 1.2051910728049526e-05,
      "loss": 0.5541,
      "step": 14733
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.61026630936758,
      "learning_rate": 1.2050939939806625e-05,
      "loss": 0.7506,
      "step": 14734
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.25785050093847195,
      "learning_rate": 1.2049969131385908e-05,
      "loss": 0.1791,
      "step": 14735
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4513028615769196,
      "learning_rate": 1.204899830279692e-05,
      "loss": 0.361,
      "step": 14736
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.28622715914059904,
      "learning_rate": 1.2048027454049209e-05,
      "loss": 0.1987,
      "step": 14737
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40983583754956787,
      "learning_rate": 1.204705658515233e-05,
      "loss": 0.2864,
      "step": 14738
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4513550619336201,
      "learning_rate": 1.2046085696115834e-05,
      "loss": 0.2556,
      "step": 14739
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.37525352668046946,
      "learning_rate": 1.204511478694928e-05,
      "loss": 0.261,
      "step": 14740
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.61691972358027,
      "learning_rate": 1.204414385766221e-05,
      "loss": 0.3522,
      "step": 14741
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33866843498209465,
      "learning_rate": 1.2043172908264178e-05,
      "loss": 0.3144,
      "step": 14742
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0244200635552325,
      "learning_rate": 1.2042201938764741e-05,
      "loss": 0.4723,
      "step": 14743
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23469821176046016,
      "learning_rate": 1.2041230949173448e-05,
      "loss": 0.073,
      "step": 14744
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3928392880897915,
      "learning_rate": 1.2040259939499855e-05,
      "loss": 0.2982,
      "step": 14745
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19277277598172576,
      "learning_rate": 1.2039288909753513e-05,
      "loss": 0.0934,
      "step": 14746
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.43950851060902113,
      "learning_rate": 1.2038317859943975e-05,
      "loss": 0.3463,
      "step": 14747
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3424900703295468,
      "learning_rate": 1.2037346790080795e-05,
      "loss": 0.2049,
      "step": 14748
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3577418517001228,
      "learning_rate": 1.2036375700173528e-05,
      "loss": 0.2926,
      "step": 14749
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7823714809583442,
      "learning_rate": 1.2035404590231725e-05,
      "loss": 0.3609,
      "step": 14750
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3088257233820266,
      "learning_rate": 1.2034433460264942e-05,
      "loss": 0.8386,
      "step": 14751
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5380527243887786,
      "learning_rate": 1.2033462310282733e-05,
      "loss": 0.0288,
      "step": 14752
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6325817145462622,
      "learning_rate": 1.2032491140294656e-05,
      "loss": 0.3194,
      "step": 14753
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24547868748301513,
      "learning_rate": 1.203151995031026e-05,
      "loss": 0.2331,
      "step": 14754
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.452431432161344,
      "learning_rate": 1.2030548740339102e-05,
      "loss": 0.2769,
      "step": 14755
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8512186201166458,
      "learning_rate": 1.202957751039074e-05,
      "loss": 0.3315,
      "step": 14756
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.379538569418611,
      "learning_rate": 1.2028606260474722e-05,
      "loss": 0.2078,
      "step": 14757
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3556562339685365,
      "learning_rate": 1.202763499060061e-05,
      "loss": 0.2814,
      "step": 14758
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7524711125641225,
      "learning_rate": 1.202666370077796e-05,
      "loss": 0.4677,
      "step": 14759
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.35045240990883886,
      "learning_rate": 1.2025692391016326e-05,
      "loss": 0.3173,
      "step": 14760
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8429847860666412,
      "learning_rate": 1.202472106132526e-05,
      "loss": 0.4811,
      "step": 14761
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.592988539493073,
      "learning_rate": 1.2023749711714325e-05,
      "loss": 0.3205,
      "step": 14762
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3319452942989301,
      "learning_rate": 1.2022778342193072e-05,
      "loss": 0.234,
      "step": 14763
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8381019135237678,
      "learning_rate": 1.202180695277106e-05,
      "loss": 0.5551,
      "step": 14764
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4665255681331968,
      "learning_rate": 1.202083554345785e-05,
      "loss": 0.3107,
      "step": 14765
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38427685376547044,
      "learning_rate": 1.2019864114262991e-05,
      "loss": 0.2815,
      "step": 14766
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29905192932471747,
      "learning_rate": 1.2018892665196045e-05,
      "loss": 0.212,
      "step": 14767
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.24999281683799549,
      "learning_rate": 1.2017921196266568e-05,
      "loss": 0.1388,
      "step": 14768
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9523497055570799,
      "learning_rate": 1.2016949707484121e-05,
      "loss": 0.6228,
      "step": 14769
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0584871753375176,
      "learning_rate": 1.2015978198858255e-05,
      "loss": 0.2501,
      "step": 14770
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8824034728125065,
      "learning_rate": 1.2015006670398535e-05,
      "loss": 0.4268,
      "step": 14771
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.245026246665668,
      "learning_rate": 1.2014035122114515e-05,
      "loss": 0.2225,
      "step": 14772
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4653988666781993,
      "learning_rate": 1.2013063554015757e-05,
      "loss": 0.3532,
      "step": 14773
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.633961954798812,
      "learning_rate": 1.2012091966111816e-05,
      "loss": 0.3563,
      "step": 14774
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5009830548025989,
      "learning_rate": 1.2011120358412253e-05,
      "loss": 0.2675,
      "step": 14775
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19205676869188631,
      "learning_rate": 1.2010148730926624e-05,
      "loss": 0.128,
      "step": 14776
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.48459397687866795,
      "learning_rate": 1.2009177083664492e-05,
      "loss": 0.3823,
      "step": 14777
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3195687158030425,
      "learning_rate": 1.2008205416635417e-05,
      "loss": 0.2611,
      "step": 14778
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3451575107495697,
      "learning_rate": 1.2007233729848956e-05,
      "loss": 0.595,
      "step": 14779
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4910261752859439,
      "learning_rate": 1.2006262023314663e-05,
      "loss": 0.1626,
      "step": 14780
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38948464447708225,
      "learning_rate": 1.2005290297042108e-05,
      "loss": 0.3215,
      "step": 14781
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8035134095243222,
      "learning_rate": 1.2004318551040847e-05,
      "loss": 0.3485,
      "step": 14782
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3758345036025907,
      "learning_rate": 1.2003346785320442e-05,
      "loss": 0.2249,
      "step": 14783
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4167036209377098,
      "learning_rate": 1.2002374999890449e-05,
      "loss": 0.3139,
      "step": 14784
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26021319888685057,
      "learning_rate": 1.2001403194760435e-05,
      "loss": 0.1823,
      "step": 14785
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.36541035667444227,
      "learning_rate": 1.2000431369939958e-05,
      "loss": 0.2392,
      "step": 14786
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.966609501383773,
      "learning_rate": 1.1999459525438576e-05,
      "loss": 0.4829,
      "step": 14787
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4505644485052905,
      "learning_rate": 1.1998487661265857e-05,
      "loss": 0.2522,
      "step": 14788
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3182861452834118,
      "learning_rate": 1.1997515777431355e-05,
      "loss": 0.2293,
      "step": 14789
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.284091298505103,
      "learning_rate": 1.1996543873944638e-05,
      "loss": 0.2757,
      "step": 14790
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6124189291100862,
      "learning_rate": 1.1995571950815264e-05,
      "loss": 0.3162,
      "step": 14791
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8020381251158568,
      "learning_rate": 1.1994600008052798e-05,
      "loss": 0.5092,
      "step": 14792
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9088528321756612,
      "learning_rate": 1.1993628045666799e-05,
      "loss": 0.3492,
      "step": 14793
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6614835678236716,
      "learning_rate": 1.1992656063666833e-05,
      "loss": 0.3863,
      "step": 14794
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.34272610404773307,
      "learning_rate": 1.199168406206246e-05,
      "loss": 0.2289,
      "step": 14795
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29233579680402605,
      "learning_rate": 1.1990712040863244e-05,
      "loss": 0.2656,
      "step": 14796
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2819564930197051,
      "learning_rate": 1.198974000007875e-05,
      "loss": 0.1849,
      "step": 14797
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7770588192604486,
      "learning_rate": 1.1988767939718537e-05,
      "loss": 0.2485,
      "step": 14798
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3329187552360907,
      "learning_rate": 1.1987795859792169e-05,
      "loss": 0.2782,
      "step": 14799
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6997498628247667,
      "learning_rate": 1.1986823760309213e-05,
      "loss": 0.3556,
      "step": 14800
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3269297513396955,
      "learning_rate": 1.1985851641279231e-05,
      "loss": 0.2836,
      "step": 14801
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0745577638684312,
      "learning_rate": 1.1984879502711788e-05,
      "loss": 0.2406,
      "step": 14802
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6373772018483813,
      "learning_rate": 1.1983907344616444e-05,
      "loss": 0.3636,
      "step": 14803
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.250417822145508,
      "learning_rate": 1.1982935167002772e-05,
      "loss": 0.1773,
      "step": 14804
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.285734956679633,
      "learning_rate": 1.198196296988033e-05,
      "loss": 0.8795,
      "step": 14805
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33077939463919415,
      "learning_rate": 1.198099075325868e-05,
      "loss": 0.1574,
      "step": 14806
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3155493105247492,
      "learning_rate": 1.1980018517147394e-05,
      "loss": 0.3014,
      "step": 14807
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.27424703356955843,
      "learning_rate": 1.1979046261556035e-05,
      "loss": 0.1893,
      "step": 14808
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7567647941723404,
      "learning_rate": 1.1978073986494165e-05,
      "loss": 0.3411,
      "step": 14809
    },
    {
      "epoch": 0.45,
      "grad_norm": 2.029084934144848,
      "learning_rate": 1.1977101691971356e-05,
      "loss": 0.8199,
      "step": 14810
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8215363484398837,
      "learning_rate": 1.197612937799717e-05,
      "loss": 0.3384,
      "step": 14811
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.40408182132242,
      "learning_rate": 1.197515704458117e-05,
      "loss": 0.2583,
      "step": 14812
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.29632363067203016,
      "learning_rate": 1.1974184691732927e-05,
      "loss": 0.2543,
      "step": 14813
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.4616000672751544,
      "learning_rate": 1.1973212319462003e-05,
      "loss": 0.3239,
      "step": 14814
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8090881547927223,
      "learning_rate": 1.197223992777797e-05,
      "loss": 0.385,
      "step": 14815
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6798270249386181,
      "learning_rate": 1.197126751669039e-05,
      "loss": 0.3809,
      "step": 14816
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23856134994839548,
      "learning_rate": 1.1970295086208832e-05,
      "loss": 0.1644,
      "step": 14817
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.49953225609607355,
      "learning_rate": 1.1969322636342862e-05,
      "loss": 0.2552,
      "step": 14818
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3084785714946816,
      "learning_rate": 1.1968350167102047e-05,
      "loss": 0.236,
      "step": 14819
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0877020399215596,
      "learning_rate": 1.1967377678495958e-05,
      "loss": 0.611,
      "step": 14820
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3614810343638659,
      "learning_rate": 1.196640517053416e-05,
      "loss": 0.0741,
      "step": 14821
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33251730401633667,
      "learning_rate": 1.1965432643226219e-05,
      "loss": 0.2437,
      "step": 14822
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3628072568025573,
      "learning_rate": 1.1964460096581705e-05,
      "loss": 0.7948,
      "step": 14823
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.104696523992738,
      "learning_rate": 1.1963487530610187e-05,
      "loss": 0.3073,
      "step": 14824
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.48052662818673225,
      "learning_rate": 1.1962514945321234e-05,
      "loss": 0.3901,
      "step": 14825
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2922462636819743,
      "learning_rate": 1.1961542340724408e-05,
      "loss": 0.1682,
      "step": 14826
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.26697888279333404,
      "learning_rate": 1.196056971682929e-05,
      "loss": 0.1845,
      "step": 14827
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.0136950386770127,
      "learning_rate": 1.1959597073645437e-05,
      "loss": 0.2583,
      "step": 14828
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.9952245393460877,
      "learning_rate": 1.1958624411182425e-05,
      "loss": 0.58,
      "step": 14829
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3984381983195757,
      "learning_rate": 1.1957651729449819e-05,
      "loss": 0.1241,
      "step": 14830
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.38140145610535214,
      "learning_rate": 1.1956679028457196e-05,
      "loss": 0.3343,
      "step": 14831
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2620259370634125,
      "learning_rate": 1.1955706308214117e-05,
      "loss": 0.2577,
      "step": 14832
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8492535022016144,
      "learning_rate": 1.1954733568730157e-05,
      "loss": 0.5461,
      "step": 14833
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.5725316015197204,
      "learning_rate": 1.1953760810014888e-05,
      "loss": 0.3539,
      "step": 14834
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3642686755497079,
      "learning_rate": 1.1952788032077872e-05,
      "loss": 0.2606,
      "step": 14835
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23116440293424792,
      "learning_rate": 1.1951815234928688e-05,
      "loss": 0.1424,
      "step": 14836
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.2791934119899204,
      "learning_rate": 1.1950842418576905e-05,
      "loss": 0.2329,
      "step": 14837
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.3780995110035192,
      "learning_rate": 1.1949869583032087e-05,
      "loss": 0.7425,
      "step": 14838
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.30560049512467197,
      "learning_rate": 1.1948896728303813e-05,
      "loss": 0.1184,
      "step": 14839
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3847635060547558,
      "learning_rate": 1.1947923854401654e-05,
      "loss": 0.3142,
      "step": 14840
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.8726475754708529,
      "learning_rate": 1.1946950961335176e-05,
      "loss": 0.3399,
      "step": 14841
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7163946017451348,
      "learning_rate": 1.1945978049113957e-05,
      "loss": 0.4643,
      "step": 14842
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.23784897402436922,
      "learning_rate": 1.1945005117747563e-05,
      "loss": 0.2143,
      "step": 14843
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3429197115448619,
      "learning_rate": 1.194403216724557e-05,
      "loss": 0.2663,
      "step": 14844
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.2949413529917173,
      "learning_rate": 1.1943059197617547e-05,
      "loss": 0.2795,
      "step": 14845
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3247764946118536,
      "learning_rate": 1.1942086208873069e-05,
      "loss": 0.2009,
      "step": 14846
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.877661514885644,
      "learning_rate": 1.194111320102171e-05,
      "loss": 0.3606,
      "step": 14847
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.6204659041239757,
      "learning_rate": 1.194014017407304e-05,
      "loss": 0.3475,
      "step": 14848
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33393582859349474,
      "learning_rate": 1.193916712803663e-05,
      "loss": 0.22,
      "step": 14849
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.33423913900111407,
      "learning_rate": 1.193819406292206e-05,
      "loss": 0.2733,
      "step": 14850
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.7550657852956333,
      "learning_rate": 1.1937220978738894e-05,
      "loss": 0.5232,
      "step": 14851
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.748718517185068,
      "learning_rate": 1.1936247875496713e-05,
      "loss": 0.3253,
      "step": 14852
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.586443679745712,
      "learning_rate": 1.1935274753205088e-05,
      "loss": 0.3008,
      "step": 14853
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.19905597491836527,
      "learning_rate": 1.1934301611873595e-05,
      "loss": 0.1494,
      "step": 14854
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.3122668717303449,
      "learning_rate": 1.1933328451511805e-05,
      "loss": 0.292,
      "step": 14855
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6128476476047033,
      "learning_rate": 1.1932355272129292e-05,
      "loss": 0.091,
      "step": 14856
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.4125074974919702,
      "learning_rate": 1.1931382073735636e-05,
      "loss": 0.5867,
      "step": 14857
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25401231261065804,
      "learning_rate": 1.1930408856340403e-05,
      "loss": 0.1897,
      "step": 14858
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.664049364304391,
      "learning_rate": 1.1929435619953174e-05,
      "loss": 0.4208,
      "step": 14859
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9364534693253407,
      "learning_rate": 1.1928462364583523e-05,
      "loss": 0.491,
      "step": 14860
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31370753324780676,
      "learning_rate": 1.1927489090241027e-05,
      "loss": 0.2965,
      "step": 14861
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3039310917916149,
      "learning_rate": 1.1926515796935256e-05,
      "loss": 0.1846,
      "step": 14862
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23847726296447805,
      "learning_rate": 1.1925542484675789e-05,
      "loss": 0.1544,
      "step": 14863
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5079785374244233,
      "learning_rate": 1.1924569153472203e-05,
      "loss": 0.3021,
      "step": 14864
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5663806558580554,
      "learning_rate": 1.1923595803334069e-05,
      "loss": 0.3217,
      "step": 14865
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3812242728919642,
      "learning_rate": 1.1922622434270972e-05,
      "loss": 0.3269,
      "step": 14866
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.253058931995983,
      "learning_rate": 1.192164904629248e-05,
      "loss": 0.218,
      "step": 14867
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8307430920751003,
      "learning_rate": 1.192067563940817e-05,
      "loss": 0.6079,
      "step": 14868
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6655219867990291,
      "learning_rate": 1.1919702213627624e-05,
      "loss": 0.4122,
      "step": 14869
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.967803221440221,
      "learning_rate": 1.1918728768960418e-05,
      "loss": 0.6481,
      "step": 14870
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3165945685144514,
      "learning_rate": 1.1917755305416121e-05,
      "loss": 0.163,
      "step": 14871
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5030387137147403,
      "learning_rate": 1.1916781823004317e-05,
      "loss": 0.3359,
      "step": 14872
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20415235752803476,
      "learning_rate": 1.1915808321734588e-05,
      "loss": 0.1691,
      "step": 14873
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.8443728428118995,
      "learning_rate": 1.1914834801616504e-05,
      "loss": 0.6978,
      "step": 14874
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5230466139611172,
      "learning_rate": 1.1913861262659645e-05,
      "loss": 0.2419,
      "step": 14875
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3311184110551585,
      "learning_rate": 1.191288770487359e-05,
      "loss": 0.2241,
      "step": 14876
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6959469605164393,
      "learning_rate": 1.1911914128267914e-05,
      "loss": 0.4522,
      "step": 14877
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2854443148373262,
      "learning_rate": 1.1910940532852197e-05,
      "loss": 0.2501,
      "step": 14878
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1539945325017416,
      "learning_rate": 1.1909966918636022e-05,
      "loss": 0.6742,
      "step": 14879
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27552819357311603,
      "learning_rate": 1.1908993285628963e-05,
      "loss": 0.0758,
      "step": 14880
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34441777505590565,
      "learning_rate": 1.1908019633840598e-05,
      "loss": 0.3147,
      "step": 14881
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20275750601839956,
      "learning_rate": 1.1907045963280507e-05,
      "loss": 0.1096,
      "step": 14882
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8316298395873201,
      "learning_rate": 1.1906072273958276e-05,
      "loss": 0.4531,
      "step": 14883
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3069774305099637,
      "learning_rate": 1.1905098565883473e-05,
      "loss": 0.2799,
      "step": 14884
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3743360052954626,
      "learning_rate": 1.1904124839065685e-05,
      "loss": 0.2191,
      "step": 14885
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37530107967989723,
      "learning_rate": 1.1903151093514493e-05,
      "loss": 0.2751,
      "step": 14886
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.331133261997577,
      "learning_rate": 1.1902177329239472e-05,
      "loss": 0.7864,
      "step": 14887
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.914018913240693,
      "learning_rate": 1.1901203546250205e-05,
      "loss": 0.3726,
      "step": 14888
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32423033988703887,
      "learning_rate": 1.1900229744556271e-05,
      "loss": 0.172,
      "step": 14889
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5202379078352848,
      "learning_rate": 1.1899255924167252e-05,
      "loss": 0.3555,
      "step": 14890
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3050675346647345,
      "learning_rate": 1.1898282085092727e-05,
      "loss": 0.2447,
      "step": 14891
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4859391481370999,
      "learning_rate": 1.1897308227342282e-05,
      "loss": 0.3085,
      "step": 14892
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3686699757618564,
      "learning_rate": 1.1896334350925494e-05,
      "loss": 0.2065,
      "step": 14893
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33741312610993607,
      "learning_rate": 1.1895360455851941e-05,
      "loss": 0.2905,
      "step": 14894
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9672131448959586,
      "learning_rate": 1.189438654213121e-05,
      "loss": 0.391,
      "step": 14895
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2599403178401551,
      "learning_rate": 1.1893412609772881e-05,
      "loss": 0.9102,
      "step": 14896
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28843199527472946,
      "learning_rate": 1.1892438658786533e-05,
      "loss": 0.2512,
      "step": 14897
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6937516723170334,
      "learning_rate": 1.1891464689181755e-05,
      "loss": 0.329,
      "step": 14898
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32639223748825824,
      "learning_rate": 1.1890490700968124e-05,
      "loss": 0.2371,
      "step": 14899
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4250006383636831,
      "learning_rate": 1.1889516694155224e-05,
      "loss": 0.262,
      "step": 14900
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5485141842470278,
      "learning_rate": 1.1888542668752634e-05,
      "loss": 0.3541,
      "step": 14901
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21557992930879585,
      "learning_rate": 1.1887568624769944e-05,
      "loss": 0.1899,
      "step": 14902
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8753271490362244,
      "learning_rate": 1.1886594562216729e-05,
      "loss": 0.3039,
      "step": 14903
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32932893845808675,
      "learning_rate": 1.1885620481102576e-05,
      "loss": 0.2552,
      "step": 14904
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3171734988827055,
      "learning_rate": 1.188464638143707e-05,
      "loss": 0.8264,
      "step": 14905
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2377551066311336,
      "learning_rate": 1.1883672263229793e-05,
      "loss": 0.0702,
      "step": 14906
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8256851208920272,
      "learning_rate": 1.1882698126490327e-05,
      "loss": 0.3966,
      "step": 14907
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3102943645226864,
      "learning_rate": 1.1881723971228258e-05,
      "loss": 0.2186,
      "step": 14908
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3049147734219631,
      "learning_rate": 1.188074979745317e-05,
      "loss": 0.3026,
      "step": 14909
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6700716186377782,
      "learning_rate": 1.1879775605174647e-05,
      "loss": 0.3309,
      "step": 14910
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43234156179356314,
      "learning_rate": 1.1878801394402272e-05,
      "loss": 0.2262,
      "step": 14911
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21970869130706128,
      "learning_rate": 1.1877827165145633e-05,
      "loss": 0.138,
      "step": 14912
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.49839524798863677,
      "learning_rate": 1.1876852917414307e-05,
      "loss": 0.3593,
      "step": 14913
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4128525050481776,
      "learning_rate": 1.1875878651217889e-05,
      "loss": 0.2835,
      "step": 14914
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3933257835596079,
      "learning_rate": 1.1874904366565957e-05,
      "loss": 0.2593,
      "step": 14915
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7384487489934396,
      "learning_rate": 1.1873930063468103e-05,
      "loss": 0.4597,
      "step": 14916
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32615135713510823,
      "learning_rate": 1.1872955741933905e-05,
      "loss": 0.229,
      "step": 14917
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7802746519297121,
      "learning_rate": 1.1871981401972952e-05,
      "loss": 0.4659,
      "step": 14918
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6979204194708619,
      "learning_rate": 1.1871007043594834e-05,
      "loss": 0.3327,
      "step": 14919
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2576473720179212,
      "learning_rate": 1.1870032666809128e-05,
      "loss": 0.2451,
      "step": 14920
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27908479786960055,
      "learning_rate": 1.1869058271625424e-05,
      "loss": 0.1855,
      "step": 14921
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5565958727555439,
      "learning_rate": 1.1868083858053316e-05,
      "loss": 0.324,
      "step": 14922
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7541312591203022,
      "learning_rate": 1.1867109426102379e-05,
      "loss": 0.6323,
      "step": 14923
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.605178269067724,
      "learning_rate": 1.1866134975782209e-05,
      "loss": 0.7948,
      "step": 14924
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5760281178766611,
      "learning_rate": 1.186516050710239e-05,
      "loss": 0.2333,
      "step": 14925
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5555953154410919,
      "learning_rate": 1.1864186020072503e-05,
      "loss": 0.3894,
      "step": 14926
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27578336049755187,
      "learning_rate": 1.1863211514702142e-05,
      "loss": 0.2424,
      "step": 14927
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.792046833796794,
      "learning_rate": 1.1862236991000898e-05,
      "loss": 0.3477,
      "step": 14928
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43336231770773437,
      "learning_rate": 1.1861262448978347e-05,
      "loss": 0.2662,
      "step": 14929
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25153075824745724,
      "learning_rate": 1.1860287888644088e-05,
      "loss": 0.0739,
      "step": 14930
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37658364265528294,
      "learning_rate": 1.1859313310007703e-05,
      "loss": 0.2821,
      "step": 14931
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31487156199054045,
      "learning_rate": 1.1858338713078783e-05,
      "loss": 0.264,
      "step": 14932
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41146353339119063,
      "learning_rate": 1.1857364097866913e-05,
      "loss": 0.2544,
      "step": 14933
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7156748384327132,
      "learning_rate": 1.1856389464381684e-05,
      "loss": 0.3378,
      "step": 14934
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3719385554909986,
      "learning_rate": 1.1855414812632686e-05,
      "loss": 0.258,
      "step": 14935
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4437191609165407,
      "learning_rate": 1.1854440142629509e-05,
      "loss": 0.3049,
      "step": 14936
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.665990872973428,
      "learning_rate": 1.1853465454381736e-05,
      "loss": 0.8735,
      "step": 14937
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3129270564768221,
      "learning_rate": 1.1852490747898962e-05,
      "loss": 0.249,
      "step": 14938
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8587443951834621,
      "learning_rate": 1.1851516023190771e-05,
      "loss": 0.3725,
      "step": 14939
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3364224657543371,
      "learning_rate": 1.1850541280266758e-05,
      "loss": 0.2483,
      "step": 14940
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28010837462656674,
      "learning_rate": 1.1849566519136515e-05,
      "loss": 0.1386,
      "step": 14941
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42578305657604687,
      "learning_rate": 1.1848591739809622e-05,
      "loss": 0.2763,
      "step": 14942
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33202908361219585,
      "learning_rate": 1.184761694229568e-05,
      "loss": 0.2589,
      "step": 14943
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3426730737348629,
      "learning_rate": 1.184664212660427e-05,
      "loss": 0.2689,
      "step": 14944
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.370021304284872,
      "learning_rate": 1.184566729274499e-05,
      "loss": 0.2658,
      "step": 14945
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.068041278298418,
      "learning_rate": 1.1844692440727428e-05,
      "loss": 0.6238,
      "step": 14946
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8750832086301219,
      "learning_rate": 1.1843717570561173e-05,
      "loss": 0.3448,
      "step": 14947
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.584462708802985,
      "learning_rate": 1.184274268225582e-05,
      "loss": 0.3197,
      "step": 14948
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33900189585698187,
      "learning_rate": 1.184176777582096e-05,
      "loss": 0.2097,
      "step": 14949
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4531917189552393,
      "learning_rate": 1.1840792851266179e-05,
      "loss": 0.3563,
      "step": 14950
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21904814805785847,
      "learning_rate": 1.1839817908601075e-05,
      "loss": 0.1738,
      "step": 14951
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5208910436229405,
      "learning_rate": 1.1838842947835232e-05,
      "loss": 0.3864,
      "step": 14952
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3586689916741779,
      "learning_rate": 1.183786796897825e-05,
      "loss": 0.1834,
      "step": 14953
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3625773078229288,
      "learning_rate": 1.183689297203972e-05,
      "loss": 0.2691,
      "step": 14954
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.61804798387123,
      "learning_rate": 1.1835917957029228e-05,
      "loss": 0.8197,
      "step": 14955
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34220617976347734,
      "learning_rate": 1.1834942923956376e-05,
      "loss": 0.2316,
      "step": 14956
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.944527094188757,
      "learning_rate": 1.1833967872830749e-05,
      "loss": 0.5253,
      "step": 14957
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.29534917109708286,
      "learning_rate": 1.1832992803661945e-05,
      "loss": 0.2183,
      "step": 14958
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5489216936565262,
      "learning_rate": 1.1832017716459552e-05,
      "loss": 0.7853,
      "step": 14959
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4670130229391171,
      "learning_rate": 1.1831042611233165e-05,
      "loss": 0.3267,
      "step": 14960
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.47433001753224246,
      "learning_rate": 1.1830067487992381e-05,
      "loss": 0.3082,
      "step": 14961
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2539301164193811,
      "learning_rate": 1.182909234674679e-05,
      "loss": 0.1968,
      "step": 14962
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24569750000052304,
      "learning_rate": 1.1828117187505987e-05,
      "loss": 0.1952,
      "step": 14963
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7423590174545407,
      "learning_rate": 1.1827142010279567e-05,
      "loss": 0.3217,
      "step": 14964
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2786418846758376,
      "learning_rate": 1.1826166815077117e-05,
      "loss": 0.6351,
      "step": 14965
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3561714619022055,
      "learning_rate": 1.1825191601908241e-05,
      "loss": 0.267,
      "step": 14966
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28658736734114,
      "learning_rate": 1.1824216370782531e-05,
      "loss": 0.2139,
      "step": 14967
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31619143097910196,
      "learning_rate": 1.1823241121709579e-05,
      "loss": 0.3298,
      "step": 14968
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5890559394508033,
      "learning_rate": 1.182226585469898e-05,
      "loss": 0.3281,
      "step": 14969
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.38587761635556456,
      "learning_rate": 1.1821290569760328e-05,
      "loss": 0.2716,
      "step": 14970
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24969995156145858,
      "learning_rate": 1.1820315266903225e-05,
      "loss": 0.1843,
      "step": 14971
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4655063921769876,
      "learning_rate": 1.1819339946137258e-05,
      "loss": 0.2725,
      "step": 14972
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2921080643688068,
      "learning_rate": 1.1818364607472026e-05,
      "loss": 0.0796,
      "step": 14973
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3083595986803536,
      "learning_rate": 1.1817389250917127e-05,
      "loss": 0.3016,
      "step": 14974
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.27146737547962874,
      "learning_rate": 1.1816413876482153e-05,
      "loss": 0.1585,
      "step": 14975
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3779333132209316,
      "learning_rate": 1.18154384841767e-05,
      "loss": 0.2856,
      "step": 14976
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9034768539376884,
      "learning_rate": 1.181446307401037e-05,
      "loss": 0.3917,
      "step": 14977
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6285868418102185,
      "learning_rate": 1.181348764599275e-05,
      "loss": 0.4539,
      "step": 14978
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3200392122645434,
      "learning_rate": 1.1812512200133443e-05,
      "loss": 0.2409,
      "step": 14979
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24894771246068112,
      "learning_rate": 1.181153673644205e-05,
      "loss": 0.0733,
      "step": 14980
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2683303050050424,
      "learning_rate": 1.1810561254928157e-05,
      "loss": 0.224,
      "step": 14981
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28379367404419326,
      "learning_rate": 1.1809585755601367e-05,
      "loss": 0.1081,
      "step": 14982
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2684297112599328,
      "learning_rate": 1.1808610238471279e-05,
      "loss": 0.6817,
      "step": 14983
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34878152920538896,
      "learning_rate": 1.1807634703547489e-05,
      "loss": 0.1477,
      "step": 14984
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3240464375760859,
      "learning_rate": 1.180665915083959e-05,
      "loss": 0.2801,
      "step": 14985
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3762259470417915,
      "learning_rate": 1.1805683580357188e-05,
      "loss": 0.2876,
      "step": 14986
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4010254611768622,
      "learning_rate": 1.1804707992109875e-05,
      "loss": 0.5206,
      "step": 14987
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0242081620731567,
      "learning_rate": 1.180373238610725e-05,
      "loss": 0.283,
      "step": 14988
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6124780810223548,
      "learning_rate": 1.1802756762358914e-05,
      "loss": 0.3143,
      "step": 14989
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1965417859627877,
      "learning_rate": 1.1801781120874462e-05,
      "loss": 0.1498,
      "step": 14990
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4653584210290241,
      "learning_rate": 1.1800805461663496e-05,
      "loss": 0.3866,
      "step": 14991
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.36463488022817836,
      "learning_rate": 1.1799829784735613e-05,
      "loss": 0.2128,
      "step": 14992
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5144986644209683,
      "learning_rate": 1.1798854090100411e-05,
      "loss": 0.2977,
      "step": 14993
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34511871962375534,
      "learning_rate": 1.1797878377767493e-05,
      "loss": 0.2912,
      "step": 14994
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6019268132515729,
      "learning_rate": 1.1796902647746453e-05,
      "loss": 0.369,
      "step": 14995
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.5360244979689095,
      "learning_rate": 1.1795926900046894e-05,
      "loss": 0.8718,
      "step": 14996
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26598469191934193,
      "learning_rate": 1.1794951134678419e-05,
      "loss": 0.2318,
      "step": 14997
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6027939609189751,
      "learning_rate": 1.179397535165062e-05,
      "loss": 0.3117,
      "step": 14998
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24265227385108457,
      "learning_rate": 1.1792999550973105e-05,
      "loss": 0.1798,
      "step": 14999
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4407516661780185,
      "learning_rate": 1.1792023732655469e-05,
      "loss": 0.3008,
      "step": 15000
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8091755136187407,
      "learning_rate": 1.1791047896707313e-05,
      "loss": 0.4111,
      "step": 15001
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5809271831991708,
      "learning_rate": 1.1790072043138237e-05,
      "loss": 0.3757,
      "step": 15002
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2663111827170401,
      "learning_rate": 1.1789096171957845e-05,
      "loss": 0.1966,
      "step": 15003
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34271492383111135,
      "learning_rate": 1.1788120283175735e-05,
      "loss": 0.3071,
      "step": 15004
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7323988275185158,
      "learning_rate": 1.1787144376801509e-05,
      "loss": 0.3808,
      "step": 15005
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.86618281417618,
      "learning_rate": 1.178616845284477e-05,
      "loss": 0.0823,
      "step": 15006
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5953277289505922,
      "learning_rate": 1.178519251131512e-05,
      "loss": 0.2797,
      "step": 15007
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24331981808178957,
      "learning_rate": 1.1784216552222156e-05,
      "loss": 0.163,
      "step": 15008
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3210814198426768,
      "learning_rate": 1.1783240575575481e-05,
      "loss": 0.2563,
      "step": 15009
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5583929551961774,
      "learning_rate": 1.1782264581384701e-05,
      "loss": 0.2736,
      "step": 15010
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0849450497646123,
      "learning_rate": 1.1781288569659414e-05,
      "loss": 0.5433,
      "step": 15011
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.37935040785805396,
      "learning_rate": 1.1780312540409227e-05,
      "loss": 0.2049,
      "step": 15012
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4873560160580994,
      "learning_rate": 1.1779336493643738e-05,
      "loss": 0.3907,
      "step": 15013
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9797685057479323,
      "learning_rate": 1.177836042937255e-05,
      "loss": 0.4094,
      "step": 15014
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.46285404433976385,
      "learning_rate": 1.1777384347605265e-05,
      "loss": 0.3655,
      "step": 15015
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3032215682831497,
      "learning_rate": 1.1776408248351491e-05,
      "loss": 0.1941,
      "step": 15016
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41159549669782924,
      "learning_rate": 1.1775432131620827e-05,
      "loss": 0.3543,
      "step": 15017
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.17004271904105658,
      "learning_rate": 1.177445599742288e-05,
      "loss": 0.0709,
      "step": 15018
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0996810800063501,
      "learning_rate": 1.1773479845767249e-05,
      "loss": 0.5297,
      "step": 15019
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0218155507160287,
      "learning_rate": 1.177250367666354e-05,
      "loss": 0.3733,
      "step": 15020
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24631346403943272,
      "learning_rate": 1.1771527490121356e-05,
      "loss": 0.1756,
      "step": 15021
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33500877921306577,
      "learning_rate": 1.17705512861503e-05,
      "loss": 0.299,
      "step": 15022
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7159589383844119,
      "learning_rate": 1.1769575064759982e-05,
      "loss": 0.4839,
      "step": 15023
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2474774381468825,
      "learning_rate": 1.1768598825959998e-05,
      "loss": 0.7214,
      "step": 15024
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.419253993057426,
      "learning_rate": 1.176762256975996e-05,
      "loss": 0.1802,
      "step": 15025
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3010417939604492,
      "learning_rate": 1.176664629616947e-05,
      "loss": 0.2781,
      "step": 15026
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2822525293562489,
      "learning_rate": 1.1765670005198129e-05,
      "loss": 0.1734,
      "step": 15027
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5769698574037968,
      "learning_rate": 1.1764693696855548e-05,
      "loss": 0.3752,
      "step": 15028
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2286863195790345,
      "learning_rate": 1.1763717371151329e-05,
      "loss": 0.3597,
      "step": 15029
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3184927977710473,
      "learning_rate": 1.1762741028095077e-05,
      "loss": 0.2818,
      "step": 15030
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8933402240590617,
      "learning_rate": 1.1761764667696402e-05,
      "loss": 0.4855,
      "step": 15031
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.524145232333531,
      "learning_rate": 1.1760788289964904e-05,
      "loss": 0.823,
      "step": 15032
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3019494011384105,
      "learning_rate": 1.1759811894910191e-05,
      "loss": 0.2552,
      "step": 15033
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4517217088771828,
      "learning_rate": 1.175883548254187e-05,
      "loss": 0.166,
      "step": 15034
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3161002160092378,
      "learning_rate": 1.1757859052869546e-05,
      "loss": 0.2806,
      "step": 15035
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.394171167718836,
      "learning_rate": 1.1756882605902828e-05,
      "loss": 0.1945,
      "step": 15036
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6245101791195546,
      "learning_rate": 1.1755906141651324e-05,
      "loss": 0.4396,
      "step": 15037
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.1783353385981263,
      "learning_rate": 1.1754929660124632e-05,
      "loss": 0.0951,
      "step": 15038
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3408608896367878,
      "learning_rate": 1.1753953161332368e-05,
      "loss": 0.2703,
      "step": 15039
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26961554131599746,
      "learning_rate": 1.1752976645284133e-05,
      "loss": 0.2455,
      "step": 15040
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4215706393006082,
      "learning_rate": 1.1752000111989536e-05,
      "loss": 0.7365,
      "step": 15041
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9383912541934566,
      "learning_rate": 1.175102356145819e-05,
      "loss": 0.2785,
      "step": 15042
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.3386392437012486,
      "learning_rate": 1.1750046993699692e-05,
      "loss": 0.7227,
      "step": 15043
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2576212797279582,
      "learning_rate": 1.1749070408723662e-05,
      "loss": 0.1918,
      "step": 15044
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32061709322840376,
      "learning_rate": 1.1748093806539698e-05,
      "loss": 0.2968,
      "step": 15045
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5202984425363426,
      "learning_rate": 1.1747117187157415e-05,
      "loss": 0.3489,
      "step": 15046
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19126334939119793,
      "learning_rate": 1.1746140550586415e-05,
      "loss": 0.1461,
      "step": 15047
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4157065684719643,
      "learning_rate": 1.1745163896836309e-05,
      "loss": 0.3155,
      "step": 15048
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4353689768468456,
      "learning_rate": 1.174418722591671e-05,
      "loss": 0.2834,
      "step": 15049
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.246609385876596,
      "learning_rate": 1.1743210537837224e-05,
      "loss": 0.6951,
      "step": 15050
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3070149923593627,
      "learning_rate": 1.1742233832607456e-05,
      "loss": 0.2493,
      "step": 15051
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9441093318237953,
      "learning_rate": 1.174125711023702e-05,
      "loss": 0.4964,
      "step": 15052
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.302290326328448,
      "learning_rate": 1.1740280370735522e-05,
      "loss": 0.2058,
      "step": 15053
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9601882424379891,
      "learning_rate": 1.1739303614112571e-05,
      "loss": 0.5602,
      "step": 15054
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7935281822580327,
      "learning_rate": 1.1738326840377782e-05,
      "loss": 0.3483,
      "step": 15055
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.23671472586167377,
      "learning_rate": 1.173735004954076e-05,
      "loss": 0.1936,
      "step": 15056
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.24877711366100935,
      "learning_rate": 1.1736373241611118e-05,
      "loss": 0.1944,
      "step": 15057
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.529083413615304,
      "learning_rate": 1.1735396416598466e-05,
      "loss": 0.3294,
      "step": 15058
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9680649926088788,
      "learning_rate": 1.173441957451241e-05,
      "loss": 0.617,
      "step": 15059
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8727925699073009,
      "learning_rate": 1.1733442715362563e-05,
      "loss": 0.2873,
      "step": 15060
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6648248650316971,
      "learning_rate": 1.1732465839158538e-05,
      "loss": 0.4674,
      "step": 15061
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2920778507841728,
      "learning_rate": 1.1731488945909946e-05,
      "loss": 0.2029,
      "step": 15062
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4310447839357216,
      "learning_rate": 1.1730512035626396e-05,
      "loss": 0.3625,
      "step": 15063
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.401971903596737,
      "learning_rate": 1.1729535108317495e-05,
      "loss": 0.2695,
      "step": 15064
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2834009198522158,
      "learning_rate": 1.1728558163992863e-05,
      "loss": 0.1759,
      "step": 15065
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2840427670569424,
      "learning_rate": 1.1727581202662101e-05,
      "loss": 0.1227,
      "step": 15066
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3686789402033797,
      "learning_rate": 1.172660422433483e-05,
      "loss": 0.2933,
      "step": 15067
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.394894986357872,
      "learning_rate": 1.1725627229020662e-05,
      "loss": 0.2823,
      "step": 15068
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4611297508334026,
      "learning_rate": 1.1724650216729203e-05,
      "loss": 0.3595,
      "step": 15069
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.605217940752255,
      "learning_rate": 1.1723673187470066e-05,
      "loss": 0.3439,
      "step": 15070
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41261658920467054,
      "learning_rate": 1.1722696141252865e-05,
      "loss": 0.3372,
      "step": 15071
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5460549298102823,
      "learning_rate": 1.1721719078087215e-05,
      "loss": 0.2754,
      "step": 15072
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8284792730604061,
      "learning_rate": 1.1720741997982723e-05,
      "loss": 0.2663,
      "step": 15073
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2882280511885751,
      "learning_rate": 1.1719764900949007e-05,
      "loss": 0.2418,
      "step": 15074
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28038703762006095,
      "learning_rate": 1.1718787786995679e-05,
      "loss": 0.1895,
      "step": 15075
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3633000287076879,
      "learning_rate": 1.1717810656132352e-05,
      "loss": 0.3199,
      "step": 15076
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0385944293725318,
      "learning_rate": 1.1716833508368634e-05,
      "loss": 0.4819,
      "step": 15077
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43605850179533984,
      "learning_rate": 1.1715856343714148e-05,
      "loss": 0.2605,
      "step": 15078
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8777742320564454,
      "learning_rate": 1.17148791621785e-05,
      "loss": 0.2944,
      "step": 15079
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3447287439514782,
      "learning_rate": 1.1713901963771303e-05,
      "loss": 0.2743,
      "step": 15080
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34512587897194863,
      "learning_rate": 1.1712924748502182e-05,
      "loss": 0.2597,
      "step": 15081
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4273336337791622,
      "learning_rate": 1.171194751638074e-05,
      "loss": 0.3415,
      "step": 15082
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.173898284431131,
      "learning_rate": 1.1710970267416592e-05,
      "loss": 0.3513,
      "step": 15083
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7480877082774652,
      "learning_rate": 1.170999300161936e-05,
      "loss": 0.4125,
      "step": 15084
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32104365963704956,
      "learning_rate": 1.1709015718998652e-05,
      "loss": 0.2141,
      "step": 15085
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28024949322070514,
      "learning_rate": 1.1708038419564085e-05,
      "loss": 0.2036,
      "step": 15086
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33101245920957906,
      "learning_rate": 1.1707061103325277e-05,
      "loss": 0.2888,
      "step": 15087
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6982020313851087,
      "learning_rate": 1.1706083770291838e-05,
      "loss": 0.3625,
      "step": 15088
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6262168684509947,
      "learning_rate": 1.1705106420473386e-05,
      "loss": 0.3246,
      "step": 15089
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3292907460781489,
      "learning_rate": 1.1704129053879533e-05,
      "loss": 0.2231,
      "step": 15090
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0303906044872302,
      "learning_rate": 1.1703151670519903e-05,
      "loss": 0.5495,
      "step": 15091
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2942414708843215,
      "learning_rate": 1.1702174270404103e-05,
      "loss": 0.2503,
      "step": 15092
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9812227974210427,
      "learning_rate": 1.1701196853541751e-05,
      "loss": 0.5528,
      "step": 15093
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2922126103567068,
      "learning_rate": 1.170021941994247e-05,
      "loss": 0.2079,
      "step": 15094
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.45793882863329394,
      "learning_rate": 1.1699241969615867e-05,
      "loss": 0.3152,
      "step": 15095
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.674776673018355,
      "learning_rate": 1.169826450257156e-05,
      "loss": 0.3357,
      "step": 15096
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43498867814625164,
      "learning_rate": 1.1697287018819173e-05,
      "loss": 0.308,
      "step": 15097
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3352031342000085,
      "learning_rate": 1.1696309518368314e-05,
      "loss": 0.1885,
      "step": 15098
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2782612959374356,
      "learning_rate": 1.1695332001228605e-05,
      "loss": 0.2508,
      "step": 15099
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.4709561170033945,
      "learning_rate": 1.1694354467409665e-05,
      "loss": 0.6988,
      "step": 15100
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9849044118287454,
      "learning_rate": 1.1693376916921107e-05,
      "loss": 0.3176,
      "step": 15101
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8777404614982623,
      "learning_rate": 1.1692399349772548e-05,
      "loss": 0.6112,
      "step": 15102
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28588553890999324,
      "learning_rate": 1.1691421765973606e-05,
      "loss": 0.2068,
      "step": 15103
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.7175480989228071,
      "learning_rate": 1.1690444165533903e-05,
      "loss": 0.8823,
      "step": 15104
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8335106569110523,
      "learning_rate": 1.1689466548463052e-05,
      "loss": 0.2693,
      "step": 15105
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4606539024100941,
      "learning_rate": 1.1688488914770672e-05,
      "loss": 0.2961,
      "step": 15106
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19707395800617927,
      "learning_rate": 1.168751126446639e-05,
      "loss": 0.124,
      "step": 15107
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5628788441016694,
      "learning_rate": 1.1686533597559809e-05,
      "loss": 0.3733,
      "step": 15108
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9424711748680207,
      "learning_rate": 1.1685555914060556e-05,
      "loss": 0.3243,
      "step": 15109
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3137488029299414,
      "learning_rate": 1.168457821397825e-05,
      "loss": 0.2815,
      "step": 15110
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32311785444525754,
      "learning_rate": 1.1683600497322511e-05,
      "loss": 0.1597,
      "step": 15111
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31793936003834944,
      "learning_rate": 1.1682622764102956e-05,
      "loss": 0.2182,
      "step": 15112
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.828922722510717,
      "learning_rate": 1.1681645014329205e-05,
      "loss": 0.5826,
      "step": 15113
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5060347925822243,
      "learning_rate": 1.1680667248010877e-05,
      "loss": 0.3396,
      "step": 15114
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.22869913359443597,
      "learning_rate": 1.1679689465157592e-05,
      "loss": 0.1869,
      "step": 15115
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.25137204524484286,
      "learning_rate": 1.1678711665778968e-05,
      "loss": 0.0719,
      "step": 15116
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28513834119755144,
      "learning_rate": 1.1677733849884626e-05,
      "loss": 0.2823,
      "step": 15117
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.1409038991558769,
      "learning_rate": 1.1676756017484188e-05,
      "loss": 0.2582,
      "step": 15118
    },
    {
      "epoch": 0.46,
      "grad_norm": 2.1961782797593004,
      "learning_rate": 1.167577816858727e-05,
      "loss": 0.6932,
      "step": 15119
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6989854159836939,
      "learning_rate": 1.1674800303203497e-05,
      "loss": 0.3289,
      "step": 15120
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3800889680131278,
      "learning_rate": 1.1673822421342489e-05,
      "loss": 0.2563,
      "step": 15121
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.33596019643325314,
      "learning_rate": 1.1672844523013863e-05,
      "loss": 0.2574,
      "step": 15122
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.44765074725348447,
      "learning_rate": 1.167186660822724e-05,
      "loss": 0.3282,
      "step": 15123
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9627391233539482,
      "learning_rate": 1.1670888676992246e-05,
      "loss": 0.3435,
      "step": 15124
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.20192799315283363,
      "learning_rate": 1.1669910729318499e-05,
      "loss": 0.1326,
      "step": 15125
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.332597494187483,
      "learning_rate": 1.1668932765215621e-05,
      "loss": 0.2467,
      "step": 15126
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8975618897600105,
      "learning_rate": 1.1667954784693237e-05,
      "loss": 0.2791,
      "step": 15127
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.32490778568817663,
      "learning_rate": 1.166697678776096e-05,
      "loss": 0.2918,
      "step": 15128
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8363637679743059,
      "learning_rate": 1.1665998774428415e-05,
      "loss": 0.286,
      "step": 15129
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3413020671728752,
      "learning_rate": 1.166502074470523e-05,
      "loss": 0.2934,
      "step": 15130
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9002899748980114,
      "learning_rate": 1.1664042698601022e-05,
      "loss": 0.3249,
      "step": 15131
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.801955682680997,
      "learning_rate": 1.1663064636125414e-05,
      "loss": 0.4732,
      "step": 15132
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.41661367953309847,
      "learning_rate": 1.1662086557288031e-05,
      "loss": 0.2557,
      "step": 15133
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3423723682573918,
      "learning_rate": 1.1661108462098492e-05,
      "loss": 0.247,
      "step": 15134
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3265756686626446,
      "learning_rate": 1.1660130350566419e-05,
      "loss": 0.2203,
      "step": 15135
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3542113443814507,
      "learning_rate": 1.165915222270144e-05,
      "loss": 0.1569,
      "step": 15136
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.98793932392246,
      "learning_rate": 1.1658174078513176e-05,
      "loss": 0.4767,
      "step": 15137
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6791382647090318,
      "learning_rate": 1.1657195918011248e-05,
      "loss": 0.3052,
      "step": 15138
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9793369053906206,
      "learning_rate": 1.1656217741205283e-05,
      "loss": 0.5438,
      "step": 15139
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.28092904965208076,
      "learning_rate": 1.1655239548104901e-05,
      "loss": 0.2306,
      "step": 15140
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.43538081366761355,
      "learning_rate": 1.1654261338719729e-05,
      "loss": 0.3679,
      "step": 15141
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34905283403402343,
      "learning_rate": 1.1653283113059389e-05,
      "loss": 0.1675,
      "step": 15142
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9320381847703909,
      "learning_rate": 1.1652304871133506e-05,
      "loss": 0.4702,
      "step": 15143
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21044129305191142,
      "learning_rate": 1.1651326612951703e-05,
      "loss": 0.1599,
      "step": 15144
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0708141943628886,
      "learning_rate": 1.1650348338523608e-05,
      "loss": 0.6045,
      "step": 15145
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31989536433354926,
      "learning_rate": 1.164937004785884e-05,
      "loss": 0.2714,
      "step": 15146
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.2190786381251972,
      "learning_rate": 1.1648391740967028e-05,
      "loss": 0.5109,
      "step": 15147
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2793966941082578,
      "learning_rate": 1.1647413417857796e-05,
      "loss": 0.2128,
      "step": 15148
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.411542427760159,
      "learning_rate": 1.1646435078540767e-05,
      "loss": 0.3388,
      "step": 15149
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8691784171576007,
      "learning_rate": 1.1645456723025572e-05,
      "loss": 0.4645,
      "step": 15150
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3965420066907513,
      "learning_rate": 1.1644478351321831e-05,
      "loss": 0.2549,
      "step": 15151
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.42482452581605845,
      "learning_rate": 1.164349996343917e-05,
      "loss": 0.3284,
      "step": 15152
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.22745265048720717,
      "learning_rate": 1.1642521559387214e-05,
      "loss": 0.1668,
      "step": 15153
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.473988737454033,
      "learning_rate": 1.164154313917559e-05,
      "loss": 0.2486,
      "step": 15154
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.837441699613856,
      "learning_rate": 1.1640564702813926e-05,
      "loss": 0.3696,
      "step": 15155
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.7902117463677475,
      "learning_rate": 1.1639586250311849e-05,
      "loss": 0.4548,
      "step": 15156
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2935353108332021,
      "learning_rate": 1.1638607781678978e-05,
      "loss": 0.1946,
      "step": 15157
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.47577751750442276,
      "learning_rate": 1.1637629296924951e-05,
      "loss": 0.3468,
      "step": 15158
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.39785909339985476,
      "learning_rate": 1.1636650796059384e-05,
      "loss": 0.2599,
      "step": 15159
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.9887314978750794,
      "learning_rate": 1.1635672279091907e-05,
      "loss": 0.6091,
      "step": 15160
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.21344352313144244,
      "learning_rate": 1.1634693746032149e-05,
      "loss": 0.0703,
      "step": 15161
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.514170631818226,
      "learning_rate": 1.1633715196889735e-05,
      "loss": 0.3455,
      "step": 15162
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.270878081780795,
      "learning_rate": 1.1632736631674297e-05,
      "loss": 0.1646,
      "step": 15163
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3929015953583817,
      "learning_rate": 1.1631758050395456e-05,
      "loss": 0.2818,
      "step": 15164
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0030272762830745,
      "learning_rate": 1.1630779453062844e-05,
      "loss": 0.5179,
      "step": 15165
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.26300580861471007,
      "learning_rate": 1.1629800839686085e-05,
      "loss": 0.0714,
      "step": 15166
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.4044951788129841,
      "learning_rate": 1.162882221027481e-05,
      "loss": 0.3167,
      "step": 15167
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0022494498344168,
      "learning_rate": 1.1627843564838646e-05,
      "loss": 0.4931,
      "step": 15168
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3282388279985036,
      "learning_rate": 1.162686490338722e-05,
      "loss": 0.2932,
      "step": 15169
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5187620563074519,
      "learning_rate": 1.1625886225930165e-05,
      "loss": 0.1662,
      "step": 15170
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.34709712779782326,
      "learning_rate": 1.1624907532477105e-05,
      "loss": 0.2707,
      "step": 15171
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.19898169510575403,
      "learning_rate": 1.1623928823037668e-05,
      "loss": 0.1314,
      "step": 15172
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6420484093179962,
      "learning_rate": 1.1622950097621488e-05,
      "loss": 0.4078,
      "step": 15173
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.8547259996787432,
      "learning_rate": 1.1621971356238189e-05,
      "loss": 0.3264,
      "step": 15174
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.6848046201473855,
      "learning_rate": 1.16209925988974e-05,
      "loss": 0.4212,
      "step": 15175
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2563522003897284,
      "learning_rate": 1.1620013825608756e-05,
      "loss": 0.234,
      "step": 15176
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.46151647805152,
      "learning_rate": 1.1619035036381885e-05,
      "loss": 0.4199,
      "step": 15177
    },
    {
      "epoch": 0.46,
      "grad_norm": 1.0737796591598534,
      "learning_rate": 1.161805623122641e-05,
      "loss": 0.5888,
      "step": 15178
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.5358437223860972,
      "learning_rate": 1.1617077410151967e-05,
      "loss": 0.2105,
      "step": 15179
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3557922783082524,
      "learning_rate": 1.1616098573168185e-05,
      "loss": 0.2892,
      "step": 15180
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.3896679949329985,
      "learning_rate": 1.161511972028469e-05,
      "loss": 0.227,
      "step": 15181
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.31628200718795446,
      "learning_rate": 1.161414085151112e-05,
      "loss": 0.3021,
      "step": 15182
    },
    {
      "epoch": 0.46,
      "grad_norm": 0.2845649403387238,
      "learning_rate": 1.16131619668571e-05,
      "loss": 0.2212,
      "step": 15183
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6480046068080786,
      "learning_rate": 1.161218306633226e-05,
      "loss": 0.3717,
      "step": 15184
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3397560533037367,
      "learning_rate": 1.1611204149946234e-05,
      "loss": 0.2327,
      "step": 15185
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6210484517779413,
      "learning_rate": 1.1610225217708653e-05,
      "loss": 0.7209,
      "step": 15186
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42036427418014033,
      "learning_rate": 1.1609246269629144e-05,
      "loss": 0.2468,
      "step": 15187
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43999585373992484,
      "learning_rate": 1.1608267305717342e-05,
      "loss": 0.3294,
      "step": 15188
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.382681507532745,
      "learning_rate": 1.1607288325982876e-05,
      "loss": 0.1935,
      "step": 15189
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41609739863519496,
      "learning_rate": 1.160630933043538e-05,
      "loss": 0.2913,
      "step": 15190
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41739432404058613,
      "learning_rate": 1.1605330319084483e-05,
      "loss": 0.2937,
      "step": 15191
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.26445502857969033,
      "learning_rate": 1.160435129193982e-05,
      "loss": 0.1968,
      "step": 15192
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8999220712673748,
      "learning_rate": 1.1603372249011018e-05,
      "loss": 0.3496,
      "step": 15193
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24521139663893943,
      "learning_rate": 1.1602393190307712e-05,
      "loss": 0.2118,
      "step": 15194
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.063118080531709,
      "learning_rate": 1.1601414115839538e-05,
      "loss": 0.6292,
      "step": 15195
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2344577439475197,
      "learning_rate": 1.1600435025616124e-05,
      "loss": 0.265,
      "step": 15196
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9454657587315201,
      "learning_rate": 1.1599455919647103e-05,
      "loss": 0.5412,
      "step": 15197
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3033091521310811,
      "learning_rate": 1.1598476797942106e-05,
      "loss": 0.2051,
      "step": 15198
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4310302788008192,
      "learning_rate": 1.1597497660510772e-05,
      "loss": 0.3611,
      "step": 15199
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2854165940680437,
      "learning_rate": 1.1596518507362728e-05,
      "loss": 0.2398,
      "step": 15200
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.47896506316915827,
      "learning_rate": 1.1595539338507611e-05,
      "loss": 0.2854,
      "step": 15201
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.30783929718988146,
      "learning_rate": 1.1594560153955055e-05,
      "loss": 0.167,
      "step": 15202
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35729741043736335,
      "learning_rate": 1.1593580953714688e-05,
      "loss": 0.307,
      "step": 15203
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.26286554619326297,
      "learning_rate": 1.1592601737796147e-05,
      "loss": 0.1348,
      "step": 15204
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37935852962430544,
      "learning_rate": 1.1591622506209067e-05,
      "loss": 0.2525,
      "step": 15205
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.429028983155883,
      "learning_rate": 1.1590643258963081e-05,
      "loss": 0.3723,
      "step": 15206
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2768662433142448,
      "learning_rate": 1.1589663996067824e-05,
      "loss": 0.2008,
      "step": 15207
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6726369278387044,
      "learning_rate": 1.1588684717532932e-05,
      "loss": 0.4248,
      "step": 15208
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.026558399213301,
      "learning_rate": 1.1587705423368034e-05,
      "loss": 0.2331,
      "step": 15209
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44830305134807363,
      "learning_rate": 1.1586726113582768e-05,
      "loss": 0.2905,
      "step": 15210
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.25617576845061324,
      "learning_rate": 1.1585746788186765e-05,
      "loss": 0.2162,
      "step": 15211
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3633546744039771,
      "learning_rate": 1.1584767447189668e-05,
      "loss": 0.295,
      "step": 15212
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6057739897955284,
      "learning_rate": 1.1583788090601105e-05,
      "loss": 0.134,
      "step": 15213
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.025919801480743,
      "learning_rate": 1.1582808718430715e-05,
      "loss": 0.6509,
      "step": 15214
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7110044148013689,
      "learning_rate": 1.1581829330688135e-05,
      "loss": 0.3565,
      "step": 15215
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6500203980560111,
      "learning_rate": 1.158084992738299e-05,
      "loss": 0.3911,
      "step": 15216
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33343730518438613,
      "learning_rate": 1.157987050852493e-05,
      "loss": 0.2231,
      "step": 15217
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.29738978001449085,
      "learning_rate": 1.1578891074123581e-05,
      "loss": 0.2245,
      "step": 15218
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3967308992185896,
      "learning_rate": 1.157791162418858e-05,
      "loss": 0.8424,
      "step": 15219
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3816374308224231,
      "learning_rate": 1.157693215872957e-05,
      "loss": 0.1466,
      "step": 15220
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35821931121100853,
      "learning_rate": 1.1575952677756182e-05,
      "loss": 0.2905,
      "step": 15221
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2046047105805808,
      "learning_rate": 1.157497318127805e-05,
      "loss": 0.0861,
      "step": 15222
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3318031558757784,
      "learning_rate": 1.1573993669304815e-05,
      "loss": 0.3211,
      "step": 15223
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9539630338857099,
      "learning_rate": 1.157301414184611e-05,
      "loss": 0.3347,
      "step": 15224
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5502378729871248,
      "learning_rate": 1.157203459891158e-05,
      "loss": 0.3617,
      "step": 15225
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33148506655343524,
      "learning_rate": 1.157105504051085e-05,
      "loss": 0.243,
      "step": 15226
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0004395159759265,
      "learning_rate": 1.1570075466653567e-05,
      "loss": 0.4917,
      "step": 15227
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0831391527045064,
      "learning_rate": 1.1569095877349365e-05,
      "loss": 0.4976,
      "step": 15228
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33021487762523366,
      "learning_rate": 1.1568116272607882e-05,
      "loss": 0.3201,
      "step": 15229
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.280107797371832,
      "learning_rate": 1.1567136652438752e-05,
      "loss": 0.1889,
      "step": 15230
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19571839552783787,
      "learning_rate": 1.156615701685162e-05,
      "loss": 0.0882,
      "step": 15231
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1300566369482894,
      "learning_rate": 1.1565177365856118e-05,
      "loss": 0.5646,
      "step": 15232
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.049038436534038,
      "learning_rate": 1.1564197699461885e-05,
      "loss": 0.3646,
      "step": 15233
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39679178012829547,
      "learning_rate": 1.1563218017678564e-05,
      "loss": 0.2595,
      "step": 15234
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.256572641928898,
      "learning_rate": 1.1562238320515786e-05,
      "loss": 0.2436,
      "step": 15235
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8971969274809835,
      "learning_rate": 1.1561258607983193e-05,
      "loss": 0.8806,
      "step": 15236
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.318285444761307,
      "learning_rate": 1.1560278880090426e-05,
      "loss": 0.5111,
      "step": 15237
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.833146485023083,
      "learning_rate": 1.1559299136847123e-05,
      "loss": 0.532,
      "step": 15238
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.29767187301311404,
      "learning_rate": 1.155831937826292e-05,
      "loss": 0.2096,
      "step": 15239
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6954483142136857,
      "learning_rate": 1.1557339604347458e-05,
      "loss": 0.2297,
      "step": 15240
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3811541025094958,
      "learning_rate": 1.1556359815110379e-05,
      "loss": 0.2816,
      "step": 15241
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4528320517713913,
      "learning_rate": 1.1555380010561317e-05,
      "loss": 0.2847,
      "step": 15242
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3221283330778035,
      "learning_rate": 1.1554400190709914e-05,
      "loss": 0.132,
      "step": 15243
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28699813814996333,
      "learning_rate": 1.1553420355565813e-05,
      "loss": 0.2003,
      "step": 15244
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9594697019766829,
      "learning_rate": 1.1552440505138651e-05,
      "loss": 0.5355,
      "step": 15245
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37934575889105515,
      "learning_rate": 1.155146063943807e-05,
      "loss": 0.2706,
      "step": 15246
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.45288729024329094,
      "learning_rate": 1.1550480758473705e-05,
      "loss": 0.3244,
      "step": 15247
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3224311460596401,
      "learning_rate": 1.1549500862255203e-05,
      "loss": 0.1755,
      "step": 15248
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4781139150401698,
      "learning_rate": 1.15485209507922e-05,
      "loss": 0.3705,
      "step": 15249
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6358318606775092,
      "learning_rate": 1.154754102409434e-05,
      "loss": 0.3595,
      "step": 15250
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43498928262134195,
      "learning_rate": 1.1546561082171261e-05,
      "loss": 0.216,
      "step": 15251
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.16723285866201293,
      "learning_rate": 1.1545581125032608e-05,
      "loss": 0.0713,
      "step": 15252
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.30469509077993817,
      "learning_rate": 1.1544601152688015e-05,
      "loss": 0.3034,
      "step": 15253
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4112799385094239,
      "learning_rate": 1.1543621165147132e-05,
      "loss": 0.23,
      "step": 15254
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.5836746975607918,
      "learning_rate": 1.1542641162419592e-05,
      "loss": 0.8048,
      "step": 15255
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9310521738183607,
      "learning_rate": 1.1541661144515042e-05,
      "loss": 0.4944,
      "step": 15256
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28218058145780783,
      "learning_rate": 1.1540681111443121e-05,
      "loss": 0.1988,
      "step": 15257
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6038944608207042,
      "learning_rate": 1.1539701063213478e-05,
      "loss": 0.4285,
      "step": 15258
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28448035130588234,
      "learning_rate": 1.1538720999835743e-05,
      "loss": 0.2205,
      "step": 15259
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40138373106159264,
      "learning_rate": 1.1537740921319567e-05,
      "loss": 0.2826,
      "step": 15260
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.20660239256439084,
      "learning_rate": 1.1536760827674589e-05,
      "loss": 0.0907,
      "step": 15261
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39420774564211575,
      "learning_rate": 1.1535780718910451e-05,
      "loss": 0.3179,
      "step": 15262
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9524778320981785,
      "learning_rate": 1.15348005950368e-05,
      "loss": 0.2905,
      "step": 15263
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.515424725799311,
      "learning_rate": 1.1533820456063273e-05,
      "loss": 0.965,
      "step": 15264
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2660195459962951,
      "learning_rate": 1.153284030199952e-05,
      "loss": 0.2282,
      "step": 15265
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5003101445680069,
      "learning_rate": 1.1531860132855177e-05,
      "loss": 0.289,
      "step": 15266
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7051870418610565,
      "learning_rate": 1.153087994863989e-05,
      "loss": 0.3408,
      "step": 15267
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.753490622758025,
      "learning_rate": 1.1529899749363303e-05,
      "loss": 0.4652,
      "step": 15268
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.18277136216649753,
      "learning_rate": 1.1528919535035055e-05,
      "loss": 0.0714,
      "step": 15269
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2606036971612911,
      "learning_rate": 1.1527939305664799e-05,
      "loss": 0.0743,
      "step": 15270
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37079498778835457,
      "learning_rate": 1.1526959061262172e-05,
      "loss": 0.3055,
      "step": 15271
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2979353568008005,
      "learning_rate": 1.1525978801836817e-05,
      "loss": 0.2406,
      "step": 15272
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3708538326375646,
      "learning_rate": 1.1524998527398382e-05,
      "loss": 0.8281,
      "step": 15273
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6591675523750243,
      "learning_rate": 1.152401823795651e-05,
      "loss": 0.3418,
      "step": 15274
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6367658628927708,
      "learning_rate": 1.1523037933520841e-05,
      "loss": 0.3195,
      "step": 15275
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.32611508086227853,
      "learning_rate": 1.1522057614101028e-05,
      "loss": 0.268,
      "step": 15276
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.305741139394572,
      "learning_rate": 1.1521077279706711e-05,
      "loss": 0.284,
      "step": 15277
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.266114695305191,
      "learning_rate": 1.1520096930347537e-05,
      "loss": 0.2498,
      "step": 15278
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.31082814422578076,
      "learning_rate": 1.1519116566033147e-05,
      "loss": 0.1705,
      "step": 15279
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2699277939862616,
      "learning_rate": 1.1518136186773188e-05,
      "loss": 0.1837,
      "step": 15280
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.158027462657327,
      "learning_rate": 1.1517155792577306e-05,
      "loss": 0.5679,
      "step": 15281
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8360973576338597,
      "learning_rate": 1.1516175383455144e-05,
      "loss": 0.5015,
      "step": 15282
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2852313957426117,
      "learning_rate": 1.1515194959416353e-05,
      "loss": 0.2675,
      "step": 15283
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3686169373730448,
      "learning_rate": 1.1514214520470578e-05,
      "loss": 0.245,
      "step": 15284
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.378883488455702,
      "learning_rate": 1.1513234066627457e-05,
      "loss": 0.2921,
      "step": 15285
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.0990650800147015,
      "learning_rate": 1.151225359789664e-05,
      "loss": 0.8363,
      "step": 15286
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38419133274956013,
      "learning_rate": 1.151127311428778e-05,
      "loss": 0.1443,
      "step": 15287
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3056590856183133,
      "learning_rate": 1.1510292615810514e-05,
      "loss": 0.2427,
      "step": 15288
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2544870843137827,
      "learning_rate": 1.1509312102474494e-05,
      "loss": 0.2154,
      "step": 15289
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.6195753185720019,
      "learning_rate": 1.1508331574289366e-05,
      "loss": 0.771,
      "step": 15290
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7333723446549426,
      "learning_rate": 1.1507351031264772e-05,
      "loss": 0.3817,
      "step": 15291
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5829116219222039,
      "learning_rate": 1.1506370473410365e-05,
      "loss": 0.4162,
      "step": 15292
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3058474340577559,
      "learning_rate": 1.150538990073579e-05,
      "loss": 0.1902,
      "step": 15293
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5751065504708855,
      "learning_rate": 1.1504409313250693e-05,
      "loss": 0.3659,
      "step": 15294
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3077924266917655,
      "learning_rate": 1.1503428710964721e-05,
      "loss": 0.2471,
      "step": 15295
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9277392387641418,
      "learning_rate": 1.1502448093887524e-05,
      "loss": 0.3251,
      "step": 15296
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3509204538237223,
      "learning_rate": 1.150146746202875e-05,
      "loss": 0.205,
      "step": 15297
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3183370133578978,
      "learning_rate": 1.1500486815398042e-05,
      "loss": 0.2198,
      "step": 15298
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4304887613620842,
      "learning_rate": 1.149950615400505e-05,
      "loss": 0.2567,
      "step": 15299
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36069865283707203,
      "learning_rate": 1.1498525477859428e-05,
      "loss": 0.2631,
      "step": 15300
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7964789695303597,
      "learning_rate": 1.1497544786970815e-05,
      "loss": 0.5196,
      "step": 15301
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28013831066909184,
      "learning_rate": 1.1496564081348867e-05,
      "loss": 0.1223,
      "step": 15302
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37268740528210026,
      "learning_rate": 1.1495583361003229e-05,
      "loss": 0.2992,
      "step": 15303
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.833808781304206,
      "learning_rate": 1.1494602625943547e-05,
      "loss": 0.3244,
      "step": 15304
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4330710448542647,
      "learning_rate": 1.1493621876179473e-05,
      "loss": 0.6553,
      "step": 15305
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.22683697822631474,
      "learning_rate": 1.1492641111720657e-05,
      "loss": 0.1856,
      "step": 15306
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4031020904853219,
      "learning_rate": 1.1491660332576744e-05,
      "loss": 0.2803,
      "step": 15307
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2712881493005985,
      "learning_rate": 1.149067953875739e-05,
      "loss": 0.1686,
      "step": 15308
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8453390255335147,
      "learning_rate": 1.1489698730272238e-05,
      "loss": 0.3596,
      "step": 15309
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6685562173684452,
      "learning_rate": 1.148871790713094e-05,
      "loss": 0.4587,
      "step": 15310
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3035094044280071,
      "learning_rate": 1.1487737069343143e-05,
      "loss": 0.1813,
      "step": 15311
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.48227620064402155,
      "learning_rate": 1.14867562169185e-05,
      "loss": 0.3655,
      "step": 15312
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2616532206048681,
      "learning_rate": 1.1485775349866663e-05,
      "loss": 0.2154,
      "step": 15313
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1060535376727698,
      "learning_rate": 1.1484794468197276e-05,
      "loss": 0.5516,
      "step": 15314
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17836294073498687,
      "learning_rate": 1.1483813571919996e-05,
      "loss": 0.0678,
      "step": 15315
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3295973167026047,
      "learning_rate": 1.1482832661044467e-05,
      "loss": 0.2636,
      "step": 15316
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41847991581488075,
      "learning_rate": 1.1481851735580342e-05,
      "loss": 0.1799,
      "step": 15317
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5569484320289778,
      "learning_rate": 1.1480870795537273e-05,
      "loss": 0.3737,
      "step": 15318
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4510583120174164,
      "learning_rate": 1.1479889840924911e-05,
      "loss": 0.2894,
      "step": 15319
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6516998806006528,
      "learning_rate": 1.1478908871752906e-05,
      "loss": 0.4085,
      "step": 15320
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.32236483423152573,
      "learning_rate": 1.1477927888030908e-05,
      "loss": 0.2193,
      "step": 15321
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0265971914521783,
      "learning_rate": 1.147694688976857e-05,
      "loss": 0.3929,
      "step": 15322
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2447733156831895,
      "learning_rate": 1.1475965876975539e-05,
      "loss": 0.5349,
      "step": 15323
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.210552747573189,
      "learning_rate": 1.1474984849661472e-05,
      "loss": 0.1853,
      "step": 15324
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8375352683121885,
      "learning_rate": 1.1474003807836019e-05,
      "loss": 0.461,
      "step": 15325
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3198639517906748,
      "learning_rate": 1.1473022751508834e-05,
      "loss": 0.2664,
      "step": 15326
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3698691338673669,
      "learning_rate": 1.1472041680689561e-05,
      "loss": 0.3014,
      "step": 15327
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.920880886533291,
      "learning_rate": 1.1471060595387861e-05,
      "loss": 0.4689,
      "step": 15328
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8106165137386144,
      "learning_rate": 1.1470079495613383e-05,
      "loss": 0.4589,
      "step": 15329
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28229313734940686,
      "learning_rate": 1.1469098381375781e-05,
      "loss": 0.2201,
      "step": 15330
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.30985156449162615,
      "learning_rate": 1.1468117252684701e-05,
      "loss": 0.3207,
      "step": 15331
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.203615931612182,
      "learning_rate": 1.1467136109549804e-05,
      "loss": 0.2715,
      "step": 15332
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.1131432062441187,
      "learning_rate": 1.1466154951980737e-05,
      "loss": 0.564,
      "step": 15333
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.277223673968392,
      "learning_rate": 1.1465173779987158e-05,
      "loss": 0.212,
      "step": 15334
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3196797167644499,
      "learning_rate": 1.1464192593578715e-05,
      "loss": 0.2835,
      "step": 15335
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.33545916283714994,
      "learning_rate": 1.1463211392765063e-05,
      "loss": 0.3193,
      "step": 15336
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.15040534626323382,
      "learning_rate": 1.1462230177555857e-05,
      "loss": 0.0854,
      "step": 15337
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8456445781704499,
      "learning_rate": 1.1461248947960747e-05,
      "loss": 0.4973,
      "step": 15338
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2618347671993929,
      "learning_rate": 1.1460267703989393e-05,
      "loss": 0.2062,
      "step": 15339
    },
    {
      "epoch": 0.47,
      "grad_norm": 2.1356944203261845,
      "learning_rate": 1.1459286445651446e-05,
      "loss": 0.8311,
      "step": 15340
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.234600528179566,
      "learning_rate": 1.1458305172956554e-05,
      "loss": 0.2575,
      "step": 15341
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3169769044759863,
      "learning_rate": 1.1457323885914378e-05,
      "loss": 0.33,
      "step": 15342
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2906107134086857,
      "learning_rate": 1.145634258453457e-05,
      "loss": 0.2034,
      "step": 15343
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.814044459012045,
      "learning_rate": 1.145536126882678e-05,
      "loss": 0.4973,
      "step": 15344
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6012767501721429,
      "learning_rate": 1.1454379938800676e-05,
      "loss": 0.3417,
      "step": 15345
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44499673109271815,
      "learning_rate": 1.1453398594465897e-05,
      "loss": 0.2926,
      "step": 15346
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37464586981639586,
      "learning_rate": 1.1452417235832108e-05,
      "loss": 0.161,
      "step": 15347
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3421392986882932,
      "learning_rate": 1.145143586290896e-05,
      "loss": 0.218,
      "step": 15348
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.281919979154083,
      "learning_rate": 1.1450454475706106e-05,
      "loss": 0.276,
      "step": 15349
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.084193563751699,
      "learning_rate": 1.1449473074233202e-05,
      "loss": 0.3275,
      "step": 15350
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5734303953689858,
      "learning_rate": 1.1448491658499908e-05,
      "loss": 0.4252,
      "step": 15351
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.26991698303569506,
      "learning_rate": 1.1447510228515875e-05,
      "loss": 0.068,
      "step": 15352
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43154308034217914,
      "learning_rate": 1.1446528784290762e-05,
      "loss": 0.3405,
      "step": 15353
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2982179079123793,
      "learning_rate": 1.1445547325834221e-05,
      "loss": 0.2428,
      "step": 15354
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.45445617492853685,
      "learning_rate": 1.1444565853155911e-05,
      "loss": 0.3007,
      "step": 15355
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9931188638164123,
      "learning_rate": 1.1443584366265487e-05,
      "loss": 0.4837,
      "step": 15356
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3921163369612088,
      "learning_rate": 1.1442602865172603e-05,
      "loss": 0.2662,
      "step": 15357
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.27381395867610947,
      "learning_rate": 1.144162134988692e-05,
      "loss": 0.1565,
      "step": 15358
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44201567049071866,
      "learning_rate": 1.1440639820418091e-05,
      "loss": 0.3606,
      "step": 15359
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40989975709665905,
      "learning_rate": 1.1439658276775774e-05,
      "loss": 0.2683,
      "step": 15360
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3575801603564642,
      "learning_rate": 1.1438676718969623e-05,
      "loss": 0.2196,
      "step": 15361
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4954411650982365,
      "learning_rate": 1.1437695147009297e-05,
      "loss": 0.3479,
      "step": 15362
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2485787920618656,
      "learning_rate": 1.1436713560904456e-05,
      "loss": 0.2619,
      "step": 15363
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3438508724627343,
      "learning_rate": 1.1435731960664751e-05,
      "loss": 0.7354,
      "step": 15364
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3535166154186721,
      "learning_rate": 1.1434750346299843e-05,
      "loss": 0.2461,
      "step": 15365
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34888593064084117,
      "learning_rate": 1.143376871781939e-05,
      "loss": 0.3059,
      "step": 15366
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2481094081516471,
      "learning_rate": 1.1432787075233049e-05,
      "loss": 0.1689,
      "step": 15367
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39390592887339265,
      "learning_rate": 1.1431805418550479e-05,
      "loss": 0.2714,
      "step": 15368
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5107768157207587,
      "learning_rate": 1.1430823747781332e-05,
      "loss": 0.3414,
      "step": 15369
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5624147814887458,
      "learning_rate": 1.1429842062935273e-05,
      "loss": 0.2965,
      "step": 15370
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3081445562184631,
      "learning_rate": 1.1428860364021956e-05,
      "loss": 0.2072,
      "step": 15371
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4505681946714034,
      "learning_rate": 1.1427878651051042e-05,
      "loss": 0.3401,
      "step": 15372
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37464154192662114,
      "learning_rate": 1.1426896924032187e-05,
      "loss": 0.2583,
      "step": 15373
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2789703607109685,
      "learning_rate": 1.1425915182975052e-05,
      "loss": 0.1637,
      "step": 15374
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6269733760897854,
      "learning_rate": 1.1424933427889293e-05,
      "loss": 0.3482,
      "step": 15375
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.21382427505093035,
      "learning_rate": 1.142395165878457e-05,
      "loss": 0.1608,
      "step": 15376
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4744683131099725,
      "learning_rate": 1.1422969875670539e-05,
      "loss": 0.3497,
      "step": 15377
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35355345187815873,
      "learning_rate": 1.1421988078556867e-05,
      "loss": 0.2566,
      "step": 15378
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9790323848451639,
      "learning_rate": 1.1421006267453207e-05,
      "loss": 0.4664,
      "step": 15379
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3284022249578694,
      "learning_rate": 1.142002444236922e-05,
      "loss": 0.2134,
      "step": 15380
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3737827612218843,
      "learning_rate": 1.1419042603314565e-05,
      "loss": 0.587,
      "step": 15381
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7596888374370829,
      "learning_rate": 1.14180607502989e-05,
      "loss": 0.6171,
      "step": 15382
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4984142064805285,
      "learning_rate": 1.1417078883331886e-05,
      "loss": 0.3373,
      "step": 15383
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2622897356874186,
      "learning_rate": 1.141609700242319e-05,
      "loss": 0.1998,
      "step": 15384
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3844861136234201,
      "learning_rate": 1.1415115107582462e-05,
      "loss": 0.2522,
      "step": 15385
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5746283082858313,
      "learning_rate": 1.1414133198819366e-05,
      "loss": 0.3332,
      "step": 15386
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.19010831481360413,
      "learning_rate": 1.141315127614356e-05,
      "loss": 0.0687,
      "step": 15387
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7122769637096691,
      "learning_rate": 1.141216933956471e-05,
      "loss": 0.422,
      "step": 15388
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.30844580894269397,
      "learning_rate": 1.1411187389092471e-05,
      "loss": 0.2129,
      "step": 15389
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.32616035235788904,
      "learning_rate": 1.1410205424736508e-05,
      "loss": 0.2939,
      "step": 15390
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0093087086701316,
      "learning_rate": 1.1409223446506481e-05,
      "loss": 0.4374,
      "step": 15391
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.4095680210199597,
      "learning_rate": 1.1408241454412049e-05,
      "loss": 0.8426,
      "step": 15392
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2750561291215378,
      "learning_rate": 1.140725944846287e-05,
      "loss": 0.201,
      "step": 15393
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7249018864628419,
      "learning_rate": 1.1406277428668616e-05,
      "loss": 0.5007,
      "step": 15394
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6184753570347539,
      "learning_rate": 1.1405295395038935e-05,
      "loss": 0.3343,
      "step": 15395
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2530662639657925,
      "learning_rate": 1.14043133475835e-05,
      "loss": 0.2405,
      "step": 15396
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.17737472471468513,
      "learning_rate": 1.1403331286311969e-05,
      "loss": 0.0704,
      "step": 15397
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.40376267964551477,
      "learning_rate": 1.1402349211234003e-05,
      "loss": 0.3143,
      "step": 15398
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3160949838578142,
      "learning_rate": 1.1401367122359261e-05,
      "loss": 0.0836,
      "step": 15399
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.374192389741502,
      "learning_rate": 1.1400385019697408e-05,
      "loss": 0.8137,
      "step": 15400
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3108967653219553,
      "learning_rate": 1.139940290325811e-05,
      "loss": 0.28,
      "step": 15401
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.32541421296122913,
      "learning_rate": 1.1398420773051022e-05,
      "loss": 0.0734,
      "step": 15402
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3743555228155434,
      "learning_rate": 1.1397438629085813e-05,
      "loss": 0.3137,
      "step": 15403
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7687678556349316,
      "learning_rate": 1.1396456471372144e-05,
      "loss": 0.3451,
      "step": 15404
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28928828114805094,
      "learning_rate": 1.1395474299919672e-05,
      "loss": 0.1584,
      "step": 15405
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35847874563424026,
      "learning_rate": 1.1394492114738069e-05,
      "loss": 0.0661,
      "step": 15406
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3576592391964655,
      "learning_rate": 1.1393509915836993e-05,
      "loss": 0.2924,
      "step": 15407
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.31041489744842193,
      "learning_rate": 1.1392527703226106e-05,
      "loss": 0.2541,
      "step": 15408
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2566705187340592,
      "learning_rate": 1.1391545476915076e-05,
      "loss": 0.8364,
      "step": 15409
    },
    {
      "epoch": 0.47,
      "grad_norm": 3.862837193100732,
      "learning_rate": 1.1390563236913563e-05,
      "loss": 0.3633,
      "step": 15410
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5516506568006149,
      "learning_rate": 1.138958098323123e-05,
      "loss": 0.4066,
      "step": 15411
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3306581689515956,
      "learning_rate": 1.1388598715877741e-05,
      "loss": 0.2656,
      "step": 15412
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4560832823033577,
      "learning_rate": 1.1387616434862762e-05,
      "loss": 0.3963,
      "step": 15413
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37332689904679933,
      "learning_rate": 1.1386634140195959e-05,
      "loss": 0.2542,
      "step": 15414
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24355873712468165,
      "learning_rate": 1.1385651831886988e-05,
      "loss": 0.0717,
      "step": 15415
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.35039359140512627,
      "learning_rate": 1.1384669509945523e-05,
      "loss": 0.3089,
      "step": 15416
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2474270418969453,
      "learning_rate": 1.1383687174381221e-05,
      "loss": 0.3235,
      "step": 15417
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.435289901299793,
      "learning_rate": 1.1382704825203749e-05,
      "loss": 0.8879,
      "step": 15418
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3478471922157705,
      "learning_rate": 1.1381722462422772e-05,
      "loss": 0.2633,
      "step": 15419
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6047316501081997,
      "learning_rate": 1.1380740086047958e-05,
      "loss": 0.337,
      "step": 15420
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3456391392615422,
      "learning_rate": 1.1379757696088964e-05,
      "loss": 0.255,
      "step": 15421
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.156018369168329,
      "learning_rate": 1.1378775292555465e-05,
      "loss": 0.6605,
      "step": 15422
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.1712311377725489,
      "learning_rate": 1.137779287545712e-05,
      "loss": 0.0706,
      "step": 15423
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7921201127513054,
      "learning_rate": 1.1376810444803592e-05,
      "loss": 0.4358,
      "step": 15424
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28346712631330057,
      "learning_rate": 1.137582800060455e-05,
      "loss": 0.2022,
      "step": 15425
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.34007025327611823,
      "learning_rate": 1.137484554286966e-05,
      "loss": 0.3004,
      "step": 15426
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.63947790203738,
      "learning_rate": 1.1373863071608589e-05,
      "loss": 0.3749,
      "step": 15427
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5140365576838158,
      "learning_rate": 1.1372880586830998e-05,
      "loss": 0.362,
      "step": 15428
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9638022638393763,
      "learning_rate": 1.1371898088546561e-05,
      "loss": 0.4843,
      "step": 15429
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2946010019419113,
      "learning_rate": 1.1370915576764936e-05,
      "loss": 0.2224,
      "step": 15430
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4419155195591363,
      "learning_rate": 1.1369933051495791e-05,
      "loss": 0.3589,
      "step": 15431
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.24019411015461697,
      "learning_rate": 1.1368950512748795e-05,
      "loss": 0.1582,
      "step": 15432
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4972328551668951,
      "learning_rate": 1.1367967960533614e-05,
      "loss": 0.2894,
      "step": 15433
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2664742062798144,
      "learning_rate": 1.1366985394859913e-05,
      "loss": 0.1564,
      "step": 15434
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5334025660061061,
      "learning_rate": 1.1366002815737363e-05,
      "loss": 0.3772,
      "step": 15435
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7410131127603785,
      "learning_rate": 1.1365020223175626e-05,
      "loss": 0.3626,
      "step": 15436
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.45712811111805685,
      "learning_rate": 1.136403761718437e-05,
      "loss": 0.3389,
      "step": 15437
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2666021210933947,
      "learning_rate": 1.1363054997773262e-05,
      "loss": 0.2173,
      "step": 15438
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3743179728320615,
      "learning_rate": 1.1362072364951971e-05,
      "loss": 0.3359,
      "step": 15439
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8567372582057979,
      "learning_rate": 1.1361089718730167e-05,
      "loss": 0.3493,
      "step": 15440
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2360724747885299,
      "learning_rate": 1.1360107059117513e-05,
      "loss": 0.272,
      "step": 15441
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.22456323537782935,
      "learning_rate": 1.1359124386123677e-05,
      "loss": 0.177,
      "step": 15442
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2696888432465001,
      "learning_rate": 1.1358141699758329e-05,
      "loss": 0.1853,
      "step": 15443
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.44050723179912377,
      "learning_rate": 1.1357159000031135e-05,
      "loss": 0.3135,
      "step": 15444
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6091900487379756,
      "learning_rate": 1.1356176286951765e-05,
      "loss": 0.3502,
      "step": 15445
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6868424661754589,
      "learning_rate": 1.1355193560529887e-05,
      "loss": 0.4768,
      "step": 15446
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3846754260499856,
      "learning_rate": 1.135421082077517e-05,
      "loss": 0.1525,
      "step": 15447
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3401276867116862,
      "learning_rate": 1.1353228067697279e-05,
      "loss": 0.2811,
      "step": 15448
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36866411596645643,
      "learning_rate": 1.1352245301305887e-05,
      "loss": 0.211,
      "step": 15449
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4427567170350201,
      "learning_rate": 1.1351262521610658e-05,
      "loss": 0.3562,
      "step": 15450
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.23115174057744087,
      "learning_rate": 1.1350279728621265e-05,
      "loss": 0.1049,
      "step": 15451
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36743507274210624,
      "learning_rate": 1.1349296922347372e-05,
      "loss": 0.2549,
      "step": 15452
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4526676758905256,
      "learning_rate": 1.1348314102798659e-05,
      "loss": 0.2715,
      "step": 15453
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6554824462214773,
      "learning_rate": 1.1347331269984786e-05,
      "loss": 0.381,
      "step": 15454
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3459435628237493,
      "learning_rate": 1.1346348423915422e-05,
      "loss": 0.2973,
      "step": 15455
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4264772009935151,
      "learning_rate": 1.1345365564600242e-05,
      "loss": 0.1491,
      "step": 15456
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37879303728626595,
      "learning_rate": 1.1344382692048912e-05,
      "loss": 0.3135,
      "step": 15457
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9576423652863211,
      "learning_rate": 1.13433998062711e-05,
      "loss": 0.3496,
      "step": 15458
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9144545565063174,
      "learning_rate": 1.1342416907276484e-05,
      "loss": 0.5941,
      "step": 15459
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.28102070272573015,
      "learning_rate": 1.1341433995074726e-05,
      "loss": 0.1706,
      "step": 15460
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.38104319646428225,
      "learning_rate": 1.1340451069675497e-05,
      "loss": 0.2722,
      "step": 15461
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3110152720349212,
      "learning_rate": 1.1339468131088474e-05,
      "loss": 0.2092,
      "step": 15462
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.6858459155415821,
      "learning_rate": 1.1338485179323318e-05,
      "loss": 0.4564,
      "step": 15463
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9107618673262812,
      "learning_rate": 1.1337502214389704e-05,
      "loss": 0.4725,
      "step": 15464
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.3451882135610138,
      "learning_rate": 1.1336519236297306e-05,
      "loss": 0.9153,
      "step": 15465
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3662333443771821,
      "learning_rate": 1.1335536245055792e-05,
      "loss": 0.1865,
      "step": 15466
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.31727474045268944,
      "learning_rate": 1.1334553240674833e-05,
      "loss": 0.2191,
      "step": 15467
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.8210264174312565,
      "learning_rate": 1.1333570223164099e-05,
      "loss": 0.8234,
      "step": 15468
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2915730515576717,
      "learning_rate": 1.1332587192533264e-05,
      "loss": 0.176,
      "step": 15469
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37573887270508394,
      "learning_rate": 1.1331604148791996e-05,
      "loss": 0.2748,
      "step": 15470
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43681104071437554,
      "learning_rate": 1.1330621091949967e-05,
      "loss": 0.3012,
      "step": 15471
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.461431478205465,
      "learning_rate": 1.1329638022016853e-05,
      "loss": 0.2355,
      "step": 15472
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2953321896162672,
      "learning_rate": 1.1328654939002322e-05,
      "loss": 0.2671,
      "step": 15473
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8333431331667421,
      "learning_rate": 1.1327671842916046e-05,
      "loss": 0.4447,
      "step": 15474
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2791778020517464,
      "learning_rate": 1.1326688733767696e-05,
      "loss": 0.2113,
      "step": 15475
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2103361858514414,
      "learning_rate": 1.1325705611566949e-05,
      "loss": 0.6661,
      "step": 15476
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.150562849036767,
      "learning_rate": 1.1324722476323469e-05,
      "loss": 0.2636,
      "step": 15477
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39400973183715643,
      "learning_rate": 1.1323739328046934e-05,
      "loss": 0.3542,
      "step": 15478
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2915403503688982,
      "learning_rate": 1.1322756166747024e-05,
      "loss": 0.1887,
      "step": 15479
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2913960241536286,
      "learning_rate": 1.1321772992433395e-05,
      "loss": 0.2346,
      "step": 15480
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.42994412893444234,
      "learning_rate": 1.1320789805115731e-05,
      "loss": 0.2994,
      "step": 15481
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.0255475619836834,
      "learning_rate": 1.1319806604803704e-05,
      "loss": 0.6321,
      "step": 15482
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.39518155976833264,
      "learning_rate": 1.1318823391506984e-05,
      "loss": 0.2234,
      "step": 15483
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.27904726715286626,
      "learning_rate": 1.1317840165235243e-05,
      "loss": 0.2016,
      "step": 15484
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3400709551044767,
      "learning_rate": 1.1316856925998161e-05,
      "loss": 0.3069,
      "step": 15485
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.8195096776689119,
      "learning_rate": 1.1315873673805406e-05,
      "loss": 0.3122,
      "step": 15486
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.5942244125590463,
      "learning_rate": 1.1314890408666653e-05,
      "loss": 0.4567,
      "step": 15487
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.37455225967720696,
      "learning_rate": 1.1313907130591575e-05,
      "loss": 0.2022,
      "step": 15488
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.36302906562719994,
      "learning_rate": 1.1312923839589844e-05,
      "loss": 0.3276,
      "step": 15489
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4008481423527884,
      "learning_rate": 1.1311940535671137e-05,
      "loss": 0.2617,
      "step": 15490
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4750740011793473,
      "learning_rate": 1.131095721884513e-05,
      "loss": 0.3258,
      "step": 15491
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2543796316285515,
      "learning_rate": 1.1309973889121496e-05,
      "loss": 0.1535,
      "step": 15492
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2527795960610012,
      "learning_rate": 1.1308990546509904e-05,
      "loss": 0.1769,
      "step": 15493
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.4442721712533362,
      "learning_rate": 1.1308007191020031e-05,
      "loss": 0.2734,
      "step": 15494
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7068384808865402,
      "learning_rate": 1.1307023822661558e-05,
      "loss": 0.3181,
      "step": 15495
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.41480363271316517,
      "learning_rate": 1.130604044144415e-05,
      "loss": 0.3292,
      "step": 15496
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3571572047950481,
      "learning_rate": 1.1305057047377487e-05,
      "loss": 0.2187,
      "step": 15497
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3751961629064749,
      "learning_rate": 1.1304073640471244e-05,
      "loss": 0.3253,
      "step": 15498
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9214857629574024,
      "learning_rate": 1.1303090220735098e-05,
      "loss": 0.0588,
      "step": 15499
    },
    {
      "epoch": 0.47,
      "grad_norm": 1.2420322740032395,
      "learning_rate": 1.1302106788178719e-05,
      "loss": 0.853,
      "step": 15500
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.183693501734471,
      "learning_rate": 1.1301123342811788e-05,
      "loss": 0.0868,
      "step": 15501
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3609132987255939,
      "learning_rate": 1.1300139884643972e-05,
      "loss": 0.2907,
      "step": 15502
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.31851347620793174,
      "learning_rate": 1.1299156413684955e-05,
      "loss": 0.24,
      "step": 15503
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.9030927105453906,
      "learning_rate": 1.1298172929944411e-05,
      "loss": 0.4809,
      "step": 15504
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.7717498265773243,
      "learning_rate": 1.1297189433432015e-05,
      "loss": 0.3304,
      "step": 15505
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3920734877773623,
      "learning_rate": 1.1296205924157439e-05,
      "loss": 0.1278,
      "step": 15506
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.3530057849489297,
      "learning_rate": 1.1295222402130364e-05,
      "loss": 0.2855,
      "step": 15507
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.45789607104702507,
      "learning_rate": 1.1294238867360466e-05,
      "loss": 0.2818,
      "step": 15508
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.43371381954619354,
      "learning_rate": 1.129325531985742e-05,
      "loss": 0.3609,
      "step": 15509
    },
    {
      "epoch": 0.47,
      "grad_norm": 0.2344142564550371,
      "learning_rate": 1.1292271759630901e-05,
      "loss": 0.1038,
      "step": 15510
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6227827078601601,
      "learning_rate": 1.129128818669059e-05,
      "loss": 0.3219,
      "step": 15511
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2186473656384753,
      "learning_rate": 1.1290304601046158e-05,
      "loss": 0.1685,
      "step": 15512
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9732349218212679,
      "learning_rate": 1.1289321002707285e-05,
      "loss": 0.4691,
      "step": 15513
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33479751460293383,
      "learning_rate": 1.1288337391683649e-05,
      "loss": 0.2664,
      "step": 15514
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.952942958051848,
      "learning_rate": 1.1287353767984925e-05,
      "loss": 0.4371,
      "step": 15515
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2643732090525035,
      "learning_rate": 1.1286370131620792e-05,
      "loss": 0.1962,
      "step": 15516
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1505911877601123,
      "learning_rate": 1.1285386482600927e-05,
      "loss": 0.6612,
      "step": 15517
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8479522998085782,
      "learning_rate": 1.1284402820935005e-05,
      "loss": 0.4886,
      "step": 15518
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2520197501554168,
      "learning_rate": 1.1283419146632706e-05,
      "loss": 0.1949,
      "step": 15519
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3939997247213615,
      "learning_rate": 1.1282435459703706e-05,
      "loss": 0.2831,
      "step": 15520
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21563377969647396,
      "learning_rate": 1.1281451760157687e-05,
      "loss": 0.2087,
      "step": 15521
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8129257039288313,
      "learning_rate": 1.128046804800432e-05,
      "loss": 0.5972,
      "step": 15522
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6156694331484818,
      "learning_rate": 1.1279484323253291e-05,
      "loss": 0.3638,
      "step": 15523
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7358730022024638,
      "learning_rate": 1.1278500585914274e-05,
      "loss": 0.5168,
      "step": 15524
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.9401358101783337,
      "learning_rate": 1.1277516835996945e-05,
      "loss": 0.2547,
      "step": 15525
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5376640386344274,
      "learning_rate": 1.1276533073510984e-05,
      "loss": 0.8985,
      "step": 15526
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3135576918525197,
      "learning_rate": 1.1275549298466074e-05,
      "loss": 0.2745,
      "step": 15527
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.5324128678607987,
      "learning_rate": 1.127456551087189e-05,
      "loss": 0.794,
      "step": 15528
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3089446841908918,
      "learning_rate": 1.127358171073811e-05,
      "loss": 0.2174,
      "step": 15529
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.47917198476933975,
      "learning_rate": 1.1272597898074414e-05,
      "loss": 0.3142,
      "step": 15530
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37257962566078306,
      "learning_rate": 1.1271614072890482e-05,
      "loss": 0.2355,
      "step": 15531
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2198925111982337,
      "learning_rate": 1.127063023519599e-05,
      "loss": 0.1977,
      "step": 15532
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6286448497884737,
      "learning_rate": 1.126964638500062e-05,
      "loss": 0.3162,
      "step": 15533
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3279097047865881,
      "learning_rate": 1.1268662522314053e-05,
      "loss": 0.2257,
      "step": 15534
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.155164650932588,
      "learning_rate": 1.1267678647145966e-05,
      "loss": 0.7026,
      "step": 15535
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7839631022186359,
      "learning_rate": 1.126669475950604e-05,
      "loss": 0.4431,
      "step": 15536
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3716194687946335,
      "learning_rate": 1.1265710859403955e-05,
      "loss": 0.3258,
      "step": 15537
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2880155970283989,
      "learning_rate": 1.1264726946849387e-05,
      "loss": 0.0722,
      "step": 15538
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2982970149633607,
      "learning_rate": 1.126374302185202e-05,
      "loss": 0.2799,
      "step": 15539
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5419247881051423,
      "learning_rate": 1.1262759084421532e-05,
      "loss": 0.326,
      "step": 15540
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32635664616483395,
      "learning_rate": 1.126177513456761e-05,
      "loss": 0.197,
      "step": 15541
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8948273031705124,
      "learning_rate": 1.1260791172299925e-05,
      "loss": 0.3216,
      "step": 15542
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.388951780267079,
      "learning_rate": 1.1259807197628161e-05,
      "loss": 0.2941,
      "step": 15543
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26966959915205985,
      "learning_rate": 1.1258823210562002e-05,
      "loss": 0.245,
      "step": 15544
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8619781986512492,
      "learning_rate": 1.1257839211111123e-05,
      "loss": 0.4636,
      "step": 15545
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.807176225729465,
      "learning_rate": 1.125685519928521e-05,
      "loss": 0.5499,
      "step": 15546
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38394700442206875,
      "learning_rate": 1.1255871175093942e-05,
      "loss": 0.2183,
      "step": 15547
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4371672262302863,
      "learning_rate": 1.1254887138547e-05,
      "loss": 0.3472,
      "step": 15548
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.155026179698651,
      "learning_rate": 1.1253903089654066e-05,
      "loss": 0.2891,
      "step": 15549
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27556644631081717,
      "learning_rate": 1.125291902842482e-05,
      "loss": 0.2595,
      "step": 15550
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.22060655400714022,
      "learning_rate": 1.1251934954868943e-05,
      "loss": 0.0736,
      "step": 15551
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3289559453625627,
      "learning_rate": 1.125095086899612e-05,
      "loss": 0.2782,
      "step": 15552
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8538271451554458,
      "learning_rate": 1.1249966770816028e-05,
      "loss": 0.3284,
      "step": 15553
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3732407121808208,
      "learning_rate": 1.1248982660338355e-05,
      "loss": 0.9329,
      "step": 15554
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.46694882883359856,
      "learning_rate": 1.124799853757278e-05,
      "loss": 0.2949,
      "step": 15555
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3425359451783914,
      "learning_rate": 1.1247014402528982e-05,
      "loss": 0.27,
      "step": 15556
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3232701109235427,
      "learning_rate": 1.1246030255216647e-05,
      "loss": 0.2147,
      "step": 15557
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8358344597540812,
      "learning_rate": 1.1245046095645455e-05,
      "loss": 0.3279,
      "step": 15558
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.304086311278886,
      "learning_rate": 1.124406192382509e-05,
      "loss": 0.1762,
      "step": 15559
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6718977712138151,
      "learning_rate": 1.1243077739765238e-05,
      "loss": 0.2122,
      "step": 15560
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0675156294266475,
      "learning_rate": 1.1242093543475576e-05,
      "loss": 0.4843,
      "step": 15561
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23674296217277918,
      "learning_rate": 1.1241109334965787e-05,
      "loss": 0.2288,
      "step": 15562
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3524809576421033,
      "learning_rate": 1.1240125114245559e-05,
      "loss": 0.9457,
      "step": 15563
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6138700076821709,
      "learning_rate": 1.1239140881324571e-05,
      "loss": 0.3333,
      "step": 15564
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9366934512410684,
      "learning_rate": 1.1238156636212505e-05,
      "loss": 0.504,
      "step": 15565
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27247598129281836,
      "learning_rate": 1.1237172378919049e-05,
      "loss": 0.1967,
      "step": 15566
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.296831624123601,
      "learning_rate": 1.1236188109453884e-05,
      "loss": 0.6593,
      "step": 15567
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.22134556472881786,
      "learning_rate": 1.1235203827826696e-05,
      "loss": 0.1976,
      "step": 15568
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4853493481518956,
      "learning_rate": 1.1234219534047162e-05,
      "loss": 0.2859,
      "step": 15569
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3272387837244998,
      "learning_rate": 1.1233235228124972e-05,
      "loss": 0.1899,
      "step": 15570
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4268346847955159,
      "learning_rate": 1.1232250910069806e-05,
      "loss": 0.2763,
      "step": 15571
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8197184616484443,
      "learning_rate": 1.1231266579891352e-05,
      "loss": 0.4643,
      "step": 15572
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.384281767827988,
      "learning_rate": 1.1230282237599293e-05,
      "loss": 0.2798,
      "step": 15573
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.37755042577460884,
      "learning_rate": 1.1229297883203312e-05,
      "loss": 0.3069,
      "step": 15574
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29190572893179334,
      "learning_rate": 1.1228313516713093e-05,
      "loss": 0.2058,
      "step": 15575
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2395435647214148,
      "learning_rate": 1.1227329138138324e-05,
      "loss": 0.7586,
      "step": 15576
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26718153976625264,
      "learning_rate": 1.1226344747488685e-05,
      "loss": 0.1114,
      "step": 15577
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4807387809677065,
      "learning_rate": 1.1225360344773861e-05,
      "loss": 0.2763,
      "step": 15578
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.25155467198902837,
      "learning_rate": 1.122437593000354e-05,
      "loss": 0.1756,
      "step": 15579
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.707635470252358,
      "learning_rate": 1.1223391503187409e-05,
      "loss": 0.9101,
      "step": 15580
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4244465446612333,
      "learning_rate": 1.1222407064335146e-05,
      "loss": 0.263,
      "step": 15581
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9761668254801342,
      "learning_rate": 1.122142261345644e-05,
      "loss": 0.52,
      "step": 15582
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29751551933365394,
      "learning_rate": 1.122043815056098e-05,
      "loss": 0.162,
      "step": 15583
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.602898154607855,
      "learning_rate": 1.1219453675658444e-05,
      "loss": 0.384,
      "step": 15584
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4449294664675409,
      "learning_rate": 1.1218469188758522e-05,
      "loss": 0.2257,
      "step": 15585
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30648362491246456,
      "learning_rate": 1.1217484689870902e-05,
      "loss": 0.2535,
      "step": 15586
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32502536559865797,
      "learning_rate": 1.1216500179005267e-05,
      "loss": 0.196,
      "step": 15587
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28278589543755456,
      "learning_rate": 1.1215515656171299e-05,
      "loss": 0.0734,
      "step": 15588
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.42798217345051914,
      "learning_rate": 1.1214531121378692e-05,
      "loss": 0.3565,
      "step": 15589
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.24231189411502,
      "learning_rate": 1.1213546574637126e-05,
      "loss": 0.374,
      "step": 15590
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.324572208302887,
      "learning_rate": 1.121256201595629e-05,
      "loss": 0.324,
      "step": 15591
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0868555385953047,
      "learning_rate": 1.121157744534587e-05,
      "loss": 0.5008,
      "step": 15592
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40042755042498523,
      "learning_rate": 1.1210592862815552e-05,
      "loss": 0.2775,
      "step": 15593
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4669417293876511,
      "learning_rate": 1.1209608268375022e-05,
      "loss": 0.2375,
      "step": 15594
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4489752977556009,
      "learning_rate": 1.1208623662033971e-05,
      "loss": 0.2253,
      "step": 15595
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1854728964740318,
      "learning_rate": 1.120763904380208e-05,
      "loss": 0.0701,
      "step": 15596
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5468485320044221,
      "learning_rate": 1.120665441368904e-05,
      "loss": 0.3612,
      "step": 15597
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29214702416858107,
      "learning_rate": 1.1205669771704536e-05,
      "loss": 0.2458,
      "step": 15598
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8365838958461459,
      "learning_rate": 1.1204685117858258e-05,
      "loss": 0.3444,
      "step": 15599
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3273343704777831,
      "learning_rate": 1.1203700452159891e-05,
      "loss": 0.8645,
      "step": 15600
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.453305035497865,
      "learning_rate": 1.120271577461912e-05,
      "loss": 0.2214,
      "step": 15601
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3320631446348775,
      "learning_rate": 1.120173108524564e-05,
      "loss": 0.2932,
      "step": 15602
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3725762088654963,
      "learning_rate": 1.1200746384049131e-05,
      "loss": 0.2345,
      "step": 15603
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.49932076969621086,
      "learning_rate": 1.1199761671039281e-05,
      "loss": 0.3732,
      "step": 15604
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35886718337445234,
      "learning_rate": 1.1198776946225787e-05,
      "loss": 0.1847,
      "step": 15605
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5132299680464489,
      "learning_rate": 1.119779220961833e-05,
      "loss": 0.3441,
      "step": 15606
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34940058301648863,
      "learning_rate": 1.1196807461226596e-05,
      "loss": 0.227,
      "step": 15607
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.47701346524947214,
      "learning_rate": 1.1195822701060277e-05,
      "loss": 0.2745,
      "step": 15608
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2854688003551444,
      "learning_rate": 1.1194837929129064e-05,
      "loss": 0.2586,
      "step": 15609
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8302549138879777,
      "learning_rate": 1.1193853145442638e-05,
      "loss": 0.5518,
      "step": 15610
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.33541190960141737,
      "learning_rate": 1.1192868350010695e-05,
      "loss": 0.1733,
      "step": 15611
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4118366865365575,
      "learning_rate": 1.119188354284292e-05,
      "loss": 0.2196,
      "step": 15612
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3133898039500598,
      "learning_rate": 1.1190898723949002e-05,
      "loss": 0.707,
      "step": 15613
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30220619087174755,
      "learning_rate": 1.1189913893338629e-05,
      "loss": 0.2261,
      "step": 15614
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8201542125862508,
      "learning_rate": 1.1188929051021495e-05,
      "loss": 0.4111,
      "step": 15615
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2674323650383034,
      "learning_rate": 1.1187944197007284e-05,
      "loss": 0.2126,
      "step": 15616
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.429019942641876,
      "learning_rate": 1.1186959331305685e-05,
      "loss": 0.3018,
      "step": 15617
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8189085684381582,
      "learning_rate": 1.1185974453926391e-05,
      "loss": 0.626,
      "step": 15618
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6060727819496392,
      "learning_rate": 1.1184989564879093e-05,
      "loss": 0.8383,
      "step": 15619
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29736776529336373,
      "learning_rate": 1.1184004664173475e-05,
      "loss": 0.1772,
      "step": 15620
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5112120309816408,
      "learning_rate": 1.1183019751819229e-05,
      "loss": 0.353,
      "step": 15621
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31277269490704485,
      "learning_rate": 1.1182034827826047e-05,
      "loss": 0.2597,
      "step": 15622
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8336996349515409,
      "learning_rate": 1.1181049892203618e-05,
      "loss": 0.4475,
      "step": 15623
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3596455374350944,
      "learning_rate": 1.1180064944961631e-05,
      "loss": 0.2385,
      "step": 15624
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32068917099213456,
      "learning_rate": 1.1179079986109776e-05,
      "loss": 0.2522,
      "step": 15625
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.409144545558145,
      "learning_rate": 1.1178095015657743e-05,
      "loss": 0.2864,
      "step": 15626
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2808529070830649,
      "learning_rate": 1.1177110033615223e-05,
      "loss": 0.246,
      "step": 15627
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.471762703839556,
      "learning_rate": 1.1176125039991911e-05,
      "loss": 0.3179,
      "step": 15628
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27580337264055743,
      "learning_rate": 1.1175140034797491e-05,
      "loss": 0.1939,
      "step": 15629
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.6596390288308913,
      "learning_rate": 1.1174155018041659e-05,
      "loss": 0.823,
      "step": 15630
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7260342155619739,
      "learning_rate": 1.1173169989734101e-05,
      "loss": 0.3297,
      "step": 15631
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.524115224996844,
      "learning_rate": 1.1172184949884513e-05,
      "loss": 0.4015,
      "step": 15632
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.24378143016092432,
      "learning_rate": 1.117119989850258e-05,
      "loss": 0.2071,
      "step": 15633
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35123301605189217,
      "learning_rate": 1.1170214835598001e-05,
      "loss": 0.2758,
      "step": 15634
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38712276880550406,
      "learning_rate": 1.1169229761180462e-05,
      "loss": 0.011,
      "step": 15635
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.239429256564199,
      "learning_rate": 1.1168244675259654e-05,
      "loss": 0.7721,
      "step": 15636
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1813939300120692,
      "learning_rate": 1.1167259577845275e-05,
      "loss": 0.0894,
      "step": 15637
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31339592310699965,
      "learning_rate": 1.1166274468947009e-05,
      "loss": 0.0765,
      "step": 15638
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3662663868694067,
      "learning_rate": 1.116528934857455e-05,
      "loss": 0.3007,
      "step": 15639
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3015208202107503,
      "learning_rate": 1.1164304216737593e-05,
      "loss": 0.2585,
      "step": 15640
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8237937508287443,
      "learning_rate": 1.1163319073445826e-05,
      "loss": 0.4235,
      "step": 15641
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5437439243312259,
      "learning_rate": 1.1162333918708948e-05,
      "loss": 0.1466,
      "step": 15642
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35171898878838975,
      "learning_rate": 1.1161348752536643e-05,
      "loss": 0.2882,
      "step": 15643
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8613922771759526,
      "learning_rate": 1.1160363574938607e-05,
      "loss": 0.4496,
      "step": 15644
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31618104951770676,
      "learning_rate": 1.1159378385924533e-05,
      "loss": 0.2985,
      "step": 15645
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23656343850138822,
      "learning_rate": 1.115839318550411e-05,
      "loss": 0.1051,
      "step": 15646
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3426477063725068,
      "learning_rate": 1.1157407973687036e-05,
      "loss": 0.2021,
      "step": 15647
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30961088928040087,
      "learning_rate": 1.1156422750483003e-05,
      "loss": 0.2168,
      "step": 15648
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8693903548431745,
      "learning_rate": 1.11554375159017e-05,
      "loss": 0.4854,
      "step": 15649
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8119606134907947,
      "learning_rate": 1.1154452269952822e-05,
      "loss": 0.3342,
      "step": 15650
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2645813321465124,
      "learning_rate": 1.1153467012646067e-05,
      "loss": 0.2216,
      "step": 15651
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34123369870647313,
      "learning_rate": 1.115248174399112e-05,
      "loss": 0.3008,
      "step": 15652
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0026326444539786,
      "learning_rate": 1.1151496463997678e-05,
      "loss": 0.4445,
      "step": 15653
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.502277198893703,
      "learning_rate": 1.1150511172675435e-05,
      "loss": 0.8429,
      "step": 15654
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1869300256825623,
      "learning_rate": 1.1149525870034089e-05,
      "loss": 0.0895,
      "step": 15655
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36968640450395795,
      "learning_rate": 1.1148540556083325e-05,
      "loss": 0.2656,
      "step": 15656
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.35050009880098904,
      "learning_rate": 1.1147555230832843e-05,
      "loss": 0.2626,
      "step": 15657
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5422799439416207,
      "learning_rate": 1.1146569894292334e-05,
      "loss": 0.3792,
      "step": 15658
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0428837248508231,
      "learning_rate": 1.1145584546471493e-05,
      "loss": 0.3664,
      "step": 15659
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6775185344515746,
      "learning_rate": 1.1144599187380014e-05,
      "loss": 0.4071,
      "step": 15660
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28329923567209603,
      "learning_rate": 1.1143613817027595e-05,
      "loss": 0.2094,
      "step": 15661
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4793339024263017,
      "learning_rate": 1.1142628435423923e-05,
      "loss": 0.8175,
      "step": 15662
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28206051200913856,
      "learning_rate": 1.1141643042578697e-05,
      "loss": 0.2655,
      "step": 15663
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.1848136136413239,
      "learning_rate": 1.1140657638501614e-05,
      "loss": 0.1087,
      "step": 15664
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6455009963445779,
      "learning_rate": 1.1139672223202363e-05,
      "loss": 0.3074,
      "step": 15665
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32586649761213954,
      "learning_rate": 1.113868679669064e-05,
      "loss": 0.2459,
      "step": 15666
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7332691650341582,
      "learning_rate": 1.1137701358976142e-05,
      "loss": 0.4562,
      "step": 15667
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2860424895982792,
      "learning_rate": 1.1136715910068567e-05,
      "loss": 0.2434,
      "step": 15668
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7824581034087785,
      "learning_rate": 1.1135730449977608e-05,
      "loss": 0.8947,
      "step": 15669
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2973921297953024,
      "learning_rate": 1.1134744978712954e-05,
      "loss": 0.162,
      "step": 15670
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.483006363177069,
      "learning_rate": 1.1133759496284311e-05,
      "loss": 0.4258,
      "step": 15671
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7619710678553349,
      "learning_rate": 1.1132774002701366e-05,
      "loss": 0.4959,
      "step": 15672
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.244978533694326,
      "learning_rate": 1.1131788497973815e-05,
      "loss": 0.6473,
      "step": 15673
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3147347745949176,
      "learning_rate": 1.1130802982111359e-05,
      "loss": 0.2444,
      "step": 15674
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2885089967246428,
      "learning_rate": 1.1129817455123696e-05,
      "loss": 0.2744,
      "step": 15675
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2635048373944194,
      "learning_rate": 1.1128831917020509e-05,
      "loss": 0.2211,
      "step": 15676
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21772021876075695,
      "learning_rate": 1.1127846367811507e-05,
      "loss": 0.0975,
      "step": 15677
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1953450947661932,
      "learning_rate": 1.1126860807506379e-05,
      "loss": 0.5949,
      "step": 15678
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2605274246116837,
      "learning_rate": 1.1125875236114823e-05,
      "loss": 0.1883,
      "step": 15679
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2765888154731408,
      "learning_rate": 1.112488965364654e-05,
      "loss": 0.9068,
      "step": 15680
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3131333220517146,
      "learning_rate": 1.1123904060111219e-05,
      "loss": 0.2392,
      "step": 15681
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8409414633500244,
      "learning_rate": 1.1122918455518562e-05,
      "loss": 0.533,
      "step": 15682
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3889785506428138,
      "learning_rate": 1.112193283987826e-05,
      "loss": 0.2471,
      "step": 15683
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3771018948264855,
      "learning_rate": 1.1120947213200018e-05,
      "loss": 0.3201,
      "step": 15684
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1675391042582464,
      "learning_rate": 1.1119961575493526e-05,
      "loss": 0.2459,
      "step": 15685
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2526107129741762,
      "learning_rate": 1.1118975926768483e-05,
      "loss": 0.2186,
      "step": 15686
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3131751272222014,
      "learning_rate": 1.1117990267034589e-05,
      "loss": 0.1949,
      "step": 15687
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34644111899439045,
      "learning_rate": 1.111700459630154e-05,
      "loss": 0.2737,
      "step": 15688
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.055478242052181,
      "learning_rate": 1.1116018914579028e-05,
      "loss": 0.3681,
      "step": 15689
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9249571639129746,
      "learning_rate": 1.1115033221876757e-05,
      "loss": 0.473,
      "step": 15690
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6654772104593238,
      "learning_rate": 1.111404751820442e-05,
      "loss": 0.4006,
      "step": 15691
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29387534931328707,
      "learning_rate": 1.111306180357172e-05,
      "loss": 0.1846,
      "step": 15692
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3496558798493015,
      "learning_rate": 1.1112076077988352e-05,
      "loss": 0.2813,
      "step": 15693
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.370073626612338,
      "learning_rate": 1.1111090341464013e-05,
      "loss": 0.2649,
      "step": 15694
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2951618072419424,
      "learning_rate": 1.11101045940084e-05,
      "loss": 0.1826,
      "step": 15695
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1716318270844717,
      "learning_rate": 1.1109118835631212e-05,
      "loss": 0.2813,
      "step": 15696
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6840420486389561,
      "learning_rate": 1.1108133066342151e-05,
      "loss": 0.371,
      "step": 15697
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3077542073194459,
      "learning_rate": 1.110714728615091e-05,
      "loss": 0.2389,
      "step": 15698
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32862292951287797,
      "learning_rate": 1.1106161495067189e-05,
      "loss": 0.2978,
      "step": 15699
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6902769565105318,
      "learning_rate": 1.110517569310069e-05,
      "loss": 0.3333,
      "step": 15700
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5901385452147239,
      "learning_rate": 1.1104189880261109e-05,
      "loss": 0.4526,
      "step": 15701
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.25473351235716923,
      "learning_rate": 1.1103204056558142e-05,
      "loss": 0.1882,
      "step": 15702
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.29643971335837743,
      "learning_rate": 1.1102218222001492e-05,
      "loss": 0.1336,
      "step": 15703
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43123742670588766,
      "learning_rate": 1.1101232376600857e-05,
      "loss": 0.2321,
      "step": 15704
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30379899934127647,
      "learning_rate": 1.1100246520365935e-05,
      "loss": 0.2264,
      "step": 15705
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38172041693623576,
      "learning_rate": 1.1099260653306425e-05,
      "loss": 0.2693,
      "step": 15706
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3900761062012514,
      "learning_rate": 1.1098274775432028e-05,
      "loss": 0.2704,
      "step": 15707
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9624254608124081,
      "learning_rate": 1.109728888675244e-05,
      "loss": 0.516,
      "step": 15708
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7306812170282674,
      "learning_rate": 1.1096302987277366e-05,
      "loss": 0.3732,
      "step": 15709
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3294396775483902,
      "learning_rate": 1.1095317077016501e-05,
      "loss": 0.2968,
      "step": 15710
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.27796185327558676,
      "learning_rate": 1.1094331155979545e-05,
      "loss": 0.1937,
      "step": 15711
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1895421859461839,
      "learning_rate": 1.1093345224176201e-05,
      "loss": 0.5577,
      "step": 15712
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.19060384067341982,
      "learning_rate": 1.1092359281616168e-05,
      "loss": 0.0735,
      "step": 15713
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.0487735465307189,
      "learning_rate": 1.1091373328309143e-05,
      "loss": 0.616,
      "step": 15714
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3050464335046028,
      "learning_rate": 1.1090387364264827e-05,
      "loss": 0.1844,
      "step": 15715
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4833396371937733,
      "learning_rate": 1.1089401389492924e-05,
      "loss": 0.2586,
      "step": 15716
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3409703120142784,
      "learning_rate": 1.1088415404003128e-05,
      "loss": 0.3245,
      "step": 15717
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8273697339692837,
      "learning_rate": 1.1087429407805145e-05,
      "loss": 0.3994,
      "step": 15718
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8961979096314476,
      "learning_rate": 1.1086443400908675e-05,
      "loss": 0.3892,
      "step": 15719
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3854584226903026,
      "learning_rate": 1.1085457383323417e-05,
      "loss": 0.2111,
      "step": 15720
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9904133001876068,
      "learning_rate": 1.1084471355059072e-05,
      "loss": 0.622,
      "step": 15721
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2167544188975852,
      "learning_rate": 1.108348531612534e-05,
      "loss": 0.2049,
      "step": 15722
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4631688006398323,
      "learning_rate": 1.1082499266531924e-05,
      "loss": 0.269,
      "step": 15723
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26675479512464667,
      "learning_rate": 1.1081513206288521e-05,
      "loss": 0.1193,
      "step": 15724
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3764377388142444,
      "learning_rate": 1.108052713540484e-05,
      "loss": 0.3177,
      "step": 15725
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.541016549189424,
      "learning_rate": 1.1079541053890577e-05,
      "loss": 0.339,
      "step": 15726
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3060257681412883,
      "learning_rate": 1.107855496175543e-05,
      "loss": 0.9048,
      "step": 15727
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3003023976736692,
      "learning_rate": 1.1077568859009107e-05,
      "loss": 0.25,
      "step": 15728
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2706357924503837,
      "learning_rate": 1.1076582745661307e-05,
      "loss": 0.1963,
      "step": 15729
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.732528087562634,
      "learning_rate": 1.107559662172173e-05,
      "loss": 0.7785,
      "step": 15730
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8208519942841973,
      "learning_rate": 1.1074610487200081e-05,
      "loss": 0.283,
      "step": 15731
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.237555948091536,
      "learning_rate": 1.107362434210606e-05,
      "loss": 0.1422,
      "step": 15732
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38199704232436044,
      "learning_rate": 1.1072638186449369e-05,
      "loss": 0.2387,
      "step": 15733
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31337624351587706,
      "learning_rate": 1.1071652020239711e-05,
      "loss": 0.2926,
      "step": 15734
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3577942972460293,
      "learning_rate": 1.1070665843486788e-05,
      "loss": 0.2814,
      "step": 15735
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8798177074928057,
      "learning_rate": 1.1069679656200303e-05,
      "loss": 0.6018,
      "step": 15736
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5892092154787842,
      "learning_rate": 1.1068693458389956e-05,
      "loss": 0.242,
      "step": 15737
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3502059274686315,
      "learning_rate": 1.106770725006545e-05,
      "loss": 0.2734,
      "step": 15738
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8170643280492299,
      "learning_rate": 1.1066721031236494e-05,
      "loss": 0.0342,
      "step": 15739
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4769594986890506,
      "learning_rate": 1.1065734801912782e-05,
      "loss": 0.3446,
      "step": 15740
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2227890241997801,
      "learning_rate": 1.1064748562104017e-05,
      "loss": 0.153,
      "step": 15741
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3583394058768064,
      "learning_rate": 1.1063762311819913e-05,
      "loss": 0.1948,
      "step": 15742
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5849513166487939,
      "learning_rate": 1.1062776051070159e-05,
      "loss": 0.3855,
      "step": 15743
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6604837246595071,
      "learning_rate": 1.1061789779864467e-05,
      "loss": 0.3637,
      "step": 15744
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1189491337369757,
      "learning_rate": 1.1060803498212536e-05,
      "loss": 0.7368,
      "step": 15745
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3426212556369462,
      "learning_rate": 1.1059817206124075e-05,
      "loss": 0.246,
      "step": 15746
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8857607347474261,
      "learning_rate": 1.1058830903608779e-05,
      "loss": 0.3911,
      "step": 15747
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32185854095490535,
      "learning_rate": 1.1057844590676357e-05,
      "loss": 0.2178,
      "step": 15748
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.1382392441840155,
      "learning_rate": 1.1056858267336515e-05,
      "loss": 0.5648,
      "step": 15749
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4337679998459041,
      "learning_rate": 1.105587193359895e-05,
      "loss": 0.2198,
      "step": 15750
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6675146675299591,
      "learning_rate": 1.1054885589473373e-05,
      "loss": 0.4059,
      "step": 15751
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.25542361864757535,
      "learning_rate": 1.1053899234969482e-05,
      "loss": 0.2141,
      "step": 15752
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4031108302586292,
      "learning_rate": 1.1052912870096985e-05,
      "loss": 0.3646,
      "step": 15753
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8062501781028589,
      "learning_rate": 1.1051926494865582e-05,
      "loss": 0.5682,
      "step": 15754
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2881455781556704,
      "learning_rate": 1.1050940109284984e-05,
      "loss": 0.1068,
      "step": 15755
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.36980967974909057,
      "learning_rate": 1.1049953713364888e-05,
      "loss": 0.2502,
      "step": 15756
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4150776104366136,
      "learning_rate": 1.1048967307115004e-05,
      "loss": 0.23,
      "step": 15757
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.48264071247157125,
      "learning_rate": 1.1047980890545033e-05,
      "loss": 0.3374,
      "step": 15758
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38399370166701735,
      "learning_rate": 1.1046994463664681e-05,
      "loss": 0.2848,
      "step": 15759
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6418377593331295,
      "learning_rate": 1.1046008026483655e-05,
      "loss": 0.3758,
      "step": 15760
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.21640245472994776,
      "learning_rate": 1.1045021579011654e-05,
      "loss": 0.1649,
      "step": 15761
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.7101028552760134,
      "learning_rate": 1.1044035121258391e-05,
      "loss": 0.938,
      "step": 15762
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2362078292755095,
      "learning_rate": 1.1043048653233568e-05,
      "loss": 0.1568,
      "step": 15763
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32062348197560775,
      "learning_rate": 1.1042062174946886e-05,
      "loss": 0.287,
      "step": 15764
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.28634329142921955,
      "learning_rate": 1.1041075686408053e-05,
      "loss": 0.1962,
      "step": 15765
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.934209630342761,
      "learning_rate": 1.1040089187626778e-05,
      "loss": 0.8767,
      "step": 15766
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7916941876890707,
      "learning_rate": 1.103910267861276e-05,
      "loss": 0.3015,
      "step": 15767
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7459729844707947,
      "learning_rate": 1.1038116159375707e-05,
      "loss": 0.4409,
      "step": 15768
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2979454098709363,
      "learning_rate": 1.1037129629925329e-05,
      "loss": 0.2181,
      "step": 15769
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30177858622057274,
      "learning_rate": 1.1036143090271327e-05,
      "loss": 0.2062,
      "step": 15770
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4788089890164301,
      "learning_rate": 1.1035156540423409e-05,
      "loss": 0.3662,
      "step": 15771
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.25343584770245414,
      "learning_rate": 1.103416998039128e-05,
      "loss": 0.1569,
      "step": 15772
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40499950672508417,
      "learning_rate": 1.1033183410184648e-05,
      "loss": 0.2241,
      "step": 15773
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.32019809732210436,
      "learning_rate": 1.1032196829813213e-05,
      "loss": 0.0754,
      "step": 15774
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.40358739210053735,
      "learning_rate": 1.103121023928669e-05,
      "loss": 0.3041,
      "step": 15775
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34377138075104774,
      "learning_rate": 1.1030223638614781e-05,
      "loss": 0.2407,
      "step": 15776
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9673293066029817,
      "learning_rate": 1.1029237027807191e-05,
      "loss": 0.4379,
      "step": 15777
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2526009112715704,
      "learning_rate": 1.102825040687363e-05,
      "loss": 0.272,
      "step": 15778
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4017382715932245,
      "learning_rate": 1.1027263775823805e-05,
      "loss": 0.2977,
      "step": 15779
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.455496485472091,
      "learning_rate": 1.1026277134667415e-05,
      "loss": 0.2984,
      "step": 15780
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4555613007915675,
      "learning_rate": 1.102529048341418e-05,
      "loss": 0.3645,
      "step": 15781
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26776533447731854,
      "learning_rate": 1.1024303822073797e-05,
      "loss": 0.1649,
      "step": 15782
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2315706945741355,
      "learning_rate": 1.1023317150655975e-05,
      "loss": 0.1344,
      "step": 15783
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5084746155562588,
      "learning_rate": 1.1022330469170421e-05,
      "loss": 0.3493,
      "step": 15784
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.2953381033031939,
      "learning_rate": 1.1021343777626846e-05,
      "loss": 0.3718,
      "step": 15785
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9201045444984374,
      "learning_rate": 1.1020357076034954e-05,
      "loss": 0.4655,
      "step": 15786
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3214862910924922,
      "learning_rate": 1.1019370364404451e-05,
      "loss": 0.2051,
      "step": 15787
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3409847995100185,
      "learning_rate": 1.101838364274505e-05,
      "loss": 0.2916,
      "step": 15788
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.34407492153705654,
      "learning_rate": 1.1017396911066456e-05,
      "loss": 0.2311,
      "step": 15789
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.3486695686720962,
      "learning_rate": 1.1016410169378374e-05,
      "loss": 0.9422,
      "step": 15790
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.16351715531781102,
      "learning_rate": 1.1015423417690516e-05,
      "loss": 0.0694,
      "step": 15791
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5617378137868939,
      "learning_rate": 1.1014436656012589e-05,
      "loss": 0.3015,
      "step": 15792
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.357227282346366,
      "learning_rate": 1.1013449884354297e-05,
      "loss": 0.2162,
      "step": 15793
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44813331851075466,
      "learning_rate": 1.1012463102725354e-05,
      "loss": 0.3397,
      "step": 15794
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4175256769388038,
      "learning_rate": 1.1011476311135466e-05,
      "loss": 0.4194,
      "step": 15795
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.49946919788721117,
      "learning_rate": 1.1010489509594343e-05,
      "loss": 0.1672,
      "step": 15796
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5541410065531341,
      "learning_rate": 1.1009502698111688e-05,
      "loss": 0.3632,
      "step": 15797
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.39164377488448393,
      "learning_rate": 1.1008515876697214e-05,
      "loss": 0.2809,
      "step": 15798
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.43316586352070663,
      "learning_rate": 1.100752904536063e-05,
      "loss": 0.3575,
      "step": 15799
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.185983410189111,
      "learning_rate": 1.1006542204111644e-05,
      "loss": 0.1478,
      "step": 15800
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8312013747429948,
      "learning_rate": 1.1005555352959966e-05,
      "loss": 0.3638,
      "step": 15801
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.31663627574229253,
      "learning_rate": 1.10045684919153e-05,
      "loss": 0.2356,
      "step": 15802
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6231155703316213,
      "learning_rate": 1.100358162098736e-05,
      "loss": 0.4159,
      "step": 15803
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7537131572296326,
      "learning_rate": 1.1002594740185855e-05,
      "loss": 0.3351,
      "step": 15804
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4369701985697385,
      "learning_rate": 1.100160784952049e-05,
      "loss": 0.3306,
      "step": 15805
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26223399790163116,
      "learning_rate": 1.100062094900098e-05,
      "loss": 0.1928,
      "step": 15806
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5000831269519049,
      "learning_rate": 1.0999634038637032e-05,
      "loss": 0.3562,
      "step": 15807
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9361589826290603,
      "learning_rate": 1.0998647118438357e-05,
      "loss": 0.4707,
      "step": 15808
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3271873572056677,
      "learning_rate": 1.0997660188414661e-05,
      "loss": 0.1739,
      "step": 15809
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5456337184076039,
      "learning_rate": 1.0996673248575653e-05,
      "loss": 0.4394,
      "step": 15810
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.26364143955387376,
      "learning_rate": 1.099568629893105e-05,
      "loss": 0.2241,
      "step": 15811
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6017624573142145,
      "learning_rate": 1.0994699339490555e-05,
      "loss": 0.3138,
      "step": 15812
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.20028787318656668,
      "learning_rate": 1.0993712370263882e-05,
      "loss": 0.068,
      "step": 15813
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.4821480163863745,
      "learning_rate": 1.0992725391260741e-05,
      "loss": 0.7408,
      "step": 15814
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2814894578418351,
      "learning_rate": 1.0991738402490836e-05,
      "loss": 0.1868,
      "step": 15815
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.406337125493442,
      "learning_rate": 1.0990751403963886e-05,
      "loss": 0.7946,
      "step": 15816
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4147014973801733,
      "learning_rate": 1.0989764395689599e-05,
      "loss": 0.2589,
      "step": 15817
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4606155608181689,
      "learning_rate": 1.0988777377677682e-05,
      "loss": 0.3469,
      "step": 15818
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.5448442934448395,
      "learning_rate": 1.0987790349937846e-05,
      "loss": 0.2413,
      "step": 15819
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.38355491368107547,
      "learning_rate": 1.0986803312479805e-05,
      "loss": 0.2903,
      "step": 15820
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.305395365570561,
      "learning_rate": 1.0985816265313272e-05,
      "loss": 0.1939,
      "step": 15821
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.286831633080841,
      "learning_rate": 1.0984829208447949e-05,
      "loss": 0.1233,
      "step": 15822
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.44092491907744447,
      "learning_rate": 1.0983842141893555e-05,
      "loss": 0.3104,
      "step": 15823
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2956744250076413,
      "learning_rate": 1.0982855065659798e-05,
      "loss": 0.0761,
      "step": 15824
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3832242790597317,
      "learning_rate": 1.0981867979756388e-05,
      "loss": 0.3222,
      "step": 15825
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.8389827279043601,
      "learning_rate": 1.0980880884193039e-05,
      "loss": 0.5457,
      "step": 15826
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7440571435883588,
      "learning_rate": 1.0979893778979463e-05,
      "loss": 0.4141,
      "step": 15827
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.48121072371248264,
      "learning_rate": 1.0978906664125366e-05,
      "loss": 0.2321,
      "step": 15828
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.3147612551710705,
      "learning_rate": 1.0977919539640465e-05,
      "loss": 0.2616,
      "step": 15829
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.23002885166548212,
      "learning_rate": 1.097693240553447e-05,
      "loss": 0.2047,
      "step": 15830
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.4380362321088256,
      "learning_rate": 1.0975945261817092e-05,
      "loss": 0.26,
      "step": 15831
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.9929918756392548,
      "learning_rate": 1.0974958108498044e-05,
      "loss": 0.3024,
      "step": 15832
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.6870708008898492,
      "learning_rate": 1.0973970945587036e-05,
      "loss": 0.4392,
      "step": 15833
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.30340371138973204,
      "learning_rate": 1.0972983773093781e-05,
      "loss": 0.2356,
      "step": 15834
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.2909016950178257,
      "learning_rate": 1.0971996591027993e-05,
      "loss": 0.2423,
      "step": 15835
    },
    {
      "epoch": 0.48,
      "grad_norm": 0.7779220335559297,
      "learning_rate": 1.0971009399399382e-05,
      "loss": 0.4702,
      "step": 15836
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.517377373746042,
      "learning_rate": 1.0970022198217663e-05,
      "loss": 0.2827,
      "step": 15837
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3686150010875224,
      "learning_rate": 1.0969034987492542e-05,
      "loss": 0.328,
      "step": 15838
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29215314773677764,
      "learning_rate": 1.096804776723374e-05,
      "loss": 0.1064,
      "step": 15839
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4678096883454755,
      "learning_rate": 1.0967060537450966e-05,
      "loss": 0.3084,
      "step": 15840
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28486070522162116,
      "learning_rate": 1.096607329815393e-05,
      "loss": 0.2261,
      "step": 15841
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5802173641323183,
      "learning_rate": 1.0965086049352345e-05,
      "loss": 0.3498,
      "step": 15842
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3290864337217036,
      "learning_rate": 1.096409879105593e-05,
      "loss": 0.2253,
      "step": 15843
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3927885780815674,
      "learning_rate": 1.096311152327439e-05,
      "loss": 0.7658,
      "step": 15844
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7310024085242818,
      "learning_rate": 1.0962124246017447e-05,
      "loss": 0.3669,
      "step": 15845
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0180257351433877,
      "learning_rate": 1.0961136959294805e-05,
      "loss": 0.5032,
      "step": 15846
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3187520265139962,
      "learning_rate": 1.0960149663116181e-05,
      "loss": 0.1673,
      "step": 15847
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2863351399431007,
      "learning_rate": 1.0959162357491292e-05,
      "loss": 0.2233,
      "step": 15848
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2933527352481781,
      "learning_rate": 1.0958175042429843e-05,
      "loss": 0.1871,
      "step": 15849
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.944825833646358,
      "learning_rate": 1.095718771794156e-05,
      "loss": 0.4224,
      "step": 15850
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9006337048348344,
      "learning_rate": 1.0956200384036145e-05,
      "loss": 0.3694,
      "step": 15851
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28651492232966413,
      "learning_rate": 1.0955213040723317e-05,
      "loss": 0.2204,
      "step": 15852
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2974944483444368,
      "learning_rate": 1.0954225688012787e-05,
      "loss": 0.3146,
      "step": 15853
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.49137518779130657,
      "learning_rate": 1.0953238325914271e-05,
      "loss": 0.3589,
      "step": 15854
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2202083523954304,
      "learning_rate": 1.0952250954437482e-05,
      "loss": 0.6289,
      "step": 15855
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.31347754097353214,
      "learning_rate": 1.0951263573592138e-05,
      "loss": 0.1822,
      "step": 15856
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5235415248320447,
      "learning_rate": 1.0950276183387948e-05,
      "loss": 0.3693,
      "step": 15857
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19719776604785258,
      "learning_rate": 1.0949288783834629e-05,
      "loss": 0.142,
      "step": 15858
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.48541257714726443,
      "learning_rate": 1.0948301374941891e-05,
      "loss": 0.3573,
      "step": 15859
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2501665473932913,
      "learning_rate": 1.0947313956719458e-05,
      "loss": 0.115,
      "step": 15860
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2879067065991767,
      "learning_rate": 1.0946326529177034e-05,
      "loss": 0.2513,
      "step": 15861
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.478323014531505,
      "learning_rate": 1.0945339092324338e-05,
      "loss": 0.3933,
      "step": 15862
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.840694597821864,
      "learning_rate": 1.094435164617109e-05,
      "loss": 0.4851,
      "step": 15863
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43692223212422077,
      "learning_rate": 1.0943364190726997e-05,
      "loss": 0.3606,
      "step": 15864
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25463847668341355,
      "learning_rate": 1.0942376726001775e-05,
      "loss": 0.1951,
      "step": 15865
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4441525530705714,
      "learning_rate": 1.0941389252005143e-05,
      "loss": 0.3549,
      "step": 15866
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2567128297056952,
      "learning_rate": 1.0940401768746812e-05,
      "loss": 0.0991,
      "step": 15867
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42280367622084114,
      "learning_rate": 1.09394142762365e-05,
      "loss": 0.2444,
      "step": 15868
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3578179306689281,
      "learning_rate": 1.0938426774483921e-05,
      "loss": 0.2056,
      "step": 15869
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4099898755897194,
      "learning_rate": 1.093743926349879e-05,
      "loss": 0.3126,
      "step": 15870
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33528749590427964,
      "learning_rate": 1.0936451743290824e-05,
      "loss": 0.2399,
      "step": 15871
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5268307022794261,
      "learning_rate": 1.0935464213869735e-05,
      "loss": 0.3987,
      "step": 15872
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9286875881976198,
      "learning_rate": 1.0934476675245248e-05,
      "loss": 0.3176,
      "step": 15873
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.30117945641389204,
      "learning_rate": 1.0933489127427063e-05,
      "loss": 0.0742,
      "step": 15874
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35747596625321304,
      "learning_rate": 1.093250157042491e-05,
      "loss": 0.2993,
      "step": 15875
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2246682722187292,
      "learning_rate": 1.09315140042485e-05,
      "loss": 0.1456,
      "step": 15876
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.48271848364930603,
      "learning_rate": 1.0930526428907548e-05,
      "loss": 0.3417,
      "step": 15877
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6102743674810304,
      "learning_rate": 1.0929538844411772e-05,
      "loss": 0.2311,
      "step": 15878
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3407217720042149,
      "learning_rate": 1.0928551250770885e-05,
      "loss": 0.2874,
      "step": 15879
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6890917618126962,
      "learning_rate": 1.0927563647994605e-05,
      "loss": 0.3614,
      "step": 15880
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0730897168499929,
      "learning_rate": 1.0926576036092649e-05,
      "loss": 0.6788,
      "step": 15881
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3751797676908634,
      "learning_rate": 1.0925588415074733e-05,
      "loss": 0.2641,
      "step": 15882
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38796397203849914,
      "learning_rate": 1.0924600784950576e-05,
      "loss": 0.2805,
      "step": 15883
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33416019399869695,
      "learning_rate": 1.092361314572989e-05,
      "loss": 0.2163,
      "step": 15884
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.45079185589506304,
      "learning_rate": 1.0922625497422391e-05,
      "loss": 0.2836,
      "step": 15885
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39946022234556366,
      "learning_rate": 1.0921637840037804e-05,
      "loss": 0.1613,
      "step": 15886
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7577503581223093,
      "learning_rate": 1.0920650173585838e-05,
      "loss": 0.3045,
      "step": 15887
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32702925855785875,
      "learning_rate": 1.0919662498076213e-05,
      "loss": 0.257,
      "step": 15888
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3102474851199257,
      "learning_rate": 1.0918674813518645e-05,
      "loss": 0.2794,
      "step": 15889
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4797042690831272,
      "learning_rate": 1.0917687119922855e-05,
      "loss": 0.7831,
      "step": 15890
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8773544426102057,
      "learning_rate": 1.0916699417298554e-05,
      "loss": 0.3614,
      "step": 15891
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.40357582743890275,
      "learning_rate": 1.0915711705655465e-05,
      "loss": 0.2995,
      "step": 15892
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39001617208228717,
      "learning_rate": 1.0914723985003302e-05,
      "loss": 0.2032,
      "step": 15893
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9863618117534634,
      "learning_rate": 1.091373625535178e-05,
      "loss": 0.7565,
      "step": 15894
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25982635811641297,
      "learning_rate": 1.0912748516710626e-05,
      "loss": 0.2219,
      "step": 15895
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.736878601029231,
      "learning_rate": 1.091176076908955e-05,
      "loss": 0.3541,
      "step": 15896
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.27649719071469375,
      "learning_rate": 1.0910773012498271e-05,
      "loss": 0.1815,
      "step": 15897
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0187252362880643,
      "learning_rate": 1.0909785246946505e-05,
      "loss": 0.6077,
      "step": 15898
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2400302863300285,
      "learning_rate": 1.0908797472443975e-05,
      "loss": 0.1629,
      "step": 15899
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29913680686451166,
      "learning_rate": 1.0907809689000395e-05,
      "loss": 0.2533,
      "step": 15900
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5691269566049713,
      "learning_rate": 1.0906821896625486e-05,
      "loss": 0.245,
      "step": 15901
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3666501168433036,
      "learning_rate": 1.0905834095328965e-05,
      "loss": 0.2304,
      "step": 15902
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0927757502480513,
      "learning_rate": 1.0904846285120547e-05,
      "loss": 0.5088,
      "step": 15903
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7459204702519888,
      "learning_rate": 1.0903858466009956e-05,
      "loss": 0.3489,
      "step": 15904
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3565821033807141,
      "learning_rate": 1.0902870638006908e-05,
      "loss": 0.2062,
      "step": 15905
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3015492948661252,
      "learning_rate": 1.090188280112112e-05,
      "loss": 0.2106,
      "step": 15906
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32219390320068336,
      "learning_rate": 1.0900894955362313e-05,
      "loss": 0.3257,
      "step": 15907
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18515038348494847,
      "learning_rate": 1.0899907100740206e-05,
      "loss": 0.1524,
      "step": 15908
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.9001422329536484,
      "learning_rate": 1.0898919237264518e-05,
      "loss": 0.787,
      "step": 15909
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2468873474641439,
      "learning_rate": 1.0897931364944965e-05,
      "loss": 0.0755,
      "step": 15910
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35630031700547116,
      "learning_rate": 1.0896943483791267e-05,
      "loss": 0.2952,
      "step": 15911
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42506643833608254,
      "learning_rate": 1.0895955593813146e-05,
      "loss": 0.2917,
      "step": 15912
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4809125125103653,
      "learning_rate": 1.0894967695020316e-05,
      "loss": 0.2691,
      "step": 15913
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1127455183262367,
      "learning_rate": 1.0893979787422502e-05,
      "loss": 0.5956,
      "step": 15914
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2701194502202573,
      "learning_rate": 1.0892991871029422e-05,
      "loss": 0.1918,
      "step": 15915
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2505876649538623,
      "learning_rate": 1.0892003945850789e-05,
      "loss": 0.8624,
      "step": 15916
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22562753425635093,
      "learning_rate": 1.0891016011896329e-05,
      "loss": 0.1052,
      "step": 15917
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24767379225672279,
      "learning_rate": 1.0890028069175764e-05,
      "loss": 0.2339,
      "step": 15918
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3433318210341794,
      "learning_rate": 1.0889040117698807e-05,
      "loss": 0.0728,
      "step": 15919
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34481252730395096,
      "learning_rate": 1.088805215747518e-05,
      "loss": 0.3045,
      "step": 15920
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6871927669527046,
      "learning_rate": 1.0887064188514607e-05,
      "loss": 0.3264,
      "step": 15921
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5942325123388787,
      "learning_rate": 1.0886076210826804e-05,
      "loss": 0.3909,
      "step": 15922
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4538493026355918,
      "learning_rate": 1.088508822442149e-05,
      "loss": 0.1577,
      "step": 15923
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32121366239818006,
      "learning_rate": 1.0884100229308388e-05,
      "loss": 0.2462,
      "step": 15924
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2873228169078555,
      "learning_rate": 1.0883112225497217e-05,
      "loss": 0.2342,
      "step": 15925
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25520288586782114,
      "learning_rate": 1.0882124212997696e-05,
      "loss": 0.1564,
      "step": 15926
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42997630609052845,
      "learning_rate": 1.0881136191819547e-05,
      "loss": 0.293,
      "step": 15927
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39654073566139614,
      "learning_rate": 1.0880148161972492e-05,
      "loss": 0.0729,
      "step": 15928
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3794922683274986,
      "learning_rate": 1.0879160123466248e-05,
      "loss": 0.3027,
      "step": 15929
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4677788210082907,
      "learning_rate": 1.0878172076310539e-05,
      "loss": 0.3024,
      "step": 15930
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4513416330791746,
      "learning_rate": 1.0877184020515083e-05,
      "loss": 0.3526,
      "step": 15931
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8919359001810729,
      "learning_rate": 1.0876195956089602e-05,
      "loss": 0.3628,
      "step": 15932
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7123460991631171,
      "learning_rate": 1.0875207883043817e-05,
      "loss": 0.3355,
      "step": 15933
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3259127726611395,
      "learning_rate": 1.0874219801387447e-05,
      "loss": 0.243,
      "step": 15934
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5202813567291078,
      "learning_rate": 1.0873231711130218e-05,
      "loss": 0.3237,
      "step": 15935
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.30604413636141137,
      "learning_rate": 1.0872243612281847e-05,
      "loss": 0.2617,
      "step": 15936
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4063214287557909,
      "learning_rate": 1.0871255504852054e-05,
      "loss": 0.259,
      "step": 15937
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2748826901052315,
      "learning_rate": 1.0870267388850566e-05,
      "loss": 0.2315,
      "step": 15938
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4713106203610372,
      "learning_rate": 1.08692792642871e-05,
      "loss": 0.3345,
      "step": 15939
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.590212170857547,
      "learning_rate": 1.0868291131171381e-05,
      "loss": 0.7917,
      "step": 15940
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33340524338675764,
      "learning_rate": 1.0867302989513124e-05,
      "loss": 0.2105,
      "step": 15941
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.37850243436852676,
      "learning_rate": 1.0866314839322057e-05,
      "loss": 0.2621,
      "step": 15942
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2728827692586501,
      "learning_rate": 1.0865326680607897e-05,
      "loss": 0.2292,
      "step": 15943
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5825605873009676,
      "learning_rate": 1.0864338513380372e-05,
      "loss": 0.8251,
      "step": 15944
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21176727811269266,
      "learning_rate": 1.08633503376492e-05,
      "loss": 0.0918,
      "step": 15945
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5450247229853762,
      "learning_rate": 1.0862362153424102e-05,
      "loss": 0.4022,
      "step": 15946
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28840088155533455,
      "learning_rate": 1.0861373960714801e-05,
      "loss": 0.2177,
      "step": 15947
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8513106249053067,
      "learning_rate": 1.0860385759531021e-05,
      "loss": 0.5049,
      "step": 15948
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32985903251869386,
      "learning_rate": 1.085939754988248e-05,
      "loss": 0.2428,
      "step": 15949
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0070197878261025,
      "learning_rate": 1.0858409331778903e-05,
      "loss": 0.5582,
      "step": 15950
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.31110652209683537,
      "learning_rate": 1.0857421105230018e-05,
      "loss": 0.1772,
      "step": 15951
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9974707738055483,
      "learning_rate": 1.0856432870245537e-05,
      "loss": 0.4277,
      "step": 15952
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9404910832056713,
      "learning_rate": 1.0855444626835187e-05,
      "loss": 0.4249,
      "step": 15953
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2377568322429101,
      "learning_rate": 1.0854456375008698e-05,
      "loss": 0.2114,
      "step": 15954
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4865077631299814,
      "learning_rate": 1.0853468114775778e-05,
      "loss": 0.3242,
      "step": 15955
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3396038834101549,
      "learning_rate": 1.085247984614616e-05,
      "loss": 0.2113,
      "step": 15956
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4133807087981057,
      "learning_rate": 1.0851491569129569e-05,
      "loss": 0.2969,
      "step": 15957
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0412386901116026,
      "learning_rate": 1.0850503283735717e-05,
      "loss": 0.2522,
      "step": 15958
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3230699279498504,
      "learning_rate": 1.0849514989974341e-05,
      "loss": 0.3201,
      "step": 15959
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.27316981186615963,
      "learning_rate": 1.0848526687855152e-05,
      "loss": 0.0718,
      "step": 15960
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4074814502208836,
      "learning_rate": 1.084753837738788e-05,
      "loss": 0.3154,
      "step": 15961
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7211716836258301,
      "learning_rate": 1.0846550058582246e-05,
      "loss": 0.4023,
      "step": 15962
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6504628057051564,
      "learning_rate": 1.0845561731447973e-05,
      "loss": 0.4353,
      "step": 15963
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6772274185362648,
      "learning_rate": 1.0844573395994788e-05,
      "loss": 0.3293,
      "step": 15964
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39497128320155517,
      "learning_rate": 1.0843585052232413e-05,
      "loss": 0.3024,
      "step": 15965
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21228817462284802,
      "learning_rate": 1.0842596700170568e-05,
      "loss": 0.2018,
      "step": 15966
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21146087366266653,
      "learning_rate": 1.0841608339818981e-05,
      "loss": 0.0666,
      "step": 15967
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5033023429963694,
      "learning_rate": 1.0840619971187372e-05,
      "loss": 0.7499,
      "step": 15968
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29171785235323405,
      "learning_rate": 1.0839631594285467e-05,
      "loss": 0.181,
      "step": 15969
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4742386341774756,
      "learning_rate": 1.0838643209122993e-05,
      "loss": 0.362,
      "step": 15970
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0471101049656504,
      "learning_rate": 1.0837654815709668e-05,
      "loss": 0.2953,
      "step": 15971
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.30775793990527367,
      "learning_rate": 1.0836666414055223e-05,
      "loss": 0.3089,
      "step": 15972
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.621094160852911,
      "learning_rate": 1.0835678004169374e-05,
      "loss": 0.2922,
      "step": 15973
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3509386685609508,
      "learning_rate": 1.0834689586061855e-05,
      "loss": 0.2933,
      "step": 15974
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.23829540317615713,
      "learning_rate": 1.0833701159742381e-05,
      "loss": 0.0714,
      "step": 15975
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5023901220322543,
      "learning_rate": 1.083271272522068e-05,
      "loss": 0.2858,
      "step": 15976
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2853165375603278,
      "learning_rate": 1.0831724282506481e-05,
      "loss": 0.2582,
      "step": 15977
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6377882809319059,
      "learning_rate": 1.0830735831609504e-05,
      "loss": 0.3885,
      "step": 15978
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3186293390550223,
      "learning_rate": 1.0829747372539474e-05,
      "loss": 0.2193,
      "step": 15979
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.805053024260271,
      "learning_rate": 1.0828758905306116e-05,
      "loss": 0.3706,
      "step": 15980
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7243513014777065,
      "learning_rate": 1.0827770429919156e-05,
      "loss": 0.4228,
      "step": 15981
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6241506710624235,
      "learning_rate": 1.0826781946388317e-05,
      "loss": 0.2017,
      "step": 15982
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9044580682243688,
      "learning_rate": 1.0825793454723325e-05,
      "loss": 0.5723,
      "step": 15983
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.24961021408656003,
      "learning_rate": 1.0824804954933907e-05,
      "loss": 0.222,
      "step": 15984
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26510138748960266,
      "learning_rate": 1.0823816447029785e-05,
      "loss": 0.1837,
      "step": 15985
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.896565838452202,
      "learning_rate": 1.0822827931020683e-05,
      "loss": 0.4145,
      "step": 15986
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8810896465121998,
      "learning_rate": 1.0821839406916334e-05,
      "loss": 0.4146,
      "step": 15987
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2825374211750545,
      "learning_rate": 1.0820850874726456e-05,
      "loss": 0.2101,
      "step": 15988
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9922335868648281,
      "learning_rate": 1.0819862334460775e-05,
      "loss": 0.4556,
      "step": 15989
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3483162978275282,
      "learning_rate": 1.0818873786129022e-05,
      "loss": 0.251,
      "step": 15990
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7832228647790405,
      "learning_rate": 1.0817885229740922e-05,
      "loss": 0.8872,
      "step": 15991
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2814627393516033,
      "learning_rate": 1.0816896665306191e-05,
      "loss": 0.1576,
      "step": 15992
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.268136692997195,
      "learning_rate": 1.0815908092834568e-05,
      "loss": 0.1608,
      "step": 15993
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4630287576801482,
      "learning_rate": 1.0814919512335768e-05,
      "loss": 0.2453,
      "step": 15994
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.38041931059024525,
      "learning_rate": 1.0813930923819522e-05,
      "loss": 0.2596,
      "step": 15995
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5261319485557604,
      "learning_rate": 1.081294232729556e-05,
      "loss": 0.3399,
      "step": 15996
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29860096810194436,
      "learning_rate": 1.0811953722773602e-05,
      "loss": 0.2151,
      "step": 15997
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9722658402360084,
      "learning_rate": 1.0810965110263374e-05,
      "loss": 0.5723,
      "step": 15998
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8687923800004609,
      "learning_rate": 1.0809976489774605e-05,
      "loss": 0.359,
      "step": 15999
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7111717125631885,
      "learning_rate": 1.0808987861317022e-05,
      "loss": 0.7538,
      "step": 16000
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25273622555899233,
      "learning_rate": 1.0807999224900349e-05,
      "loss": 0.1811,
      "step": 16001
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.47784304596339405,
      "learning_rate": 1.0807010580534316e-05,
      "loss": 0.3053,
      "step": 16002
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.181722529764978,
      "learning_rate": 1.0806021928228647e-05,
      "loss": 0.1331,
      "step": 16003
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0205898384715728,
      "learning_rate": 1.0805033267993065e-05,
      "loss": 0.6702,
      "step": 16004
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5255991293030837,
      "learning_rate": 1.0804044599837304e-05,
      "loss": 0.2238,
      "step": 16005
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.338190864179667,
      "learning_rate": 1.0803055923771088e-05,
      "loss": 0.2213,
      "step": 16006
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8009666051195115,
      "learning_rate": 1.0802067239804144e-05,
      "loss": 0.4435,
      "step": 16007
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2886502202898307,
      "learning_rate": 1.0801078547946195e-05,
      "loss": 0.2575,
      "step": 16008
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.191717755461131,
      "learning_rate": 1.0800089848206976e-05,
      "loss": 0.6344,
      "step": 16009
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3386751968920034,
      "learning_rate": 1.079910114059621e-05,
      "loss": 0.0721,
      "step": 16010
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4315303056746997,
      "learning_rate": 1.0798112425123619e-05,
      "loss": 0.3088,
      "step": 16011
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.15765559015653588,
      "learning_rate": 1.0797123701798937e-05,
      "loss": 0.0858,
      "step": 16012
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3442612228913831,
      "learning_rate": 1.0796134970631893e-05,
      "loss": 0.3244,
      "step": 16013
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1124639377527707,
      "learning_rate": 1.079514623163221e-05,
      "loss": 0.3427,
      "step": 16014
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39006080100004203,
      "learning_rate": 1.0794157484809616e-05,
      "loss": 0.2789,
      "step": 16015
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4362435252220861,
      "learning_rate": 1.0793168730173842e-05,
      "loss": 0.2802,
      "step": 16016
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1824425906613343,
      "learning_rate": 1.0792179967734608e-05,
      "loss": 0.7693,
      "step": 16017
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0518041973301753,
      "learning_rate": 1.0791191197501649e-05,
      "loss": 0.2701,
      "step": 16018
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26096596622644086,
      "learning_rate": 1.0790202419484693e-05,
      "loss": 0.1924,
      "step": 16019
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3558303636297382,
      "learning_rate": 1.078921363369346e-05,
      "loss": 0.3043,
      "step": 16020
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.17211537204385566,
      "learning_rate": 1.0788224840137686e-05,
      "loss": 0.0708,
      "step": 16021
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7121840196122313,
      "learning_rate": 1.07872360388271e-05,
      "loss": 0.8364,
      "step": 16022
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6125140107365696,
      "learning_rate": 1.0786247229771425e-05,
      "loss": 0.3093,
      "step": 16023
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3887091610161569,
      "learning_rate": 1.078525841298039e-05,
      "loss": 0.2893,
      "step": 16024
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7693503309773756,
      "learning_rate": 1.0784269588463726e-05,
      "loss": 0.3526,
      "step": 16025
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3385816305019963,
      "learning_rate": 1.0783280756231158e-05,
      "loss": 0.2999,
      "step": 16026
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9274023680984506,
      "learning_rate": 1.0782291916292416e-05,
      "loss": 0.2216,
      "step": 16027
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5616977012287988,
      "learning_rate": 1.0781303068657232e-05,
      "loss": 0.315,
      "step": 16028
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32449334743878067,
      "learning_rate": 1.078031421333533e-05,
      "loss": 0.2196,
      "step": 16029
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4347657032259891,
      "learning_rate": 1.0779325350336436e-05,
      "loss": 0.2621,
      "step": 16030
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.31484673297471283,
      "learning_rate": 1.0778336479670286e-05,
      "loss": 0.2742,
      "step": 16031
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5880876967861408,
      "learning_rate": 1.0777347601346605e-05,
      "loss": 0.253,
      "step": 16032
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5417214840667983,
      "learning_rate": 1.0776358715375124e-05,
      "loss": 0.3701,
      "step": 16033
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.44359554153258796,
      "learning_rate": 1.0775369821765569e-05,
      "loss": 0.289,
      "step": 16034
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.42877270521588795,
      "learning_rate": 1.077438092052767e-05,
      "loss": 0.2724,
      "step": 16035
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3633135425538131,
      "learning_rate": 1.077339201167116e-05,
      "loss": 0.2522,
      "step": 16036
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9349232920624269,
      "learning_rate": 1.0772403095205762e-05,
      "loss": 0.4343,
      "step": 16037
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.23882947835453963,
      "learning_rate": 1.0771414171141206e-05,
      "loss": 0.2086,
      "step": 16038
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.6967948412947178,
      "learning_rate": 1.0770425239487227e-05,
      "loss": 0.7792,
      "step": 16039
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5222403583058166,
      "learning_rate": 1.0769436300253551e-05,
      "loss": 0.3494,
      "step": 16040
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.673162811625369,
      "learning_rate": 1.0768447353449907e-05,
      "loss": 0.8321,
      "step": 16041
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29841804573564507,
      "learning_rate": 1.0767458399086026e-05,
      "loss": 0.1787,
      "step": 16042
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.36301413105296587,
      "learning_rate": 1.0766469437171634e-05,
      "loss": 0.2992,
      "step": 16043
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.21396095970631368,
      "learning_rate": 1.0765480467716465e-05,
      "loss": 0.1498,
      "step": 16044
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0679924347022731,
      "learning_rate": 1.076449149073025e-05,
      "loss": 0.3176,
      "step": 16045
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6237157484082544,
      "learning_rate": 1.0763502506222712e-05,
      "loss": 0.3437,
      "step": 16046
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3300777803530959,
      "learning_rate": 1.0762513514203588e-05,
      "loss": 0.2093,
      "step": 16047
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7540103383033074,
      "learning_rate": 1.0761524514682602e-05,
      "loss": 0.4603,
      "step": 16048
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2918404608837762,
      "learning_rate": 1.0760535507669489e-05,
      "loss": 0.2517,
      "step": 16049
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6493805903301808,
      "learning_rate": 1.0759546493173976e-05,
      "loss": 0.4872,
      "step": 16050
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29068147367459135,
      "learning_rate": 1.0758557471205795e-05,
      "loss": 0.2011,
      "step": 16051
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5706510162480707,
      "learning_rate": 1.075756844177468e-05,
      "loss": 0.9071,
      "step": 16052
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.18074467057898908,
      "learning_rate": 1.0756579404890355e-05,
      "loss": 0.0877,
      "step": 16053
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4723038556064676,
      "learning_rate": 1.075559036056255e-05,
      "loss": 0.5894,
      "step": 16054
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.23946839664878317,
      "learning_rate": 1.0754601308801002e-05,
      "loss": 0.2185,
      "step": 16055
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4002068519531668,
      "learning_rate": 1.0753612249615437e-05,
      "loss": 0.3071,
      "step": 16056
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9456175923351366,
      "learning_rate": 1.0752623183015586e-05,
      "loss": 0.3198,
      "step": 16057
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5629477362493879,
      "learning_rate": 1.0751634109011182e-05,
      "loss": 0.3498,
      "step": 16058
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3027727016350583,
      "learning_rate": 1.0750645027611953e-05,
      "loss": 0.8904,
      "step": 16059
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.25918865185875145,
      "learning_rate": 1.0749655938827632e-05,
      "loss": 0.0735,
      "step": 16060
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2878373308610321,
      "learning_rate": 1.0748666842667947e-05,
      "loss": 0.2653,
      "step": 16061
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26232736305773297,
      "learning_rate": 1.0747677739142635e-05,
      "loss": 0.1772,
      "step": 16062
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43831715127765924,
      "learning_rate": 1.0746688628261419e-05,
      "loss": 0.2576,
      "step": 16063
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6131737824469276,
      "learning_rate": 1.0745699510034038e-05,
      "loss": 0.1665,
      "step": 16064
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3332179595774341,
      "learning_rate": 1.0744710384470218e-05,
      "loss": 0.2555,
      "step": 16065
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9428737775840066,
      "learning_rate": 1.0743721251579694e-05,
      "loss": 0.3587,
      "step": 16066
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.32157864746390025,
      "learning_rate": 1.0742732111372195e-05,
      "loss": 0.318,
      "step": 16067
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9577163452089691,
      "learning_rate": 1.0741742963857453e-05,
      "loss": 0.3121,
      "step": 16068
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5989731835357999,
      "learning_rate": 1.0740753809045199e-05,
      "loss": 0.318,
      "step": 16069
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35575789571376626,
      "learning_rate": 1.0739764646945166e-05,
      "loss": 0.2567,
      "step": 16070
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.16699757874835458,
      "learning_rate": 1.0738775477567085e-05,
      "loss": 0.0714,
      "step": 16071
    },
    {
      "epoch": 0.49,
      "grad_norm": 2.014563742119368,
      "learning_rate": 1.0737786300920686e-05,
      "loss": 0.9078,
      "step": 16072
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2735758743028432,
      "learning_rate": 1.0736797117015705e-05,
      "loss": 0.2206,
      "step": 16073
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.35481648471960714,
      "learning_rate": 1.073580792586187e-05,
      "loss": 0.2899,
      "step": 16074
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.011206348172139,
      "learning_rate": 1.0734818727468916e-05,
      "loss": 0.3633,
      "step": 16075
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0114104518217224,
      "learning_rate": 1.0733829521846573e-05,
      "loss": 0.5169,
      "step": 16076
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8881738531908148,
      "learning_rate": 1.0732840309004571e-05,
      "loss": 0.4064,
      "step": 16077
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39212311210364087,
      "learning_rate": 1.073185108895265e-05,
      "loss": 0.2846,
      "step": 16078
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3242941808007154,
      "learning_rate": 1.0730861861700535e-05,
      "loss": 0.2283,
      "step": 16079
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4514645649672602,
      "learning_rate": 1.0729872627257957e-05,
      "loss": 0.3392,
      "step": 16080
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.19755666214796078,
      "learning_rate": 1.0728883385634656e-05,
      "loss": 0.0984,
      "step": 16081
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6598318427440604,
      "learning_rate": 1.072789413684036e-05,
      "loss": 0.3969,
      "step": 16082
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3285131594745115,
      "learning_rate": 1.07269048808848e-05,
      "loss": 0.2096,
      "step": 16083
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7756942470173046,
      "learning_rate": 1.0725915617777713e-05,
      "loss": 0.3384,
      "step": 16084
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33277908327176536,
      "learning_rate": 1.072492634752883e-05,
      "loss": 0.3176,
      "step": 16085
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8401470841283557,
      "learning_rate": 1.072393707014788e-05,
      "loss": 0.432,
      "step": 16086
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9418761819628791,
      "learning_rate": 1.0722947785644599e-05,
      "loss": 0.5119,
      "step": 16087
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3025451901583966,
      "learning_rate": 1.072195849402872e-05,
      "loss": 0.2172,
      "step": 16088
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5984982109130375,
      "learning_rate": 1.0720969195309978e-05,
      "loss": 0.8295,
      "step": 16089
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.23381827690470394,
      "learning_rate": 1.07199798894981e-05,
      "loss": 0.2155,
      "step": 16090
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0295588008483134,
      "learning_rate": 1.0718990576602828e-05,
      "loss": 0.5324,
      "step": 16091
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7596581232160717,
      "learning_rate": 1.0718001256633887e-05,
      "loss": 0.2137,
      "step": 16092
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29183027706881826,
      "learning_rate": 1.0717011929601014e-05,
      "loss": 0.2552,
      "step": 16093
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8986494817954578,
      "learning_rate": 1.0716022595513944e-05,
      "loss": 0.2792,
      "step": 16094
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4811904554184656,
      "learning_rate": 1.0715033254382404e-05,
      "loss": 0.8502,
      "step": 16095
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2328520644964136,
      "learning_rate": 1.0714043906216132e-05,
      "loss": 0.1644,
      "step": 16096
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2474407459975876,
      "learning_rate": 1.0713054551024865e-05,
      "loss": 0.2391,
      "step": 16097
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2613522863172866,
      "learning_rate": 1.0712065188818331e-05,
      "loss": 0.6793,
      "step": 16098
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.46506594556496644,
      "learning_rate": 1.0711075819606262e-05,
      "loss": 0.1647,
      "step": 16099
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.593386547397781,
      "learning_rate": 1.0710086443398398e-05,
      "loss": 0.4416,
      "step": 16100
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.264130701331256,
      "learning_rate": 1.070909706020447e-05,
      "loss": 0.1844,
      "step": 16101
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3427485939114862,
      "learning_rate": 1.0708107670034211e-05,
      "loss": 0.2634,
      "step": 16102
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39048403486151756,
      "learning_rate": 1.0707118272897357e-05,
      "loss": 0.2454,
      "step": 16103
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5824733526622763,
      "learning_rate": 1.070612886880364e-05,
      "loss": 0.7237,
      "step": 16104
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34403225014559763,
      "learning_rate": 1.0705139457762793e-05,
      "loss": 0.1557,
      "step": 16105
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4396363707587744,
      "learning_rate": 1.0704150039784553e-05,
      "loss": 0.3191,
      "step": 16106
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.5936581880040964,
      "learning_rate": 1.0703160614878655e-05,
      "loss": 0.0319,
      "step": 16107
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39069434621255833,
      "learning_rate": 1.0702171183054831e-05,
      "loss": 0.3309,
      "step": 16108
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3328091026526056,
      "learning_rate": 1.0701181744322814e-05,
      "loss": 0.263,
      "step": 16109
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.29759774626922453,
      "learning_rate": 1.0700192298692343e-05,
      "loss": 0.2077,
      "step": 16110
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4466676796238576,
      "learning_rate": 1.0699202846173149e-05,
      "loss": 0.2252,
      "step": 16111
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34173348473811094,
      "learning_rate": 1.0698213386774965e-05,
      "loss": 0.1493,
      "step": 16112
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.1858762353363743,
      "learning_rate": 1.0697223920507529e-05,
      "loss": 0.7374,
      "step": 16113
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33287661459050066,
      "learning_rate": 1.0696234447380577e-05,
      "loss": 0.2056,
      "step": 16114
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3436459298059857,
      "learning_rate": 1.069524496740384e-05,
      "loss": 0.2712,
      "step": 16115
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34341962805592446,
      "learning_rate": 1.0694255480587055e-05,
      "loss": 0.2632,
      "step": 16116
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7549163722147664,
      "learning_rate": 1.0693265986939957e-05,
      "loss": 0.4108,
      "step": 16117
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9760429154632493,
      "learning_rate": 1.0692276486472276e-05,
      "loss": 0.2928,
      "step": 16118
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.31839888836722247,
      "learning_rate": 1.0691286979193752e-05,
      "loss": 0.1899,
      "step": 16119
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.22981022027952594,
      "learning_rate": 1.0690297465114121e-05,
      "loss": 0.1645,
      "step": 16120
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3317895001259787,
      "learning_rate": 1.0689307944243114e-05,
      "loss": 0.2958,
      "step": 16121
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.2563945457853345,
      "learning_rate": 1.0688318416590471e-05,
      "loss": 0.3555,
      "step": 16122
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.39536448449544476,
      "learning_rate": 1.0687328882165928e-05,
      "loss": 0.2623,
      "step": 16123
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33205712121938585,
      "learning_rate": 1.0686339340979213e-05,
      "loss": 0.2896,
      "step": 16124
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.715935505454854,
      "learning_rate": 1.0685349793040065e-05,
      "loss": 0.3269,
      "step": 16125
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.47362614449043383,
      "learning_rate": 1.068436023835822e-05,
      "loss": 0.3492,
      "step": 16126
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3554650384874627,
      "learning_rate": 1.0683370676943417e-05,
      "loss": 0.2543,
      "step": 16127
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3855238733906965,
      "learning_rate": 1.0682381108805386e-05,
      "loss": 0.2868,
      "step": 16128
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2034389700122651,
      "learning_rate": 1.0681391533953867e-05,
      "loss": 0.1282,
      "step": 16129
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.3245208540634248,
      "learning_rate": 1.0680401952398596e-05,
      "loss": 0.7469,
      "step": 16130
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9145169038650632,
      "learning_rate": 1.0679412364149302e-05,
      "loss": 0.5667,
      "step": 16131
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.33981749010956314,
      "learning_rate": 1.0678422769215727e-05,
      "loss": 0.2855,
      "step": 16132
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2898534838815672,
      "learning_rate": 1.0677433167607604e-05,
      "loss": 0.2154,
      "step": 16133
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.009985225126437,
      "learning_rate": 1.0676443559334672e-05,
      "loss": 0.5774,
      "step": 16134
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5132503386917544,
      "learning_rate": 1.0675453944406668e-05,
      "loss": 0.3348,
      "step": 16135
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.0112524306786745,
      "learning_rate": 1.0674464322833327e-05,
      "loss": 0.2812,
      "step": 16136
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6472503531167614,
      "learning_rate": 1.0673474694624379e-05,
      "loss": 0.2828,
      "step": 16137
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2344084273060313,
      "learning_rate": 1.0672485059789566e-05,
      "loss": 0.1702,
      "step": 16138
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2849515947905374,
      "learning_rate": 1.0671495418338625e-05,
      "loss": 0.2474,
      "step": 16139
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.8745319787148252,
      "learning_rate": 1.0670505770281292e-05,
      "loss": 0.4844,
      "step": 16140
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7012689825452845,
      "learning_rate": 1.0669516115627301e-05,
      "loss": 0.3963,
      "step": 16141
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2800231137215079,
      "learning_rate": 1.0668526454386392e-05,
      "loss": 0.196,
      "step": 16142
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.7361412621473843,
      "learning_rate": 1.0667536786568299e-05,
      "loss": 0.4366,
      "step": 16143
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28205753608474365,
      "learning_rate": 1.0666547112182758e-05,
      "loss": 0.2679,
      "step": 16144
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.7963838377505799,
      "learning_rate": 1.0665557431239509e-05,
      "loss": 0.755,
      "step": 16145
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.43111914525125633,
      "learning_rate": 1.0664567743748287e-05,
      "loss": 0.1293,
      "step": 16146
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.34264402080927936,
      "learning_rate": 1.0663578049718825e-05,
      "loss": 0.2899,
      "step": 16147
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.17904875356213565,
      "learning_rate": 1.066258834916087e-05,
      "loss": 0.1062,
      "step": 16148
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.4524826000047413,
      "learning_rate": 1.066159864208415e-05,
      "loss": 0.8459,
      "step": 16149
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.28202621490702673,
      "learning_rate": 1.0660608928498404e-05,
      "loss": 0.2571,
      "step": 16150
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2759221977084223,
      "learning_rate": 1.065961920841337e-05,
      "loss": 0.1911,
      "step": 16151
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.6956977757395821,
      "learning_rate": 1.0658629481838784e-05,
      "loss": 0.4275,
      "step": 16152
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9937246154924129,
      "learning_rate": 1.0657639748784389e-05,
      "loss": 0.3449,
      "step": 16153
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.5345286233643554,
      "learning_rate": 1.0656650009259917e-05,
      "loss": 0.7071,
      "step": 16154
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2828611589464925,
      "learning_rate": 1.0655660263275103e-05,
      "loss": 0.0731,
      "step": 16155
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.2850137941350817,
      "learning_rate": 1.0654670510839688e-05,
      "loss": 0.2716,
      "step": 16156
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.26369801082915917,
      "learning_rate": 1.065368075196341e-05,
      "loss": 0.1656,
      "step": 16157
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.503718694753189,
      "learning_rate": 1.0652690986656005e-05,
      "loss": 0.3117,
      "step": 16158
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4434401246669143,
      "learning_rate": 1.0651701214927212e-05,
      "loss": 0.3015,
      "step": 16159
    },
    {
      "epoch": 0.49,
      "grad_norm": 1.967224214917152,
      "learning_rate": 1.0650711436786767e-05,
      "loss": 0.3019,
      "step": 16160
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.9046158390928756,
      "learning_rate": 1.064972165224441e-05,
      "loss": 0.3496,
      "step": 16161
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.4833576488050897,
      "learning_rate": 1.0648731861309877e-05,
      "loss": 0.4023,
      "step": 16162
    },
    {
      "epoch": 0.49,
      "grad_norm": 0.3756644239747899,
      "learning_rate": 1.064774206399291e-05,
      "loss": 0.2626,
      "step": 16163
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.38888949689064123,
      "learning_rate": 1.0646752260303236e-05,
      "loss": 0.1114,
      "step": 16164
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3991031394221115,
      "learning_rate": 1.0645762450250605e-05,
      "loss": 0.2914,
      "step": 16165
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1678790244884316,
      "learning_rate": 1.0644772633844751e-05,
      "loss": 0.0872,
      "step": 16166
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.447003860037341,
      "learning_rate": 1.0643782811095413e-05,
      "loss": 0.3617,
      "step": 16167
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4887953090609302,
      "learning_rate": 1.0642792982012324e-05,
      "loss": 0.2889,
      "step": 16168
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6857926772451501,
      "learning_rate": 1.064180314660523e-05,
      "loss": 0.3554,
      "step": 16169
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3382970942183838,
      "learning_rate": 1.0640813304883862e-05,
      "loss": 0.2464,
      "step": 16170
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.1314569520290612,
      "learning_rate": 1.0639823456857963e-05,
      "loss": 0.7156,
      "step": 16171
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0110467667740404,
      "learning_rate": 1.063883360253727e-05,
      "loss": 0.2515,
      "step": 16172
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9472426133369353,
      "learning_rate": 1.0637843741931525e-05,
      "loss": 0.5093,
      "step": 16173
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25925755772372766,
      "learning_rate": 1.0636853875050459e-05,
      "loss": 0.22,
      "step": 16174
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3340632421902797,
      "learning_rate": 1.0635864001903816e-05,
      "loss": 0.2399,
      "step": 16175
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.656878057753719,
      "learning_rate": 1.0634874122501335e-05,
      "loss": 0.3411,
      "step": 16176
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4772953260620549,
      "learning_rate": 1.0633884236852751e-05,
      "loss": 0.3365,
      "step": 16177
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40768108503810785,
      "learning_rate": 1.0632894344967808e-05,
      "loss": 0.3101,
      "step": 16178
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.26618922467789785,
      "learning_rate": 1.0631904446856243e-05,
      "loss": 0.1744,
      "step": 16179
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4643335876075556,
      "learning_rate": 1.0630914542527793e-05,
      "loss": 0.3082,
      "step": 16180
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37276427355275193,
      "learning_rate": 1.0629924631992197e-05,
      "loss": 0.2351,
      "step": 16181
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7878441560477497,
      "learning_rate": 1.0628934715259195e-05,
      "loss": 0.459,
      "step": 16182
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28723750900415734,
      "learning_rate": 1.0627944792338526e-05,
      "loss": 0.2082,
      "step": 16183
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.81053234793209,
      "learning_rate": 1.0626954863239931e-05,
      "loss": 0.909,
      "step": 16184
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4468194968301965,
      "learning_rate": 1.0625964927973148e-05,
      "loss": 0.2989,
      "step": 16185
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43514924845532327,
      "learning_rate": 1.0624974986547918e-05,
      "loss": 0.3599,
      "step": 16186
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28857563390892665,
      "learning_rate": 1.0623985038973974e-05,
      "loss": 0.1921,
      "step": 16187
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5699525922763535,
      "learning_rate": 1.062299508526106e-05,
      "loss": 0.8153,
      "step": 16188
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20048627213014075,
      "learning_rate": 1.062200512541892e-05,
      "loss": 0.1041,
      "step": 16189
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3083806605979846,
      "learning_rate": 1.0621015159457284e-05,
      "loss": 0.2314,
      "step": 16190
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35002080175950756,
      "learning_rate": 1.0620025187385899e-05,
      "loss": 0.2761,
      "step": 16191
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3142169717480714,
      "learning_rate": 1.0619035209214504e-05,
      "loss": 0.1997,
      "step": 16192
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.438868735854478,
      "learning_rate": 1.0618045224952834e-05,
      "loss": 0.3339,
      "step": 16193
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.776314875825636,
      "learning_rate": 1.061705523461063e-05,
      "loss": 0.3349,
      "step": 16194
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.490422714790661,
      "learning_rate": 1.061606523819764e-05,
      "loss": 0.7715,
      "step": 16195
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2683902749404531,
      "learning_rate": 1.061507523572359e-05,
      "loss": 0.0695,
      "step": 16196
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.39083973883423784,
      "learning_rate": 1.0614085227198232e-05,
      "loss": 0.3177,
      "step": 16197
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.17516531620297845,
      "learning_rate": 1.0613095212631302e-05,
      "loss": 0.1516,
      "step": 16198
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.999499729434849,
      "learning_rate": 1.0612105192032539e-05,
      "loss": 0.5495,
      "step": 16199
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7113703027160289,
      "learning_rate": 1.061111516541168e-05,
      "loss": 0.0499,
      "step": 16200
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3710211889530208,
      "learning_rate": 1.0610125132778471e-05,
      "loss": 0.2461,
      "step": 16201
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45895115270949827,
      "learning_rate": 1.0609135094142652e-05,
      "loss": 0.2899,
      "step": 16202
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7821952951002057,
      "learning_rate": 1.060814504951396e-05,
      "loss": 0.3027,
      "step": 16203
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33714025527461355,
      "learning_rate": 1.0607154998902138e-05,
      "loss": 0.3185,
      "step": 16204
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34388819494619083,
      "learning_rate": 1.0606164942316925e-05,
      "loss": 0.1876,
      "step": 16205
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5295669362146883,
      "learning_rate": 1.060517487976806e-05,
      "loss": 0.3988,
      "step": 16206
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24310636020320575,
      "learning_rate": 1.0604184811265286e-05,
      "loss": 0.0716,
      "step": 16207
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36364082850596746,
      "learning_rate": 1.0603194736818346e-05,
      "loss": 0.2133,
      "step": 16208
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2547852597019363,
      "learning_rate": 1.0602204656436974e-05,
      "loss": 0.224,
      "step": 16209
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35695519954691873,
      "learning_rate": 1.0601214570130915e-05,
      "loss": 0.2885,
      "step": 16210
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9504284634235948,
      "learning_rate": 1.060022447790991e-05,
      "loss": 0.3619,
      "step": 16211
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.726664464009545,
      "learning_rate": 1.0599234379783703e-05,
      "loss": 0.459,
      "step": 16212
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8473798784080288,
      "learning_rate": 1.0598244275762023e-05,
      "loss": 0.4482,
      "step": 16213
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5696740250440768,
      "learning_rate": 1.0597254165854625e-05,
      "loss": 0.3247,
      "step": 16214
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.372643503396287,
      "learning_rate": 1.0596264050071243e-05,
      "loss": 0.2277,
      "step": 16215
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2193272220858657,
      "learning_rate": 1.0595273928421617e-05,
      "loss": 0.1999,
      "step": 16216
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42561296354720773,
      "learning_rate": 1.0594283800915493e-05,
      "loss": 0.2873,
      "step": 16217
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7273037697659454,
      "learning_rate": 1.059329366756261e-05,
      "loss": 0.2634,
      "step": 16218
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5226778316284839,
      "learning_rate": 1.0592303528372705e-05,
      "loss": 0.3416,
      "step": 16219
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5062423816667053,
      "learning_rate": 1.0591313383355523e-05,
      "loss": 0.2979,
      "step": 16220
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4259444071750577,
      "learning_rate": 1.0590323232520807e-05,
      "loss": 0.3439,
      "step": 16221
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35648140930663674,
      "learning_rate": 1.0589333075878296e-05,
      "loss": 0.2954,
      "step": 16222
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0113507152114862,
      "learning_rate": 1.0588342913437735e-05,
      "loss": 0.4068,
      "step": 16223
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28560499532689104,
      "learning_rate": 1.0587352745208864e-05,
      "loss": 0.2153,
      "step": 16224
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42695246526092695,
      "learning_rate": 1.0586362571201418e-05,
      "loss": 0.2367,
      "step": 16225
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29753375852843994,
      "learning_rate": 1.0585372391425143e-05,
      "loss": 0.1692,
      "step": 16226
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34145609549716366,
      "learning_rate": 1.0584382205889785e-05,
      "loss": 0.3225,
      "step": 16227
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28281589352243347,
      "learning_rate": 1.0583392014605085e-05,
      "loss": 0.2091,
      "step": 16228
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36657222219959795,
      "learning_rate": 1.0582401817580778e-05,
      "loss": 0.2866,
      "step": 16229
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6622090790382587,
      "learning_rate": 1.0581411614826615e-05,
      "loss": 0.8036,
      "step": 16230
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.790967427783055,
      "learning_rate": 1.0580421406352332e-05,
      "loss": 0.5469,
      "step": 16231
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7458119046808772,
      "learning_rate": 1.0579431192167668e-05,
      "loss": 0.4199,
      "step": 16232
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23807120374262636,
      "learning_rate": 1.0578440972282374e-05,
      "loss": 0.2182,
      "step": 16233
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.45608489842779676,
      "learning_rate": 1.0577450746706186e-05,
      "loss": 0.3766,
      "step": 16234
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28038397573374985,
      "learning_rate": 1.0576460515448846e-05,
      "loss": 0.1663,
      "step": 16235
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.501963963818201,
      "learning_rate": 1.0575470278520098e-05,
      "loss": 0.44,
      "step": 16236
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28864919665290173,
      "learning_rate": 1.0574480035929688e-05,
      "loss": 0.1889,
      "step": 16237
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.47324615946235904,
      "learning_rate": 1.057348978768735e-05,
      "loss": 0.2658,
      "step": 16238
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4015652171963936,
      "learning_rate": 1.0572499533802831e-05,
      "loss": 0.2541,
      "step": 16239
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.470563409147354,
      "learning_rate": 1.0571509274285874e-05,
      "loss": 0.3729,
      "step": 16240
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31474510522023447,
      "learning_rate": 1.0570519009146221e-05,
      "loss": 0.1514,
      "step": 16241
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37758867955372616,
      "learning_rate": 1.0569528738393614e-05,
      "loss": 0.2238,
      "step": 16242
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.886111170760862,
      "learning_rate": 1.0568538462037797e-05,
      "loss": 0.456,
      "step": 16243
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5740557109958506,
      "learning_rate": 1.0567548180088511e-05,
      "loss": 0.3349,
      "step": 16244
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2605606507700823,
      "learning_rate": 1.0566557892555499e-05,
      "loss": 0.2356,
      "step": 16245
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2312721022364303,
      "learning_rate": 1.0565567599448502e-05,
      "loss": 0.0702,
      "step": 16246
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2744530972829374,
      "learning_rate": 1.0564577300777269e-05,
      "loss": 0.2381,
      "step": 16247
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.065663813625205,
      "learning_rate": 1.0563586996551535e-05,
      "loss": 0.3486,
      "step": 16248
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3733687472518987,
      "learning_rate": 1.0562596686781051e-05,
      "loss": 0.7198,
      "step": 16249
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3716041804338265,
      "learning_rate": 1.0561606371475552e-05,
      "loss": 0.0685,
      "step": 16250
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33650682897705964,
      "learning_rate": 1.0560616050644785e-05,
      "loss": 0.2951,
      "step": 16251
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28113760519412184,
      "learning_rate": 1.0559625724298493e-05,
      "loss": 0.2318,
      "step": 16252
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5652253974214745,
      "learning_rate": 1.0558635392446417e-05,
      "loss": 0.4306,
      "step": 16253
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5602794937017651,
      "learning_rate": 1.0557645055098307e-05,
      "loss": 0.2895,
      "step": 16254
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.17936542423240726,
      "learning_rate": 1.0556654712263901e-05,
      "loss": 0.116,
      "step": 16255
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31246095684381,
      "learning_rate": 1.055566436395294e-05,
      "loss": 0.2298,
      "step": 16256
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3110722345245823,
      "learning_rate": 1.0554674010175173e-05,
      "loss": 0.2426,
      "step": 16257
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4556905380797944,
      "learning_rate": 1.0553683650940336e-05,
      "loss": 0.7776,
      "step": 16258
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41726868095020353,
      "learning_rate": 1.0552693286258178e-05,
      "loss": 0.2058,
      "step": 16259
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35311036775081883,
      "learning_rate": 1.0551702916138445e-05,
      "loss": 0.2777,
      "step": 16260
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8391849712961632,
      "learning_rate": 1.0550712540590876e-05,
      "loss": 0.3322,
      "step": 16261
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7068566867080254,
      "learning_rate": 1.0549722159625216e-05,
      "loss": 0.4453,
      "step": 16262
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2809409309648563,
      "learning_rate": 1.0548731773251207e-05,
      "loss": 0.2313,
      "step": 16263
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3070968598408545,
      "learning_rate": 1.0547741381478594e-05,
      "loss": 0.1518,
      "step": 16264
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3605251597848719,
      "learning_rate": 1.0546750984317122e-05,
      "loss": 0.2148,
      "step": 16265
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4428347174736228,
      "learning_rate": 1.0545760581776532e-05,
      "loss": 0.2632,
      "step": 16266
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.789031037719066,
      "learning_rate": 1.0544770173866574e-05,
      "loss": 0.6003,
      "step": 16267
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2855287903111234,
      "learning_rate": 1.0543779760596985e-05,
      "loss": 0.1974,
      "step": 16268
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5577234322877135,
      "learning_rate": 1.0542789341977513e-05,
      "loss": 0.3841,
      "step": 16269
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4437758882932322,
      "learning_rate": 1.05417989180179e-05,
      "loss": 0.2759,
      "step": 16270
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9962897329262889,
      "learning_rate": 1.0540808488727891e-05,
      "loss": 0.4863,
      "step": 16271
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29059347105885813,
      "learning_rate": 1.0539818054117231e-05,
      "loss": 0.1002,
      "step": 16272
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.263015691742613,
      "learning_rate": 1.053882761419566e-05,
      "loss": 0.475,
      "step": 16273
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.238044265442355,
      "learning_rate": 1.0537837168972931e-05,
      "loss": 0.1643,
      "step": 16274
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3180871551467499,
      "learning_rate": 1.053684671845878e-05,
      "loss": 0.2703,
      "step": 16275
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8930202163328094,
      "learning_rate": 1.0535856262662955e-05,
      "loss": 0.619,
      "step": 16276
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.183971790556153,
      "learning_rate": 1.05348658015952e-05,
      "loss": 0.6173,
      "step": 16277
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.332782690074579,
      "learning_rate": 1.0533875335265255e-05,
      "loss": 0.2059,
      "step": 16278
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4573125644106012,
      "learning_rate": 1.0532884863682872e-05,
      "loss": 0.3677,
      "step": 16279
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34897203355020373,
      "learning_rate": 1.0531894386857794e-05,
      "loss": 0.28,
      "step": 16280
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36292834500110877,
      "learning_rate": 1.0530903904799765e-05,
      "loss": 0.2119,
      "step": 16281
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5368194202312611,
      "learning_rate": 1.0529913417518524e-05,
      "loss": 0.2804,
      "step": 16282
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23040985776585654,
      "learning_rate": 1.0528922925023826e-05,
      "loss": 0.173,
      "step": 16283
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5232874129404754,
      "learning_rate": 1.0527932427325405e-05,
      "loss": 0.3248,
      "step": 16284
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8722592896256942,
      "learning_rate": 1.0526941924433012e-05,
      "loss": 0.4794,
      "step": 16285
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.44971962874462124,
      "learning_rate": 1.0525951416356396e-05,
      "loss": 0.3695,
      "step": 16286
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2507379258383869,
      "learning_rate": 1.0524960903105292e-05,
      "loss": 0.1834,
      "step": 16287
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36917602629297747,
      "learning_rate": 1.052397038468945e-05,
      "loss": 0.3611,
      "step": 16288
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6832059315219743,
      "learning_rate": 1.0522979861118616e-05,
      "loss": 0.0268,
      "step": 16289
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.026996001901711,
      "learning_rate": 1.0521989332402534e-05,
      "loss": 0.6392,
      "step": 16290
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2834507895174005,
      "learning_rate": 1.0520998798550947e-05,
      "loss": 0.0964,
      "step": 16291
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.37617095817330615,
      "learning_rate": 1.0520008259573604e-05,
      "loss": 0.2916,
      "step": 16292
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2415185442823119,
      "learning_rate": 1.051901771548025e-05,
      "loss": 0.1879,
      "step": 16293
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7390223434191915,
      "learning_rate": 1.0518027166280632e-05,
      "loss": 0.4295,
      "step": 16294
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5076175403190616,
      "learning_rate": 1.0517036611984485e-05,
      "loss": 0.3994,
      "step": 16295
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36010437002106266,
      "learning_rate": 1.0516046052601567e-05,
      "loss": 0.0724,
      "step": 16296
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36833701870175195,
      "learning_rate": 1.0515055488141615e-05,
      "loss": 0.3065,
      "step": 16297
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3879104292536954,
      "learning_rate": 1.0514064918614377e-05,
      "loss": 0.2684,
      "step": 16298
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4745696271741637,
      "learning_rate": 1.0513074344029604e-05,
      "loss": 0.3585,
      "step": 16299
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5069781321500465,
      "learning_rate": 1.0512083764397034e-05,
      "loss": 0.1386,
      "step": 16300
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3204662663287165,
      "learning_rate": 1.0511093179726415e-05,
      "loss": 0.2804,
      "step": 16301
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2810661972983747,
      "learning_rate": 1.0510102590027491e-05,
      "loss": 0.1447,
      "step": 16302
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7016061173640182,
      "learning_rate": 1.0509111995310014e-05,
      "loss": 0.4839,
      "step": 16303
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36899210894678336,
      "learning_rate": 1.0508121395583723e-05,
      "loss": 0.256,
      "step": 16304
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3581777765165731,
      "learning_rate": 1.0507130790858368e-05,
      "loss": 0.2938,
      "step": 16305
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3098620234682837,
      "learning_rate": 1.0506140181143693e-05,
      "loss": 0.246,
      "step": 16306
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3668733866018604,
      "learning_rate": 1.0505149566449444e-05,
      "loss": 0.132,
      "step": 16307
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6133217691590378,
      "learning_rate": 1.0504158946785368e-05,
      "loss": 0.7724,
      "step": 16308
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2289593819657389,
      "learning_rate": 1.050316832216121e-05,
      "loss": 0.0921,
      "step": 16309
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3552493086558124,
      "learning_rate": 1.0502177692586714e-05,
      "loss": 0.2976,
      "step": 16310
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28287220983410033,
      "learning_rate": 1.0501187058071632e-05,
      "loss": 0.2545,
      "step": 16311
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9148979932099766,
      "learning_rate": 1.0500196418625708e-05,
      "loss": 0.5283,
      "step": 16312
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.600295523489238,
      "learning_rate": 1.0499205774258685e-05,
      "loss": 0.3289,
      "step": 16313
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43398919292932686,
      "learning_rate": 1.0498215124980311e-05,
      "loss": 0.2725,
      "step": 16314
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23897341431810207,
      "learning_rate": 1.0497224470800332e-05,
      "loss": 0.1481,
      "step": 16315
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.328119305218902,
      "learning_rate": 1.0496233811728495e-05,
      "loss": 0.7057,
      "step": 16316
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34156443149512616,
      "learning_rate": 1.0495243147774547e-05,
      "loss": 0.247,
      "step": 16317
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8891023457041776,
      "learning_rate": 1.0494252478948237e-05,
      "loss": 0.3752,
      "step": 16318
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29100255677737513,
      "learning_rate": 1.0493261805259305e-05,
      "loss": 0.2052,
      "step": 16319
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8714887753629074,
      "learning_rate": 1.0492271126717501e-05,
      "loss": 0.3632,
      "step": 16320
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8055517371446165,
      "learning_rate": 1.0491280443332572e-05,
      "loss": 0.4174,
      "step": 16321
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2909100873715404,
      "learning_rate": 1.0490289755114267e-05,
      "loss": 0.2396,
      "step": 16322
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6062992771641705,
      "learning_rate": 1.0489299062072326e-05,
      "loss": 0.2759,
      "step": 16323
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.321797080107486,
      "learning_rate": 1.0488308364216504e-05,
      "loss": 0.2202,
      "step": 16324
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.27211870865272697,
      "learning_rate": 1.0487317661556542e-05,
      "loss": 0.1624,
      "step": 16325
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0878375137742198,
      "learning_rate": 1.0486326954102187e-05,
      "loss": 0.3056,
      "step": 16326
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.2839081797824687,
      "learning_rate": 1.0485336241863187e-05,
      "loss": 0.7696,
      "step": 16327
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29533254523356717,
      "learning_rate": 1.048434552484929e-05,
      "loss": 0.1614,
      "step": 16328
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.27263274771602514,
      "learning_rate": 1.0483354803070244e-05,
      "loss": 0.2779,
      "step": 16329
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0579095109574952,
      "learning_rate": 1.0482364076535794e-05,
      "loss": 0.3944,
      "step": 16330
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.5537533750322523,
      "learning_rate": 1.0481373345255688e-05,
      "loss": 0.8326,
      "step": 16331
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3135780824343752,
      "learning_rate": 1.0480382609239672e-05,
      "loss": 0.1643,
      "step": 16332
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3569427850403833,
      "learning_rate": 1.0479391868497493e-05,
      "loss": 0.2976,
      "step": 16333
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20981178410072066,
      "learning_rate": 1.0478401123038897e-05,
      "loss": 0.1655,
      "step": 16334
    },
    {
      "epoch": 0.5,
      "grad_norm": 8.858111250855446,
      "learning_rate": 1.0477410372873637e-05,
      "loss": 0.0579,
      "step": 16335
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2747951829458082,
      "learning_rate": 1.0476419618011454e-05,
      "loss": 0.7021,
      "step": 16336
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2545784724927881,
      "learning_rate": 1.0475428858462103e-05,
      "loss": 0.1836,
      "step": 16337
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7769680014612793,
      "learning_rate": 1.0474438094235322e-05,
      "loss": 0.4389,
      "step": 16338
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8316784691642032,
      "learning_rate": 1.0473447325340865e-05,
      "loss": 0.3364,
      "step": 16339
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3288944655905963,
      "learning_rate": 1.0472456551788475e-05,
      "loss": 0.3215,
      "step": 16340
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2661892030040534,
      "learning_rate": 1.0471465773587902e-05,
      "loss": 0.1613,
      "step": 16341
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3799156773049123,
      "learning_rate": 1.0470474990748895e-05,
      "loss": 0.2967,
      "step": 16342
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.177557743175347,
      "learning_rate": 1.0469484203281199e-05,
      "loss": 0.0719,
      "step": 16343
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.530614315611682,
      "learning_rate": 1.0468493411194564e-05,
      "loss": 0.6626,
      "step": 16344
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8903006072438217,
      "learning_rate": 1.0467502614498739e-05,
      "loss": 0.3076,
      "step": 16345
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3587245025124344,
      "learning_rate": 1.0466511813203467e-05,
      "loss": 0.2803,
      "step": 16346
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29216584805591805,
      "learning_rate": 1.0465521007318498e-05,
      "loss": 0.2456,
      "step": 16347
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6360416115333728,
      "learning_rate": 1.0464530196853582e-05,
      "loss": 0.3385,
      "step": 16348
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2061849813353704,
      "learning_rate": 1.0463539381818462e-05,
      "loss": 0.7866,
      "step": 16349
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4259298641107592,
      "learning_rate": 1.0462548562222892e-05,
      "loss": 0.1263,
      "step": 16350
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3686821411517725,
      "learning_rate": 1.0461557738076615e-05,
      "loss": 0.2733,
      "step": 16351
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21802847682207038,
      "learning_rate": 1.0460566909389383e-05,
      "loss": 0.1836,
      "step": 16352
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4090951194805763,
      "learning_rate": 1.0459576076170939e-05,
      "loss": 0.2418,
      "step": 16353
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7720125494251007,
      "learning_rate": 1.0458585238431036e-05,
      "loss": 0.3291,
      "step": 16354
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.670902617316652,
      "learning_rate": 1.0457594396179422e-05,
      "loss": 0.339,
      "step": 16355
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33238963092401475,
      "learning_rate": 1.0456603549425843e-05,
      "loss": 0.2668,
      "step": 16356
    },
    {
      "epoch": 0.5,
      "grad_norm": 2.0485340155324896,
      "learning_rate": 1.0455612698180046e-05,
      "loss": 0.7843,
      "step": 16357
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2858490488136302,
      "learning_rate": 1.0454621842451784e-05,
      "loss": 0.2319,
      "step": 16358
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8906397301325996,
      "learning_rate": 1.0453630982250801e-05,
      "loss": 0.4992,
      "step": 16359
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29494979136568683,
      "learning_rate": 1.0452640117586847e-05,
      "loss": 0.2114,
      "step": 16360
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2594218922335452,
      "learning_rate": 1.0451649248469674e-05,
      "loss": 0.1061,
      "step": 16361
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.7243604925268248,
      "learning_rate": 1.0450658374909024e-05,
      "loss": 0.888,
      "step": 16362
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5222955513930259,
      "learning_rate": 1.044966749691465e-05,
      "loss": 0.2926,
      "step": 16363
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.40369958488145763,
      "learning_rate": 1.04486766144963e-05,
      "loss": 0.2477,
      "step": 16364
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3404564848571773,
      "learning_rate": 1.0447685727663718e-05,
      "loss": 0.2523,
      "step": 16365
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.9872125429468286,
      "learning_rate": 1.0446694836426658e-05,
      "loss": 0.7633,
      "step": 16366
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8600037680439532,
      "learning_rate": 1.0445703940794867e-05,
      "loss": 0.547,
      "step": 16367
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9505828218396346,
      "learning_rate": 1.0444713040778096e-05,
      "loss": 0.5966,
      "step": 16368
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2966420876291012,
      "learning_rate": 1.0443722136386092e-05,
      "loss": 0.1999,
      "step": 16369
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4941940510672945,
      "learning_rate": 1.0442731227628602e-05,
      "loss": 0.346,
      "step": 16370
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.20053293895197602,
      "learning_rate": 1.0441740314515378e-05,
      "loss": 0.1594,
      "step": 16371
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8136859480882337,
      "learning_rate": 1.0440749397056164e-05,
      "loss": 0.4589,
      "step": 16372
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3037532855325145,
      "learning_rate": 1.0439758475260715e-05,
      "loss": 0.1725,
      "step": 16373
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.43900527658788907,
      "learning_rate": 1.0438767549138777e-05,
      "loss": 0.2346,
      "step": 16374
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9591659901413557,
      "learning_rate": 1.0437776618700101e-05,
      "loss": 0.6476,
      "step": 16375
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28720497652584975,
      "learning_rate": 1.0436785683954433e-05,
      "loss": 0.2605,
      "step": 16376
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.657914421007686,
      "learning_rate": 1.0435794744911526e-05,
      "loss": 0.3681,
      "step": 16377
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3480277278383817,
      "learning_rate": 1.0434803801581124e-05,
      "loss": 0.2253,
      "step": 16378
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.48934918894219626,
      "learning_rate": 1.0433812853972977e-05,
      "loss": 0.2721,
      "step": 16379
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.372496644295297,
      "learning_rate": 1.043282190209684e-05,
      "loss": 0.2423,
      "step": 16380
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.42543811735266934,
      "learning_rate": 1.0431830945962458e-05,
      "loss": 0.3395,
      "step": 16381
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29103686188960226,
      "learning_rate": 1.0430839985579582e-05,
      "loss": 0.1759,
      "step": 16382
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3799881805535334,
      "learning_rate": 1.0429849020957959e-05,
      "loss": 0.303,
      "step": 16383
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9137701225393681,
      "learning_rate": 1.0428858052107342e-05,
      "loss": 0.3158,
      "step": 16384
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.4821743764095752,
      "learning_rate": 1.0427867079037475e-05,
      "loss": 0.8685,
      "step": 16385
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9643706920938967,
      "learning_rate": 1.0426876101758113e-05,
      "loss": 0.3313,
      "step": 16386
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3003355214165793,
      "learning_rate": 1.0425885120279002e-05,
      "loss": 0.1828,
      "step": 16387
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3183639629883584,
      "learning_rate": 1.0424894134609898e-05,
      "loss": 0.284,
      "step": 16388
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0618684438134802,
      "learning_rate": 1.0423903144760538e-05,
      "loss": 0.3128,
      "step": 16389
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3535920981275174,
      "learning_rate": 1.0422912150740685e-05,
      "loss": 0.2912,
      "step": 16390
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25049168794872784,
      "learning_rate": 1.0421921152560082e-05,
      "loss": 0.1326,
      "step": 16391
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5228788224715744,
      "learning_rate": 1.0420930150228479e-05,
      "loss": 0.3075,
      "step": 16392
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9769774627587426,
      "learning_rate": 1.0419939143755625e-05,
      "loss": 0.3576,
      "step": 16393
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3628872614822882,
      "learning_rate": 1.0418948133151274e-05,
      "loss": 0.3308,
      "step": 16394
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5459122108278406,
      "learning_rate": 1.0417957118425176e-05,
      "loss": 0.2411,
      "step": 16395
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34450436163878323,
      "learning_rate": 1.0416966099587072e-05,
      "loss": 0.2559,
      "step": 16396
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.009367776668158,
      "learning_rate": 1.0415975076646722e-05,
      "loss": 0.3092,
      "step": 16397
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6221349346179762,
      "learning_rate": 1.0414984049613871e-05,
      "loss": 0.4004,
      "step": 16398
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29879421447444365,
      "learning_rate": 1.0413993018498273e-05,
      "loss": 0.2314,
      "step": 16399
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21919222554891762,
      "learning_rate": 1.0413001983309674e-05,
      "loss": 0.1266,
      "step": 16400
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5610464442047889,
      "learning_rate": 1.0412010944057827e-05,
      "loss": 0.3512,
      "step": 16401
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31740303838382083,
      "learning_rate": 1.0411019900752478e-05,
      "loss": 0.1362,
      "step": 16402
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.394027391265002,
      "learning_rate": 1.041002885340338e-05,
      "loss": 0.8698,
      "step": 16403
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9439533423468728,
      "learning_rate": 1.0409037802020283e-05,
      "loss": 0.3179,
      "step": 16404
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6204185442431575,
      "learning_rate": 1.040804674661294e-05,
      "loss": 0.3076,
      "step": 16405
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.242457684718326,
      "learning_rate": 1.0407055687191097e-05,
      "loss": 0.2459,
      "step": 16406
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.647122252276206,
      "learning_rate": 1.0406064623764509e-05,
      "loss": 0.4252,
      "step": 16407
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2972633813307294,
      "learning_rate": 1.040507355634292e-05,
      "loss": 0.0984,
      "step": 16408
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.38328791328104006,
      "learning_rate": 1.0404082484936085e-05,
      "loss": 0.229,
      "step": 16409
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.297799877890094,
      "learning_rate": 1.0403091409553755e-05,
      "loss": 0.2064,
      "step": 16410
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.460857731924156,
      "learning_rate": 1.0402100330205676e-05,
      "loss": 0.3744,
      "step": 16411
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.41306520640091077,
      "learning_rate": 1.0401109246901604e-05,
      "loss": 0.2923,
      "step": 16412
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8239067993915421,
      "learning_rate": 1.0400118159651287e-05,
      "loss": 0.3529,
      "step": 16413
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36758506689850023,
      "learning_rate": 1.0399127068464478e-05,
      "loss": 0.248,
      "step": 16414
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4634439331154578,
      "learning_rate": 1.0398135973350922e-05,
      "loss": 0.3012,
      "step": 16415
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.965379358962342,
      "learning_rate": 1.0397144874320376e-05,
      "loss": 0.7457,
      "step": 16416
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28320540811659317,
      "learning_rate": 1.0396153771382587e-05,
      "loss": 0.2213,
      "step": 16417
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36879069613007337,
      "learning_rate": 1.0395162664547305e-05,
      "loss": 0.1899,
      "step": 16418
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23500137013510433,
      "learning_rate": 1.0394171553824287e-05,
      "loss": 0.1633,
      "step": 16419
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6726870438624915,
      "learning_rate": 1.0393180439223276e-05,
      "loss": 0.8353,
      "step": 16420
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8566063860254148,
      "learning_rate": 1.0392189320754027e-05,
      "loss": 0.6178,
      "step": 16421
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.713768177835151,
      "learning_rate": 1.0391198198426288e-05,
      "loss": 0.4251,
      "step": 16422
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25303680341959794,
      "learning_rate": 1.0390207072249818e-05,
      "loss": 0.1832,
      "step": 16423
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.46376227154701505,
      "learning_rate": 1.0389215942234358e-05,
      "loss": 0.3724,
      "step": 16424
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36739196434736215,
      "learning_rate": 1.0388224808389667e-05,
      "loss": 0.281,
      "step": 16425
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3230138041771411,
      "learning_rate": 1.0387233670725492e-05,
      "loss": 0.1298,
      "step": 16426
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31897815961955883,
      "learning_rate": 1.0386242529251582e-05,
      "loss": 0.1902,
      "step": 16427
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.34069041176056014,
      "learning_rate": 1.038525138397769e-05,
      "loss": 0.2373,
      "step": 16428
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.3813480470274617,
      "learning_rate": 1.038426023491357e-05,
      "loss": 0.7784,
      "step": 16429
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2840725688595699,
      "learning_rate": 1.0383269082068974e-05,
      "loss": 0.2602,
      "step": 16430
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8566022736333274,
      "learning_rate": 1.0382277925453647e-05,
      "loss": 0.4388,
      "step": 16431
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2418896861156964,
      "learning_rate": 1.0381286765077344e-05,
      "loss": 0.0732,
      "step": 16432
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.36133783128054536,
      "learning_rate": 1.0380295600949821e-05,
      "loss": 0.3325,
      "step": 16433
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0207842656522774,
      "learning_rate": 1.037930443308082e-05,
      "loss": 0.2827,
      "step": 16434
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3224214987641011,
      "learning_rate": 1.0378313261480098e-05,
      "loss": 0.2815,
      "step": 16435
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1567046590317094,
      "learning_rate": 1.0377322086157408e-05,
      "loss": 0.0688,
      "step": 16436
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4525712396567134,
      "learning_rate": 1.0376330907122496e-05,
      "loss": 0.2975,
      "step": 16437
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25185735225274725,
      "learning_rate": 1.0375339724385121e-05,
      "loss": 0.1733,
      "step": 16438
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6469298090574283,
      "learning_rate": 1.0374348537955027e-05,
      "loss": 0.3505,
      "step": 16439
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6767273085499002,
      "learning_rate": 1.0373357347841968e-05,
      "loss": 0.448,
      "step": 16440
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.24832872325046404,
      "learning_rate": 1.0372366154055697e-05,
      "loss": 0.2106,
      "step": 16441
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3538470825926202,
      "learning_rate": 1.0371374956605967e-05,
      "loss": 0.3206,
      "step": 16442
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8648572124414311,
      "learning_rate": 1.0370383755502526e-05,
      "loss": 0.0236,
      "step": 16443
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0003653841668807,
      "learning_rate": 1.0369392550755128e-05,
      "loss": 0.5277,
      "step": 16444
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.28506028776343706,
      "learning_rate": 1.036840134237353e-05,
      "loss": 0.12,
      "step": 16445
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3584878687939789,
      "learning_rate": 1.0367410130367472e-05,
      "loss": 0.3097,
      "step": 16446
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8362506196319085,
      "learning_rate": 1.0366418914746715e-05,
      "loss": 0.4846,
      "step": 16447
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.32646493857597575,
      "learning_rate": 1.0365427695521006e-05,
      "loss": 0.3248,
      "step": 16448
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8384753464076834,
      "learning_rate": 1.0364436472700101e-05,
      "loss": 0.3601,
      "step": 16449
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6346311140315956,
      "learning_rate": 1.0363445246293748e-05,
      "loss": 0.3399,
      "step": 16450
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3137512480630387,
      "learning_rate": 1.0362454016311705e-05,
      "loss": 0.2212,
      "step": 16451
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2708927783194006,
      "learning_rate": 1.0361462782763718e-05,
      "loss": 0.0723,
      "step": 16452
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.35631799158579996,
      "learning_rate": 1.036047154565954e-05,
      "loss": 0.314,
      "step": 16453
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4288341780787734,
      "learning_rate": 1.0359480305008924e-05,
      "loss": 0.131,
      "step": 16454
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5321547045970751,
      "learning_rate": 1.0358489060821622e-05,
      "loss": 0.3506,
      "step": 16455
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.26397937410667294,
      "learning_rate": 1.0357497813107391e-05,
      "loss": 0.1741,
      "step": 16456
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.7343557919246748,
      "learning_rate": 1.0356506561875976e-05,
      "loss": 0.4434,
      "step": 16457
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.31442469355185443,
      "learning_rate": 1.0355515307137132e-05,
      "loss": 0.2509,
      "step": 16458
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.0192442919014784,
      "learning_rate": 1.0354524048900614e-05,
      "loss": 0.5175,
      "step": 16459
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29299549711835027,
      "learning_rate": 1.0353532787176169e-05,
      "loss": 0.2166,
      "step": 16460
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.4034844685754517,
      "learning_rate": 1.0352541521973549e-05,
      "loss": 0.2653,
      "step": 16461
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.8847333553900382,
      "learning_rate": 1.0351550253302515e-05,
      "loss": 0.2682,
      "step": 16462
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.6195443944101091,
      "learning_rate": 1.035055898117281e-05,
      "loss": 0.8568,
      "step": 16463
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3122467857785842,
      "learning_rate": 1.0349567705594194e-05,
      "loss": 0.1885,
      "step": 16464
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.26047436305338806,
      "learning_rate": 1.0348576426576416e-05,
      "loss": 0.2553,
      "step": 16465
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5788165775707399,
      "learning_rate": 1.0347585144129227e-05,
      "loss": 0.4113,
      "step": 16466
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.933063099554907,
      "learning_rate": 1.034659385826238e-05,
      "loss": 0.4999,
      "step": 16467
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3778042515966785,
      "learning_rate": 1.0345602568985626e-05,
      "loss": 0.2434,
      "step": 16468
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.21898063693087527,
      "learning_rate": 1.0344611276308725e-05,
      "loss": 0.1565,
      "step": 16469
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.8899304842772076,
      "learning_rate": 1.0343619980241427e-05,
      "loss": 0.874,
      "step": 16470
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2995567222791473,
      "learning_rate": 1.0342628680793476e-05,
      "loss": 0.2324,
      "step": 16471
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.093509105079744,
      "learning_rate": 1.0341637377974637e-05,
      "loss": 0.5763,
      "step": 16472
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2800092762682217,
      "learning_rate": 1.0340646071794651e-05,
      "loss": 0.1926,
      "step": 16473
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6572961658812448,
      "learning_rate": 1.033965476226328e-05,
      "loss": 0.4279,
      "step": 16474
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5550472811138121,
      "learning_rate": 1.0338663449390276e-05,
      "loss": 0.3626,
      "step": 16475
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5061119091860494,
      "learning_rate": 1.0337672133185388e-05,
      "loss": 0.417,
      "step": 16476
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.2827280935175321,
      "learning_rate": 1.033668081365837e-05,
      "loss": 0.1912,
      "step": 16477
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.30903782843674943,
      "learning_rate": 1.0335689490818974e-05,
      "loss": 0.2216,
      "step": 16478
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.29511176821855784,
      "learning_rate": 1.0334698164676955e-05,
      "loss": 0.1733,
      "step": 16479
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.9936447125717055,
      "learning_rate": 1.0333706835242066e-05,
      "loss": 0.2923,
      "step": 16480
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.6219698839769735,
      "learning_rate": 1.0332715502524057e-05,
      "loss": 0.4231,
      "step": 16481
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.23963726429767454,
      "learning_rate": 1.0331724166532688e-05,
      "loss": 0.0754,
      "step": 16482
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.25467114195575097,
      "learning_rate": 1.0330732827277705e-05,
      "loss": 0.2444,
      "step": 16483
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3600241681323128,
      "learning_rate": 1.0329741484768862e-05,
      "loss": 0.2982,
      "step": 16484
    },
    {
      "epoch": 0.5,
      "grad_norm": 1.2090030812980044,
      "learning_rate": 1.0328750139015918e-05,
      "loss": 0.7345,
      "step": 16485
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.5563965800638301,
      "learning_rate": 1.0327758790028619e-05,
      "loss": 0.1614,
      "step": 16486
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.3299917186795754,
      "learning_rate": 1.032676743781672e-05,
      "loss": 0.2813,
      "step": 16487
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.13963001532313749,
      "learning_rate": 1.0325776082389978e-05,
      "loss": 0.068,
      "step": 16488
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.33126097518469527,
      "learning_rate": 1.0324784723758143e-05,
      "loss": 0.2861,
      "step": 16489
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.844140047007777,
      "learning_rate": 1.032379336193097e-05,
      "loss": 0.3063,
      "step": 16490
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35746897542046746,
      "learning_rate": 1.032280199691821e-05,
      "loss": 0.2051,
      "step": 16491
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4096095574363647,
      "learning_rate": 1.0321810628729618e-05,
      "loss": 0.3861,
      "step": 16492
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.868832676233558,
      "learning_rate": 1.0320819257374945e-05,
      "loss": 0.4053,
      "step": 16493
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4761445973057157,
      "learning_rate": 1.0319827882863948e-05,
      "loss": 0.3345,
      "step": 16494
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35944763658360174,
      "learning_rate": 1.0318836505206385e-05,
      "loss": 0.2422,
      "step": 16495
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33800088683611484,
      "learning_rate": 1.0317845124411995e-05,
      "loss": 0.2681,
      "step": 16496
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19024162830788738,
      "learning_rate": 1.0316853740490542e-05,
      "loss": 0.0854,
      "step": 16497
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.591582732121397,
      "learning_rate": 1.031586235345178e-05,
      "loss": 0.7743,
      "step": 16498
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7866027674275476,
      "learning_rate": 1.031487096330546e-05,
      "loss": 0.3468,
      "step": 16499
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42784049552092424,
      "learning_rate": 1.0313879570061334e-05,
      "loss": 0.2684,
      "step": 16500
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.25189518594253213,
      "learning_rate": 1.031288817372916e-05,
      "loss": 0.2347,
      "step": 16501
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.724940643103052,
      "learning_rate": 1.0311896774318688e-05,
      "loss": 0.7977,
      "step": 16502
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7388955912314458,
      "learning_rate": 1.0310905371839673e-05,
      "loss": 0.5997,
      "step": 16503
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.43823322619798993,
      "learning_rate": 1.0309913966301868e-05,
      "loss": 0.1552,
      "step": 16504
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5492790557667597,
      "learning_rate": 1.0308922557715026e-05,
      "loss": 0.3249,
      "step": 16505
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.21237832686898658,
      "learning_rate": 1.0307931146088904e-05,
      "loss": 0.1275,
      "step": 16506
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3515584843787464,
      "learning_rate": 1.0306939731433255e-05,
      "loss": 0.3026,
      "step": 16507
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7688751500121271,
      "learning_rate": 1.0305948313757831e-05,
      "loss": 0.343,
      "step": 16508
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9845694815122518,
      "learning_rate": 1.0304956893072386e-05,
      "loss": 0.4694,
      "step": 16509
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30828756976743577,
      "learning_rate": 1.0303965469386676e-05,
      "loss": 0.2021,
      "step": 16510
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.342854293405729,
      "learning_rate": 1.0302974042710454e-05,
      "loss": 0.8629,
      "step": 16511
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2837234377974518,
      "learning_rate": 1.0301982613053472e-05,
      "loss": 0.2444,
      "step": 16512
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6594654393847958,
      "learning_rate": 1.0300991180425486e-05,
      "loss": 0.7667,
      "step": 16513
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34613792927340875,
      "learning_rate": 1.029999974483625e-05,
      "loss": 0.1641,
      "step": 16514
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32866195204640086,
      "learning_rate": 1.0299008306295519e-05,
      "loss": 0.2364,
      "step": 16515
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.492342661342346,
      "learning_rate": 1.0298016864813045e-05,
      "loss": 0.244,
      "step": 16516
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9222942608726994,
      "learning_rate": 1.0297025420398586e-05,
      "loss": 0.4831,
      "step": 16517
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28682838243355663,
      "learning_rate": 1.0296033973061888e-05,
      "loss": 0.1734,
      "step": 16518
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2547551630920599,
      "learning_rate": 1.029504252281271e-05,
      "loss": 0.2342,
      "step": 16519
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.163341497625586,
      "learning_rate": 1.029405106966081e-05,
      "loss": 0.7317,
      "step": 16520
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0274505295936514,
      "learning_rate": 1.0293059613615942e-05,
      "loss": 0.4962,
      "step": 16521
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.654086472551487,
      "learning_rate": 1.029206815468785e-05,
      "loss": 0.7908,
      "step": 16522
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.27036764021464293,
      "learning_rate": 1.0291076692886296e-05,
      "loss": 0.1934,
      "step": 16523
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3278079994075239,
      "learning_rate": 1.0290085228221038e-05,
      "loss": 0.2635,
      "step": 16524
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.430949604229622,
      "learning_rate": 1.0289093760701825e-05,
      "loss": 0.2723,
      "step": 16525
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4950392137990552,
      "learning_rate": 1.0288102290338413e-05,
      "loss": 0.2764,
      "step": 16526
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.427394749747434,
      "learning_rate": 1.0287110817140556e-05,
      "loss": 0.1539,
      "step": 16527
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35912657657338726,
      "learning_rate": 1.0286119341118006e-05,
      "loss": 0.296,
      "step": 16528
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8461365394939563,
      "learning_rate": 1.028512786228052e-05,
      "loss": 0.4739,
      "step": 16529
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46592703544500463,
      "learning_rate": 1.0284136380637852e-05,
      "loss": 0.3906,
      "step": 16530
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32274698713554495,
      "learning_rate": 1.0283144896199758e-05,
      "loss": 0.2165,
      "step": 16531
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3073268910052048,
      "learning_rate": 1.028215340897599e-05,
      "loss": 0.1677,
      "step": 16532
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5429887955126632,
      "learning_rate": 1.0281161918976306e-05,
      "loss": 0.3607,
      "step": 16533
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5511309973864792,
      "learning_rate": 1.0280170426210458e-05,
      "loss": 0.3465,
      "step": 16534
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2291612357453954,
      "learning_rate": 1.0279178930688201e-05,
      "loss": 0.1888,
      "step": 16535
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3152767896493605,
      "learning_rate": 1.0278187432419288e-05,
      "loss": 0.1906,
      "step": 16536
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34066310613896794,
      "learning_rate": 1.0277195931413477e-05,
      "loss": 0.2752,
      "step": 16537
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9428362617689303,
      "learning_rate": 1.027620442768052e-05,
      "loss": 0.4026,
      "step": 16538
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.226398448654789,
      "learning_rate": 1.0275212921230176e-05,
      "loss": 0.7476,
      "step": 16539
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8473662404289309,
      "learning_rate": 1.0274221412072196e-05,
      "loss": 0.0322,
      "step": 16540
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5482462111177654,
      "learning_rate": 1.0273229900216335e-05,
      "loss": 0.3462,
      "step": 16541
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5668714073378506,
      "learning_rate": 1.0272238385672346e-05,
      "loss": 0.2415,
      "step": 16542
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4041025039954522,
      "learning_rate": 1.0271246868449989e-05,
      "loss": 0.3394,
      "step": 16543
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1024541693315526,
      "learning_rate": 1.0270255348559012e-05,
      "loss": 0.2921,
      "step": 16544
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.13749797406959696,
      "learning_rate": 1.026926382600918e-05,
      "loss": 0.0714,
      "step": 16545
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34788284008236436,
      "learning_rate": 1.0268272300810237e-05,
      "loss": 0.2628,
      "step": 16546
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8484299908252303,
      "learning_rate": 1.0267280772971947e-05,
      "loss": 0.5945,
      "step": 16547
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.48317298749490883,
      "learning_rate": 1.0266289242504057e-05,
      "loss": 0.3616,
      "step": 16548
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3633004671363672,
      "learning_rate": 1.0265297709416326e-05,
      "loss": 0.2593,
      "step": 16549
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42330395114626634,
      "learning_rate": 1.0264306173718513e-05,
      "loss": 0.3202,
      "step": 16550
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3801324384469541,
      "learning_rate": 1.0263314635420363e-05,
      "loss": 0.2929,
      "step": 16551
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.146299196249348,
      "learning_rate": 1.0262323094531641e-05,
      "loss": 0.9072,
      "step": 16552
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.22312370080912827,
      "learning_rate": 1.02613315510621e-05,
      "loss": 0.1013,
      "step": 16553
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3139232878843767,
      "learning_rate": 1.0260340005021488e-05,
      "loss": 0.2826,
      "step": 16554
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30187285398175495,
      "learning_rate": 1.0259348456419567e-05,
      "loss": 0.1978,
      "step": 16555
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3508069878100926,
      "learning_rate": 1.025835690526609e-05,
      "loss": 0.6942,
      "step": 16556
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8730918301851057,
      "learning_rate": 1.0257365351570817e-05,
      "loss": 0.4734,
      "step": 16557
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5985770138410014,
      "learning_rate": 1.0256373795343495e-05,
      "loss": 0.3241,
      "step": 16558
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6917566765632546,
      "learning_rate": 1.0255382236593882e-05,
      "loss": 0.3819,
      "step": 16559
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28640231490826623,
      "learning_rate": 1.025439067533174e-05,
      "loss": 0.2255,
      "step": 16560
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40605530767491804,
      "learning_rate": 1.0253399111566812e-05,
      "loss": 0.3613,
      "step": 16561
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32969444685632415,
      "learning_rate": 1.0252407545308864e-05,
      "loss": 0.1333,
      "step": 16562
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3722519971737789,
      "learning_rate": 1.0251415976567649e-05,
      "loss": 0.2225,
      "step": 16563
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2944951643714638,
      "learning_rate": 1.025042440535292e-05,
      "loss": 0.1976,
      "step": 16564
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2437021495651779,
      "learning_rate": 1.024943283167443e-05,
      "loss": 0.8336,
      "step": 16565
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28892779809797453,
      "learning_rate": 1.0248441255541943e-05,
      "loss": 0.2593,
      "step": 16566
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.69940165727582,
      "learning_rate": 1.0247449676965205e-05,
      "loss": 0.4589,
      "step": 16567
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36707512292959366,
      "learning_rate": 1.0246458095953977e-05,
      "loss": 0.176,
      "step": 16568
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32954766493445636,
      "learning_rate": 1.0245466512518012e-05,
      "loss": 0.3122,
      "step": 16569
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.289409732412113,
      "learning_rate": 1.0244474926667069e-05,
      "loss": 0.2112,
      "step": 16570
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2808219243384809,
      "learning_rate": 1.02434833384109e-05,
      "loss": 0.1016,
      "step": 16571
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2592743330339438,
      "learning_rate": 1.0242491747759262e-05,
      "loss": 0.2218,
      "step": 16572
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2582122623845014,
      "learning_rate": 1.0241500154721911e-05,
      "loss": 0.1718,
      "step": 16573
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3805820803000421,
      "learning_rate": 1.02405085593086e-05,
      "loss": 0.7955,
      "step": 16574
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8571104714959485,
      "learning_rate": 1.0239516961529087e-05,
      "loss": 0.3468,
      "step": 16575
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6015348022231927,
      "learning_rate": 1.023852536139313e-05,
      "loss": 0.4304,
      "step": 16576
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.29421773239914806,
      "learning_rate": 1.023753375891048e-05,
      "loss": 0.186,
      "step": 16577
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2533378212933735,
      "learning_rate": 1.0236542154090893e-05,
      "loss": 0.2616,
      "step": 16578
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1525410210703315,
      "learning_rate": 1.0235550546944132e-05,
      "loss": 0.3002,
      "step": 16579
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2624492445884754,
      "learning_rate": 1.0234558937479942e-05,
      "loss": 0.628,
      "step": 16580
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20114578453662763,
      "learning_rate": 1.0233567325708084e-05,
      "loss": 0.0951,
      "step": 16581
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34868530347201443,
      "learning_rate": 1.0232575711638314e-05,
      "loss": 0.2825,
      "step": 16582
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0318094491210823,
      "learning_rate": 1.0231584095280392e-05,
      "loss": 0.4815,
      "step": 16583
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3152011383185871,
      "learning_rate": 1.0230592476644067e-05,
      "loss": 0.2412,
      "step": 16584
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.032760210081002,
      "learning_rate": 1.0229600855739095e-05,
      "loss": 0.469,
      "step": 16585
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20558467216894824,
      "learning_rate": 1.0228609232575237e-05,
      "loss": 0.0739,
      "step": 16586
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3689728601456445,
      "learning_rate": 1.0227617607162246e-05,
      "loss": 0.3131,
      "step": 16587
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.482768925463565,
      "learning_rate": 1.0226625979509877e-05,
      "loss": 0.3208,
      "step": 16588
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4579532066975625,
      "learning_rate": 1.0225634349627886e-05,
      "loss": 0.3541,
      "step": 16589
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2564848394025831,
      "learning_rate": 1.0224642717526035e-05,
      "loss": 0.1757,
      "step": 16590
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6713814450380564,
      "learning_rate": 1.022365108321407e-05,
      "loss": 0.3466,
      "step": 16591
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3170782406010548,
      "learning_rate": 1.0222659446701753e-05,
      "loss": 0.2392,
      "step": 16592
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7422044400151405,
      "learning_rate": 1.0221667807998839e-05,
      "loss": 0.4425,
      "step": 16593
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7132332458476055,
      "learning_rate": 1.0220676167115084e-05,
      "loss": 0.349,
      "step": 16594
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.36209956597804016,
      "learning_rate": 1.0219684524060244e-05,
      "loss": 0.2348,
      "step": 16595
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2461075508765046,
      "learning_rate": 1.0218692878844078e-05,
      "loss": 0.2116,
      "step": 16596
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3866061367050623,
      "learning_rate": 1.0217701231476339e-05,
      "loss": 0.2402,
      "step": 16597
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6039213256543898,
      "learning_rate": 1.0216709581966781e-05,
      "loss": 0.7298,
      "step": 16598
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.23654946943095295,
      "learning_rate": 1.0215717930325166e-05,
      "loss": 0.1027,
      "step": 16599
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3839260356216029,
      "learning_rate": 1.0214726276561245e-05,
      "loss": 0.2681,
      "step": 16600
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4125104785403485,
      "learning_rate": 1.0213734620684774e-05,
      "loss": 0.2513,
      "step": 16601
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5720504173895017,
      "learning_rate": 1.0212742962705516e-05,
      "loss": 0.3562,
      "step": 16602
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8978122108078905,
      "learning_rate": 1.0211751302633222e-05,
      "loss": 0.3844,
      "step": 16603
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38183113847438843,
      "learning_rate": 1.0210759640477646e-05,
      "loss": 0.2677,
      "step": 16604
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2916182601536773,
      "learning_rate": 1.020976797624855e-05,
      "loss": 0.2076,
      "step": 16605
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3106226172442323,
      "learning_rate": 1.0208776309955685e-05,
      "loss": 0.6746,
      "step": 16606
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.404499266554699,
      "learning_rate": 1.0207784641608811e-05,
      "loss": 0.2357,
      "step": 16607
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46537376566305433,
      "learning_rate": 1.0206792971217688e-05,
      "loss": 0.3713,
      "step": 16608
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2906620537240577,
      "learning_rate": 1.0205801298792064e-05,
      "loss": 0.1758,
      "step": 16609
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46042485081020784,
      "learning_rate": 1.0204809624341696e-05,
      "loss": 0.2636,
      "step": 16610
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7758833293713019,
      "learning_rate": 1.0203817947876347e-05,
      "loss": 0.4,
      "step": 16611
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.20987217016852225,
      "learning_rate": 1.020282626940577e-05,
      "loss": 0.1466,
      "step": 16612
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3327349428180361,
      "learning_rate": 1.0201834588939719e-05,
      "loss": 0.2359,
      "step": 16613
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31148909672273206,
      "learning_rate": 1.0200842906487954e-05,
      "loss": 0.2044,
      "step": 16614
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.27943388327295704,
      "learning_rate": 1.0199851222060232e-05,
      "loss": 0.2336,
      "step": 16615
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8498091013174675,
      "learning_rate": 1.0198859535666308e-05,
      "loss": 0.1693,
      "step": 16616
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6086757781588058,
      "learning_rate": 1.0197867847315937e-05,
      "loss": 0.443,
      "step": 16617
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.23105158055734118,
      "learning_rate": 1.0196876157018878e-05,
      "loss": 0.0678,
      "step": 16618
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3739855149095329,
      "learning_rate": 1.0195884464784886e-05,
      "loss": 0.3008,
      "step": 16619
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3193899144690126,
      "learning_rate": 1.0194892770623717e-05,
      "loss": 0.2691,
      "step": 16620
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5094520228898942,
      "learning_rate": 1.0193901074545132e-05,
      "loss": 0.803,
      "step": 16621
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19969549628592703,
      "learning_rate": 1.0192909376558881e-05,
      "loss": 0.0872,
      "step": 16622
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2616716719279202,
      "learning_rate": 1.0191917676674724e-05,
      "loss": 0.1805,
      "step": 16623
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.16979393629231,
      "learning_rate": 1.019092597490242e-05,
      "loss": 0.512,
      "step": 16624
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40967249312453996,
      "learning_rate": 1.0189934271251725e-05,
      "loss": 0.2496,
      "step": 16625
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3974773231423578,
      "learning_rate": 1.0188942565732389e-05,
      "loss": 0.3205,
      "step": 16626
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3464564492130161,
      "learning_rate": 1.0187950858354179e-05,
      "loss": 0.2118,
      "step": 16627
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4121969042250506,
      "learning_rate": 1.0186959149126846e-05,
      "loss": 0.3452,
      "step": 16628
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9735236668912981,
      "learning_rate": 1.0185967438060145e-05,
      "loss": 0.3109,
      "step": 16629
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2817071092396688,
      "learning_rate": 1.0184975725163832e-05,
      "loss": 0.7457,
      "step": 16630
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.26791427238647625,
      "learning_rate": 1.0183984010447674e-05,
      "loss": 0.2116,
      "step": 16631
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31711560637571773,
      "learning_rate": 1.0182992293921417e-05,
      "loss": 0.265,
      "step": 16632
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.13852273938701412,
      "learning_rate": 1.0182000575594819e-05,
      "loss": 0.072,
      "step": 16633
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0912182867155105,
      "learning_rate": 1.0181008855477645e-05,
      "loss": 0.5991,
      "step": 16634
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5197843799671794,
      "learning_rate": 1.0180017133579644e-05,
      "loss": 0.3392,
      "step": 16635
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.35894236193925083,
      "learning_rate": 1.0179025409910575e-05,
      "loss": 0.2089,
      "step": 16636
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5590743274033521,
      "learning_rate": 1.0178033684480194e-05,
      "loss": 0.3699,
      "step": 16637
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2784038660479472,
      "learning_rate": 1.0177041957298261e-05,
      "loss": 0.2405,
      "step": 16638
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3719609233556878,
      "learning_rate": 1.017605022837453e-05,
      "loss": 0.8462,
      "step": 16639
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4095518580765554,
      "learning_rate": 1.0175058497718759e-05,
      "loss": 0.1277,
      "step": 16640
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5160147071649114,
      "learning_rate": 1.0174066765340707e-05,
      "loss": 0.3617,
      "step": 16641
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.16440567224211303,
      "learning_rate": 1.0173075031250126e-05,
      "loss": 0.1128,
      "step": 16642
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3229078218475609,
      "learning_rate": 1.0172083295456777e-05,
      "loss": 0.2875,
      "step": 16643
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.64165456814884,
      "learning_rate": 1.0171091557970417e-05,
      "loss": 0.3359,
      "step": 16644
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6279531733824798,
      "learning_rate": 1.0170099818800799e-05,
      "loss": 0.4364,
      "step": 16645
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30474842462179147,
      "learning_rate": 1.0169108077957686e-05,
      "loss": 0.1977,
      "step": 16646
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3015262888861794,
      "learning_rate": 1.0168116335450831e-05,
      "loss": 0.8456,
      "step": 16647
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0476230554580286,
      "learning_rate": 1.0167124591289992e-05,
      "loss": 0.4609,
      "step": 16648
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33477989444490847,
      "learning_rate": 1.0166132845484923e-05,
      "loss": 0.2505,
      "step": 16649
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4063367796485724,
      "learning_rate": 1.0165141098045388e-05,
      "loss": 0.2901,
      "step": 16650
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19411646839132257,
      "learning_rate": 1.0164149348981142e-05,
      "loss": 0.1291,
      "step": 16651
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0152741764701376,
      "learning_rate": 1.0163157598301938e-05,
      "loss": 0.432,
      "step": 16652
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7645588087221213,
      "learning_rate": 1.0162165846017538e-05,
      "loss": 0.3371,
      "step": 16653
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7852204113669543,
      "learning_rate": 1.0161174092137699e-05,
      "loss": 0.4318,
      "step": 16654
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2707435235805365,
      "learning_rate": 1.0160182336672171e-05,
      "loss": 0.2213,
      "step": 16655
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.48634611008137946,
      "learning_rate": 1.0159190579630716e-05,
      "loss": 0.3709,
      "step": 16656
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.907700905935243,
      "learning_rate": 1.0158198821023095e-05,
      "loss": 0.5593,
      "step": 16657
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9716673748114887,
      "learning_rate": 1.015720706085906e-05,
      "loss": 0.5933,
      "step": 16658
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2772046548189495,
      "learning_rate": 1.0156215299148373e-05,
      "loss": 0.1847,
      "step": 16659
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46335487795225105,
      "learning_rate": 1.0155223535900787e-05,
      "loss": 0.269,
      "step": 16660
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2415015859436831,
      "learning_rate": 1.0154231771126062e-05,
      "loss": 0.209,
      "step": 16661
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.0050231006236656,
      "learning_rate": 1.015324000483395e-05,
      "loss": 0.3691,
      "step": 16662
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6303928324964105,
      "learning_rate": 1.0152248237034215e-05,
      "loss": 0.3532,
      "step": 16663
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3104609202020212,
      "learning_rate": 1.0151256467736611e-05,
      "loss": 0.2152,
      "step": 16664
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.431771563236974,
      "learning_rate": 1.0150264696950898e-05,
      "loss": 0.7722,
      "step": 16665
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39598548228475305,
      "learning_rate": 1.014927292468683e-05,
      "loss": 0.282,
      "step": 16666
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.44702109377827415,
      "learning_rate": 1.0148281150954169e-05,
      "loss": 0.3572,
      "step": 16667
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2868090087737907,
      "learning_rate": 1.0147289375762665e-05,
      "loss": 0.0693,
      "step": 16668
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.270491777602033,
      "learning_rate": 1.0146297599122082e-05,
      "loss": 0.2254,
      "step": 16669
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5572870356801907,
      "learning_rate": 1.0145305821042175e-05,
      "loss": 0.3593,
      "step": 16670
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39022627125452664,
      "learning_rate": 1.0144314041532699e-05,
      "loss": 0.264,
      "step": 16671
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5107734856798709,
      "learning_rate": 1.0143322260603417e-05,
      "loss": 0.1536,
      "step": 16672
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.325552890236257,
      "learning_rate": 1.0142330478264082e-05,
      "loss": 0.263,
      "step": 16673
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.307146146510712,
      "learning_rate": 1.0141338694524455e-05,
      "loss": 0.2589,
      "step": 16674
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8615559884626854,
      "learning_rate": 1.0140346909394287e-05,
      "loss": 0.4611,
      "step": 16675
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6982498388189717,
      "learning_rate": 1.013935512288334e-05,
      "loss": 0.745,
      "step": 16676
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2968663910291305,
      "learning_rate": 1.0138363335001376e-05,
      "loss": 0.2044,
      "step": 16677
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.39267730023027964,
      "learning_rate": 1.0137371545758147e-05,
      "loss": 0.3399,
      "step": 16678
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.27925422140311357,
      "learning_rate": 1.0136379755163411e-05,
      "loss": 0.2497,
      "step": 16679
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.45086020538327365,
      "learning_rate": 1.0135387963226925e-05,
      "loss": 0.2897,
      "step": 16680
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.18049645927564817,
      "learning_rate": 1.0134396169958446e-05,
      "loss": 0.0663,
      "step": 16681
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33090226925756255,
      "learning_rate": 1.0133404375367737e-05,
      "loss": 0.285,
      "step": 16682
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.980434610814656,
      "learning_rate": 1.013241257946455e-05,
      "loss": 0.4775,
      "step": 16683
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1483829911534416,
      "learning_rate": 1.0131420782258646e-05,
      "loss": 0.6481,
      "step": 16684
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2789016633957947,
      "learning_rate": 1.013042898375978e-05,
      "loss": 0.2409,
      "step": 16685
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46909127033511105,
      "learning_rate": 1.0129437183977709e-05,
      "loss": 0.32,
      "step": 16686
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32630043769440936,
      "learning_rate": 1.0128445382922197e-05,
      "loss": 0.2246,
      "step": 16687
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5443964566894258,
      "learning_rate": 1.012745358060299e-05,
      "loss": 0.3236,
      "step": 16688
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2413750968472417,
      "learning_rate": 1.0126461777029857e-05,
      "loss": 0.1379,
      "step": 16689
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31594020364746844,
      "learning_rate": 1.0125469972212553e-05,
      "loss": 0.1889,
      "step": 16690
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5084608631489526,
      "learning_rate": 1.0124478166160835e-05,
      "loss": 0.3485,
      "step": 16691
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3004161800903638,
      "learning_rate": 1.0123486358884454e-05,
      "loss": 0.24,
      "step": 16692
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1610615232591308,
      "learning_rate": 1.0122494550393178e-05,
      "loss": 0.7539,
      "step": 16693
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6176878021688813,
      "learning_rate": 1.0121502740696758e-05,
      "loss": 0.3171,
      "step": 16694
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9227023710845121,
      "learning_rate": 1.0120510929804954e-05,
      "loss": 0.452,
      "step": 16695
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.29849200649917196,
      "learning_rate": 1.0119519117727528e-05,
      "loss": 0.224,
      "step": 16696
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3381777006811848,
      "learning_rate": 1.011852730447423e-05,
      "loss": 0.2954,
      "step": 16697
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.15615505341047767,
      "learning_rate": 1.0117535490054822e-05,
      "loss": 0.0691,
      "step": 16698
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.1256190997707645,
      "learning_rate": 1.0116543674479061e-05,
      "loss": 0.882,
      "step": 16699
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30626157596342957,
      "learning_rate": 1.0115551857756704e-05,
      "loss": 0.1867,
      "step": 16700
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.43810328288498435,
      "learning_rate": 1.011456003989751e-05,
      "loss": 0.2765,
      "step": 16701
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3365473114566892,
      "learning_rate": 1.0113568220911236e-05,
      "loss": 0.3071,
      "step": 16702
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8303781980966904,
      "learning_rate": 1.0112576400807645e-05,
      "loss": 0.3228,
      "step": 16703
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.699274054713495,
      "learning_rate": 1.0111584579596486e-05,
      "loss": 0.3558,
      "step": 16704
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32227112181765893,
      "learning_rate": 1.0110592757287522e-05,
      "loss": 0.2441,
      "step": 16705
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9362320344591601,
      "learning_rate": 1.0109600933890513e-05,
      "loss": 0.5779,
      "step": 16706
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3070128835308792,
      "learning_rate": 1.0108609109415211e-05,
      "loss": 0.0963,
      "step": 16707
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2605609797748125,
      "learning_rate": 1.0107617283871376e-05,
      "loss": 0.2595,
      "step": 16708
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2872769641781516,
      "learning_rate": 1.0106625457268769e-05,
      "loss": 0.188,
      "step": 16709
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.4102238338524606,
      "learning_rate": 1.0105633629617146e-05,
      "loss": 0.7069,
      "step": 16710
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8860332598980788,
      "learning_rate": 1.0104641800926263e-05,
      "loss": 0.407,
      "step": 16711
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7659711397865158,
      "learning_rate": 1.0103649971205878e-05,
      "loss": 0.4352,
      "step": 16712
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4436705432792249,
      "learning_rate": 1.0102658140465755e-05,
      "loss": 0.1932,
      "step": 16713
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.379251979785297,
      "learning_rate": 1.0101666308715642e-05,
      "loss": 0.3092,
      "step": 16714
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2822788936509968,
      "learning_rate": 1.0100674475965305e-05,
      "loss": 0.2318,
      "step": 16715
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.16669162852822192,
      "learning_rate": 1.0099682642224502e-05,
      "loss": 0.0725,
      "step": 16716
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.41085064696708873,
      "learning_rate": 1.0098690807502987e-05,
      "loss": 0.2588,
      "step": 16717
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19947503855774823,
      "learning_rate": 1.0097698971810517e-05,
      "loss": 0.0676,
      "step": 16718
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.37620810322225234,
      "learning_rate": 1.0096707135156853e-05,
      "loss": 0.32,
      "step": 16719
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.482091405991403,
      "learning_rate": 1.0095715297551754e-05,
      "loss": 0.307,
      "step": 16720
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.47313997167127175,
      "learning_rate": 1.0094723459004974e-05,
      "loss": 0.3434,
      "step": 16721
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5777894573979635,
      "learning_rate": 1.0093731619526277e-05,
      "loss": 0.141,
      "step": 16722
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7735031997593846,
      "learning_rate": 1.0092739779125415e-05,
      "loss": 0.2896,
      "step": 16723
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.009370821089483,
      "learning_rate": 1.0091747937812149e-05,
      "loss": 0.1815,
      "step": 16724
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.43183102727321293,
      "learning_rate": 1.0090756095596235e-05,
      "loss": 0.2614,
      "step": 16725
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2809491731896402,
      "learning_rate": 1.0089764252487434e-05,
      "loss": 0.2186,
      "step": 16726
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6111077465824826,
      "learning_rate": 1.0088772408495504e-05,
      "loss": 0.337,
      "step": 16727
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34689237550984053,
      "learning_rate": 1.0087780563630201e-05,
      "loss": 0.2474,
      "step": 16728
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.570002210805995,
      "learning_rate": 1.0086788717901286e-05,
      "loss": 0.3348,
      "step": 16729
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.8285980030566773,
      "learning_rate": 1.0085796871318512e-05,
      "loss": 0.9333,
      "step": 16730
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.24076041661180042,
      "learning_rate": 1.0084805023891639e-05,
      "loss": 0.0741,
      "step": 16731
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.32776440788015343,
      "learning_rate": 1.008381317563043e-05,
      "loss": 0.2575,
      "step": 16732
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30404799948814215,
      "learning_rate": 1.0082821326544637e-05,
      "loss": 0.2446,
      "step": 16733
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2226610925110628,
      "learning_rate": 1.0081829476644018e-05,
      "loss": 0.5904,
      "step": 16734
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8134143750986212,
      "learning_rate": 1.008083762593834e-05,
      "loss": 0.4441,
      "step": 16735
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.413581230408582,
      "learning_rate": 1.0079845774437351e-05,
      "loss": 0.3116,
      "step": 16736
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.47960748368473105,
      "learning_rate": 1.0078853922150813e-05,
      "loss": 0.2801,
      "step": 16737
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7199447127119398,
      "learning_rate": 1.0077862069088484e-05,
      "loss": 0.4399,
      "step": 16738
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2949138811337786,
      "learning_rate": 1.0076870215260123e-05,
      "loss": 0.2554,
      "step": 16739
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.24589450800406093,
      "learning_rate": 1.0075878360675488e-05,
      "loss": 0.1499,
      "step": 16740
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2868180992852543,
      "learning_rate": 1.007488650534434e-05,
      "loss": 0.2035,
      "step": 16741
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9598574820733474,
      "learning_rate": 1.007389464927643e-05,
      "loss": 0.2652,
      "step": 16742
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3090773586574749,
      "learning_rate": 1.0072902792481518e-05,
      "loss": 0.5427,
      "step": 16743
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28882621197987135,
      "learning_rate": 1.0071910934969367e-05,
      "loss": 0.25,
      "step": 16744
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6270286331286606,
      "learning_rate": 1.0070919076749734e-05,
      "loss": 0.3496,
      "step": 16745
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3355438322815997,
      "learning_rate": 1.0069927217832373e-05,
      "loss": 0.233,
      "step": 16746
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7229676260877312,
      "learning_rate": 1.0068935358227047e-05,
      "loss": 0.4303,
      "step": 16747
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.19279696144791905,
      "learning_rate": 1.006794349794351e-05,
      "loss": 0.1008,
      "step": 16748
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.6987939939908228,
      "learning_rate": 1.0066951636991527e-05,
      "loss": 0.8166,
      "step": 16749
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2512895331109368,
      "learning_rate": 1.0065959775380847e-05,
      "loss": 0.1998,
      "step": 16750
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3455415906966538,
      "learning_rate": 1.0064967913121233e-05,
      "loss": 0.2957,
      "step": 16751
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1863642924937474,
      "learning_rate": 1.0063976050222445e-05,
      "loss": 0.2976,
      "step": 16752
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5325232224082697,
      "learning_rate": 1.006298418669424e-05,
      "loss": 0.86,
      "step": 16753
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.42621231957065553,
      "learning_rate": 1.0061992322546376e-05,
      "loss": 0.2369,
      "step": 16754
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3056288553300082,
      "learning_rate": 1.0061000457788612e-05,
      "loss": 0.2556,
      "step": 16755
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31820121421345654,
      "learning_rate": 1.0060008592430704e-05,
      "loss": 0.3081,
      "step": 16756
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9037133600841586,
      "learning_rate": 1.0059016726482409e-05,
      "loss": 0.4857,
      "step": 16757
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2790760357506778,
      "learning_rate": 1.0058024859953493e-05,
      "loss": 0.1716,
      "step": 16758
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2759024959602181,
      "learning_rate": 1.0057032992853706e-05,
      "loss": 0.192,
      "step": 16759
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3432205492642864,
      "learning_rate": 1.0056041125192812e-05,
      "loss": 0.6554,
      "step": 16760
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1513666163021796,
      "learning_rate": 1.0055049256980565e-05,
      "loss": 0.0738,
      "step": 16761
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30158736269052083,
      "learning_rate": 1.0054057388226726e-05,
      "loss": 0.3018,
      "step": 16762
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3898999425496446,
      "learning_rate": 1.0053065518941052e-05,
      "loss": 0.2338,
      "step": 16763
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3295275757185249,
      "learning_rate": 1.0052073649133302e-05,
      "loss": 0.2841,
      "step": 16764
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9197933159044618,
      "learning_rate": 1.0051081778813236e-05,
      "loss": 0.3139,
      "step": 16765
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.1242713255079784,
      "learning_rate": 1.0050089907990607e-05,
      "loss": 0.6825,
      "step": 16766
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.18524372324453867,
      "learning_rate": 1.0049098036675183e-05,
      "loss": 0.1315,
      "step": 16767
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.3043777130397102,
      "learning_rate": 1.0048106164876714e-05,
      "loss": 0.1911,
      "step": 16768
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.46195775185906746,
      "learning_rate": 1.0047114292604957e-05,
      "loss": 0.3404,
      "step": 16769
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8835185023583862,
      "learning_rate": 1.0046122419869677e-05,
      "loss": 0.2651,
      "step": 16770
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8321376985785311,
      "learning_rate": 1.0045130546680631e-05,
      "loss": 0.42,
      "step": 16771
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6891786896728026,
      "learning_rate": 1.0044138673047572e-05,
      "loss": 0.2334,
      "step": 16772
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.33672913860978015,
      "learning_rate": 1.0043146798980266e-05,
      "loss": 0.2893,
      "step": 16773
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2863933712452528,
      "learning_rate": 1.0042154924488465e-05,
      "loss": 0.2494,
      "step": 16774
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2786790462822855,
      "learning_rate": 1.0041163049581933e-05,
      "loss": 0.7227,
      "step": 16775
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.24850445672276758,
      "learning_rate": 1.0040171174270422e-05,
      "loss": 0.0992,
      "step": 16776
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2920288728781188,
      "learning_rate": 1.0039179298563692e-05,
      "loss": 0.1584,
      "step": 16777
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.34730645298092283,
      "learning_rate": 1.0038187422471509e-05,
      "loss": 0.218,
      "step": 16778
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9168481846259521,
      "learning_rate": 1.0037195546003622e-05,
      "loss": 0.4306,
      "step": 16779
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2914277107224579,
      "learning_rate": 1.0036203669169793e-05,
      "loss": 0.2351,
      "step": 16780
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.5620039279584971,
      "learning_rate": 1.0035211791979782e-05,
      "loss": 0.2477,
      "step": 16781
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38092228331547134,
      "learning_rate": 1.0034219914443344e-05,
      "loss": 0.2927,
      "step": 16782
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.8296888775086696,
      "learning_rate": 1.003322803657024e-05,
      "loss": 0.62,
      "step": 16783
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.5319962993180591,
      "learning_rate": 1.0032236158370228e-05,
      "loss": 0.7313,
      "step": 16784
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.30466826781629264,
      "learning_rate": 1.0031244279853066e-05,
      "loss": 0.2467,
      "step": 16785
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.40084547281498767,
      "learning_rate": 1.0030252401028513e-05,
      "loss": 0.2827,
      "step": 16786
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.21311593465359788,
      "learning_rate": 1.0029260521906326e-05,
      "loss": 0.1313,
      "step": 16787
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.3945354613149912,
      "learning_rate": 1.0028268642496264e-05,
      "loss": 0.5232,
      "step": 16788
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.9758715525944646,
      "learning_rate": 1.0027276762808085e-05,
      "loss": 0.3619,
      "step": 16789
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7805408864663883,
      "learning_rate": 1.0026284882851549e-05,
      "loss": 0.4264,
      "step": 16790
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.28149177351755816,
      "learning_rate": 1.0025293002636417e-05,
      "loss": 0.2063,
      "step": 16791
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.431121335356275,
      "learning_rate": 1.0024301122172442e-05,
      "loss": 0.3201,
      "step": 16792
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4023915187913334,
      "learning_rate": 1.0023309241469382e-05,
      "loss": 0.2878,
      "step": 16793
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.1938707737254749,
      "learning_rate": 1.0022317360537001e-05,
      "loss": 0.0708,
      "step": 16794
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.38878623640209337,
      "learning_rate": 1.0021325479385053e-05,
      "loss": 0.2525,
      "step": 16795
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.4273848392942441,
      "learning_rate": 1.0020333598023297e-05,
      "loss": 0.199,
      "step": 16796
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6682797788413566,
      "learning_rate": 1.0019341716461495e-05,
      "loss": 0.4154,
      "step": 16797
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.284436450459122,
      "learning_rate": 1.0018349834709402e-05,
      "loss": 0.2425,
      "step": 16798
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7863851830463631,
      "learning_rate": 1.0017357952776777e-05,
      "loss": 0.4376,
      "step": 16799
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2668667051221946,
      "learning_rate": 1.0016366070673378e-05,
      "loss": 0.1956,
      "step": 16800
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.402284801298838,
      "learning_rate": 1.0015374188408963e-05,
      "loss": 0.8912,
      "step": 16801
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.028088700315922,
      "learning_rate": 1.0014382305993296e-05,
      "loss": 0.3312,
      "step": 16802
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.43812489419615314,
      "learning_rate": 1.0013390423436126e-05,
      "loss": 0.3401,
      "step": 16803
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.308203474418121,
      "learning_rate": 1.0012398540747221e-05,
      "loss": 0.1847,
      "step": 16804
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.29754203552696473,
      "learning_rate": 1.0011406657936333e-05,
      "loss": 0.2691,
      "step": 16805
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.7034556569664251,
      "learning_rate": 1.0010414775013222e-05,
      "loss": 0.3403,
      "step": 16806
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.23052676528657662,
      "learning_rate": 1.000942289198765e-05,
      "loss": 0.0682,
      "step": 16807
    },
    {
      "epoch": 0.51,
      "grad_norm": 2.3839278242786577,
      "learning_rate": 1.0008431008869369e-05,
      "loss": 0.8813,
      "step": 16808
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.2673410986816453,
      "learning_rate": 1.0007439125668141e-05,
      "loss": 0.1836,
      "step": 16809
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.31648727344136784,
      "learning_rate": 1.0006447242393726e-05,
      "loss": 0.294,
      "step": 16810
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2862510146366632,
      "learning_rate": 1.0005455359055882e-05,
      "loss": 0.376,
      "step": 16811
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.2418931713644246,
      "learning_rate": 1.0004463475664366e-05,
      "loss": 0.7707,
      "step": 16812
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.23150064238912493,
      "learning_rate": 1.0003471592228936e-05,
      "loss": 0.1317,
      "step": 16813
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.44589339527098115,
      "learning_rate": 1.0002479708759349e-05,
      "loss": 0.3757,
      "step": 16814
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.6920771986211964,
      "learning_rate": 1.0001487825265368e-05,
      "loss": 0.3308,
      "step": 16815
    },
    {
      "epoch": 0.51,
      "grad_norm": 0.24862910860673934,
      "learning_rate": 1.000049594175675e-05,
      "loss": 0.24,
      "step": 16816
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5866781302934728,
      "learning_rate": 9.999504058243252e-06,
      "loss": 0.2482,
      "step": 16817
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35113242764377584,
      "learning_rate": 9.998512174734635e-06,
      "loss": 0.2617,
      "step": 16818
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8527534796712994,
      "learning_rate": 9.997520291240654e-06,
      "loss": 0.481,
      "step": 16819
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8426879968145625,
      "learning_rate": 9.996528407771066e-06,
      "loss": 0.3027,
      "step": 16820
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31827818350871373,
      "learning_rate": 9.995536524335637e-06,
      "loss": 0.2915,
      "step": 16821
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5511573198651871,
      "learning_rate": 9.994544640944121e-06,
      "loss": 0.1941,
      "step": 16822
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3638605315378204,
      "learning_rate": 9.993552757606276e-06,
      "loss": 0.2816,
      "step": 16823
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5665494926976525,
      "learning_rate": 9.99256087433186e-06,
      "loss": 0.3168,
      "step": 16824
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35976311199706745,
      "learning_rate": 9.991568991130636e-06,
      "loss": 0.1849,
      "step": 16825
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3584031492355384,
      "learning_rate": 9.990577108012353e-06,
      "loss": 0.2292,
      "step": 16826
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6099550518999308,
      "learning_rate": 9.98958522498678e-06,
      "loss": 0.3724,
      "step": 16827
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.256243999144625,
      "learning_rate": 9.988593342063672e-06,
      "loss": 0.2445,
      "step": 16828
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.498539111235783,
      "learning_rate": 9.987601459252782e-06,
      "loss": 0.7556,
      "step": 16829
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7161832472257593,
      "learning_rate": 9.986609576563877e-06,
      "loss": 0.3108,
      "step": 16830
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6622092153562984,
      "learning_rate": 9.985617694006706e-06,
      "loss": 0.4379,
      "step": 16831
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3203048864950119,
      "learning_rate": 9.984625811591038e-06,
      "loss": 0.2373,
      "step": 16832
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3009680908148372,
      "learning_rate": 9.983633929326625e-06,
      "loss": 0.239,
      "step": 16833
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.540811063141269,
      "learning_rate": 9.982642047223225e-06,
      "loss": 0.6802,
      "step": 16834
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.16324347293434507,
      "learning_rate": 9.9816501652906e-06,
      "loss": 0.1012,
      "step": 16835
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38031492855374077,
      "learning_rate": 9.980658283538508e-06,
      "loss": 0.2522,
      "step": 16836
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48747568409762654,
      "learning_rate": 9.979666401976703e-06,
      "loss": 0.2615,
      "step": 16837
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4535996533313458,
      "learning_rate": 9.97867452061495e-06,
      "loss": 0.6894,
      "step": 16838
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3300383347741455,
      "learning_rate": 9.977682639463004e-06,
      "loss": 0.2649,
      "step": 16839
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7773284496482242,
      "learning_rate": 9.976690758530618e-06,
      "loss": 0.3871,
      "step": 16840
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3133898886745819,
      "learning_rate": 9.975698877827561e-06,
      "loss": 0.2263,
      "step": 16841
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.401637162426855,
      "learning_rate": 9.974706997363587e-06,
      "loss": 0.7098,
      "step": 16842
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18974644635407098,
      "learning_rate": 9.973715117148451e-06,
      "loss": 0.0693,
      "step": 16843
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.44407931273901,
      "learning_rate": 9.972723237191917e-06,
      "loss": 0.8072,
      "step": 16844
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3051183024223205,
      "learning_rate": 9.971731357503741e-06,
      "loss": 0.1918,
      "step": 16845
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3155003688338348,
      "learning_rate": 9.970739478093675e-06,
      "loss": 0.2369,
      "step": 16846
    },
    {
      "epoch": 0.52,
      "grad_norm": 2.4638766845057285,
      "learning_rate": 9.96974759897149e-06,
      "loss": 0.7048,
      "step": 16847
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8764070532292672,
      "learning_rate": 9.968755720146938e-06,
      "loss": 0.3353,
      "step": 16848
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5587293381609949,
      "learning_rate": 9.967763841629775e-06,
      "loss": 0.3118,
      "step": 16849
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31404905586337,
      "learning_rate": 9.966771963429763e-06,
      "loss": 0.215,
      "step": 16850
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4509909824494744,
      "learning_rate": 9.96578008555666e-06,
      "loss": 0.3423,
      "step": 16851
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2518225343224286,
      "learning_rate": 9.964788208020222e-06,
      "loss": 0.1612,
      "step": 16852
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4283284842004385,
      "learning_rate": 9.96379633083021e-06,
      "loss": 0.2755,
      "step": 16853
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2836283564049612,
      "learning_rate": 9.962804453996383e-06,
      "loss": 0.0694,
      "step": 16854
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3884893828354693,
      "learning_rate": 9.961812577528496e-06,
      "loss": 0.3323,
      "step": 16855
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8347198009146113,
      "learning_rate": 9.960820701436311e-06,
      "loss": 0.3612,
      "step": 16856
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3537777873695128,
      "learning_rate": 9.95982882572958e-06,
      "loss": 0.296,
      "step": 16857
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0308524579577574,
      "learning_rate": 9.95883695041807e-06,
      "loss": 0.2172,
      "step": 16858
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.26478651708362116,
      "learning_rate": 9.957845075511538e-06,
      "loss": 0.1847,
      "step": 16859
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2936777292316155,
      "learning_rate": 9.956853201019737e-06,
      "loss": 0.5306,
      "step": 16860
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.23784006127344787,
      "learning_rate": 9.955861326952431e-06,
      "loss": 0.0726,
      "step": 16861
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.45743135197337637,
      "learning_rate": 9.954869453319374e-06,
      "loss": 0.2843,
      "step": 16862
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2761137972080804,
      "learning_rate": 9.953877580130325e-06,
      "loss": 0.2061,
      "step": 16863
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35203696066111345,
      "learning_rate": 9.952885707395044e-06,
      "loss": 0.3261,
      "step": 16864
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8543533809509808,
      "learning_rate": 9.951893835123293e-06,
      "loss": 0.3709,
      "step": 16865
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6693872160758704,
      "learning_rate": 9.950901963324819e-06,
      "loss": 0.4655,
      "step": 16866
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8730672355044945,
      "learning_rate": 9.949910092009394e-06,
      "loss": 0.1727,
      "step": 16867
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3828615130058399,
      "learning_rate": 9.94891822118677e-06,
      "loss": 0.2896,
      "step": 16868
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42207428287157317,
      "learning_rate": 9.9479263508667e-06,
      "loss": 0.2392,
      "step": 16869
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.47440870330379314,
      "learning_rate": 9.946934481058951e-06,
      "loss": 0.362,
      "step": 16870
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27360688818727036,
      "learning_rate": 9.945942611773279e-06,
      "loss": 0.1367,
      "step": 16871
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2900930455927132,
      "learning_rate": 9.944950743019436e-06,
      "loss": 0.1815,
      "step": 16872
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48180978877429403,
      "learning_rate": 9.943958874807192e-06,
      "loss": 0.3415,
      "step": 16873
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8695035443335878,
      "learning_rate": 9.942967007146298e-06,
      "loss": 0.3562,
      "step": 16874
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3122859924471666,
      "learning_rate": 9.94197514004651e-06,
      "loss": 0.2999,
      "step": 16875
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.236944797431989,
      "learning_rate": 9.940983273517593e-06,
      "loss": 0.099,
      "step": 16876
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5214377169115479,
      "learning_rate": 9.939991407569301e-06,
      "loss": 0.3509,
      "step": 16877
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43286303359674005,
      "learning_rate": 9.93899954221139e-06,
      "loss": 0.2085,
      "step": 16878
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4248806420669035,
      "learning_rate": 9.938007677453625e-06,
      "loss": 0.2026,
      "step": 16879
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27104511580957347,
      "learning_rate": 9.937015813305763e-06,
      "loss": 0.259,
      "step": 16880
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8741530647366734,
      "learning_rate": 9.936023949777557e-06,
      "loss": 0.4558,
      "step": 16881
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2799006625723781,
      "learning_rate": 9.93503208687877e-06,
      "loss": 0.2282,
      "step": 16882
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5804296646387722,
      "learning_rate": 9.934040224619154e-06,
      "loss": 0.4141,
      "step": 16883
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4039651055186835,
      "learning_rate": 9.933048363008478e-06,
      "loss": 0.1602,
      "step": 16884
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6219271816677368,
      "learning_rate": 9.932056502056491e-06,
      "loss": 0.2616,
      "step": 16885
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35673565486175596,
      "learning_rate": 9.931064641772956e-06,
      "loss": 0.2449,
      "step": 16886
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.26628782950854785,
      "learning_rate": 9.93007278216763e-06,
      "loss": 0.2321,
      "step": 16887
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.265896992706183,
      "learning_rate": 9.929080923250271e-06,
      "loss": 0.7014,
      "step": 16888
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8980689524963313,
      "learning_rate": 9.928089065030633e-06,
      "loss": 0.4752,
      "step": 16889
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5882935363968123,
      "learning_rate": 9.927097207518483e-06,
      "loss": 0.3766,
      "step": 16890
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2756533128408103,
      "learning_rate": 9.926105350723575e-06,
      "loss": 0.2281,
      "step": 16891
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7565579039956625,
      "learning_rate": 9.925113494655665e-06,
      "loss": 0.5555,
      "step": 16892
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27961492300009433,
      "learning_rate": 9.924121639324515e-06,
      "loss": 0.2441,
      "step": 16893
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32378004075111294,
      "learning_rate": 9.92312978473988e-06,
      "loss": 0.2221,
      "step": 16894
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.26262882103128826,
      "learning_rate": 9.922137930911515e-06,
      "loss": 0.1839,
      "step": 16895
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7885241793955753,
      "learning_rate": 9.921146077849189e-06,
      "loss": 0.7458,
      "step": 16896
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9709354126414889,
      "learning_rate": 9.920154225562652e-06,
      "loss": 0.2821,
      "step": 16897
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.42212694770635606,
      "learning_rate": 9.919162374061663e-06,
      "loss": 0.365,
      "step": 16898
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2950016210303023,
      "learning_rate": 9.918170523355983e-06,
      "loss": 0.2041,
      "step": 16899
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3313220828604053,
      "learning_rate": 9.917178673455368e-06,
      "loss": 0.2287,
      "step": 16900
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7120563198620335,
      "learning_rate": 9.916186824369573e-06,
      "loss": 0.4698,
      "step": 16901
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.20410349119450696,
      "learning_rate": 9.915194976108363e-06,
      "loss": 0.1236,
      "step": 16902
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48331653781882333,
      "learning_rate": 9.914203128681493e-06,
      "loss": 0.2362,
      "step": 16903
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24139886536128016,
      "learning_rate": 9.913211282098718e-06,
      "loss": 0.0668,
      "step": 16904
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2859292562535256,
      "learning_rate": 9.9122194363698e-06,
      "loss": 0.2835,
      "step": 16905
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3616452984987081,
      "learning_rate": 9.911227591504498e-06,
      "loss": 0.212,
      "step": 16906
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5874930157569757,
      "learning_rate": 9.910235747512567e-06,
      "loss": 0.4069,
      "step": 16907
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5798822098085478,
      "learning_rate": 9.909243904403769e-06,
      "loss": 0.2158,
      "step": 16908
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33873773202785845,
      "learning_rate": 9.908252062187851e-06,
      "loss": 0.2633,
      "step": 16909
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4027113979329595,
      "learning_rate": 9.907260220874588e-06,
      "loss": 0.2933,
      "step": 16910
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4757717087604649,
      "learning_rate": 9.906268380473726e-06,
      "loss": 0.3885,
      "step": 16911
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.1946095947159518,
      "learning_rate": 9.905276540995026e-06,
      "loss": 0.0708,
      "step": 16912
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34566767153513583,
      "learning_rate": 9.904284702448249e-06,
      "loss": 0.1799,
      "step": 16913
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5597062293937805,
      "learning_rate": 9.90329286484315e-06,
      "loss": 0.3357,
      "step": 16914
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8374442038901112,
      "learning_rate": 9.902301028189485e-06,
      "loss": 0.3091,
      "step": 16915
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4410903078531614,
      "learning_rate": 9.901309192497017e-06,
      "loss": 0.3276,
      "step": 16916
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32509415043580003,
      "learning_rate": 9.900317357775501e-06,
      "loss": 0.255,
      "step": 16917
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3377397641658503,
      "learning_rate": 9.899325524034695e-06,
      "loss": 0.2608,
      "step": 16918
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7997137293070723,
      "learning_rate": 9.89833369128436e-06,
      "loss": 0.5389,
      "step": 16919
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2223620958901864,
      "learning_rate": 9.89734185953425e-06,
      "loss": 0.6431,
      "step": 16920
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18849103507237203,
      "learning_rate": 9.896350028794122e-06,
      "loss": 0.0896,
      "step": 16921
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39195905958658545,
      "learning_rate": 9.89535819907374e-06,
      "loss": 0.2597,
      "step": 16922
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3071433832257586,
      "learning_rate": 9.89436637038286e-06,
      "loss": 0.2243,
      "step": 16923
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48157588888472,
      "learning_rate": 9.893374542731233e-06,
      "loss": 0.3333,
      "step": 16924
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6408593472860475,
      "learning_rate": 9.892382716128626e-06,
      "loss": 0.3185,
      "step": 16925
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.25806385650319313,
      "learning_rate": 9.891390890584794e-06,
      "loss": 0.1179,
      "step": 16926
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5313597305372504,
      "learning_rate": 9.89039906610949e-06,
      "loss": 0.348,
      "step": 16927
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4275189681163694,
      "learning_rate": 9.88940724271248e-06,
      "loss": 0.2816,
      "step": 16928
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3169317048379586,
      "learning_rate": 9.888415420403517e-06,
      "loss": 0.2886,
      "step": 16929
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21019680917629724,
      "learning_rate": 9.887423599192358e-06,
      "loss": 0.1019,
      "step": 16930
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9480490225325404,
      "learning_rate": 9.886431779088766e-06,
      "loss": 0.4797,
      "step": 16931
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2840403967961668,
      "learning_rate": 9.88543996010249e-06,
      "loss": 0.2082,
      "step": 16932
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7476007182975146,
      "learning_rate": 9.884448142243298e-06,
      "loss": 0.4198,
      "step": 16933
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8000181789904902,
      "learning_rate": 9.883456325520944e-06,
      "loss": 0.25,
      "step": 16934
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.7211314487827525,
      "learning_rate": 9.882464509945178e-06,
      "loss": 0.7666,
      "step": 16935
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30299838435243226,
      "learning_rate": 9.881472695525772e-06,
      "loss": 0.1825,
      "step": 16936
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.492139236276713,
      "learning_rate": 9.880480882272476e-06,
      "loss": 0.3869,
      "step": 16937
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1431149119835267,
      "learning_rate": 9.879489070195046e-06,
      "loss": 0.361,
      "step": 16938
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.158367249866925,
      "learning_rate": 9.878497259303245e-06,
      "loss": 0.069,
      "step": 16939
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3863103483956737,
      "learning_rate": 9.877505449606827e-06,
      "loss": 0.3195,
      "step": 16940
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2666844327002134,
      "learning_rate": 9.876513641115546e-06,
      "loss": 0.2321,
      "step": 16941
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6340633814909128,
      "learning_rate": 9.87552183383917e-06,
      "loss": 0.3976,
      "step": 16942
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7960836281571636,
      "learning_rate": 9.87453002778745e-06,
      "loss": 0.2751,
      "step": 16943
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4679198600746883,
      "learning_rate": 9.873538222970144e-06,
      "loss": 0.8821,
      "step": 16944
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27889786643624986,
      "learning_rate": 9.872546419397012e-06,
      "loss": 0.1928,
      "step": 16945
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.48467879488993415,
      "learning_rate": 9.87155461707781e-06,
      "loss": 0.4148,
      "step": 16946
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39887102917548656,
      "learning_rate": 9.870562816022291e-06,
      "loss": 0.2294,
      "step": 16947
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.335075474834433,
      "learning_rate": 9.869571016240224e-06,
      "loss": 0.651,
      "step": 16948
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31805290638846007,
      "learning_rate": 9.868579217741359e-06,
      "loss": 0.1386,
      "step": 16949
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.365973297419077,
      "learning_rate": 9.867587420535451e-06,
      "loss": 0.3028,
      "step": 16950
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0067261172137856,
      "learning_rate": 9.866595624632268e-06,
      "loss": 0.3651,
      "step": 16951
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21628987083967954,
      "learning_rate": 9.865603830041557e-06,
      "loss": 0.2014,
      "step": 16952
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7466796395495802,
      "learning_rate": 9.864612036773076e-06,
      "loss": 0.4873,
      "step": 16953
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3759477807636384,
      "learning_rate": 9.863620244836592e-06,
      "loss": 0.1749,
      "step": 16954
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5233872069873058,
      "learning_rate": 9.862628454241858e-06,
      "loss": 0.3988,
      "step": 16955
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.924726778485354,
      "learning_rate": 9.861636664998625e-06,
      "loss": 0.322,
      "step": 16956
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.45352983013195386,
      "learning_rate": 9.860644877116661e-06,
      "loss": 0.329,
      "step": 16957
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35451691993551415,
      "learning_rate": 9.859653090605715e-06,
      "loss": 0.2049,
      "step": 16958
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3073034689318191,
      "learning_rate": 9.85866130547555e-06,
      "loss": 0.2368,
      "step": 16959
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.541059009504092,
      "learning_rate": 9.857669521735921e-06,
      "loss": 0.2465,
      "step": 16960
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4109056908965328,
      "learning_rate": 9.856677739396587e-06,
      "loss": 0.2137,
      "step": 16961
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8850704661999745,
      "learning_rate": 9.855685958467304e-06,
      "loss": 0.483,
      "step": 16962
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6065985143916537,
      "learning_rate": 9.85469417895783e-06,
      "loss": 0.3696,
      "step": 16963
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.271288271341008,
      "learning_rate": 9.85370240087792e-06,
      "loss": 0.2245,
      "step": 16964
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37568819905378364,
      "learning_rate": 9.852710624237338e-06,
      "loss": 0.237,
      "step": 16965
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2044703625388193,
      "learning_rate": 9.851718849045836e-06,
      "loss": 0.5985,
      "step": 16966
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4325958074240171,
      "learning_rate": 9.850727075313171e-06,
      "loss": 0.2953,
      "step": 16967
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3469869808892646,
      "learning_rate": 9.849735303049106e-06,
      "loss": 0.2514,
      "step": 16968
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2497935552677829,
      "learning_rate": 9.848743532263392e-06,
      "loss": 0.1323,
      "step": 16969
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5090567946996314,
      "learning_rate": 9.847751762965786e-06,
      "loss": 0.4006,
      "step": 16970
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.35032484503680983,
      "learning_rate": 9.846759995166052e-06,
      "loss": 0.241,
      "step": 16971
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3967973553357777,
      "learning_rate": 9.845768228873944e-06,
      "loss": 0.2817,
      "step": 16972
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4370144813505216,
      "learning_rate": 9.844776464099213e-06,
      "loss": 0.25,
      "step": 16973
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6168162204003984,
      "learning_rate": 9.84378470085163e-06,
      "loss": 0.8296,
      "step": 16974
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3702761024725099,
      "learning_rate": 9.842792939140941e-06,
      "loss": 0.2912,
      "step": 16975
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36104167906969176,
      "learning_rate": 9.841801178976907e-06,
      "loss": 0.297,
      "step": 16976
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3017310337395726,
      "learning_rate": 9.840809420369286e-06,
      "loss": 0.1837,
      "step": 16977
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7284280386519562,
      "learning_rate": 9.839817663327835e-06,
      "loss": 0.2831,
      "step": 16978
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.33800731479642354,
      "learning_rate": 9.838825907862306e-06,
      "loss": 0.2123,
      "step": 16979
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0650713300179424,
      "learning_rate": 9.837834153982466e-06,
      "loss": 0.6085,
      "step": 16980
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5884972250916184,
      "learning_rate": 9.836842401698064e-06,
      "loss": 0.2709,
      "step": 16981
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.28974708891426604,
      "learning_rate": 9.83585065101886e-06,
      "loss": 0.2242,
      "step": 16982
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4639880427879447,
      "learning_rate": 9.834858901954614e-06,
      "loss": 0.3343,
      "step": 16983
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5901637243382474,
      "learning_rate": 9.833867154515075e-06,
      "loss": 0.3256,
      "step": 16984
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9167838070357135,
      "learning_rate": 9.832875408710011e-06,
      "loss": 0.4932,
      "step": 16985
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3030451983702193,
      "learning_rate": 9.831883664549174e-06,
      "loss": 0.2001,
      "step": 16986
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0976539221473398,
      "learning_rate": 9.830891922042317e-06,
      "loss": 0.5785,
      "step": 16987
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18547876899010562,
      "learning_rate": 9.829900181199203e-06,
      "loss": 0.1648,
      "step": 16988
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3991314278592353,
      "learning_rate": 9.828908442029589e-06,
      "loss": 0.8169,
      "step": 16989
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.347092696646403,
      "learning_rate": 9.827916704543225e-06,
      "loss": 0.1592,
      "step": 16990
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3211222523873901,
      "learning_rate": 9.826924968749876e-06,
      "loss": 0.2189,
      "step": 16991
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7106828669642867,
      "learning_rate": 9.825933234659298e-06,
      "loss": 0.4743,
      "step": 16992
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5086729362481549,
      "learning_rate": 9.824941502281243e-06,
      "loss": 0.2658,
      "step": 16993
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.47062932897007154,
      "learning_rate": 9.823949771625473e-06,
      "loss": 0.3154,
      "step": 16994
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2625257547506642,
      "learning_rate": 9.822958042701744e-06,
      "loss": 0.1869,
      "step": 16995
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1725562652261288,
      "learning_rate": 9.821966315519807e-06,
      "loss": 0.5257,
      "step": 16996
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.19868886108479814,
      "learning_rate": 9.820974590089428e-06,
      "loss": 0.0611,
      "step": 16997
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.307600014626211,
      "learning_rate": 9.81998286642036e-06,
      "loss": 0.8286,
      "step": 16998
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3212182113256155,
      "learning_rate": 9.818991144522358e-06,
      "loss": 0.2013,
      "step": 16999
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.29280712910121093,
      "learning_rate": 9.817999424405183e-06,
      "loss": 0.2742,
      "step": 17000
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36253785605391137,
      "learning_rate": 9.81700770607859e-06,
      "loss": 0.1463,
      "step": 17001
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6167017203479528,
      "learning_rate": 9.816015989552329e-06,
      "loss": 0.4248,
      "step": 17002
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5999113562928369,
      "learning_rate": 9.81502427483617e-06,
      "loss": 0.2074,
      "step": 17003
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2858586767553072,
      "learning_rate": 9.814032561939862e-06,
      "loss": 0.2035,
      "step": 17004
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.218922931129377,
      "learning_rate": 9.813040850873158e-06,
      "loss": 0.6422,
      "step": 17005
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21701968605787092,
      "learning_rate": 9.812049141645824e-06,
      "loss": 0.1877,
      "step": 17006
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2799177876017036,
      "learning_rate": 9.811057434267613e-06,
      "loss": 0.7838,
      "step": 17007
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27869460101458415,
      "learning_rate": 9.81006572874828e-06,
      "loss": 0.1123,
      "step": 17008
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3731037484078084,
      "learning_rate": 9.809074025097582e-06,
      "loss": 0.2796,
      "step": 17009
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8108003351767521,
      "learning_rate": 9.808082323325274e-06,
      "loss": 0.3416,
      "step": 17010
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3010937408767566,
      "learning_rate": 9.80709062344112e-06,
      "loss": 0.2248,
      "step": 17011
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4122064213691765,
      "learning_rate": 9.806098925454873e-06,
      "loss": 0.2373,
      "step": 17012
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7359272205654255,
      "learning_rate": 9.805107229376284e-06,
      "loss": 0.3957,
      "step": 17013
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3163376158092613,
      "learning_rate": 9.804115535215118e-06,
      "loss": 0.2066,
      "step": 17014
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9808012261308358,
      "learning_rate": 9.803123842981127e-06,
      "loss": 0.4973,
      "step": 17015
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7657242150624571,
      "learning_rate": 9.802132152684063e-06,
      "loss": 0.4619,
      "step": 17016
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.25966862312200634,
      "learning_rate": 9.801140464333695e-06,
      "loss": 0.1819,
      "step": 17017
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3123432286390946,
      "learning_rate": 9.800148777939771e-06,
      "loss": 0.2785,
      "step": 17018
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6461724108977318,
      "learning_rate": 9.799157093512046e-06,
      "loss": 0.3307,
      "step": 17019
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4179355483244492,
      "learning_rate": 9.798165411060283e-06,
      "loss": 0.2239,
      "step": 17020
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32794959839655186,
      "learning_rate": 9.797173730594234e-06,
      "loss": 0.0988,
      "step": 17021
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.429187431994261,
      "learning_rate": 9.796182052123653e-06,
      "loss": 0.2374,
      "step": 17022
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34569980315766324,
      "learning_rate": 9.795190375658306e-06,
      "loss": 0.2493,
      "step": 17023
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.46117190854343526,
      "learning_rate": 9.794198701207943e-06,
      "loss": 0.3027,
      "step": 17024
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8472259667976773,
      "learning_rate": 9.793207028782317e-06,
      "loss": 0.5535,
      "step": 17025
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5473709820204529,
      "learning_rate": 9.79221535839119e-06,
      "loss": 0.3557,
      "step": 17026
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2836440131295865,
      "learning_rate": 9.79122369004432e-06,
      "loss": 0.2222,
      "step": 17027
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8716908993340998,
      "learning_rate": 9.790232023751452e-06,
      "loss": 0.5346,
      "step": 17028
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18948735125385155,
      "learning_rate": 9.789240359522356e-06,
      "loss": 0.1457,
      "step": 17029
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4204796905414927,
      "learning_rate": 9.788248697366783e-06,
      "loss": 0.2513,
      "step": 17030
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36720480522514115,
      "learning_rate": 9.787257037294487e-06,
      "loss": 0.2396,
      "step": 17031
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4496407770498269,
      "learning_rate": 9.786265379315229e-06,
      "loss": 0.2047,
      "step": 17032
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5229238798105171,
      "learning_rate": 9.785273723438757e-06,
      "loss": 0.8704,
      "step": 17033
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40302056812180787,
      "learning_rate": 9.784282069674838e-06,
      "loss": 0.303,
      "step": 17034
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38147940729285107,
      "learning_rate": 9.783290418033222e-06,
      "loss": 0.2621,
      "step": 17035
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3187939599598715,
      "learning_rate": 9.782298768523663e-06,
      "loss": 0.215,
      "step": 17036
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6832332368668492,
      "learning_rate": 9.781307121155923e-06,
      "loss": 0.4324,
      "step": 17037
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.18988246626554373,
      "learning_rate": 9.780315475939757e-06,
      "loss": 0.1019,
      "step": 17038
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.326669509207214,
      "learning_rate": 9.779323832884918e-06,
      "loss": 0.5997,
      "step": 17039
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.25356793890856494,
      "learning_rate": 9.778332192001163e-06,
      "loss": 0.07,
      "step": 17040
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36130131787117176,
      "learning_rate": 9.777340553298252e-06,
      "loss": 0.2908,
      "step": 17041
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.28016283896539007,
      "learning_rate": 9.776348916785932e-06,
      "loss": 0.2462,
      "step": 17042
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6987978134904635,
      "learning_rate": 9.77535728247397e-06,
      "loss": 0.3504,
      "step": 17043
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7444416701457556,
      "learning_rate": 9.774365650372116e-06,
      "loss": 0.4225,
      "step": 17044
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2437952350045514,
      "learning_rate": 9.773374020490124e-06,
      "loss": 0.1771,
      "step": 17045
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4646799572812927,
      "learning_rate": 9.772382392837757e-06,
      "loss": 0.7779,
      "step": 17046
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38826908720228837,
      "learning_rate": 9.771390767424766e-06,
      "loss": 0.2381,
      "step": 17047
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2302025059241122,
      "learning_rate": 9.770399144260905e-06,
      "loss": 0.2002,
      "step": 17048
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30155502529757533,
      "learning_rate": 9.769407523355936e-06,
      "loss": 0.1589,
      "step": 17049
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5185021113707662,
      "learning_rate": 9.768415904719612e-06,
      "loss": 0.3375,
      "step": 17050
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2649300340076977,
      "learning_rate": 9.767424288361685e-06,
      "loss": 0.339,
      "step": 17051
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9723063368059449,
      "learning_rate": 9.766432674291919e-06,
      "loss": 0.4585,
      "step": 17052
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2735777773433146,
      "learning_rate": 9.765441062520063e-06,
      "loss": 0.2238,
      "step": 17053
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34286528212623585,
      "learning_rate": 9.76444945305587e-06,
      "loss": 0.3048,
      "step": 17054
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8373498602756848,
      "learning_rate": 9.763457845909109e-06,
      "loss": 0.4804,
      "step": 17055
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3470948610299232,
      "learning_rate": 9.762466241089525e-06,
      "loss": 0.1315,
      "step": 17056
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.40187525971193616,
      "learning_rate": 9.761474638606874e-06,
      "loss": 0.2213,
      "step": 17057
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.34390370618289967,
      "learning_rate": 9.760483038470916e-06,
      "loss": 0.1716,
      "step": 17058
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5791101269291733,
      "learning_rate": 9.7594914406914e-06,
      "loss": 0.3306,
      "step": 17059
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32487836699333417,
      "learning_rate": 9.758499845278092e-06,
      "loss": 0.2529,
      "step": 17060
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8637581089751498,
      "learning_rate": 9.757508252240741e-06,
      "loss": 0.4166,
      "step": 17061
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4397446265944972,
      "learning_rate": 9.756516661589103e-06,
      "loss": 0.2179,
      "step": 17062
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4659420434624087,
      "learning_rate": 9.755525073332935e-06,
      "loss": 0.3139,
      "step": 17063
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39624734942447,
      "learning_rate": 9.754533487481993e-06,
      "loss": 0.2585,
      "step": 17064
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32548945032861804,
      "learning_rate": 9.753541904046025e-06,
      "loss": 0.3042,
      "step": 17065
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2284208420081313,
      "learning_rate": 9.752550323034797e-06,
      "loss": 0.0912,
      "step": 17066
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9363266472946388,
      "learning_rate": 9.751558744458062e-06,
      "loss": 0.3917,
      "step": 17067
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.27581424565109436,
      "learning_rate": 9.750567168325571e-06,
      "loss": 0.1978,
      "step": 17068
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6437778447702923,
      "learning_rate": 9.749575594647086e-06,
      "loss": 0.3243,
      "step": 17069
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8604321797443552,
      "learning_rate": 9.748584023432356e-06,
      "loss": 0.5322,
      "step": 17070
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.29013300577223433,
      "learning_rate": 9.747592454691136e-06,
      "loss": 0.2504,
      "step": 17071
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4677968900598956,
      "learning_rate": 9.74660088843319e-06,
      "loss": 0.2705,
      "step": 17072
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.37847089093439695,
      "learning_rate": 9.745609324668266e-06,
      "loss": 0.2397,
      "step": 17073
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4967427946435712,
      "learning_rate": 9.744617763406118e-06,
      "loss": 0.7254,
      "step": 17074
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.200439146489219,
      "learning_rate": 9.743626204656508e-06,
      "loss": 0.0892,
      "step": 17075
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5793969471865281,
      "learning_rate": 9.742634648429188e-06,
      "loss": 0.3349,
      "step": 17076
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.23108585552775926,
      "learning_rate": 9.74164309473391e-06,
      "loss": 0.2024,
      "step": 17077
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8537075153428705,
      "learning_rate": 9.740651543580437e-06,
      "loss": 0.5536,
      "step": 17078
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.7153421876611918,
      "learning_rate": 9.739659994978516e-06,
      "loss": 0.3446,
      "step": 17079
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3303752598712133,
      "learning_rate": 9.738668448937904e-06,
      "loss": 0.8061,
      "step": 17080
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.25785951232345894,
      "learning_rate": 9.73767690546836e-06,
      "loss": 0.1865,
      "step": 17081
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.5248666296236526,
      "learning_rate": 9.736685364579638e-06,
      "loss": 0.8193,
      "step": 17082
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.276343990143535,
      "learning_rate": 9.735693826281492e-06,
      "loss": 0.2184,
      "step": 17083
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.21690912651819422,
      "learning_rate": 9.734702290583676e-06,
      "loss": 0.1016,
      "step": 17084
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5561216407935977,
      "learning_rate": 9.733710757495942e-06,
      "loss": 0.3705,
      "step": 17085
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32513418548281503,
      "learning_rate": 9.732719227028055e-06,
      "loss": 0.2106,
      "step": 17086
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8072335359737972,
      "learning_rate": 9.731727699189764e-06,
      "loss": 0.4406,
      "step": 17087
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.39006818381848823,
      "learning_rate": 9.730736173990822e-06,
      "loss": 0.2656,
      "step": 17088
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4419190705231263,
      "learning_rate": 9.72974465144099e-06,
      "loss": 0.3409,
      "step": 17089
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24830952488472718,
      "learning_rate": 9.728753131550016e-06,
      "loss": 0.0698,
      "step": 17090
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.38798398076646623,
      "learning_rate": 9.727761614327655e-06,
      "loss": 0.3262,
      "step": 17091
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8335849792686929,
      "learning_rate": 9.726770099783669e-06,
      "loss": 0.0605,
      "step": 17092
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.43485122097618384,
      "learning_rate": 9.725778587927809e-06,
      "loss": 0.2982,
      "step": 17093
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.47560583531497264,
      "learning_rate": 9.724787078769825e-06,
      "loss": 0.2421,
      "step": 17094
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.4928113334411802,
      "learning_rate": 9.723795572319483e-06,
      "loss": 0.2725,
      "step": 17095
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.26829976428045915,
      "learning_rate": 9.722804068586527e-06,
      "loss": 0.2632,
      "step": 17096
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.29013428143628556,
      "learning_rate": 9.721812567580713e-06,
      "loss": 0.1106,
      "step": 17097
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.4505906470561136,
      "learning_rate": 9.720821069311802e-06,
      "loss": 0.7921,
      "step": 17098
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.30821691024674935,
      "learning_rate": 9.719829573789547e-06,
      "loss": 0.1699,
      "step": 17099
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5030051239047928,
      "learning_rate": 9.718838081023697e-06,
      "loss": 0.3688,
      "step": 17100
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3371560844348931,
      "learning_rate": 9.717846591024013e-06,
      "loss": 0.2408,
      "step": 17101
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2478843107336777,
      "learning_rate": 9.716855103800247e-06,
      "loss": 0.6663,
      "step": 17102
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5155296452609162,
      "learning_rate": 9.71586361936215e-06,
      "loss": 0.2915,
      "step": 17103
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32289545292320704,
      "learning_rate": 9.714872137719484e-06,
      "loss": 0.2559,
      "step": 17104
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2514106874865049,
      "learning_rate": 9.713880658882e-06,
      "loss": 0.14,
      "step": 17105
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.44990247270325356,
      "learning_rate": 9.712889182859448e-06,
      "loss": 0.2837,
      "step": 17106
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31546255013208724,
      "learning_rate": 9.71189770966159e-06,
      "loss": 0.2664,
      "step": 17107
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6370534739787601,
      "learning_rate": 9.710906239298178e-06,
      "loss": 0.3097,
      "step": 17108
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3608999185366603,
      "learning_rate": 9.709914771778964e-06,
      "loss": 0.2197,
      "step": 17109
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.2585538244769754,
      "learning_rate": 9.708923307113705e-06,
      "loss": 0.3196,
      "step": 17110
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.601132619015178,
      "learning_rate": 9.70793184531215e-06,
      "loss": 0.4116,
      "step": 17111
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31723307248203403,
      "learning_rate": 9.706940386384063e-06,
      "loss": 0.2091,
      "step": 17112
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.5196204584826645,
      "learning_rate": 9.705948930339193e-06,
      "loss": 0.3274,
      "step": 17113
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.280126758421581,
      "learning_rate": 9.70495747718729e-06,
      "loss": 0.2097,
      "step": 17114
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2766236465079267,
      "learning_rate": 9.703966026938115e-06,
      "loss": 0.1692,
      "step": 17115
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.898154480521586,
      "learning_rate": 9.70297457960142e-06,
      "loss": 0.6154,
      "step": 17116
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0286066324100374,
      "learning_rate": 9.701983135186955e-06,
      "loss": 0.4842,
      "step": 17117
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31856381308827453,
      "learning_rate": 9.700991693704483e-06,
      "loss": 0.219,
      "step": 17118
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3313069193380417,
      "learning_rate": 9.700000255163751e-06,
      "loss": 0.2948,
      "step": 17119
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.8262874342322865,
      "learning_rate": 9.699008819574516e-06,
      "loss": 0.3338,
      "step": 17120
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9169108671600066,
      "learning_rate": 9.698017386946532e-06,
      "loss": 0.5751,
      "step": 17121
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31168467811037975,
      "learning_rate": 9.697025957289551e-06,
      "loss": 0.1713,
      "step": 17122
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.24365878301084876,
      "learning_rate": 9.696034530613325e-06,
      "loss": 0.1328,
      "step": 17123
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3291518913627277,
      "learning_rate": 9.695043106927617e-06,
      "loss": 0.2706,
      "step": 17124
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9237575633608328,
      "learning_rate": 9.694051686242172e-06,
      "loss": 0.4762,
      "step": 17125
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.6210937243963165,
      "learning_rate": 9.693060268566747e-06,
      "loss": 0.311,
      "step": 17126
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.32876922213903226,
      "learning_rate": 9.692068853911098e-06,
      "loss": 0.2321,
      "step": 17127
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9372425322952914,
      "learning_rate": 9.691077442284979e-06,
      "loss": 0.6093,
      "step": 17128
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.9829116450102447,
      "learning_rate": 9.690086033698136e-06,
      "loss": 0.3374,
      "step": 17129
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.31631243370410755,
      "learning_rate": 9.689094628160332e-06,
      "loss": 0.2867,
      "step": 17130
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.2865559306703378,
      "learning_rate": 9.688103225681317e-06,
      "loss": 0.1874,
      "step": 17131
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.1799336568528138,
      "learning_rate": 9.687111826270844e-06,
      "loss": 0.4855,
      "step": 17132
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.236263104771181,
      "learning_rate": 9.68612042993867e-06,
      "loss": 0.0889,
      "step": 17133
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.3471581944870814,
      "learning_rate": 9.685129036694546e-06,
      "loss": 0.7242,
      "step": 17134
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.375748262760295,
      "learning_rate": 9.684137646548221e-06,
      "loss": 0.1561,
      "step": 17135
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3199119260244291,
      "learning_rate": 9.68314625950946e-06,
      "loss": 0.2149,
      "step": 17136
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.334350166594115,
      "learning_rate": 9.682154875588004e-06,
      "loss": 0.2975,
      "step": 17137
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.0415163798559168,
      "learning_rate": 9.681163494793619e-06,
      "loss": 0.3732,
      "step": 17138
    },
    {
      "epoch": 0.52,
      "grad_norm": 1.6927716786741909,
      "learning_rate": 9.680172117136055e-06,
      "loss": 0.7505,
      "step": 17139
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.3768289067709654,
      "learning_rate": 9.679180742625055e-06,
      "loss": 0.0721,
      "step": 17140
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.36477995243829375,
      "learning_rate": 9.678189371270385e-06,
      "loss": 0.2613,
      "step": 17141
    },
    {
      "epoch": 0.52,
      "grad_norm": 0.23306470637577179,
      "learning_rate": 9.677198003081795e-06,
      "loss": 0.194,
      "step": 17142
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2649914008253476,
      "learning_rate": 9.676206638069032e-06,
      "loss": 0.7878,
      "step": 17143
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3538056339654122,
      "learning_rate": 9.675215276241858e-06,
      "loss": 0.1514,
      "step": 17144
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.33812468967428033,
      "learning_rate": 9.674223917610024e-06,
      "loss": 0.2548,
      "step": 17145
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4454323640800633,
      "learning_rate": 9.67323256218328e-06,
      "loss": 0.2344,
      "step": 17146
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9329855991856092,
      "learning_rate": 9.672241209971384e-06,
      "loss": 0.5099,
      "step": 17147
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3000527994450311,
      "learning_rate": 9.671249860984087e-06,
      "loss": 0.2452,
      "step": 17148
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34150695932759934,
      "learning_rate": 9.670258515231137e-06,
      "loss": 0.1751,
      "step": 17149
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4841789467643885,
      "learning_rate": 9.669267172722297e-06,
      "loss": 0.3584,
      "step": 17150
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22470406788772337,
      "learning_rate": 9.668275833467315e-06,
      "loss": 0.0957,
      "step": 17151
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3964486415717638,
      "learning_rate": 9.667284497475942e-06,
      "loss": 0.7437,
      "step": 17152
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5132933027343641,
      "learning_rate": 9.666293164757937e-06,
      "loss": 0.2867,
      "step": 17153
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3494319586311496,
      "learning_rate": 9.66530183532305e-06,
      "loss": 0.2626,
      "step": 17154
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2749086264378582,
      "learning_rate": 9.664310509181028e-06,
      "loss": 0.2484,
      "step": 17155
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4509791352134127,
      "learning_rate": 9.663319186341634e-06,
      "loss": 0.3064,
      "step": 17156
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9498339082739076,
      "learning_rate": 9.662327866814617e-06,
      "loss": 0.308,
      "step": 17157
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42379860225338833,
      "learning_rate": 9.661336550609727e-06,
      "loss": 0.3065,
      "step": 17158
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4382419676989337,
      "learning_rate": 9.660345237736723e-06,
      "loss": 0.2332,
      "step": 17159
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5079193160762525,
      "learning_rate": 9.659353928205349e-06,
      "loss": 0.3924,
      "step": 17160
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3855376049503233,
      "learning_rate": 9.658362622025369e-06,
      "loss": 0.2995,
      "step": 17161
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4607934411468998,
      "learning_rate": 9.657371319206527e-06,
      "loss": 0.2372,
      "step": 17162
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35588789355183564,
      "learning_rate": 9.656380019758578e-06,
      "loss": 0.2705,
      "step": 17163
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6023683006413588,
      "learning_rate": 9.655388723691277e-06,
      "loss": 0.3406,
      "step": 17164
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29746492955859494,
      "learning_rate": 9.654397431014376e-06,
      "loss": 0.1779,
      "step": 17165
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2968783571661903,
      "learning_rate": 9.653406141737622e-06,
      "loss": 0.2439,
      "step": 17166
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8217685286734028,
      "learning_rate": 9.652414855870778e-06,
      "loss": 0.3054,
      "step": 17167
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3218276119465516,
      "learning_rate": 9.651423573423589e-06,
      "loss": 0.2162,
      "step": 17168
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3179201503908806,
      "learning_rate": 9.650432294405809e-06,
      "loss": 0.7042,
      "step": 17169
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7517391631943656,
      "learning_rate": 9.649441018827191e-06,
      "loss": 0.3582,
      "step": 17170
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7553802897482362,
      "learning_rate": 9.64844974669749e-06,
      "loss": 0.4588,
      "step": 17171
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.25247600419293703,
      "learning_rate": 9.647458478026451e-06,
      "loss": 0.1874,
      "step": 17172
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3261001385715806,
      "learning_rate": 9.646467212823836e-06,
      "loss": 0.2951,
      "step": 17173
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.199203938037499,
      "learning_rate": 9.645475951099392e-06,
      "loss": 0.0877,
      "step": 17174
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4406395276863643,
      "learning_rate": 9.64448469286287e-06,
      "loss": 0.0221,
      "step": 17175
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6056210177966757,
      "learning_rate": 9.643493438124027e-06,
      "loss": 0.3135,
      "step": 17176
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35022097433376653,
      "learning_rate": 9.642502186892612e-06,
      "loss": 0.2218,
      "step": 17177
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3199561822521293,
      "learning_rate": 9.641510939178378e-06,
      "loss": 0.3019,
      "step": 17178
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7299979492357337,
      "learning_rate": 9.640519694991078e-06,
      "loss": 0.3247,
      "step": 17179
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7096137824969202,
      "learning_rate": 9.639528454340465e-06,
      "loss": 0.3877,
      "step": 17180
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26127257400696713,
      "learning_rate": 9.638537217236284e-06,
      "loss": 0.1793,
      "step": 17181
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1597389301613283,
      "learning_rate": 9.637545983688298e-06,
      "loss": 0.7398,
      "step": 17182
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.15077154487002217,
      "learning_rate": 9.636554753706255e-06,
      "loss": 0.0687,
      "step": 17183
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30698268760232555,
      "learning_rate": 9.635563527299902e-06,
      "loss": 0.2845,
      "step": 17184
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.42758237988213615,
      "learning_rate": 9.634572304478998e-06,
      "loss": 0.1042,
      "step": 17185
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3597795291986942,
      "learning_rate": 9.633581085253287e-06,
      "loss": 0.2781,
      "step": 17186
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0110521731490931,
      "learning_rate": 9.63258986963253e-06,
      "loss": 0.305,
      "step": 17187
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9415081539699152,
      "learning_rate": 9.631598657626474e-06,
      "loss": 0.3424,
      "step": 17188
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3681051336799074,
      "learning_rate": 9.63060744924487e-06,
      "loss": 0.2857,
      "step": 17189
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.33799592736844897,
      "learning_rate": 9.629616244497475e-06,
      "loss": 0.1668,
      "step": 17190
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3212011449902041,
      "learning_rate": 9.628625043394036e-06,
      "loss": 0.2812,
      "step": 17191
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20919563113192158,
      "learning_rate": 9.627633845944303e-06,
      "loss": 0.0962,
      "step": 17192
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4809498310984163,
      "learning_rate": 9.626642652158034e-06,
      "loss": 0.2437,
      "step": 17193
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8019798918774299,
      "learning_rate": 9.625651462044978e-06,
      "loss": 0.1526,
      "step": 17194
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3608062794502234,
      "learning_rate": 9.624660275614884e-06,
      "loss": 0.2589,
      "step": 17195
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3079344264606696,
      "learning_rate": 9.623669092877507e-06,
      "loss": 0.2443,
      "step": 17196
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8286620749779718,
      "learning_rate": 9.622677913842597e-06,
      "loss": 0.4136,
      "step": 17197
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8030521225296253,
      "learning_rate": 9.621686738519902e-06,
      "loss": 0.3313,
      "step": 17198
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6072893896629216,
      "learning_rate": 9.620695566919183e-06,
      "loss": 0.3288,
      "step": 17199
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.33482386857145385,
      "learning_rate": 9.619704399050184e-06,
      "loss": 0.2397,
      "step": 17200
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.23307277212106228,
      "learning_rate": 9.618713234922657e-06,
      "loss": 0.1926,
      "step": 17201
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4866311934397812,
      "learning_rate": 9.617722074546356e-06,
      "loss": 0.2272,
      "step": 17202
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5242261044837437,
      "learning_rate": 9.616730917931032e-06,
      "loss": 0.1435,
      "step": 17203
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4116057751964581,
      "learning_rate": 9.61573976508643e-06,
      "loss": 0.3036,
      "step": 17204
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.053021941479201,
      "learning_rate": 9.614748616022313e-06,
      "loss": 0.375,
      "step": 17205
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9272985165974512,
      "learning_rate": 9.613757470748423e-06,
      "loss": 0.5506,
      "step": 17206
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30199929992685814,
      "learning_rate": 9.612766329274511e-06,
      "loss": 0.2602,
      "step": 17207
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37398331470914853,
      "learning_rate": 9.611775191610337e-06,
      "loss": 0.2643,
      "step": 17208
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41577757946618327,
      "learning_rate": 9.610784057765645e-06,
      "loss": 0.2813,
      "step": 17209
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.6821200815848427,
      "learning_rate": 9.609792927750185e-06,
      "loss": 0.8241,
      "step": 17210
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18421955963716008,
      "learning_rate": 9.608801801573714e-06,
      "loss": 0.0896,
      "step": 17211
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6425928716998098,
      "learning_rate": 9.607810679245975e-06,
      "loss": 0.3415,
      "step": 17212
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29604633232109945,
      "learning_rate": 9.606819560776727e-06,
      "loss": 0.2274,
      "step": 17213
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3001490271037095,
      "learning_rate": 9.605828446175718e-06,
      "loss": 0.2319,
      "step": 17214
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6989215537089224,
      "learning_rate": 9.604837335452695e-06,
      "loss": 0.4727,
      "step": 17215
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7840115249044293,
      "learning_rate": 9.603846228617417e-06,
      "loss": 0.6186,
      "step": 17216
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5986277299821833,
      "learning_rate": 9.60285512567963e-06,
      "loss": 0.3001,
      "step": 17217
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35942991353431986,
      "learning_rate": 9.601864026649078e-06,
      "loss": 0.2179,
      "step": 17218
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4550581860955139,
      "learning_rate": 9.600872931535526e-06,
      "loss": 0.3498,
      "step": 17219
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18046818339753687,
      "learning_rate": 9.599881840348715e-06,
      "loss": 0.1231,
      "step": 17220
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8929151559457335,
      "learning_rate": 9.598890753098398e-06,
      "loss": 0.4609,
      "step": 17221
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2905191483771221,
      "learning_rate": 9.597899669794326e-06,
      "loss": 0.2026,
      "step": 17222
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5918878483206848,
      "learning_rate": 9.59690859044625e-06,
      "loss": 0.4198,
      "step": 17223
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6652151638562798,
      "learning_rate": 9.595917515063915e-06,
      "loss": 0.3758,
      "step": 17224
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3502384089148027,
      "learning_rate": 9.594926443657084e-06,
      "loss": 0.3074,
      "step": 17225
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3001750035165427,
      "learning_rate": 9.593935376235497e-06,
      "loss": 0.1203,
      "step": 17226
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3246099144797512,
      "learning_rate": 9.592944312808905e-06,
      "loss": 0.2393,
      "step": 17227
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0703085696553791,
      "learning_rate": 9.591953253387064e-06,
      "loss": 0.5296,
      "step": 17228
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.15492559919502658,
      "learning_rate": 9.59096219797972e-06,
      "loss": 0.0694,
      "step": 17229
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6104903815319201,
      "learning_rate": 9.589971146596621e-06,
      "loss": 0.437,
      "step": 17230
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.27527486645322335,
      "learning_rate": 9.588980099247528e-06,
      "loss": 0.1913,
      "step": 17231
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30774810903510913,
      "learning_rate": 9.58798905594218e-06,
      "loss": 0.2901,
      "step": 17232
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1000874943496295,
      "learning_rate": 9.58699801669033e-06,
      "loss": 0.2141,
      "step": 17233
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3347441490111578,
      "learning_rate": 9.586006981501732e-06,
      "loss": 0.8081,
      "step": 17234
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28654447472108857,
      "learning_rate": 9.585015950386134e-06,
      "loss": 0.1789,
      "step": 17235
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5321766809466225,
      "learning_rate": 9.58402492335328e-06,
      "loss": 0.3489,
      "step": 17236
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2142927663504124,
      "learning_rate": 9.58303390041293e-06,
      "loss": 0.2366,
      "step": 17237
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3262413974367957,
      "learning_rate": 9.582042881574826e-06,
      "loss": 0.2896,
      "step": 17238
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5498876256295772,
      "learning_rate": 9.581051866848729e-06,
      "loss": 0.2748,
      "step": 17239
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28252309508488593,
      "learning_rate": 9.580060856244379e-06,
      "loss": 0.1875,
      "step": 17240
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4396946403472804,
      "learning_rate": 9.579069849771523e-06,
      "loss": 0.2678,
      "step": 17241
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2915845162904332,
      "learning_rate": 9.578078847439922e-06,
      "loss": 0.141,
      "step": 17242
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3092028355331444,
      "learning_rate": 9.577087849259319e-06,
      "loss": 0.2751,
      "step": 17243
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2415737048441335,
      "learning_rate": 9.57609685523946e-06,
      "loss": 0.1144,
      "step": 17244
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3841401583164964,
      "learning_rate": 9.575105865390106e-06,
      "loss": 0.2898,
      "step": 17245
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.356807137822953,
      "learning_rate": 9.574114879721e-06,
      "loss": 0.2166,
      "step": 17246
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7705464294629134,
      "learning_rate": 9.573123898241889e-06,
      "loss": 0.4511,
      "step": 17247
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3720841324003366,
      "learning_rate": 9.572132920962529e-06,
      "loss": 0.2647,
      "step": 17248
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6630769385835107,
      "learning_rate": 9.571141947892663e-06,
      "loss": 0.3468,
      "step": 17249
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18717516013523222,
      "learning_rate": 9.570150979042043e-06,
      "loss": 0.178,
      "step": 17250
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45448383931664066,
      "learning_rate": 9.56916001442042e-06,
      "loss": 0.308,
      "step": 17251
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7685564256039205,
      "learning_rate": 9.568169054037545e-06,
      "loss": 0.5607,
      "step": 17252
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35669358846778354,
      "learning_rate": 9.567178097903162e-06,
      "loss": 0.1285,
      "step": 17253
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38877476344344075,
      "learning_rate": 9.566187146027024e-06,
      "loss": 0.3147,
      "step": 17254
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2921512536264145,
      "learning_rate": 9.565196198418883e-06,
      "loss": 0.2466,
      "step": 17255
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8592276265602826,
      "learning_rate": 9.564205255088478e-06,
      "loss": 0.5193,
      "step": 17256
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6852372116922598,
      "learning_rate": 9.56321431604557e-06,
      "loss": 0.3266,
      "step": 17257
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3677809617698546,
      "learning_rate": 9.562223381299904e-06,
      "loss": 0.2613,
      "step": 17258
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18610582587355412,
      "learning_rate": 9.561232450861225e-06,
      "loss": 0.1175,
      "step": 17259
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8183106214014018,
      "learning_rate": 9.560241524739288e-06,
      "loss": 0.8909,
      "step": 17260
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26523057806940514,
      "learning_rate": 9.55925060294384e-06,
      "loss": 0.2556,
      "step": 17261
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7190386927584684,
      "learning_rate": 9.558259685484625e-06,
      "loss": 0.3787,
      "step": 17262
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29122157371133855,
      "learning_rate": 9.5572687723714e-06,
      "loss": 0.2196,
      "step": 17263
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2169151049136928,
      "learning_rate": 9.556277863613911e-06,
      "loss": 0.6346,
      "step": 17264
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7772104262255581,
      "learning_rate": 9.555286959221907e-06,
      "loss": 0.3392,
      "step": 17265
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4615208966175749,
      "learning_rate": 9.554296059205136e-06,
      "loss": 0.3869,
      "step": 17266
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2383558163776907,
      "learning_rate": 9.553305163573342e-06,
      "loss": 0.1706,
      "step": 17267
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3484269886638004,
      "learning_rate": 9.552314272336283e-06,
      "loss": 0.23,
      "step": 17268
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.25993909036389756,
      "learning_rate": 9.551323385503706e-06,
      "loss": 0.1738,
      "step": 17269
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8019310979066598,
      "learning_rate": 9.550332503085353e-06,
      "loss": 0.5392,
      "step": 17270
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6592162225518603,
      "learning_rate": 9.54934162509098e-06,
      "loss": 0.3027,
      "step": 17271
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2916437268700786,
      "learning_rate": 9.548350751530331e-06,
      "loss": 0.1998,
      "step": 17272
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4232770312763517,
      "learning_rate": 9.547359882413153e-06,
      "loss": 0.3498,
      "step": 17273
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4135966653508684,
      "learning_rate": 9.5463690177492e-06,
      "loss": 0.2813,
      "step": 17274
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0786646074813933,
      "learning_rate": 9.54537815754822e-06,
      "loss": 0.5099,
      "step": 17275
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2986847402678946,
      "learning_rate": 9.544387301819954e-06,
      "loss": 0.0694,
      "step": 17276
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.39941167415111817,
      "learning_rate": 9.54339645057416e-06,
      "loss": 0.3138,
      "step": 17277
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.259954617894003,
      "learning_rate": 9.542405603820581e-06,
      "loss": 0.1633,
      "step": 17278
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3540119779275657,
      "learning_rate": 9.541414761568966e-06,
      "loss": 0.3018,
      "step": 17279
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28105390994669627,
      "learning_rate": 9.540423923829065e-06,
      "loss": 0.1298,
      "step": 17280
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26871695332496165,
      "learning_rate": 9.539433090610624e-06,
      "loss": 0.1822,
      "step": 17281
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6460763197293043,
      "learning_rate": 9.538442261923387e-06,
      "loss": 0.4122,
      "step": 17282
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9019803019511926,
      "learning_rate": 9.537451437777113e-06,
      "loss": 0.2653,
      "step": 17283
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4601691758885446,
      "learning_rate": 9.536460618181541e-06,
      "loss": 0.3263,
      "step": 17284
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2552121414083247,
      "learning_rate": 9.535469803146423e-06,
      "loss": 0.1985,
      "step": 17285
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5350072448070715,
      "learning_rate": 9.534478992681506e-06,
      "loss": 0.3631,
      "step": 17286
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0026308976384724,
      "learning_rate": 9.533488186796535e-06,
      "loss": 0.3392,
      "step": 17287
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3646086179542205,
      "learning_rate": 9.532497385501263e-06,
      "loss": 0.9327,
      "step": 17288
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.385012166505328,
      "learning_rate": 9.531506588805438e-06,
      "loss": 0.1813,
      "step": 17289
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3404866617797804,
      "learning_rate": 9.530515796718801e-06,
      "loss": 0.2846,
      "step": 17290
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4971004743960614,
      "learning_rate": 9.529525009251108e-06,
      "loss": 0.2579,
      "step": 17291
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3025202750631785,
      "learning_rate": 9.528534226412102e-06,
      "loss": 0.2519,
      "step": 17292
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9181109177128962,
      "learning_rate": 9.527543448211527e-06,
      "loss": 0.2573,
      "step": 17293
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.289758429213411,
      "learning_rate": 9.52655267465914e-06,
      "loss": 0.1627,
      "step": 17294
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.530413293226148,
      "learning_rate": 9.525561905764683e-06,
      "loss": 0.3342,
      "step": 17295
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9527265062033252,
      "learning_rate": 9.524571141537902e-06,
      "loss": 0.5316,
      "step": 17296
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34676411466476287,
      "learning_rate": 9.523580381988547e-06,
      "loss": 0.3211,
      "step": 17297
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26743452903652715,
      "learning_rate": 9.522589627126368e-06,
      "loss": 0.1357,
      "step": 17298
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5119463299761815,
      "learning_rate": 9.521598876961103e-06,
      "loss": 0.3522,
      "step": 17299
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.43960373851802864,
      "learning_rate": 9.52060813150251e-06,
      "loss": 0.2767,
      "step": 17300
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.387657974924848,
      "learning_rate": 9.519617390760333e-06,
      "loss": 0.2411,
      "step": 17301
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2918259612346396,
      "learning_rate": 9.518626654744317e-06,
      "loss": 0.2366,
      "step": 17302
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.019092266159518,
      "learning_rate": 9.51763592346421e-06,
      "loss": 0.5717,
      "step": 17303
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2671061832698854,
      "learning_rate": 9.51664519692976e-06,
      "loss": 0.1803,
      "step": 17304
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4436823776772183,
      "learning_rate": 9.515654475150711e-06,
      "loss": 0.7957,
      "step": 17305
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7661519296733416,
      "learning_rate": 9.514663758136815e-06,
      "loss": 0.3658,
      "step": 17306
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7563912780361681,
      "learning_rate": 9.513673045897818e-06,
      "loss": 0.3158,
      "step": 17307
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38847756195420824,
      "learning_rate": 9.512682338443461e-06,
      "loss": 0.2441,
      "step": 17308
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2720667340546487,
      "learning_rate": 9.5116916357835e-06,
      "loss": 0.2322,
      "step": 17309
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2786336255352127,
      "learning_rate": 9.510700937927675e-06,
      "loss": 0.1603,
      "step": 17310
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9264585054245189,
      "learning_rate": 9.509710244885737e-06,
      "loss": 0.2977,
      "step": 17311
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7656637345616388,
      "learning_rate": 9.50871955666743e-06,
      "loss": 0.3357,
      "step": 17312
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.30444424359605765,
      "learning_rate": 9.507728873282499e-06,
      "loss": 0.2166,
      "step": 17313
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4385586512933406,
      "learning_rate": 9.506738194740697e-06,
      "loss": 0.8487,
      "step": 17314
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31330127905596705,
      "learning_rate": 9.505747521051767e-06,
      "loss": 0.255,
      "step": 17315
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.915378339564538,
      "learning_rate": 9.504756852225453e-06,
      "loss": 0.42,
      "step": 17316
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.27303571950551603,
      "learning_rate": 9.503766188271506e-06,
      "loss": 0.1787,
      "step": 17317
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.630975774333196,
      "learning_rate": 9.502775529199671e-06,
      "loss": 0.8336,
      "step": 17318
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.15822836768383128,
      "learning_rate": 9.50178487501969e-06,
      "loss": 0.0708,
      "step": 17319
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3092396196407621,
      "learning_rate": 9.500794225741318e-06,
      "loss": 0.2408,
      "step": 17320
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6015375320946107,
      "learning_rate": 9.499803581374296e-06,
      "loss": 0.2951,
      "step": 17321
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36215900477982826,
      "learning_rate": 9.49881294192837e-06,
      "loss": 0.2181,
      "step": 17322
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.426708783054996,
      "learning_rate": 9.497822307413287e-06,
      "loss": 0.8099,
      "step": 17323
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8370241330905785,
      "learning_rate": 9.496831677838795e-06,
      "loss": 0.3564,
      "step": 17324
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4464815568743552,
      "learning_rate": 9.495841053214633e-06,
      "loss": 0.3528,
      "step": 17325
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2528687759162694,
      "learning_rate": 9.494850433550557e-06,
      "loss": 0.0683,
      "step": 17326
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.278134065826923,
      "learning_rate": 9.49385981885631e-06,
      "loss": 0.2753,
      "step": 17327
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.21155815276074105,
      "learning_rate": 9.492869209141634e-06,
      "loss": 0.0974,
      "step": 17328
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2501160235527096,
      "learning_rate": 9.49187860441628e-06,
      "loss": 0.6987,
      "step": 17329
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5060858001963178,
      "learning_rate": 9.49088800468999e-06,
      "loss": 0.1397,
      "step": 17330
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35906545330753875,
      "learning_rate": 9.489897409972509e-06,
      "loss": 0.2924,
      "step": 17331
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5226658152930478,
      "learning_rate": 9.488906820273589e-06,
      "loss": 0.3171,
      "step": 17332
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.530939813728696,
      "learning_rate": 9.487916235602971e-06,
      "loss": 0.2913,
      "step": 17333
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.876374206737047,
      "learning_rate": 9.4869256559704e-06,
      "loss": 0.5724,
      "step": 17334
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31045614042575675,
      "learning_rate": 9.485935081385625e-06,
      "loss": 0.1922,
      "step": 17335
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4890719238005445,
      "learning_rate": 9.48494451185839e-06,
      "loss": 0.3361,
      "step": 17336
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1633975886876082,
      "learning_rate": 9.483953947398436e-06,
      "loss": 0.0737,
      "step": 17337
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4426418352866131,
      "learning_rate": 9.482963388015516e-06,
      "loss": 0.3557,
      "step": 17338
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3454595135659563,
      "learning_rate": 9.481972833719373e-06,
      "loss": 0.1984,
      "step": 17339
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3504889587875937,
      "learning_rate": 9.480982284519751e-06,
      "loss": 0.2628,
      "step": 17340
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0690075142626667,
      "learning_rate": 9.479991740426398e-06,
      "loss": 0.354,
      "step": 17341
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0346615106209627,
      "learning_rate": 9.479001201449053e-06,
      "loss": 0.4173,
      "step": 17342
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4266460451039564,
      "learning_rate": 9.47801066759747e-06,
      "loss": 0.3034,
      "step": 17343
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.346005094036584,
      "learning_rate": 9.477020138881388e-06,
      "loss": 0.2991,
      "step": 17344
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.32994620911189076,
      "learning_rate": 9.47602961531055e-06,
      "loss": 0.2135,
      "step": 17345
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3403040277770218,
      "learning_rate": 9.475039096894711e-06,
      "loss": 0.1275,
      "step": 17346
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.44941968242112085,
      "learning_rate": 9.47404858364361e-06,
      "loss": 0.2627,
      "step": 17347
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3920323499822713,
      "learning_rate": 9.473058075566988e-06,
      "loss": 0.1237,
      "step": 17348
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5716374078376839,
      "learning_rate": 9.472067572674597e-06,
      "loss": 0.3743,
      "step": 17349
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5557635333421026,
      "learning_rate": 9.471077074976179e-06,
      "loss": 0.276,
      "step": 17350
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7018636652922917,
      "learning_rate": 9.470086582481476e-06,
      "loss": 0.3728,
      "step": 17351
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2229704879083987,
      "learning_rate": 9.469096095200238e-06,
      "loss": 0.6159,
      "step": 17352
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9303544306686485,
      "learning_rate": 9.468105613142207e-06,
      "loss": 0.4816,
      "step": 17353
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.279162778317299,
      "learning_rate": 9.467115136317128e-06,
      "loss": 0.1926,
      "step": 17354
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5291201646143305,
      "learning_rate": 9.466124664734746e-06,
      "loss": 0.2568,
      "step": 17355
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2232463714018178,
      "learning_rate": 9.465134198404806e-06,
      "loss": 0.1906,
      "step": 17356
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5857518878089116,
      "learning_rate": 9.464143737337047e-06,
      "loss": 0.6433,
      "step": 17357
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.341845657859248,
      "learning_rate": 9.463153281541223e-06,
      "loss": 0.1937,
      "step": 17358
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3886636066294883,
      "learning_rate": 9.462162831027074e-06,
      "loss": 0.2827,
      "step": 17359
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1717815490958392,
      "learning_rate": 9.461172385804341e-06,
      "loss": 0.6134,
      "step": 17360
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2708861600254309,
      "learning_rate": 9.460181945882774e-06,
      "loss": 0.4967,
      "step": 17361
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34786064343975176,
      "learning_rate": 9.459191511272114e-06,
      "loss": 0.2472,
      "step": 17362
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2818921443075066,
      "learning_rate": 9.458201081982103e-06,
      "loss": 0.2383,
      "step": 17363
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3815109955715957,
      "learning_rate": 9.45721065802249e-06,
      "loss": 0.2239,
      "step": 17364
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31363275133027213,
      "learning_rate": 9.456220239403016e-06,
      "loss": 0.1111,
      "step": 17365
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6610768458028227,
      "learning_rate": 9.455229826133429e-06,
      "loss": 0.4355,
      "step": 17366
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2614704737543614,
      "learning_rate": 9.454239418223472e-06,
      "loss": 0.1776,
      "step": 17367
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4342139282407304,
      "learning_rate": 9.45324901568288e-06,
      "loss": 0.3301,
      "step": 17368
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4586478757346505,
      "learning_rate": 9.452258618521408e-06,
      "loss": 0.2592,
      "step": 17369
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4129464393429405,
      "learning_rate": 9.451268226748797e-06,
      "loss": 0.8521,
      "step": 17370
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28978804199489916,
      "learning_rate": 9.45027784037479e-06,
      "loss": 0.1221,
      "step": 17371
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.34471100663242205,
      "learning_rate": 9.44928745940913e-06,
      "loss": 0.2198,
      "step": 17372
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.301675165493631,
      "learning_rate": 9.44829708386156e-06,
      "loss": 0.7041,
      "step": 17373
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.1850388240932362,
      "learning_rate": 9.447306713741823e-06,
      "loss": 0.1586,
      "step": 17374
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6139269736931188,
      "learning_rate": 9.446316349059667e-06,
      "loss": 0.3531,
      "step": 17375
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.35388150709036137,
      "learning_rate": 9.445325989824834e-06,
      "loss": 0.1697,
      "step": 17376
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4579229605214721,
      "learning_rate": 9.444335636047062e-06,
      "loss": 0.3529,
      "step": 17377
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9669942917664192,
      "learning_rate": 9.443345287736102e-06,
      "loss": 0.3827,
      "step": 17378
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5417569283687216,
      "learning_rate": 9.442354944901696e-06,
      "loss": 0.4201,
      "step": 17379
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3549018736545095,
      "learning_rate": 9.441364607553583e-06,
      "loss": 0.2096,
      "step": 17380
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.36161721832041016,
      "learning_rate": 9.44037427570151e-06,
      "loss": 0.2764,
      "step": 17381
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9353595929843148,
      "learning_rate": 9.43938394935522e-06,
      "loss": 0.3076,
      "step": 17382
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7379469202423646,
      "learning_rate": 9.438393628524451e-06,
      "loss": 0.4387,
      "step": 17383
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.631140418944318,
      "learning_rate": 9.437403313218954e-06,
      "loss": 0.3074,
      "step": 17384
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.22234901958699022,
      "learning_rate": 9.436413003448468e-06,
      "loss": 0.1324,
      "step": 17385
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.28184167527086973,
      "learning_rate": 9.435422699222735e-06,
      "loss": 0.2868,
      "step": 17386
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2567954933713517,
      "learning_rate": 9.434432400551501e-06,
      "loss": 0.1208,
      "step": 17387
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.5642005751734618,
      "learning_rate": 9.433442107444507e-06,
      "loss": 0.9093,
      "step": 17388
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31970449096386466,
      "learning_rate": 9.432451819911492e-06,
      "loss": 0.1436,
      "step": 17389
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3331937531045618,
      "learning_rate": 9.431461537962207e-06,
      "loss": 0.2514,
      "step": 17390
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.46797504637588283,
      "learning_rate": 9.430471261606388e-06,
      "loss": 0.2893,
      "step": 17391
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4784599682747886,
      "learning_rate": 9.42948099085378e-06,
      "loss": 0.3819,
      "step": 17392
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8684973503267959,
      "learning_rate": 9.42849072571413e-06,
      "loss": 0.3503,
      "step": 17393
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3861048893255241,
      "learning_rate": 9.42750046619717e-06,
      "loss": 0.2831,
      "step": 17394
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.20140444751713946,
      "learning_rate": 9.426510212312653e-06,
      "loss": 0.1215,
      "step": 17395
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3811958945859102,
      "learning_rate": 9.425519964070317e-06,
      "loss": 0.6266,
      "step": 17396
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3355787283603829,
      "learning_rate": 9.424529721479903e-06,
      "loss": 0.2662,
      "step": 17397
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2592329141417823,
      "learning_rate": 9.423539484551158e-06,
      "loss": 0.1691,
      "step": 17398
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.38429945357250467,
      "learning_rate": 9.42254925329382e-06,
      "loss": 0.3334,
      "step": 17399
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1883075763896207,
      "learning_rate": 9.421559027717628e-06,
      "loss": 0.2358,
      "step": 17400
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.790298987812157,
      "learning_rate": 9.420568807832334e-06,
      "loss": 0.4452,
      "step": 17401
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3584585024793745,
      "learning_rate": 9.419578593647673e-06,
      "loss": 0.1514,
      "step": 17402
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6661176029668795,
      "learning_rate": 9.418588385173388e-06,
      "loss": 0.354,
      "step": 17403
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2798121279589213,
      "learning_rate": 9.417598182419224e-06,
      "loss": 0.2399,
      "step": 17404
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2715455679642267,
      "learning_rate": 9.41660798539492e-06,
      "loss": 0.2425,
      "step": 17405
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8727231145849151,
      "learning_rate": 9.415617794110215e-06,
      "loss": 0.4787,
      "step": 17406
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.8219733086116094,
      "learning_rate": 9.414627608574858e-06,
      "loss": 0.9534,
      "step": 17407
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2723002865777599,
      "learning_rate": 9.413637428798588e-06,
      "loss": 0.191,
      "step": 17408
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6793143981297052,
      "learning_rate": 9.412647254791141e-06,
      "loss": 0.4588,
      "step": 17409
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3123162623924394,
      "learning_rate": 9.411657086562269e-06,
      "loss": 0.2645,
      "step": 17410
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9796758554786484,
      "learning_rate": 9.410666924121709e-06,
      "loss": 0.2316,
      "step": 17411
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6234610527566656,
      "learning_rate": 9.409676767479194e-06,
      "loss": 0.3587,
      "step": 17412
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.23018120880959436,
      "learning_rate": 9.40868661664448e-06,
      "loss": 0.1523,
      "step": 17413
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.323471106635992,
      "learning_rate": 9.407696471627297e-06,
      "loss": 0.8721,
      "step": 17414
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.37953180666178454,
      "learning_rate": 9.406706332437395e-06,
      "loss": 0.2527,
      "step": 17415
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4888139067431042,
      "learning_rate": 9.40571619908451e-06,
      "loss": 0.3816,
      "step": 17416
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2764599621291749,
      "learning_rate": 9.404726071578385e-06,
      "loss": 0.1931,
      "step": 17417
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5643723407796105,
      "learning_rate": 9.40373594992876e-06,
      "loss": 0.4191,
      "step": 17418
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6436295465731219,
      "learning_rate": 9.402745834145379e-06,
      "loss": 0.2995,
      "step": 17419
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.1655842434680603,
      "learning_rate": 9.401755724237975e-06,
      "loss": 0.6872,
      "step": 17420
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3045968727704801,
      "learning_rate": 9.400765620216302e-06,
      "loss": 0.1869,
      "step": 17421
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.29818868701805196,
      "learning_rate": 9.399775522090091e-06,
      "loss": 0.3009,
      "step": 17422
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2330747318145938,
      "learning_rate": 9.398785429869086e-06,
      "loss": 0.1013,
      "step": 17423
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2756126020221202,
      "learning_rate": 9.39779534356303e-06,
      "loss": 0.1474,
      "step": 17424
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.633366129335511,
      "learning_rate": 9.39680526318166e-06,
      "loss": 0.398,
      "step": 17425
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2936083856281917,
      "learning_rate": 9.395815188734714e-06,
      "loss": 0.1838,
      "step": 17426
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7523383534125516,
      "learning_rate": 9.394825120231941e-06,
      "loss": 0.4849,
      "step": 17427
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31075185215141193,
      "learning_rate": 9.39383505768308e-06,
      "loss": 0.2605,
      "step": 17428
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.218108170289673,
      "learning_rate": 9.392845001097866e-06,
      "loss": 0.5315,
      "step": 17429
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6109911593294305,
      "learning_rate": 9.391854950486043e-06,
      "loss": 0.1567,
      "step": 17430
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31789999425386706,
      "learning_rate": 9.390864905857353e-06,
      "loss": 0.2591,
      "step": 17431
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.21584166443797717,
      "learning_rate": 9.389874867221529e-06,
      "loss": 0.1432,
      "step": 17432
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3234472816905958,
      "learning_rate": 9.388884834588323e-06,
      "loss": 0.3078,
      "step": 17433
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.41024206868188523,
      "learning_rate": 9.387894807967466e-06,
      "loss": 0.1498,
      "step": 17434
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6268517882065043,
      "learning_rate": 9.386904787368703e-06,
      "loss": 0.2832,
      "step": 17435
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.384956486154931,
      "learning_rate": 9.385914772801773e-06,
      "loss": 0.2589,
      "step": 17436
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6665505132869708,
      "learning_rate": 9.384924764276413e-06,
      "loss": 0.0353,
      "step": 17437
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3057830214903403,
      "learning_rate": 9.383934761802364e-06,
      "loss": 0.6812,
      "step": 17438
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3206121522636044,
      "learning_rate": 9.382944765389371e-06,
      "loss": 0.2049,
      "step": 17439
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2851171596239593,
      "learning_rate": 9.381954775047168e-06,
      "loss": 0.2704,
      "step": 17440
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9917401712195553,
      "learning_rate": 9.380964790785501e-06,
      "loss": 0.4768,
      "step": 17441
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.3502176240343735,
      "learning_rate": 9.379974812614104e-06,
      "loss": 0.8494,
      "step": 17442
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6267170922895406,
      "learning_rate": 9.378984840542716e-06,
      "loss": 0.3312,
      "step": 17443
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3944074562097554,
      "learning_rate": 9.377994874581084e-06,
      "loss": 0.2652,
      "step": 17444
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.18712150230097802,
      "learning_rate": 9.377004914738941e-06,
      "loss": 0.1596,
      "step": 17445
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.45483711205555627,
      "learning_rate": 9.376014961026026e-06,
      "loss": 0.3334,
      "step": 17446
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.2495473896492566,
      "learning_rate": 9.375025013452085e-06,
      "loss": 0.2313,
      "step": 17447
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.7675039364424574,
      "learning_rate": 9.374035072026853e-06,
      "loss": 0.347,
      "step": 17448
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2732798844159496,
      "learning_rate": 9.373045136760069e-06,
      "loss": 0.2107,
      "step": 17449
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.8804942224302652,
      "learning_rate": 9.372055207661475e-06,
      "loss": 0.4603,
      "step": 17450
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3322053869776221,
      "learning_rate": 9.37106528474081e-06,
      "loss": 0.3056,
      "step": 17451
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.5376179541538929,
      "learning_rate": 9.370075368007804e-06,
      "loss": 0.3285,
      "step": 17452
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.4044811706791691,
      "learning_rate": 9.36908545747221e-06,
      "loss": 0.285,
      "step": 17453
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.47404222350755176,
      "learning_rate": 9.36809555314376e-06,
      "loss": 0.2181,
      "step": 17454
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.2579520150669597,
      "learning_rate": 9.367105655032194e-06,
      "loss": 0.1547,
      "step": 17455
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.3988532219329598,
      "learning_rate": 9.36611576314725e-06,
      "loss": 0.2129,
      "step": 17456
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.406116943555977,
      "learning_rate": 9.36512587749867e-06,
      "loss": 0.3174,
      "step": 17457
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31137721135069485,
      "learning_rate": 9.364135998096186e-06,
      "loss": 0.1989,
      "step": 17458
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4525684398725516,
      "learning_rate": 9.363146124949544e-06,
      "loss": 0.871,
      "step": 17459
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.6130109334760183,
      "learning_rate": 9.36215625806848e-06,
      "loss": 0.3456,
      "step": 17460
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0515117356864074,
      "learning_rate": 9.361166397462733e-06,
      "loss": 0.5263,
      "step": 17461
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.31085596912266666,
      "learning_rate": 9.36017654314204e-06,
      "loss": 0.0706,
      "step": 17462
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.271072226110151,
      "learning_rate": 9.359186695116143e-06,
      "loss": 0.2225,
      "step": 17463
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.23039883446623768,
      "learning_rate": 9.358196853394773e-06,
      "loss": 0.2041,
      "step": 17464
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.0921425000419736,
      "learning_rate": 9.357207017987678e-06,
      "loss": 0.3106,
      "step": 17465
    },
    {
      "epoch": 0.53,
      "grad_norm": 1.4450608363405464,
      "learning_rate": 9.35621718890459e-06,
      "loss": 0.8066,
      "step": 17466
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.26715353346206966,
      "learning_rate": 9.35522736615525e-06,
      "loss": 0.1838,
      "step": 17467
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.9587098724995962,
      "learning_rate": 9.354237549749398e-06,
      "loss": 0.4668,
      "step": 17468
    },
    {
      "epoch": 0.53,
      "grad_norm": 0.292090482106933,
      "learning_rate": 9.353247739696762e-06,
      "loss": 0.2401,
      "step": 17469
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.697699688091592,
      "learning_rate": 9.352257936007095e-06,
      "loss": 0.4872,
      "step": 17470
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2857258054246977,
      "learning_rate": 9.351268138690124e-06,
      "loss": 0.1732,
      "step": 17471
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5037887056382808,
      "learning_rate": 9.350278347755591e-06,
      "loss": 0.334,
      "step": 17472
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22525203159328372,
      "learning_rate": 9.349288563213234e-06,
      "loss": 0.0719,
      "step": 17473
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9633727435346338,
      "learning_rate": 9.348298785072793e-06,
      "loss": 0.5538,
      "step": 17474
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2840942807087017,
      "learning_rate": 9.347309013343995e-06,
      "loss": 0.2295,
      "step": 17475
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3307384770496202,
      "learning_rate": 9.346319248036593e-06,
      "loss": 0.2722,
      "step": 17476
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9424574574532525,
      "learning_rate": 9.345329489160315e-06,
      "loss": 0.4829,
      "step": 17477
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.021998594876286,
      "learning_rate": 9.344339736724898e-06,
      "loss": 0.3458,
      "step": 17478
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.04896975385368,
      "learning_rate": 9.343349990740088e-06,
      "loss": 0.6844,
      "step": 17479
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30184578939579104,
      "learning_rate": 9.342360251215614e-06,
      "loss": 0.1712,
      "step": 17480
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5197767105260193,
      "learning_rate": 9.341370518161216e-06,
      "loss": 0.336,
      "step": 17481
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.189299756457787,
      "learning_rate": 9.340380791586633e-06,
      "loss": 0.1582,
      "step": 17482
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7579788989238228,
      "learning_rate": 9.3393910715016e-06,
      "loss": 0.8263,
      "step": 17483
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8251540699005774,
      "learning_rate": 9.338401357915852e-06,
      "loss": 0.338,
      "step": 17484
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5944269736145257,
      "learning_rate": 9.337411650839134e-06,
      "loss": 0.3351,
      "step": 17485
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3894771831472349,
      "learning_rate": 9.336421950281176e-06,
      "loss": 0.2529,
      "step": 17486
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3246389328449542,
      "learning_rate": 9.335432256251717e-06,
      "loss": 0.3104,
      "step": 17487
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2746134901682797,
      "learning_rate": 9.334442568760494e-06,
      "loss": 0.5671,
      "step": 17488
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8554783415063234,
      "learning_rate": 9.333452887817247e-06,
      "loss": 0.4348,
      "step": 17489
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29175519292221247,
      "learning_rate": 9.332463213431704e-06,
      "loss": 0.2092,
      "step": 17490
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.23341386360586225,
      "learning_rate": 9.331473545613611e-06,
      "loss": 0.107,
      "step": 17491
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4844165058620925,
      "learning_rate": 9.3304838843727e-06,
      "loss": 0.3611,
      "step": 17492
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4464508787246368,
      "learning_rate": 9.329494229718711e-06,
      "loss": 0.2776,
      "step": 17493
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3632159668676153,
      "learning_rate": 9.328504581661379e-06,
      "loss": 0.2576,
      "step": 17494
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6435471650072859,
      "learning_rate": 9.327514940210436e-06,
      "loss": 0.2852,
      "step": 17495
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7355864812558275,
      "learning_rate": 9.326525305375625e-06,
      "loss": 0.8361,
      "step": 17496
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0809843188169441,
      "learning_rate": 9.32553567716668e-06,
      "loss": 0.4933,
      "step": 17497
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41612002059864156,
      "learning_rate": 9.324546055593335e-06,
      "loss": 0.3304,
      "step": 17498
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25581566176540105,
      "learning_rate": 9.32355644066533e-06,
      "loss": 0.2117,
      "step": 17499
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.326579373655436,
      "learning_rate": 9.3225668323924e-06,
      "loss": 0.5453,
      "step": 17500
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.15873549427846054,
      "learning_rate": 9.321577230784275e-06,
      "loss": 0.0723,
      "step": 17501
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6427161805194079,
      "learning_rate": 9.320587635850703e-06,
      "loss": 0.4264,
      "step": 17502
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2686281033925076,
      "learning_rate": 9.319598047601411e-06,
      "loss": 0.182,
      "step": 17503
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6406010398104156,
      "learning_rate": 9.318608466046135e-06,
      "loss": 0.3259,
      "step": 17504
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36211080626723186,
      "learning_rate": 9.317618891194616e-06,
      "loss": 0.3177,
      "step": 17505
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9768521431757015,
      "learning_rate": 9.316629323056588e-06,
      "loss": 0.5465,
      "step": 17506
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6703184891553351,
      "learning_rate": 9.31563976164178e-06,
      "loss": 0.3853,
      "step": 17507
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.337913860883088,
      "learning_rate": 9.314650206959938e-06,
      "loss": 0.2115,
      "step": 17508
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39842072452482113,
      "learning_rate": 9.313660659020792e-06,
      "loss": 0.25,
      "step": 17509
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2361796090169685,
      "learning_rate": 9.312671117834076e-06,
      "loss": 0.1939,
      "step": 17510
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.766190552588089,
      "learning_rate": 9.311681583409532e-06,
      "loss": 0.449,
      "step": 17511
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2799731791372632,
      "learning_rate": 9.310692055756891e-06,
      "loss": 0.0718,
      "step": 17512
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3553858834905141,
      "learning_rate": 9.309702534885882e-06,
      "loss": 0.2988,
      "step": 17513
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9110661142567199,
      "learning_rate": 9.308713020806251e-06,
      "loss": 0.4642,
      "step": 17514
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1580905368569436,
      "learning_rate": 9.30772351352773e-06,
      "loss": 0.6635,
      "step": 17515
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3279083426490804,
      "learning_rate": 9.306734013060049e-06,
      "loss": 0.186,
      "step": 17516
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2345860532875233,
      "learning_rate": 9.30574451941295e-06,
      "loss": 0.2023,
      "step": 17517
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.7730025943044287,
      "learning_rate": 9.304755032596162e-06,
      "loss": 0.8114,
      "step": 17518
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29402474763384623,
      "learning_rate": 9.303765552619426e-06,
      "loss": 0.1526,
      "step": 17519
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6282259007318088,
      "learning_rate": 9.302776079492475e-06,
      "loss": 0.4145,
      "step": 17520
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2361464003088165,
      "learning_rate": 9.301786613225035e-06,
      "loss": 0.1309,
      "step": 17521
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4945671892739079,
      "learning_rate": 9.300797153826854e-06,
      "loss": 0.3607,
      "step": 17522
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30553336946986487,
      "learning_rate": 9.29980770130766e-06,
      "loss": 0.2428,
      "step": 17523
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3234612965841164,
      "learning_rate": 9.298818255677187e-06,
      "loss": 0.8127,
      "step": 17524
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.560684172778433,
      "learning_rate": 9.297828816945172e-06,
      "loss": 0.2015,
      "step": 17525
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34730708702564217,
      "learning_rate": 9.296839385121348e-06,
      "loss": 0.2596,
      "step": 17526
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9911304780315808,
      "learning_rate": 9.295849960215445e-06,
      "loss": 0.0352,
      "step": 17527
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4249613604084356,
      "learning_rate": 9.294860542237208e-06,
      "loss": 0.2966,
      "step": 17528
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3019752299045527,
      "learning_rate": 9.293871131196364e-06,
      "loss": 0.2614,
      "step": 17529
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21087344393761703,
      "learning_rate": 9.292881727102647e-06,
      "loss": 0.1318,
      "step": 17530
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5272968640891513,
      "learning_rate": 9.291892329965792e-06,
      "loss": 0.3381,
      "step": 17531
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8304103785043224,
      "learning_rate": 9.290902939795535e-06,
      "loss": 0.5448,
      "step": 17532
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0867175400006779,
      "learning_rate": 9.289913556601604e-06,
      "loss": 0.5793,
      "step": 17533
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2470708879200068,
      "learning_rate": 9.288924180393741e-06,
      "loss": 0.2129,
      "step": 17534
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5181496411127405,
      "learning_rate": 9.287934811181674e-06,
      "loss": 0.3656,
      "step": 17535
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4332131559769205,
      "learning_rate": 9.286945448975139e-06,
      "loss": 0.2207,
      "step": 17536
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6055989113372401,
      "learning_rate": 9.285956093783871e-06,
      "loss": 0.4155,
      "step": 17537
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2781450780796258,
      "learning_rate": 9.284966745617601e-06,
      "loss": 0.1334,
      "step": 17538
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8961534124604217,
      "learning_rate": 9.283977404486059e-06,
      "loss": 0.4974,
      "step": 17539
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29751735637919813,
      "learning_rate": 9.282988070398988e-06,
      "loss": 0.206,
      "step": 17540
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32057525662019837,
      "learning_rate": 9.281998743366114e-06,
      "loss": 0.2952,
      "step": 17541
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9797658554527349,
      "learning_rate": 9.281009423397175e-06,
      "loss": 0.3818,
      "step": 17542
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0357274557302312,
      "learning_rate": 9.280020110501901e-06,
      "loss": 0.2201,
      "step": 17543
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3762965270501176,
      "learning_rate": 9.279030804690023e-06,
      "loss": 0.2518,
      "step": 17544
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4921835502712377,
      "learning_rate": 9.278041505971281e-06,
      "loss": 0.2897,
      "step": 17545
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4167214076125469,
      "learning_rate": 9.277052214355404e-06,
      "loss": 0.3367,
      "step": 17546
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3995577046241661,
      "learning_rate": 9.27606292985212e-06,
      "loss": 0.256,
      "step": 17547
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6183580664335147,
      "learning_rate": 9.275073652471173e-06,
      "loss": 0.3104,
      "step": 17548
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.23930327847781244,
      "learning_rate": 9.27408438222229e-06,
      "loss": 0.1824,
      "step": 17549
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4663509054223016,
      "learning_rate": 9.273095119115202e-06,
      "loss": 0.8261,
      "step": 17550
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.18826370079749985,
      "learning_rate": 9.272105863159644e-06,
      "loss": 0.092,
      "step": 17551
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.31402102888717137,
      "learning_rate": 9.271116614365347e-06,
      "loss": 0.2854,
      "step": 17552
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.27137336487420427,
      "learning_rate": 9.270127372742043e-06,
      "loss": 0.1883,
      "step": 17553
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7584191686963672,
      "learning_rate": 9.269138138299468e-06,
      "loss": 0.4589,
      "step": 17554
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5735962986530476,
      "learning_rate": 9.268148911047354e-06,
      "loss": 0.3224,
      "step": 17555
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22171437869761937,
      "learning_rate": 9.267159690995429e-06,
      "loss": 0.0696,
      "step": 17556
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6273675056086763,
      "learning_rate": 9.266170478153432e-06,
      "loss": 0.3691,
      "step": 17557
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.31939724449594276,
      "learning_rate": 9.26518127253109e-06,
      "loss": 0.2131,
      "step": 17558
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32876914075470626,
      "learning_rate": 9.26419207413813e-06,
      "loss": 0.3054,
      "step": 17559
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3258529045167659,
      "learning_rate": 9.263202882984299e-06,
      "loss": 0.1601,
      "step": 17560
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6418529720151827,
      "learning_rate": 9.262213699079317e-06,
      "loss": 0.4497,
      "step": 17561
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19902753490161842,
      "learning_rate": 9.261224522432918e-06,
      "loss": 0.0703,
      "step": 17562
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3266439533215981,
      "learning_rate": 9.260235353054839e-06,
      "loss": 0.2987,
      "step": 17563
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2845939031113055,
      "learning_rate": 9.259246190954806e-06,
      "loss": 0.2454,
      "step": 17564
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3564889496079218,
      "learning_rate": 9.25825703614255e-06,
      "loss": 0.6085,
      "step": 17565
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34495111996888234,
      "learning_rate": 9.257267888627808e-06,
      "loss": 0.0686,
      "step": 17566
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32051765538976623,
      "learning_rate": 9.256278748420307e-06,
      "loss": 0.271,
      "step": 17567
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7431907726363368,
      "learning_rate": 9.255289615529784e-06,
      "loss": 0.6176,
      "step": 17568
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25399337075938294,
      "learning_rate": 9.254300489965967e-06,
      "loss": 0.1238,
      "step": 17569
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33239919446687316,
      "learning_rate": 9.25331137173858e-06,
      "loss": 0.3201,
      "step": 17570
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3038241138702719,
      "learning_rate": 9.252322260857369e-06,
      "loss": 0.1942,
      "step": 17571
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.42251276752914524,
      "learning_rate": 9.251333157332056e-06,
      "loss": 0.3563,
      "step": 17572
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0929014766367822,
      "learning_rate": 9.250344061172371e-06,
      "loss": 0.2774,
      "step": 17573
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.2571072376642025,
      "learning_rate": 9.24935497238805e-06,
      "loss": 0.6149,
      "step": 17574
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2738964800001179,
      "learning_rate": 9.248365890988823e-06,
      "loss": 0.1844,
      "step": 17575
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3365118445331676,
      "learning_rate": 9.247376816984414e-06,
      "loss": 0.2639,
      "step": 17576
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40076211487230795,
      "learning_rate": 9.246387750384566e-06,
      "loss": 0.2813,
      "step": 17577
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35541812556881125,
      "learning_rate": 9.245398691199003e-06,
      "loss": 0.2227,
      "step": 17578
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.44637689396153846,
      "learning_rate": 9.24440963943745e-06,
      "loss": 0.2417,
      "step": 17579
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3765209954822979,
      "learning_rate": 9.243420595109648e-06,
      "loss": 0.2679,
      "step": 17580
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4092521588117008,
      "learning_rate": 9.242431558225324e-06,
      "loss": 0.2403,
      "step": 17581
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.31782710976561207,
      "learning_rate": 9.241442528794205e-06,
      "loss": 0.241,
      "step": 17582
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6031424359323143,
      "learning_rate": 9.240453506826025e-06,
      "loss": 0.8623,
      "step": 17583
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34200745768457735,
      "learning_rate": 9.239464492330516e-06,
      "loss": 0.1334,
      "step": 17584
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.571047973900313,
      "learning_rate": 9.2384754853174e-06,
      "loss": 0.3927,
      "step": 17585
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4383977329552726,
      "learning_rate": 9.237486485796417e-06,
      "loss": 0.2885,
      "step": 17586
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7229128746719531,
      "learning_rate": 9.236497493777293e-06,
      "loss": 0.4281,
      "step": 17587
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3145789314220869,
      "learning_rate": 9.235508509269756e-06,
      "loss": 0.2664,
      "step": 17588
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8444678145078487,
      "learning_rate": 9.234519532283539e-06,
      "loss": 0.3556,
      "step": 17589
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.23017156117730306,
      "learning_rate": 9.23353056282837e-06,
      "loss": 0.1618,
      "step": 17590
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4801267302447439,
      "learning_rate": 9.232541600913978e-06,
      "loss": 0.274,
      "step": 17591
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.08246740409128,
      "learning_rate": 9.231552646550096e-06,
      "loss": 0.3255,
      "step": 17592
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5100434425837703,
      "learning_rate": 9.23056369974645e-06,
      "loss": 0.3824,
      "step": 17593
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25182604821652854,
      "learning_rate": 9.229574760512775e-06,
      "loss": 0.1957,
      "step": 17594
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4013861909807765,
      "learning_rate": 9.228585828858796e-06,
      "loss": 0.2737,
      "step": 17595
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8363614098158598,
      "learning_rate": 9.22759690479424e-06,
      "loss": 0.3941,
      "step": 17596
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.861250738651878,
      "learning_rate": 9.226607988328844e-06,
      "loss": 0.3107,
      "step": 17597
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5811644600060446,
      "learning_rate": 9.225619079472332e-06,
      "loss": 0.2902,
      "step": 17598
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1957507783064919,
      "learning_rate": 9.224630178234433e-06,
      "loss": 0.1481,
      "step": 17599
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3016522704944245,
      "learning_rate": 9.22364128462488e-06,
      "loss": 0.2703,
      "step": 17600
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3840729294427691,
      "learning_rate": 9.2226523986534e-06,
      "loss": 0.0156,
      "step": 17601
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3071246713352593,
      "learning_rate": 9.221663520329716e-06,
      "loss": 0.8191,
      "step": 17602
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2844692749477643,
      "learning_rate": 9.220674649663566e-06,
      "loss": 0.1953,
      "step": 17603
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3667680916728289,
      "learning_rate": 9.219685786664677e-06,
      "loss": 0.6549,
      "step": 17604
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1550575282838367,
      "learning_rate": 9.218696931342773e-06,
      "loss": 0.3517,
      "step": 17605
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32313205129162687,
      "learning_rate": 9.217708083707587e-06,
      "loss": 0.2887,
      "step": 17606
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4495591666324586,
      "learning_rate": 9.216719243768847e-06,
      "loss": 0.1434,
      "step": 17607
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33160468657143816,
      "learning_rate": 9.215730411536276e-06,
      "loss": 0.2162,
      "step": 17608
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2633541721279153,
      "learning_rate": 9.214741587019612e-06,
      "loss": 0.1667,
      "step": 17609
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0020757638728264,
      "learning_rate": 9.213752770228579e-06,
      "loss": 0.2525,
      "step": 17610
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3963340675986101,
      "learning_rate": 9.212763961172903e-06,
      "loss": 0.3033,
      "step": 17611
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35010629527108805,
      "learning_rate": 9.211775159862316e-06,
      "loss": 0.1856,
      "step": 17612
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3733890770053105,
      "learning_rate": 9.210786366306544e-06,
      "loss": 0.3281,
      "step": 17613
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3291988806765325,
      "learning_rate": 9.209797580515312e-06,
      "loss": 0.3894,
      "step": 17614
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3310781667538967,
      "learning_rate": 9.208808802498355e-06,
      "loss": 0.8172,
      "step": 17615
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3577294898433764,
      "learning_rate": 9.207820032265396e-06,
      "loss": 0.0741,
      "step": 17616
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.31372327911795034,
      "learning_rate": 9.206831269826162e-06,
      "loss": 0.2577,
      "step": 17617
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.186187918074293,
      "learning_rate": 9.205842515190386e-06,
      "loss": 0.1707,
      "step": 17618
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1472191920528554,
      "learning_rate": 9.204853768367792e-06,
      "loss": 0.5282,
      "step": 17619
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9502641157769188,
      "learning_rate": 9.203865029368109e-06,
      "loss": 0.3022,
      "step": 17620
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3408428775618796,
      "learning_rate": 9.202876298201065e-06,
      "loss": 0.2123,
      "step": 17621
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4060859709025004,
      "learning_rate": 9.201887574876381e-06,
      "loss": 0.3316,
      "step": 17622
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2874823704278295,
      "learning_rate": 9.200898859403794e-06,
      "loss": 0.2664,
      "step": 17623
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6314391149654084,
      "learning_rate": 9.199910151793026e-06,
      "loss": 0.804,
      "step": 17624
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.43335298657550175,
      "learning_rate": 9.198921452053805e-06,
      "loss": 0.141,
      "step": 17625
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35008463530952205,
      "learning_rate": 9.19793276019586e-06,
      "loss": 0.2793,
      "step": 17626
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21300046328853436,
      "learning_rate": 9.196944076228917e-06,
      "loss": 0.0978,
      "step": 17627
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3354856349320534,
      "learning_rate": 9.195955400162696e-06,
      "loss": 0.5724,
      "step": 17628
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2891302790452402,
      "learning_rate": 9.194966732006936e-06,
      "loss": 0.2249,
      "step": 17629
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4785718553602848,
      "learning_rate": 9.193978071771358e-06,
      "loss": 0.2549,
      "step": 17630
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.36876737993618375,
      "learning_rate": 9.192989419465686e-06,
      "loss": 0.2817,
      "step": 17631
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0339352850151162,
      "learning_rate": 9.192000775099654e-06,
      "loss": 0.5686,
      "step": 17632
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0076751450543446,
      "learning_rate": 9.191012138682982e-06,
      "loss": 0.54,
      "step": 17633
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3675587567669835,
      "learning_rate": 9.190023510225395e-06,
      "loss": 0.0687,
      "step": 17634
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39802164945306207,
      "learning_rate": 9.18903488973663e-06,
      "loss": 0.2864,
      "step": 17635
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21316539547013275,
      "learning_rate": 9.188046277226403e-06,
      "loss": 0.1897,
      "step": 17636
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4690506959354621,
      "learning_rate": 9.187057672704444e-06,
      "loss": 0.2623,
      "step": 17637
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.516352060200457,
      "learning_rate": 9.18606907618048e-06,
      "loss": 0.315,
      "step": 17638
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5901507182408647,
      "learning_rate": 9.185080487664237e-06,
      "loss": 0.289,
      "step": 17639
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3191816746208351,
      "learning_rate": 9.184091907165435e-06,
      "loss": 0.245,
      "step": 17640
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4888903970699964,
      "learning_rate": 9.18310333469381e-06,
      "loss": 0.4075,
      "step": 17641
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3763921649276099,
      "learning_rate": 9.182114770259083e-06,
      "loss": 0.257,
      "step": 17642
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.872546322975815,
      "learning_rate": 9.18112621387098e-06,
      "loss": 0.4587,
      "step": 17643
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29652811104016874,
      "learning_rate": 9.180137665539226e-06,
      "loss": 0.1991,
      "step": 17644
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.49491815224132585,
      "learning_rate": 9.179149125273546e-06,
      "loss": 0.271,
      "step": 17645
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3908157271034034,
      "learning_rate": 9.17816059308367e-06,
      "loss": 0.2177,
      "step": 17646
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4228408097220278,
      "learning_rate": 9.177172068979318e-06,
      "loss": 0.2973,
      "step": 17647
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2928150413555226,
      "learning_rate": 9.176183552970217e-06,
      "loss": 0.206,
      "step": 17648
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3331185991423729,
      "learning_rate": 9.175195045066096e-06,
      "loss": 0.2317,
      "step": 17649
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3229844148468686,
      "learning_rate": 9.174206545276678e-06,
      "loss": 0.8767,
      "step": 17650
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0530470019434872,
      "learning_rate": 9.173218053611685e-06,
      "loss": 0.332,
      "step": 17651
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0726916931470045,
      "learning_rate": 9.172229570080849e-06,
      "loss": 0.5628,
      "step": 17652
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28735431688339363,
      "learning_rate": 9.171241094693887e-06,
      "loss": 0.194,
      "step": 17653
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21515242774398205,
      "learning_rate": 9.170252627460528e-06,
      "loss": 0.21,
      "step": 17654
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29296939798729144,
      "learning_rate": 9.169264168390498e-06,
      "loss": 0.1015,
      "step": 17655
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5913790897205863,
      "learning_rate": 9.168275717493522e-06,
      "loss": 0.4199,
      "step": 17656
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32495579101428207,
      "learning_rate": 9.16728727477932e-06,
      "loss": 0.2046,
      "step": 17657
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5094433401636616,
      "learning_rate": 9.16629884025762e-06,
      "loss": 0.3557,
      "step": 17658
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35556046063390645,
      "learning_rate": 9.165310413938151e-06,
      "loss": 0.225,
      "step": 17659
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.45727422600228956,
      "learning_rate": 9.164321995830626e-06,
      "loss": 0.3422,
      "step": 17660
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29033453950114513,
      "learning_rate": 9.16333358594478e-06,
      "loss": 0.0574,
      "step": 17661
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.28702622316221743,
      "learning_rate": 9.162345184290334e-06,
      "loss": 0.2012,
      "step": 17662
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5539059806480215,
      "learning_rate": 9.161356790877009e-06,
      "loss": 0.5471,
      "step": 17663
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4287623819153945,
      "learning_rate": 9.160368405714537e-06,
      "loss": 0.2227,
      "step": 17664
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.33929860584772814,
      "learning_rate": 9.159380028812633e-06,
      "loss": 0.3103,
      "step": 17665
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.12567438349199053,
      "learning_rate": 9.158391660181022e-06,
      "loss": 0.072,
      "step": 17666
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3921235893817322,
      "learning_rate": 9.157403299829434e-06,
      "loss": 0.295,
      "step": 17667
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.106999198764715,
      "learning_rate": 9.156414947767592e-06,
      "loss": 0.515,
      "step": 17668
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1808123054050548,
      "learning_rate": 9.155426604005213e-06,
      "loss": 0.6117,
      "step": 17669
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.40533475335140534,
      "learning_rate": 9.154438268552029e-06,
      "loss": 0.2458,
      "step": 17670
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6207083830042796,
      "learning_rate": 9.153449941417754e-06,
      "loss": 0.3071,
      "step": 17671
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2935009346010208,
      "learning_rate": 9.152461622612121e-06,
      "loss": 0.2276,
      "step": 17672
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7033432961055636,
      "learning_rate": 9.151473312144851e-06,
      "loss": 0.4353,
      "step": 17673
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3584390839558505,
      "learning_rate": 9.150485010025662e-06,
      "loss": 0.1689,
      "step": 17674
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22230133520666245,
      "learning_rate": 9.149496716264284e-06,
      "loss": 0.0962,
      "step": 17675
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3490794435515108,
      "learning_rate": 9.148508430870438e-06,
      "loss": 0.2875,
      "step": 17676
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32577725893513326,
      "learning_rate": 9.14752015385384e-06,
      "loss": 0.253,
      "step": 17677
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5955160226694787,
      "learning_rate": 9.146531885224224e-06,
      "loss": 0.9665,
      "step": 17678
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0976863330503526,
      "learning_rate": 9.145543624991309e-06,
      "loss": 0.2806,
      "step": 17679
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3841696442123716,
      "learning_rate": 9.144555373164813e-06,
      "loss": 0.2735,
      "step": 17680
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4300519680759879,
      "learning_rate": 9.143567129754465e-06,
      "loss": 0.2816,
      "step": 17681
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8916559459243163,
      "learning_rate": 9.142578894769987e-06,
      "loss": 0.5625,
      "step": 17682
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.26596641096754653,
      "learning_rate": 9.141590668221097e-06,
      "loss": 0.2204,
      "step": 17683
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2845423557573443,
      "learning_rate": 9.140602450117522e-06,
      "loss": 0.2023,
      "step": 17684
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30630246706166386,
      "learning_rate": 9.139614240468984e-06,
      "loss": 0.2079,
      "step": 17685
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7082258560455514,
      "learning_rate": 9.1386260392852e-06,
      "loss": 0.8741,
      "step": 17686
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9596983930679979,
      "learning_rate": 9.137637846575902e-06,
      "loss": 0.3818,
      "step": 17687
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3648510902774094,
      "learning_rate": 9.136649662350805e-06,
      "loss": 0.2323,
      "step": 17688
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.32220989887737783,
      "learning_rate": 9.13566148661963e-06,
      "loss": 0.2607,
      "step": 17689
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4345113955154714,
      "learning_rate": 9.134673319392105e-06,
      "loss": 0.2654,
      "step": 17690
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5942422249582283,
      "learning_rate": 9.133685160677948e-06,
      "loss": 0.4177,
      "step": 17691
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34463967329833356,
      "learning_rate": 9.132697010486878e-06,
      "loss": 0.1357,
      "step": 17692
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5382225161124046,
      "learning_rate": 9.131708868828624e-06,
      "loss": 0.2926,
      "step": 17693
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30603497637810895,
      "learning_rate": 9.130720735712901e-06,
      "loss": 0.2068,
      "step": 17694
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.471113489041178,
      "learning_rate": 9.129732611149436e-06,
      "loss": 0.368,
      "step": 17695
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.26177322361454697,
      "learning_rate": 9.128744495147948e-06,
      "loss": 0.1778,
      "step": 17696
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.0294831967756841,
      "learning_rate": 9.127756387718155e-06,
      "loss": 0.5043,
      "step": 17697
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.34021694307760053,
      "learning_rate": 9.126768288869783e-06,
      "loss": 0.1755,
      "step": 17698
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35895940041346996,
      "learning_rate": 9.125780198612555e-06,
      "loss": 0.3029,
      "step": 17699
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39318611463980924,
      "learning_rate": 9.124792116956186e-06,
      "loss": 0.2615,
      "step": 17700
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4072420737872253,
      "learning_rate": 9.123804043910403e-06,
      "loss": 0.2192,
      "step": 17701
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5175383424288847,
      "learning_rate": 9.122815979484922e-06,
      "loss": 0.3198,
      "step": 17702
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.257816218915846,
      "learning_rate": 9.121827923689463e-06,
      "loss": 0.1789,
      "step": 17703
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.347920431595766,
      "learning_rate": 9.120839876533755e-06,
      "loss": 0.8549,
      "step": 17704
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.171368362028621,
      "learning_rate": 9.119851838027513e-06,
      "loss": 0.0681,
      "step": 17705
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41280464021418334,
      "learning_rate": 9.118863808180454e-06,
      "loss": 0.3577,
      "step": 17706
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25550983588908377,
      "learning_rate": 9.117875787002309e-06,
      "loss": 0.1895,
      "step": 17707
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4913194989914686,
      "learning_rate": 9.11688777450279e-06,
      "loss": 0.3354,
      "step": 17708
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8385530554500753,
      "learning_rate": 9.115899770691613e-06,
      "loss": 0.3095,
      "step": 17709
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.3319102131290184,
      "learning_rate": 9.114911775578513e-06,
      "loss": 0.6722,
      "step": 17710
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5598158483402664,
      "learning_rate": 9.1139237891732e-06,
      "loss": 0.2363,
      "step": 17711
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3191720203012798,
      "learning_rate": 9.112935811485396e-06,
      "loss": 0.2683,
      "step": 17712
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.41836048029553957,
      "learning_rate": 9.111947842524821e-06,
      "loss": 0.2635,
      "step": 17713
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.19089352749044333,
      "learning_rate": 9.110959882301198e-06,
      "loss": 0.1405,
      "step": 17714
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7345837125213771,
      "learning_rate": 9.109971930824238e-06,
      "loss": 0.4371,
      "step": 17715
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.355899654577898,
      "learning_rate": 9.108983988103673e-06,
      "loss": 0.195,
      "step": 17716
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5333118092128516,
      "learning_rate": 9.107996054149214e-06,
      "loss": 0.3787,
      "step": 17717
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39417211407669445,
      "learning_rate": 9.107008128970583e-06,
      "loss": 0.238,
      "step": 17718
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.46301104185665076,
      "learning_rate": 9.106020212577501e-06,
      "loss": 0.2839,
      "step": 17719
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30055256431133087,
      "learning_rate": 9.105032304979685e-06,
      "loss": 0.1175,
      "step": 17720
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5265619456033319,
      "learning_rate": 9.104044406186859e-06,
      "loss": 0.3436,
      "step": 17721
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4149256673619267,
      "learning_rate": 9.103056516208736e-06,
      "loss": 0.2793,
      "step": 17722
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4048493992385545,
      "learning_rate": 9.102068635055037e-06,
      "loss": 0.2923,
      "step": 17723
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.279788840496901,
      "learning_rate": 9.101080762735485e-06,
      "loss": 0.2422,
      "step": 17724
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.639656246025642,
      "learning_rate": 9.100092899259795e-06,
      "loss": 0.4444,
      "step": 17725
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21277234754005273,
      "learning_rate": 9.099105044637689e-06,
      "loss": 0.1549,
      "step": 17726
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7785531219885118,
      "learning_rate": 9.098117198878882e-06,
      "loss": 0.2854,
      "step": 17727
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.6618600381213826,
      "learning_rate": 9.097129361993097e-06,
      "loss": 0.8574,
      "step": 17728
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.802434647843734,
      "learning_rate": 9.096141533990044e-06,
      "loss": 0.5419,
      "step": 17729
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3800804416737515,
      "learning_rate": 9.095153714879455e-06,
      "loss": 0.263,
      "step": 17730
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2717859746701329,
      "learning_rate": 9.09416590467104e-06,
      "loss": 0.2424,
      "step": 17731
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8000471792779081,
      "learning_rate": 9.093178103374518e-06,
      "loss": 0.4939,
      "step": 17732
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.49789170837440566,
      "learning_rate": 9.092190310999609e-06,
      "loss": 0.3247,
      "step": 17733
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3371787907284803,
      "learning_rate": 9.09120252755603e-06,
      "loss": 0.1297,
      "step": 17734
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3091705843186474,
      "learning_rate": 9.090214753053496e-06,
      "loss": 0.2033,
      "step": 17735
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.298799717968196,
      "learning_rate": 9.089226987501734e-06,
      "loss": 0.704,
      "step": 17736
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30007065671904176,
      "learning_rate": 9.088239230910455e-06,
      "loss": 0.2353,
      "step": 17737
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.1509169568193305,
      "learning_rate": 9.087251483289376e-06,
      "loss": 0.6384,
      "step": 17738
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2579190275583728,
      "learning_rate": 9.086263744648221e-06,
      "loss": 0.1821,
      "step": 17739
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9909352990090186,
      "learning_rate": 9.085276014996703e-06,
      "loss": 0.4706,
      "step": 17740
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6784947926247494,
      "learning_rate": 9.084288294344536e-06,
      "loss": 0.3999,
      "step": 17741
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2821437277967923,
      "learning_rate": 9.083300582701447e-06,
      "loss": 0.2523,
      "step": 17742
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6535234234275288,
      "learning_rate": 9.082312880077148e-06,
      "loss": 0.386,
      "step": 17743
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.30115423358447935,
      "learning_rate": 9.081325186481356e-06,
      "loss": 0.194,
      "step": 17744
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2524283853171737,
      "learning_rate": 9.08033750192379e-06,
      "loss": 0.1475,
      "step": 17745
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.026882109857668,
      "learning_rate": 9.079349826414162e-06,
      "loss": 0.3312,
      "step": 17746
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4393606705105664,
      "learning_rate": 9.078362159962198e-06,
      "loss": 0.3458,
      "step": 17747
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.6174313010943042,
      "learning_rate": 9.07737450257761e-06,
      "loss": 0.0717,
      "step": 17748
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.25806116848466654,
      "learning_rate": 9.076386854270112e-06,
      "loss": 0.2357,
      "step": 17749
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7326848839001534,
      "learning_rate": 9.075399215049428e-06,
      "loss": 0.3295,
      "step": 17750
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4258318452498775,
      "learning_rate": 9.074411584925269e-06,
      "loss": 0.6169,
      "step": 17751
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.20293639178572564,
      "learning_rate": 9.073423963907353e-06,
      "loss": 0.0751,
      "step": 17752
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.21729071626477062,
      "learning_rate": 9.072436352005398e-06,
      "loss": 0.1617,
      "step": 17753
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4803630408367558,
      "learning_rate": 9.07144874922912e-06,
      "loss": 0.3412,
      "step": 17754
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.039543675207161,
      "learning_rate": 9.07046115558823e-06,
      "loss": 0.2546,
      "step": 17755
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4141433791163875,
      "learning_rate": 9.069473571092454e-06,
      "loss": 0.8535,
      "step": 17756
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2679468659582268,
      "learning_rate": 9.068485995751504e-06,
      "loss": 0.167,
      "step": 17757
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4634975965321059,
      "learning_rate": 9.06749842957509e-06,
      "loss": 0.3286,
      "step": 17758
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8129299399421966,
      "learning_rate": 9.066510872572938e-06,
      "loss": 0.3343,
      "step": 17759
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3009675622912245,
      "learning_rate": 9.065523324754759e-06,
      "loss": 0.2896,
      "step": 17760
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5971488684197211,
      "learning_rate": 9.064535786130264e-06,
      "loss": 0.1921,
      "step": 17761
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3447711354492934,
      "learning_rate": 9.063548256709178e-06,
      "loss": 0.2626,
      "step": 17762
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.17540778144248334,
      "learning_rate": 9.062560736501213e-06,
      "loss": 0.0727,
      "step": 17763
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7291589716926288,
      "learning_rate": 9.061573225516082e-06,
      "loss": 0.8796,
      "step": 17764
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4175616643023779,
      "learning_rate": 9.060585723763504e-06,
      "loss": 0.283,
      "step": 17765
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.27968114395736926,
      "learning_rate": 9.059598231253193e-06,
      "loss": 0.2071,
      "step": 17766
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5928668926075387,
      "learning_rate": 9.058610747994859e-06,
      "loss": 0.3566,
      "step": 17767
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9536386115362308,
      "learning_rate": 9.057623273998227e-06,
      "loss": 0.3691,
      "step": 17768
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.7219988703640303,
      "learning_rate": 9.056635809273008e-06,
      "loss": 0.9988,
      "step": 17769
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.223841867666531,
      "learning_rate": 9.055648353828915e-06,
      "loss": 0.0906,
      "step": 17770
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5775649171011839,
      "learning_rate": 9.054660907675663e-06,
      "loss": 0.341,
      "step": 17771
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.22401396742220242,
      "learning_rate": 9.053673470822966e-06,
      "loss": 0.1813,
      "step": 17772
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4373136776812823,
      "learning_rate": 9.052686043280545e-06,
      "loss": 0.339,
      "step": 17773
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.7246493395045606,
      "learning_rate": 9.05169862505811e-06,
      "loss": 0.3381,
      "step": 17774
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8875007922529026,
      "learning_rate": 9.050711216165374e-06,
      "loss": 0.5051,
      "step": 17775
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.29638994295512644,
      "learning_rate": 9.049723816612056e-06,
      "loss": 0.2129,
      "step": 17776
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.8435164463225517,
      "learning_rate": 9.048736426407867e-06,
      "loss": 0.8348,
      "step": 17777
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2962928864563523,
      "learning_rate": 9.047749045562518e-06,
      "loss": 0.2333,
      "step": 17778
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.9339368438954802,
      "learning_rate": 9.04676167408573e-06,
      "loss": 0.0673,
      "step": 17779
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.4144046303286816,
      "learning_rate": 9.045774311987218e-06,
      "loss": 0.3052,
      "step": 17780
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.1858322539211282,
      "learning_rate": 9.044786959276685e-06,
      "loss": 0.1221,
      "step": 17781
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8324604261899233,
      "learning_rate": 9.043799615963857e-06,
      "loss": 0.5188,
      "step": 17782
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.39427667106212244,
      "learning_rate": 9.042812282058446e-06,
      "loss": 0.2547,
      "step": 17783
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.35096251558613584,
      "learning_rate": 9.041824957570157e-06,
      "loss": 0.285,
      "step": 17784
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3452542021818185,
      "learning_rate": 9.040837642508712e-06,
      "loss": 0.2387,
      "step": 17785
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4168842147874403,
      "learning_rate": 9.039850336883822e-06,
      "loss": 0.7765,
      "step": 17786
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.8900432847530002,
      "learning_rate": 9.038863040705196e-06,
      "loss": 0.2986,
      "step": 17787
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.5504698214337296,
      "learning_rate": 9.037875753982556e-06,
      "loss": 0.8201,
      "step": 17788
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.2750860767814774,
      "learning_rate": 9.036888476725612e-06,
      "loss": 0.1871,
      "step": 17789
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.305660883681943,
      "learning_rate": 9.035901208944074e-06,
      "loss": 0.2631,
      "step": 17790
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.260052313784903,
      "learning_rate": 9.034913950647658e-06,
      "loss": 0.1681,
      "step": 17791
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.800051847240306,
      "learning_rate": 9.033926701846077e-06,
      "loss": 0.3279,
      "step": 17792
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.5515614864407907,
      "learning_rate": 9.032939462549037e-06,
      "loss": 0.2931,
      "step": 17793
    },
    {
      "epoch": 0.54,
      "grad_norm": 0.3150623954334327,
      "learning_rate": 9.031952232766263e-06,
      "loss": 0.2006,
      "step": 17794
    },
    {
      "epoch": 0.54,
      "grad_norm": 1.4329315485145022,
      "learning_rate": 9.030965012507461e-06,
      "loss": 0.8346,
      "step": 17795
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33414582948012017,
      "learning_rate": 9.029977801782342e-06,
      "loss": 0.2441,
      "step": 17796
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9448836329016812,
      "learning_rate": 9.028990600600623e-06,
      "loss": 0.4711,
      "step": 17797
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3747731649896096,
      "learning_rate": 9.028003408972009e-06,
      "loss": 0.1926,
      "step": 17798
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3135976757530543,
      "learning_rate": 9.02701622690622e-06,
      "loss": 0.2276,
      "step": 17799
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4998723499582771,
      "learning_rate": 9.026029054412969e-06,
      "loss": 0.2186,
      "step": 17800
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3211384320150327,
      "learning_rate": 9.02504189150196e-06,
      "loss": 0.2917,
      "step": 17801
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5968731771907211,
      "learning_rate": 9.024054738182913e-06,
      "loss": 0.1313,
      "step": 17802
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3522890681564909,
      "learning_rate": 9.023067594465535e-06,
      "loss": 0.28,
      "step": 17803
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8548928055135543,
      "learning_rate": 9.022080460359537e-06,
      "loss": 0.6043,
      "step": 17804
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2609649370015052,
      "learning_rate": 9.021093335874637e-06,
      "loss": 0.2584,
      "step": 17805
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3771756129423267,
      "learning_rate": 9.020106221020542e-06,
      "loss": 0.6545,
      "step": 17806
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31418535910783235,
      "learning_rate": 9.019119115806963e-06,
      "loss": 0.1663,
      "step": 17807
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2680629413071946,
      "learning_rate": 9.018132020243615e-06,
      "loss": 0.2713,
      "step": 17808
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3707663787077057,
      "learning_rate": 9.017144934340207e-06,
      "loss": 0.2328,
      "step": 17809
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8480160129690937,
      "learning_rate": 9.016157858106447e-06,
      "loss": 0.5069,
      "step": 17810
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20729357908161247,
      "learning_rate": 9.015170791552053e-06,
      "loss": 0.121,
      "step": 17811
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3205098669315851,
      "learning_rate": 9.014183734686733e-06,
      "loss": 0.2568,
      "step": 17812
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.196683452988907,
      "learning_rate": 9.013196687520197e-06,
      "loss": 0.4606,
      "step": 17813
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32927827268575277,
      "learning_rate": 9.012209650062157e-06,
      "loss": 0.3115,
      "step": 17814
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7513875411319819,
      "learning_rate": 9.011222622322324e-06,
      "loss": 0.0351,
      "step": 17815
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39118972136681335,
      "learning_rate": 9.010235604310404e-06,
      "loss": 0.2693,
      "step": 17816
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.43941382722151046,
      "learning_rate": 9.009248596036117e-06,
      "loss": 0.262,
      "step": 17817
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6146489073972994,
      "learning_rate": 9.008261597509167e-06,
      "loss": 0.3312,
      "step": 17818
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23252246689185804,
      "learning_rate": 9.007274608739262e-06,
      "loss": 0.1877,
      "step": 17819
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31835796562853846,
      "learning_rate": 9.006287629736121e-06,
      "loss": 0.1885,
      "step": 17820
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36080892587836394,
      "learning_rate": 9.005300660509447e-06,
      "loss": 0.2868,
      "step": 17821
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8471871206233011,
      "learning_rate": 9.004313701068954e-06,
      "loss": 0.5592,
      "step": 17822
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1586558873481578,
      "learning_rate": 9.003326751424349e-06,
      "loss": 0.5136,
      "step": 17823
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37859934417690005,
      "learning_rate": 9.00233981158534e-06,
      "loss": 0.2674,
      "step": 17824
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.008373955659653,
      "learning_rate": 9.001352881561648e-06,
      "loss": 0.4847,
      "step": 17825
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2626233134322745,
      "learning_rate": 9.00036596136297e-06,
      "loss": 0.2143,
      "step": 17826
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6422013086662033,
      "learning_rate": 8.99937905099902e-06,
      "loss": 0.4088,
      "step": 17827
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3065608254657877,
      "learning_rate": 8.998392150479511e-06,
      "loss": 0.1263,
      "step": 17828
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6066851276903475,
      "learning_rate": 8.99740525981415e-06,
      "loss": 0.9082,
      "step": 17829
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.285355591818453,
      "learning_rate": 8.996418379012641e-06,
      "loss": 0.1957,
      "step": 17830
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4889102026055344,
      "learning_rate": 8.995431508084702e-06,
      "loss": 0.3421,
      "step": 17831
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.26558767544858536,
      "learning_rate": 8.994444647040039e-06,
      "loss": 0.2231,
      "step": 17832
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4374159663447104,
      "learning_rate": 8.993457795888358e-06,
      "loss": 0.2701,
      "step": 17833
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5692478127565906,
      "learning_rate": 8.992470954639373e-06,
      "loss": 0.3177,
      "step": 17834
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.305925727522102,
      "learning_rate": 8.99148412330279e-06,
      "loss": 0.2537,
      "step": 17835
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8289249572618147,
      "learning_rate": 8.990497301888314e-06,
      "loss": 0.5046,
      "step": 17836
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.378141274366493,
      "learning_rate": 8.989510490405663e-06,
      "loss": 0.2263,
      "step": 17837
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4569207817433683,
      "learning_rate": 8.988523688864539e-06,
      "loss": 0.3775,
      "step": 17838
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2578715498697439,
      "learning_rate": 8.987536897274647e-06,
      "loss": 0.1728,
      "step": 17839
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4432779904332665,
      "learning_rate": 8.986550115645706e-06,
      "loss": 0.2616,
      "step": 17840
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3091472623784666,
      "learning_rate": 8.985563343987418e-06,
      "loss": 0.1114,
      "step": 17841
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6607542985832761,
      "learning_rate": 8.984576582309486e-06,
      "loss": 0.4317,
      "step": 17842
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.26233132831107775,
      "learning_rate": 8.983589830621628e-06,
      "loss": 0.2114,
      "step": 17843
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33185446616358677,
      "learning_rate": 8.98260308893355e-06,
      "loss": 0.2927,
      "step": 17844
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6202663593933222,
      "learning_rate": 8.981616357254952e-06,
      "loss": 0.267,
      "step": 17845
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.003252320280151,
      "learning_rate": 8.980629635595552e-06,
      "loss": 0.3133,
      "step": 17846
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8551332991717685,
      "learning_rate": 8.979642923965048e-06,
      "loss": 0.5898,
      "step": 17847
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34442117554487167,
      "learning_rate": 8.978656222373156e-06,
      "loss": 0.18,
      "step": 17848
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31159081284533174,
      "learning_rate": 8.977669530829582e-06,
      "loss": 0.2417,
      "step": 17849
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21810911323777615,
      "learning_rate": 8.976682849344027e-06,
      "loss": 0.1893,
      "step": 17850
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7631822789780056,
      "learning_rate": 8.975696177926206e-06,
      "loss": 0.4011,
      "step": 17851
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7942916535065966,
      "learning_rate": 8.974709516585825e-06,
      "loss": 0.2027,
      "step": 17852
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3465666770573134,
      "learning_rate": 8.973722865332585e-06,
      "loss": 0.276,
      "step": 17853
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2181544374279138,
      "learning_rate": 8.972736224176199e-06,
      "loss": 0.2734,
      "step": 17854
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32065424465280196,
      "learning_rate": 8.971749593126373e-06,
      "loss": 0.2793,
      "step": 17855
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8189544873410125,
      "learning_rate": 8.970762972192809e-06,
      "loss": 0.4701,
      "step": 17856
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5873974774194283,
      "learning_rate": 8.96977636138522e-06,
      "loss": 0.3067,
      "step": 17857
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3497987890294004,
      "learning_rate": 8.968789760713312e-06,
      "loss": 0.2402,
      "step": 17858
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4414983832589382,
      "learning_rate": 8.967803170186787e-06,
      "loss": 0.169,
      "step": 17859
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.120153407432463,
      "learning_rate": 8.966816589815358e-06,
      "loss": 0.5244,
      "step": 17860
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.325405325090426,
      "learning_rate": 8.965830019608725e-06,
      "loss": 0.213,
      "step": 17861
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2757352700531272,
      "learning_rate": 8.964843459576593e-06,
      "loss": 0.2583,
      "step": 17862
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1746771701757435,
      "learning_rate": 8.963856909728675e-06,
      "loss": 0.2226,
      "step": 17863
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6629157107835,
      "learning_rate": 8.962870370074674e-06,
      "loss": 0.7728,
      "step": 17864
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.886930354100496,
      "learning_rate": 8.961883840624294e-06,
      "loss": 0.4297,
      "step": 17865
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37974304113974144,
      "learning_rate": 8.960897321387244e-06,
      "loss": 0.2804,
      "step": 17866
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.25637356387365,
      "learning_rate": 8.95991081237323e-06,
      "loss": 0.2375,
      "step": 17867
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4287950600733208,
      "learning_rate": 8.958924313591948e-06,
      "loss": 0.2264,
      "step": 17868
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8685000898758873,
      "learning_rate": 8.957937825053117e-06,
      "loss": 0.3614,
      "step": 17869
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3662114604483957,
      "learning_rate": 8.956951346766437e-06,
      "loss": 0.1954,
      "step": 17870
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2891320415505394,
      "learning_rate": 8.95596487874161e-06,
      "loss": 0.1993,
      "step": 17871
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.417090529652816,
      "learning_rate": 8.954978420988348e-06,
      "loss": 0.2417,
      "step": 17872
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32776901742341424,
      "learning_rate": 8.953991973516347e-06,
      "loss": 0.2963,
      "step": 17873
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.058763292902371,
      "learning_rate": 8.95300553633532e-06,
      "loss": 0.4593,
      "step": 17874
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.616813859698858,
      "learning_rate": 8.95201910945497e-06,
      "loss": 0.3169,
      "step": 17875
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31359266831643806,
      "learning_rate": 8.951032692885e-06,
      "loss": 0.2298,
      "step": 17876
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5982235257683048,
      "learning_rate": 8.950046286635114e-06,
      "loss": 0.4066,
      "step": 17877
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24140850768681885,
      "learning_rate": 8.949059890715022e-06,
      "loss": 0.1921,
      "step": 17878
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3810346378448686,
      "learning_rate": 8.948073505134418e-06,
      "loss": 0.3004,
      "step": 17879
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28846414209935073,
      "learning_rate": 8.947087129903018e-06,
      "loss": 0.1956,
      "step": 17880
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4599518669623117,
      "learning_rate": 8.946100765030523e-06,
      "loss": 0.2494,
      "step": 17881
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.881562839059205,
      "learning_rate": 8.945114410526629e-06,
      "loss": 0.3211,
      "step": 17882
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4417360815273361,
      "learning_rate": 8.944128066401052e-06,
      "loss": 0.8312,
      "step": 17883
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3357005043227454,
      "learning_rate": 8.94314173266349e-06,
      "loss": 0.118,
      "step": 17884
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24040564180314655,
      "learning_rate": 8.942155409323643e-06,
      "loss": 0.2335,
      "step": 17885
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.600986140663044,
      "learning_rate": 8.941169096391223e-06,
      "loss": 0.4044,
      "step": 17886
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.698384714404922,
      "learning_rate": 8.940182793875931e-06,
      "loss": 0.2892,
      "step": 17887
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.26222737680379843,
      "learning_rate": 8.939196501787464e-06,
      "loss": 0.1662,
      "step": 17888
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2618089419598093,
      "learning_rate": 8.938210220135536e-06,
      "loss": 0.1816,
      "step": 17889
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0569832046722307,
      "learning_rate": 8.937223948929845e-06,
      "loss": 0.5106,
      "step": 17890
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3032354186593061,
      "learning_rate": 8.936237688180092e-06,
      "loss": 0.2257,
      "step": 17891
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1049339901237865,
      "learning_rate": 8.935251437895984e-06,
      "loss": 0.6816,
      "step": 17892
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3099348992969606,
      "learning_rate": 8.934265198087224e-06,
      "loss": 0.1807,
      "step": 17893
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.48957171254931886,
      "learning_rate": 8.93327896876351e-06,
      "loss": 0.3377,
      "step": 17894
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.881563478299493,
      "learning_rate": 8.93229274993455e-06,
      "loss": 0.3345,
      "step": 17895
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5167438618480236,
      "learning_rate": 8.931306541610048e-06,
      "loss": 0.8651,
      "step": 17896
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20934179095033853,
      "learning_rate": 8.9303203437997e-06,
      "loss": 0.1793,
      "step": 17897
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2169908568094256,
      "learning_rate": 8.929334156513216e-06,
      "loss": 0.1549,
      "step": 17898
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9054173435926314,
      "learning_rate": 8.928347979760289e-06,
      "loss": 0.7506,
      "step": 17899
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2801696450603783,
      "learning_rate": 8.927361813550633e-06,
      "loss": 0.3186,
      "step": 17900
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8876367874613467,
      "learning_rate": 8.926375657893942e-06,
      "loss": 0.4453,
      "step": 17901
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44712632464504704,
      "learning_rate": 8.925389512799922e-06,
      "loss": 0.2049,
      "step": 17902
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38132137297373697,
      "learning_rate": 8.924403378278274e-06,
      "loss": 0.2971,
      "step": 17903
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3282490122894821,
      "learning_rate": 8.923417254338698e-06,
      "loss": 0.2564,
      "step": 17904
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0294447135209202,
      "learning_rate": 8.922431140990895e-06,
      "loss": 0.6087,
      "step": 17905
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19407618500644,
      "learning_rate": 8.921445038244573e-06,
      "loss": 0.0857,
      "step": 17906
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6007267377946698,
      "learning_rate": 8.92045894610943e-06,
      "loss": 0.2797,
      "step": 17907
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3248894887692237,
      "learning_rate": 8.919472864595164e-06,
      "loss": 0.2015,
      "step": 17908
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37115782250456525,
      "learning_rate": 8.91848679371148e-06,
      "loss": 0.3,
      "step": 17909
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0407570313144514,
      "learning_rate": 8.917500733468081e-06,
      "loss": 0.406,
      "step": 17910
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5335601271872678,
      "learning_rate": 8.916514683874662e-06,
      "loss": 0.193,
      "step": 17911
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3339002805227984,
      "learning_rate": 8.915528644940931e-06,
      "loss": 0.2856,
      "step": 17912
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9654560813544324,
      "learning_rate": 8.914542616676587e-06,
      "loss": 0.3601,
      "step": 17913
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7676334433878333,
      "learning_rate": 8.913556599091328e-06,
      "loss": 0.7354,
      "step": 17914
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2369000382540414,
      "learning_rate": 8.912570592194858e-06,
      "loss": 0.194,
      "step": 17915
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40514516593680144,
      "learning_rate": 8.911584595996875e-06,
      "loss": 0.2707,
      "step": 17916
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2668861930806154,
      "learning_rate": 8.91059861050708e-06,
      "loss": 0.1449,
      "step": 17917
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.890428269105526,
      "learning_rate": 8.909612635735176e-06,
      "loss": 0.805,
      "step": 17918
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1336715047506785,
      "learning_rate": 8.908626671690863e-06,
      "loss": 0.3763,
      "step": 17919
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.44697279949349306,
      "learning_rate": 8.907640718383834e-06,
      "loss": 0.2975,
      "step": 17920
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.25668284151601367,
      "learning_rate": 8.9066547758238e-06,
      "loss": 0.2038,
      "step": 17921
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.818795193546572,
      "learning_rate": 8.905668844020457e-06,
      "loss": 0.9796,
      "step": 17922
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8699409080065497,
      "learning_rate": 8.904682922983502e-06,
      "loss": 0.4611,
      "step": 17923
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3320155884852556,
      "learning_rate": 8.903697012722639e-06,
      "loss": 0.1216,
      "step": 17924
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3697475876978997,
      "learning_rate": 8.90271111324756e-06,
      "loss": 0.3154,
      "step": 17925
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.20926248704478992,
      "learning_rate": 8.901725224567975e-06,
      "loss": 0.0724,
      "step": 17926
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3375243462882534,
      "learning_rate": 8.900739346693578e-06,
      "loss": 0.3069,
      "step": 17927
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6569755233990818,
      "learning_rate": 8.899753479634067e-06,
      "loss": 0.3261,
      "step": 17928
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5594160892372823,
      "learning_rate": 8.898767623399148e-06,
      "loss": 0.2761,
      "step": 17929
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3482238148753619,
      "learning_rate": 8.897781777998511e-06,
      "loss": 0.2147,
      "step": 17930
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3208106350612039,
      "learning_rate": 8.896795943441858e-06,
      "loss": 0.8861,
      "step": 17931
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31152543072507116,
      "learning_rate": 8.895810119738894e-06,
      "loss": 0.2555,
      "step": 17932
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.290224585716626,
      "learning_rate": 8.894824306899314e-06,
      "loss": 0.6692,
      "step": 17933
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31448764999525136,
      "learning_rate": 8.893838504932811e-06,
      "loss": 0.0706,
      "step": 17934
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2651079926121397,
      "learning_rate": 8.892852713849095e-06,
      "loss": 0.225,
      "step": 17935
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3049663034983406,
      "learning_rate": 8.891866933657854e-06,
      "loss": 0.204,
      "step": 17936
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5829772255353441,
      "learning_rate": 8.89088116436879e-06,
      "loss": 0.3173,
      "step": 17937
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38400302694993005,
      "learning_rate": 8.889895405991603e-06,
      "loss": 0.3041,
      "step": 17938
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23974071735972552,
      "learning_rate": 8.888909658535992e-06,
      "loss": 0.2102,
      "step": 17939
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3363945027764506,
      "learning_rate": 8.887923922011651e-06,
      "loss": 0.8353,
      "step": 17940
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2424956426426785,
      "learning_rate": 8.886938196428282e-06,
      "loss": 0.2233,
      "step": 17941
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1872502934701783,
      "learning_rate": 8.885952481795582e-06,
      "loss": 0.6698,
      "step": 17942
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.22557594753451668,
      "learning_rate": 8.884966778123245e-06,
      "loss": 0.1333,
      "step": 17943
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.30521665977537704,
      "learning_rate": 8.883981085420975e-06,
      "loss": 0.2522,
      "step": 17944
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28256630033116065,
      "learning_rate": 8.882995403698467e-06,
      "loss": 0.2263,
      "step": 17945
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6416986097155648,
      "learning_rate": 8.882009732965414e-06,
      "loss": 0.4351,
      "step": 17946
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.423208334888232,
      "learning_rate": 8.88102407323152e-06,
      "loss": 0.2142,
      "step": 17947
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3186124673127162,
      "learning_rate": 8.880038424506476e-06,
      "loss": 0.2537,
      "step": 17948
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8785369815373127,
      "learning_rate": 8.879052786799985e-06,
      "loss": 0.4863,
      "step": 17949
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29473800289020724,
      "learning_rate": 8.878067160121743e-06,
      "loss": 0.2259,
      "step": 17950
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1594301773694198,
      "learning_rate": 8.87708154448144e-06,
      "loss": 0.606,
      "step": 17951
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3261202339287958,
      "learning_rate": 8.876095939888783e-06,
      "loss": 0.1823,
      "step": 17952
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.49663088234164643,
      "learning_rate": 8.875110346353464e-06,
      "loss": 0.3109,
      "step": 17953
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5414106421396072,
      "learning_rate": 8.874124763885177e-06,
      "loss": 0.3264,
      "step": 17954
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24832908309104448,
      "learning_rate": 8.873139192493624e-06,
      "loss": 0.1621,
      "step": 17955
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2418673182058038,
      "learning_rate": 8.872153632188498e-06,
      "loss": 0.2026,
      "step": 17956
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5469610828276648,
      "learning_rate": 8.871168082979491e-06,
      "loss": 0.3694,
      "step": 17957
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4354255857936569,
      "learning_rate": 8.87018254487631e-06,
      "loss": 0.2829,
      "step": 17958
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3278420435176344,
      "learning_rate": 8.869197017888643e-06,
      "loss": 0.5772,
      "step": 17959
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6258457397131372,
      "learning_rate": 8.868211502026185e-06,
      "loss": 0.278,
      "step": 17960
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9539908616413761,
      "learning_rate": 8.867225997298639e-06,
      "loss": 0.4172,
      "step": 17961
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2849047571143508,
      "learning_rate": 8.866240503715695e-06,
      "loss": 0.22,
      "step": 17962
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31357002268812223,
      "learning_rate": 8.865255021287046e-06,
      "loss": 0.2491,
      "step": 17963
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.31016846259936426,
      "learning_rate": 8.864269550022395e-06,
      "loss": 0.1766,
      "step": 17964
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9118799350884279,
      "learning_rate": 8.863284089931436e-06,
      "loss": 0.5622,
      "step": 17965
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37069639739103794,
      "learning_rate": 8.862298641023858e-06,
      "loss": 0.2361,
      "step": 17966
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4491578717078907,
      "learning_rate": 8.861313203309363e-06,
      "loss": 0.2472,
      "step": 17967
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3236341598666083,
      "learning_rate": 8.860327776797644e-06,
      "loss": 0.3003,
      "step": 17968
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2617993071752023,
      "learning_rate": 8.85934236149839e-06,
      "loss": 0.2053,
      "step": 17969
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5548669788634425,
      "learning_rate": 8.858356957421306e-06,
      "loss": 0.3128,
      "step": 17970
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28983789052286146,
      "learning_rate": 8.857371564576082e-06,
      "loss": 0.2505,
      "step": 17971
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7650685150579185,
      "learning_rate": 8.85638618297241e-06,
      "loss": 0.4582,
      "step": 17972
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.339182973109778,
      "learning_rate": 8.855400812619989e-06,
      "loss": 0.1264,
      "step": 17973
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33942583537416576,
      "learning_rate": 8.854415453528508e-06,
      "loss": 0.304,
      "step": 17974
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2754079116477004,
      "learning_rate": 8.853430105707668e-06,
      "loss": 0.1852,
      "step": 17975
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0095858307652827,
      "learning_rate": 8.85244476916716e-06,
      "loss": 0.4775,
      "step": 17976
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42381322736308996,
      "learning_rate": 8.851459443916677e-06,
      "loss": 0.2355,
      "step": 17977
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6480602286952879,
      "learning_rate": 8.850474129965915e-06,
      "loss": 0.3177,
      "step": 17978
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5605016416720033,
      "learning_rate": 8.849488827324567e-06,
      "loss": 0.3119,
      "step": 17979
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3487058584043357,
      "learning_rate": 8.848503536002322e-06,
      "loss": 0.2155,
      "step": 17980
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33714624922811837,
      "learning_rate": 8.847518256008883e-06,
      "loss": 0.2965,
      "step": 17981
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.202783953321448,
      "learning_rate": 8.846532987353938e-06,
      "loss": 0.0986,
      "step": 17982
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.3965810884027834,
      "learning_rate": 8.84554773004718e-06,
      "loss": 0.8361,
      "step": 17983
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23631101456458697,
      "learning_rate": 8.844562484098304e-06,
      "loss": 0.0687,
      "step": 17984
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.27731949010887696,
      "learning_rate": 8.843577249517004e-06,
      "loss": 0.2335,
      "step": 17985
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.32567246904673786,
      "learning_rate": 8.842592026312965e-06,
      "loss": 0.2527,
      "step": 17986
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7576554312297415,
      "learning_rate": 8.841606814495893e-06,
      "loss": 0.4769,
      "step": 17987
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6017562433062593,
      "learning_rate": 8.840621614075474e-06,
      "loss": 0.211,
      "step": 17988
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.30183094970384866,
      "learning_rate": 8.839636425061395e-06,
      "loss": 0.2284,
      "step": 17989
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2550250595767949,
      "learning_rate": 8.83865124746336e-06,
      "loss": 0.6448,
      "step": 17990
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38439550245159465,
      "learning_rate": 8.837666081291057e-06,
      "loss": 0.2291,
      "step": 17991
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.46635695889287004,
      "learning_rate": 8.836680926554174e-06,
      "loss": 0.3465,
      "step": 17992
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3019675964130014,
      "learning_rate": 8.83569578326241e-06,
      "loss": 0.1784,
      "step": 17993
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.30283116017718914,
      "learning_rate": 8.834710651425453e-06,
      "loss": 0.2179,
      "step": 17994
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3809810251512946,
      "learning_rate": 8.833725531052993e-06,
      "loss": 0.1525,
      "step": 17995
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9578108407096655,
      "learning_rate": 8.832740422154728e-06,
      "loss": 0.4281,
      "step": 17996
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33025532190241635,
      "learning_rate": 8.831755324740347e-06,
      "loss": 0.2415,
      "step": 17997
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3076751371919777,
      "learning_rate": 8.83077023881954e-06,
      "loss": 0.2568,
      "step": 17998
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42059136262454383,
      "learning_rate": 8.829785164402002e-06,
      "loss": 0.26,
      "step": 17999
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1358348766831943,
      "learning_rate": 8.82880010149742e-06,
      "loss": 0.5744,
      "step": 18000
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8349752109442646,
      "learning_rate": 8.82781505011549e-06,
      "loss": 0.6074,
      "step": 18001
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.26142163153488257,
      "learning_rate": 8.826830010265902e-06,
      "loss": 0.1857,
      "step": 18002
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.855202018145277,
      "learning_rate": 8.825844981958343e-06,
      "loss": 0.7948,
      "step": 18003
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.26490719349642594,
      "learning_rate": 8.82485996520251e-06,
      "loss": 0.2151,
      "step": 18004
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8704715412959807,
      "learning_rate": 8.823874960008094e-06,
      "loss": 0.4187,
      "step": 18005
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.15036007863647283,
      "learning_rate": 8.822889966384777e-06,
      "loss": 0.0685,
      "step": 18006
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5040583173211969,
      "learning_rate": 8.82190498434226e-06,
      "loss": 0.3376,
      "step": 18007
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4304978185319606,
      "learning_rate": 8.82092001389023e-06,
      "loss": 0.2458,
      "step": 18008
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.46484891255990574,
      "learning_rate": 8.819935055038374e-06,
      "loss": 0.3129,
      "step": 18009
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3648361010987108,
      "learning_rate": 8.818950107796387e-06,
      "loss": 0.2784,
      "step": 18010
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9415812518903851,
      "learning_rate": 8.817965172173958e-06,
      "loss": 0.4609,
      "step": 18011
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.27740368115429953,
      "learning_rate": 8.816980248180771e-06,
      "loss": 0.2013,
      "step": 18012
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5091313284774103,
      "learning_rate": 8.815995335826528e-06,
      "loss": 0.2796,
      "step": 18013
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3729121273997116,
      "learning_rate": 8.815010435120912e-06,
      "loss": 0.2111,
      "step": 18014
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2987389239286521,
      "learning_rate": 8.81402554607361e-06,
      "loss": 0.227,
      "step": 18015
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4029746867151307,
      "learning_rate": 8.813040668694319e-06,
      "loss": 0.2872,
      "step": 18016
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.41701299607633463,
      "learning_rate": 8.812055802992723e-06,
      "loss": 0.2176,
      "step": 18017
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1143106596028376,
      "learning_rate": 8.811070948978508e-06,
      "loss": 0.6841,
      "step": 18018
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9530543341658417,
      "learning_rate": 8.810086106661374e-06,
      "loss": 0.4867,
      "step": 18019
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6063266239599445,
      "learning_rate": 8.809101276051003e-06,
      "loss": 0.3552,
      "step": 18020
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3048458767672194,
      "learning_rate": 8.808116457157082e-06,
      "loss": 0.2519,
      "step": 18021
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33605986931720605,
      "learning_rate": 8.807131649989308e-06,
      "loss": 0.3129,
      "step": 18022
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2068815281246033,
      "learning_rate": 8.806146854557364e-06,
      "loss": 0.2923,
      "step": 18023
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28198755506176787,
      "learning_rate": 8.805162070870941e-06,
      "loss": 0.1492,
      "step": 18024
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.288719695838523,
      "learning_rate": 8.804177298939726e-06,
      "loss": 0.1898,
      "step": 18025
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.9796015705320724,
      "learning_rate": 8.803192538773406e-06,
      "loss": 0.8545,
      "step": 18026
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.37740458506394703,
      "learning_rate": 8.802207790381674e-06,
      "loss": 0.2382,
      "step": 18027
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4366329360804811,
      "learning_rate": 8.801223053774216e-06,
      "loss": 0.311,
      "step": 18028
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4232646703834564,
      "learning_rate": 8.800238328960718e-06,
      "loss": 0.2178,
      "step": 18029
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3775954364680688,
      "learning_rate": 8.799253615950874e-06,
      "loss": 0.2262,
      "step": 18030
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7176363909460142,
      "learning_rate": 8.798268914754367e-06,
      "loss": 0.4239,
      "step": 18031
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.19349395362670319,
      "learning_rate": 8.79728422538088e-06,
      "loss": 0.0869,
      "step": 18032
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29032375288455264,
      "learning_rate": 8.796299547840112e-06,
      "loss": 0.2789,
      "step": 18033
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.322869323828921,
      "learning_rate": 8.795314882141745e-06,
      "loss": 0.1724,
      "step": 18034
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5258900433275239,
      "learning_rate": 8.794330228295464e-06,
      "loss": 0.3594,
      "step": 18035
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0187143026859924,
      "learning_rate": 8.793345586310964e-06,
      "loss": 0.3761,
      "step": 18036
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.887022206260318,
      "learning_rate": 8.792360956197925e-06,
      "loss": 0.5482,
      "step": 18037
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.742455622440849,
      "learning_rate": 8.79137633796603e-06,
      "loss": 0.2219,
      "step": 18038
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3291111894871475,
      "learning_rate": 8.79039173162498e-06,
      "loss": 0.2737,
      "step": 18039
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2749645994923699,
      "learning_rate": 8.789407137184452e-06,
      "loss": 0.2463,
      "step": 18040
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2433181005463503,
      "learning_rate": 8.788422554654132e-06,
      "loss": 0.6815,
      "step": 18041
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.18174883791589797,
      "learning_rate": 8.787437984043714e-06,
      "loss": 0.0708,
      "step": 18042
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29243514178020236,
      "learning_rate": 8.78645342536288e-06,
      "loss": 0.1562,
      "step": 18043
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5015356563766555,
      "learning_rate": 8.78546887862131e-06,
      "loss": 0.3182,
      "step": 18044
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28826005023856915,
      "learning_rate": 8.784484343828703e-06,
      "loss": 0.253,
      "step": 18045
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6743133039681901,
      "learning_rate": 8.783499820994738e-06,
      "loss": 0.9338,
      "step": 18046
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6990429040724796,
      "learning_rate": 8.7825153101291e-06,
      "loss": 0.2921,
      "step": 18047
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5560272001915951,
      "learning_rate": 8.781530811241481e-06,
      "loss": 0.2686,
      "step": 18048
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7438326094725127,
      "learning_rate": 8.780546324341561e-06,
      "loss": 0.3368,
      "step": 18049
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.903322400332542,
      "learning_rate": 8.779561849439024e-06,
      "loss": 0.8005,
      "step": 18050
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2263484243291752,
      "learning_rate": 8.778577386543561e-06,
      "loss": 0.1785,
      "step": 18051
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3916971403427398,
      "learning_rate": 8.777592935664856e-06,
      "loss": 0.2413,
      "step": 18052
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2605498147407754,
      "learning_rate": 8.776608496812595e-06,
      "loss": 0.1394,
      "step": 18053
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.2527356979618285,
      "learning_rate": 8.775624069996463e-06,
      "loss": 0.7118,
      "step": 18054
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9071317027541471,
      "learning_rate": 8.77463965522614e-06,
      "loss": 0.3585,
      "step": 18055
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4826560834105103,
      "learning_rate": 8.77365525251132e-06,
      "loss": 0.2079,
      "step": 18056
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34412148959205546,
      "learning_rate": 8.772670861861681e-06,
      "loss": 0.2711,
      "step": 18057
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.30000395909025834,
      "learning_rate": 8.771686483286907e-06,
      "loss": 0.2447,
      "step": 18058
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9982336364227785,
      "learning_rate": 8.77070211679669e-06,
      "loss": 0.5263,
      "step": 18059
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2194853653494907,
      "learning_rate": 8.76971776240071e-06,
      "loss": 0.0912,
      "step": 18060
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5929283834763668,
      "learning_rate": 8.76873342010865e-06,
      "loss": 0.3007,
      "step": 18061
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36528369731581295,
      "learning_rate": 8.767749089930196e-06,
      "loss": 0.2307,
      "step": 18062
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4879708469432749,
      "learning_rate": 8.766764771875033e-06,
      "loss": 0.3553,
      "step": 18063
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.6875762817149671,
      "learning_rate": 8.76578046595284e-06,
      "loss": 0.323,
      "step": 18064
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.0384624585339999,
      "learning_rate": 8.76479617217331e-06,
      "loss": 0.564,
      "step": 18065
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23787025608395707,
      "learning_rate": 8.763811890546118e-06,
      "loss": 0.1765,
      "step": 18066
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.4778238035643838,
      "learning_rate": 8.762827621080953e-06,
      "loss": 0.7687,
      "step": 18067
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.45162658595358457,
      "learning_rate": 8.761843363787497e-06,
      "loss": 0.2626,
      "step": 18068
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4285406521248501,
      "learning_rate": 8.760859118675435e-06,
      "loss": 0.2692,
      "step": 18069
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.33004142330181135,
      "learning_rate": 8.759874885754443e-06,
      "loss": 0.1824,
      "step": 18070
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23763663913267918,
      "learning_rate": 8.758890665034215e-06,
      "loss": 0.1646,
      "step": 18071
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8243834076519172,
      "learning_rate": 8.75790645652443e-06,
      "loss": 0.4525,
      "step": 18072
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7648052920319071,
      "learning_rate": 8.756922260234765e-06,
      "loss": 0.3186,
      "step": 18073
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.466744358234353,
      "learning_rate": 8.755938076174912e-06,
      "loss": 0.3855,
      "step": 18074
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.29292065464499745,
      "learning_rate": 8.754953904354547e-06,
      "loss": 0.1821,
      "step": 18075
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.45365628013208215,
      "learning_rate": 8.753969744783356e-06,
      "loss": 0.352,
      "step": 18076
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5740416305873407,
      "learning_rate": 8.752985597471022e-06,
      "loss": 0.0269,
      "step": 18077
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4524106593441607,
      "learning_rate": 8.752001462427224e-06,
      "loss": 0.2398,
      "step": 18078
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.23472373173104594,
      "learning_rate": 8.751017339661648e-06,
      "loss": 0.1249,
      "step": 18079
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5130239330867438,
      "learning_rate": 8.750033229183976e-06,
      "loss": 0.3238,
      "step": 18080
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4565874325748083,
      "learning_rate": 8.749049131003882e-06,
      "loss": 0.2598,
      "step": 18081
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5195737986137735,
      "learning_rate": 8.748065045131058e-06,
      "loss": 0.2793,
      "step": 18082
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.8245156883886272,
      "learning_rate": 8.747080971575185e-06,
      "loss": 0.5357,
      "step": 18083
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2897893550125608,
      "learning_rate": 8.746096910345938e-06,
      "loss": 0.1922,
      "step": 18084
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1473927323521615,
      "learning_rate": 8.745112861453004e-06,
      "loss": 0.5394,
      "step": 18085
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1953577623880633,
      "learning_rate": 8.744128824906063e-06,
      "loss": 0.2344,
      "step": 18086
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3273932228880476,
      "learning_rate": 8.743144800714792e-06,
      "loss": 0.2852,
      "step": 18087
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.21798526297398219,
      "learning_rate": 8.742160788888878e-06,
      "loss": 0.0723,
      "step": 18088
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.39901098196742707,
      "learning_rate": 8.741176789438003e-06,
      "loss": 0.3191,
      "step": 18089
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.45647943898016907,
      "learning_rate": 8.740192802371839e-06,
      "loss": 0.2337,
      "step": 18090
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9429724072582785,
      "learning_rate": 8.739208827700077e-06,
      "loss": 0.5541,
      "step": 18091
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.36555189779837094,
      "learning_rate": 8.738224865432395e-06,
      "loss": 0.225,
      "step": 18092
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.343458623591967,
      "learning_rate": 8.737240915578468e-06,
      "loss": 0.2603,
      "step": 18093
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3180164284149507,
      "learning_rate": 8.736256978147984e-06,
      "loss": 0.2315,
      "step": 18094
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.1094370755296519,
      "learning_rate": 8.735273053150618e-06,
      "loss": 0.2688,
      "step": 18095
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.7703154014623463,
      "learning_rate": 8.734289140596048e-06,
      "loss": 0.7617,
      "step": 18096
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4589733837586892,
      "learning_rate": 8.733305240493964e-06,
      "loss": 0.1867,
      "step": 18097
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34607104702218133,
      "learning_rate": 8.732321352854038e-06,
      "loss": 0.2968,
      "step": 18098
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3175176925967153,
      "learning_rate": 8.73133747768595e-06,
      "loss": 0.2683,
      "step": 18099
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42406376103717675,
      "learning_rate": 8.730353614999383e-06,
      "loss": 0.2034,
      "step": 18100
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.9597572496515967,
      "learning_rate": 8.72936976480401e-06,
      "loss": 0.6145,
      "step": 18101
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.38421852115472077,
      "learning_rate": 8.728385927109521e-06,
      "loss": 0.2604,
      "step": 18102
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.28503680687335803,
      "learning_rate": 8.72740210192559e-06,
      "loss": 0.162,
      "step": 18103
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.324629856313688,
      "learning_rate": 8.726418289261893e-06,
      "loss": 0.7199,
      "step": 18104
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.30637213226552634,
      "learning_rate": 8.725434489128114e-06,
      "loss": 0.2419,
      "step": 18105
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.513311404663262,
      "learning_rate": 8.72445070153393e-06,
      "loss": 0.3692,
      "step": 18106
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3183523377513538,
      "learning_rate": 8.723466926489014e-06,
      "loss": 0.2211,
      "step": 18107
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7338871704632467,
      "learning_rate": 8.722483164003056e-06,
      "loss": 0.2857,
      "step": 18108
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.42981147134534514,
      "learning_rate": 8.72149941408573e-06,
      "loss": 0.2917,
      "step": 18109
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2620675039914514,
      "learning_rate": 8.72051567674671e-06,
      "loss": 0.2181,
      "step": 18110
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.34448910243012576,
      "learning_rate": 8.719531951995681e-06,
      "loss": 0.2959,
      "step": 18111
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.22876134837443685,
      "learning_rate": 8.718548239842318e-06,
      "loss": 0.1054,
      "step": 18112
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.6292552522153871,
      "learning_rate": 8.717564540296294e-06,
      "loss": 0.8063,
      "step": 18113
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.5715701258164052,
      "learning_rate": 8.716580853367298e-06,
      "loss": 0.3244,
      "step": 18114
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.7460004624792441,
      "learning_rate": 8.715597179064998e-06,
      "loss": 0.3826,
      "step": 18115
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.2833042084546855,
      "learning_rate": 8.714613517399076e-06,
      "loss": 0.2014,
      "step": 18116
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3082105281891544,
      "learning_rate": 8.713629868379213e-06,
      "loss": 0.3012,
      "step": 18117
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.40909569275075425,
      "learning_rate": 8.71264623201508e-06,
      "loss": 0.1482,
      "step": 18118
    },
    {
      "epoch": 0.55,
      "grad_norm": 1.5186217751469167,
      "learning_rate": 8.711662608316353e-06,
      "loss": 0.7826,
      "step": 18119
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.3453298115724212,
      "learning_rate": 8.710678997292717e-06,
      "loss": 0.1175,
      "step": 18120
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.24255133013948166,
      "learning_rate": 8.709695398953847e-06,
      "loss": 0.1839,
      "step": 18121
    },
    {
      "epoch": 0.55,
      "grad_norm": 0.4814236053137174,
      "learning_rate": 8.708711813309413e-06,
      "loss": 0.3651,
      "step": 18122
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4879974829643739,
      "learning_rate": 8.707728240369102e-06,
      "loss": 0.253,
      "step": 18123
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.644845164469384,
      "learning_rate": 8.706744680142586e-06,
      "loss": 0.4005,
      "step": 18124
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24572359020462253,
      "learning_rate": 8.705761132639536e-06,
      "loss": 0.1809,
      "step": 18125
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2223316329827556,
      "learning_rate": 8.70477759786964e-06,
      "loss": 0.6268,
      "step": 18126
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2529343248774564,
      "learning_rate": 8.70379407584256e-06,
      "loss": 0.0932,
      "step": 18127
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4867432490158664,
      "learning_rate": 8.702810566567988e-06,
      "loss": 0.4348,
      "step": 18128
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.25910059455001583,
      "learning_rate": 8.701827070055592e-06,
      "loss": 0.1934,
      "step": 18129
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3520629595309079,
      "learning_rate": 8.700843586315045e-06,
      "loss": 0.2705,
      "step": 18130
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9662684755015871,
      "learning_rate": 8.69986011535603e-06,
      "loss": 0.0291,
      "step": 18131
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6407331841646358,
      "learning_rate": 8.698876657188219e-06,
      "loss": 0.383,
      "step": 18132
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5523265963241795,
      "learning_rate": 8.69789321182128e-06,
      "loss": 0.2407,
      "step": 18133
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24699912983760133,
      "learning_rate": 8.696909779264903e-06,
      "loss": 0.2126,
      "step": 18134
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5211175711881245,
      "learning_rate": 8.695926359528757e-06,
      "loss": 0.3706,
      "step": 18135
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.988178729257435,
      "learning_rate": 8.694942952622515e-06,
      "loss": 0.3358,
      "step": 18136
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.294434867656829,
      "learning_rate": 8.693959558555853e-06,
      "loss": 0.8501,
      "step": 18137
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2344612488131148,
      "learning_rate": 8.692976177338447e-06,
      "loss": 0.127,
      "step": 18138
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5113545809679234,
      "learning_rate": 8.691992808979969e-06,
      "loss": 0.3231,
      "step": 18139
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29293200971869504,
      "learning_rate": 8.6910094534901e-06,
      "loss": 0.2067,
      "step": 18140
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.47102867051865266,
      "learning_rate": 8.69002611087851e-06,
      "loss": 0.3093,
      "step": 18141
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5490456004118408,
      "learning_rate": 8.689042781154871e-06,
      "loss": 0.2553,
      "step": 18142
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5017901816151442,
      "learning_rate": 8.688059464328866e-06,
      "loss": 0.2751,
      "step": 18143
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4493628015810822,
      "learning_rate": 8.68707616041016e-06,
      "loss": 0.2234,
      "step": 18144
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3302876545586941,
      "learning_rate": 8.686092869408428e-06,
      "loss": 0.7279,
      "step": 18145
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29483638021945,
      "learning_rate": 8.685109591333352e-06,
      "loss": 0.2424,
      "step": 18146
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2915742949438736,
      "learning_rate": 8.684126326194599e-06,
      "loss": 0.1135,
      "step": 18147
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39486518479726296,
      "learning_rate": 8.683143074001842e-06,
      "loss": 0.308,
      "step": 18148
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.38795289678338035,
      "learning_rate": 8.68215983476476e-06,
      "loss": 0.2277,
      "step": 18149
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3948874738483753,
      "learning_rate": 8.681176608493023e-06,
      "loss": 0.2634,
      "step": 18150
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3456889739719754,
      "learning_rate": 8.6801933951963e-06,
      "loss": 0.2139,
      "step": 18151
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36180860473180115,
      "learning_rate": 8.67921019488427e-06,
      "loss": 0.2557,
      "step": 18152
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31606450781424195,
      "learning_rate": 8.678227007566604e-06,
      "loss": 0.2288,
      "step": 18153
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3112058544281378,
      "learning_rate": 8.67724383325298e-06,
      "loss": 0.6551,
      "step": 18154
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8167593725520378,
      "learning_rate": 8.676260671953067e-06,
      "loss": 0.5971,
      "step": 18155
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5920454874869454,
      "learning_rate": 8.675277523676532e-06,
      "loss": 0.3999,
      "step": 18156
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2714487154696572,
      "learning_rate": 8.674294388433056e-06,
      "loss": 0.2156,
      "step": 18157
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9447799914551864,
      "learning_rate": 8.673311266232309e-06,
      "loss": 0.5101,
      "step": 18158
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21801852572263325,
      "learning_rate": 8.672328157083956e-06,
      "loss": 0.2097,
      "step": 18159
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19478288428171064,
      "learning_rate": 8.671345060997681e-06,
      "loss": 0.0717,
      "step": 18160
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.33569378464675736,
      "learning_rate": 8.67036197798315e-06,
      "loss": 0.2857,
      "step": 18161
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2062430058037497,
      "learning_rate": 8.669378908050033e-06,
      "loss": 0.0592,
      "step": 18162
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.166734594310601,
      "learning_rate": 8.668395851208008e-06,
      "loss": 0.6638,
      "step": 18163
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.28133159517991135,
      "learning_rate": 8.667412807466741e-06,
      "loss": 0.2438,
      "step": 18164
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6438587134316955,
      "learning_rate": 8.666429776835901e-06,
      "loss": 0.3088,
      "step": 18165
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3170634653811462,
      "learning_rate": 8.665446759325168e-06,
      "loss": 0.2069,
      "step": 18166
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7011298083809395,
      "learning_rate": 8.664463754944212e-06,
      "loss": 0.4166,
      "step": 18167
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22320841266772204,
      "learning_rate": 8.663480763702696e-06,
      "loss": 0.1266,
      "step": 18168
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.463216391029406,
      "learning_rate": 8.662497785610298e-06,
      "loss": 0.3473,
      "step": 18169
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5409226621719407,
      "learning_rate": 8.661514820676687e-06,
      "loss": 0.0721,
      "step": 18170
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2903558889201768,
      "learning_rate": 8.66053186891153e-06,
      "loss": 0.2633,
      "step": 18171
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.95837902722057,
      "learning_rate": 8.659548930324504e-06,
      "loss": 0.3952,
      "step": 18172
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8573230839831584,
      "learning_rate": 8.658566004925279e-06,
      "loss": 0.3671,
      "step": 18173
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.653313904121608,
      "learning_rate": 8.657583092723521e-06,
      "loss": 0.3815,
      "step": 18174
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29812447562408667,
      "learning_rate": 8.656600193728903e-06,
      "loss": 0.2127,
      "step": 18175
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31691927993586355,
      "learning_rate": 8.655617307951093e-06,
      "loss": 0.3275,
      "step": 18176
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2439022013720528,
      "learning_rate": 8.65463443539976e-06,
      "loss": 0.0756,
      "step": 18177
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.42903435030399284,
      "learning_rate": 8.65365157608458e-06,
      "loss": 0.2566,
      "step": 18178
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31949574730133984,
      "learning_rate": 8.652668730015217e-06,
      "loss": 0.1621,
      "step": 18179
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5353198196195419,
      "learning_rate": 8.651685897201344e-06,
      "loss": 0.3475,
      "step": 18180
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8462556829315804,
      "learning_rate": 8.65070307765263e-06,
      "loss": 0.4699,
      "step": 18181
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3252405478973568,
      "learning_rate": 8.649720271378737e-06,
      "loss": 0.2999,
      "step": 18182
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9148697522766667,
      "learning_rate": 8.648737478389345e-06,
      "loss": 0.3103,
      "step": 18183
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5007231015085974,
      "learning_rate": 8.647754698694118e-06,
      "loss": 0.2556,
      "step": 18184
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9265520083790368,
      "learning_rate": 8.646771932302725e-06,
      "loss": 0.3269,
      "step": 18185
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36593910177398237,
      "learning_rate": 8.645789179224836e-06,
      "loss": 0.1341,
      "step": 18186
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31834348696361126,
      "learning_rate": 8.644806439470118e-06,
      "loss": 0.2637,
      "step": 18187
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.26478807095158713,
      "learning_rate": 8.643823713048237e-06,
      "loss": 0.1901,
      "step": 18188
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5660315787528557,
      "learning_rate": 8.642840999968867e-06,
      "loss": 0.3422,
      "step": 18189
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0219419254317739,
      "learning_rate": 8.641858300241675e-06,
      "loss": 0.408,
      "step": 18190
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.102646962528559,
      "learning_rate": 8.640875613876324e-06,
      "loss": 0.5056,
      "step": 18191
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7214712607506362,
      "learning_rate": 8.63989294088249e-06,
      "loss": 0.2607,
      "step": 18192
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5704362831585286,
      "learning_rate": 8.638910281269837e-06,
      "loss": 0.3945,
      "step": 18193
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23754327194523245,
      "learning_rate": 8.637927635048029e-06,
      "loss": 0.2378,
      "step": 18194
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7204061447003978,
      "learning_rate": 8.63694500222674e-06,
      "loss": 0.6445,
      "step": 18195
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2001780655455616,
      "learning_rate": 8.635962382815635e-06,
      "loss": 0.1013,
      "step": 18196
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8814228617149152,
      "learning_rate": 8.634979776824377e-06,
      "loss": 0.4554,
      "step": 18197
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3208719544646363,
      "learning_rate": 8.63399718426264e-06,
      "loss": 0.2024,
      "step": 18198
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.40494021880425113,
      "learning_rate": 8.633014605140089e-06,
      "loss": 0.2714,
      "step": 18199
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5274824721191456,
      "learning_rate": 8.632032039466388e-06,
      "loss": 0.3638,
      "step": 18200
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9176684463087093,
      "learning_rate": 8.631049487251208e-06,
      "loss": 0.3737,
      "step": 18201
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.40164392910002134,
      "learning_rate": 8.63006694850421e-06,
      "loss": 0.3037,
      "step": 18202
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4562291782679547,
      "learning_rate": 8.629084423235069e-06,
      "loss": 0.2572,
      "step": 18203
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.569003575019856,
      "learning_rate": 8.628101911453444e-06,
      "loss": 0.6922,
      "step": 18204
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19152021644394598,
      "learning_rate": 8.627119413169003e-06,
      "loss": 0.1414,
      "step": 18205
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3911216953072646,
      "learning_rate": 8.626136928391414e-06,
      "loss": 0.2796,
      "step": 18206
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3304406662177766,
      "learning_rate": 8.625154457130343e-06,
      "loss": 0.2204,
      "step": 18207
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8017033454340442,
      "learning_rate": 8.62417199939545e-06,
      "loss": 0.5364,
      "step": 18208
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.71954916632903,
      "learning_rate": 8.623189555196411e-06,
      "loss": 0.3559,
      "step": 18209
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2081680208317154,
      "learning_rate": 8.622207124542886e-06,
      "loss": 0.8209,
      "step": 18210
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27983954853286763,
      "learning_rate": 8.621224707444537e-06,
      "loss": 0.1873,
      "step": 18211
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4806333218705728,
      "learning_rate": 8.620242303911038e-06,
      "loss": 0.3734,
      "step": 18212
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35492544157982564,
      "learning_rate": 8.619259913952047e-06,
      "loss": 0.2159,
      "step": 18213
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.19783863201872415,
      "learning_rate": 8.618277537577228e-06,
      "loss": 0.0915,
      "step": 18214
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6152917538590156,
      "learning_rate": 8.617295174796253e-06,
      "loss": 0.2789,
      "step": 18215
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30399873149366435,
      "learning_rate": 8.616312825618782e-06,
      "loss": 0.2062,
      "step": 18216
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6608520790412347,
      "learning_rate": 8.61533049005448e-06,
      "loss": 0.4517,
      "step": 18217
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2669083572777445,
      "learning_rate": 8.614348168113014e-06,
      "loss": 0.2459,
      "step": 18218
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4151716469750841,
      "learning_rate": 8.613365859804046e-06,
      "loss": 0.8055,
      "step": 18219
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.32028419654092827,
      "learning_rate": 8.612383565137237e-06,
      "loss": 0.0688,
      "step": 18220
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37646164300527707,
      "learning_rate": 8.61140128412226e-06,
      "loss": 0.2901,
      "step": 18221
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.000562090229169,
      "learning_rate": 8.610419016768774e-06,
      "loss": 0.263,
      "step": 18222
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24156490730761848,
      "learning_rate": 8.609436763086439e-06,
      "loss": 0.2183,
      "step": 18223
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2209807372458928,
      "learning_rate": 8.608454523084925e-06,
      "loss": 0.0714,
      "step": 18224
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3713216008360178,
      "learning_rate": 8.607472296773896e-06,
      "loss": 0.2642,
      "step": 18225
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4923745832915056,
      "learning_rate": 8.606490084163009e-06,
      "loss": 0.32,
      "step": 18226
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6540232748828622,
      "learning_rate": 8.605507885261934e-06,
      "loss": 0.41,
      "step": 18227
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5831184957048867,
      "learning_rate": 8.604525700080326e-06,
      "loss": 0.8594,
      "step": 18228
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.32185292636755497,
      "learning_rate": 8.603543528627857e-06,
      "loss": 0.1816,
      "step": 18229
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.26064132413018865,
      "learning_rate": 8.602561370914189e-06,
      "loss": 0.2542,
      "step": 18230
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.101655569902898,
      "learning_rate": 8.601579226948977e-06,
      "loss": 0.1698,
      "step": 18231
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.7995326395808213,
      "learning_rate": 8.600597096741893e-06,
      "loss": 0.7916,
      "step": 18232
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.33981598944816815,
      "learning_rate": 8.599614980302594e-06,
      "loss": 0.1914,
      "step": 18233
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3432704261200375,
      "learning_rate": 8.59863287764074e-06,
      "loss": 0.2796,
      "step": 18234
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36856920332065685,
      "learning_rate": 8.597650788766e-06,
      "loss": 0.2777,
      "step": 18235
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3122547954317279,
      "learning_rate": 8.596668713688034e-06,
      "loss": 0.2517,
      "step": 18236
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.825574667838015,
      "learning_rate": 8.595686652416502e-06,
      "loss": 0.6018,
      "step": 18237
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39701062642183865,
      "learning_rate": 8.594704604961067e-06,
      "loss": 0.272,
      "step": 18238
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.521146948859441,
      "learning_rate": 8.593722571331391e-06,
      "loss": 0.25,
      "step": 18239
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5379415274415674,
      "learning_rate": 8.592740551537131e-06,
      "loss": 0.0321,
      "step": 18240
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2992575167931452,
      "learning_rate": 8.591758545587956e-06,
      "loss": 0.2564,
      "step": 18241
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5978754441021727,
      "learning_rate": 8.590776553493525e-06,
      "loss": 0.1979,
      "step": 18242
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5287006619308676,
      "learning_rate": 8.589794575263494e-06,
      "loss": 0.3086,
      "step": 18243
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3951899240491173,
      "learning_rate": 8.588812610907532e-06,
      "loss": 0.2711,
      "step": 18244
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3984218528931669,
      "learning_rate": 8.587830660435295e-06,
      "loss": 0.2616,
      "step": 18245
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2463221663066545,
      "learning_rate": 8.586848723856442e-06,
      "loss": 0.1795,
      "step": 18246
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8612215659129703,
      "learning_rate": 8.585866801180639e-06,
      "loss": 0.386,
      "step": 18247
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22986308636695257,
      "learning_rate": 8.584884892417543e-06,
      "loss": 0.2073,
      "step": 18248
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0305918789107993,
      "learning_rate": 8.583902997576814e-06,
      "loss": 0.4849,
      "step": 18249
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5202875082283493,
      "learning_rate": 8.582921116668115e-06,
      "loss": 0.312,
      "step": 18250
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8591638424185681,
      "learning_rate": 8.581939249701105e-06,
      "loss": 0.4738,
      "step": 18251
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2781425233144932,
      "learning_rate": 8.580957396685438e-06,
      "loss": 0.1754,
      "step": 18252
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23737649003676226,
      "learning_rate": 8.579975557630784e-06,
      "loss": 0.1792,
      "step": 18253
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.46208559017524437,
      "learning_rate": 8.578993732546793e-06,
      "loss": 0.3689,
      "step": 18254
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9475421750186067,
      "learning_rate": 8.578011921443135e-06,
      "loss": 0.486,
      "step": 18255
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3230153910237769,
      "learning_rate": 8.577030124329463e-06,
      "loss": 0.1754,
      "step": 18256
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3519417558185494,
      "learning_rate": 8.576048341215432e-06,
      "loss": 0.2301,
      "step": 18257
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7824536516158418,
      "learning_rate": 8.57506657211071e-06,
      "loss": 0.5049,
      "step": 18258
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.297723779800388,
      "learning_rate": 8.574084817024953e-06,
      "loss": 0.2267,
      "step": 18259
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5924102728978945,
      "learning_rate": 8.573103075967813e-06,
      "loss": 0.4516,
      "step": 18260
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.26567197731745834,
      "learning_rate": 8.57212134894896e-06,
      "loss": 0.1894,
      "step": 18261
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.245777623207812,
      "learning_rate": 8.571139635978046e-06,
      "loss": 0.4889,
      "step": 18262
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8370927201421616,
      "learning_rate": 8.570157937064728e-06,
      "loss": 0.4771,
      "step": 18263
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.45966020625825166,
      "learning_rate": 8.56917625221867e-06,
      "loss": 0.3104,
      "step": 18264
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2566049705407987,
      "learning_rate": 8.568194581449526e-06,
      "loss": 0.2191,
      "step": 18265
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.22306392161627736,
      "learning_rate": 8.56721292476695e-06,
      "loss": 0.1688,
      "step": 18266
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.175641928143336,
      "learning_rate": 8.566231282180611e-06,
      "loss": 0.8438,
      "step": 18267
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6996567757412319,
      "learning_rate": 8.565249653700158e-06,
      "loss": 0.3385,
      "step": 18268
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6721315373536678,
      "learning_rate": 8.56426803933525e-06,
      "loss": 0.4017,
      "step": 18269
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34087509202251576,
      "learning_rate": 8.563286439095549e-06,
      "loss": 0.175,
      "step": 18270
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5502622873414588,
      "learning_rate": 8.562304852990708e-06,
      "loss": 0.3611,
      "step": 18271
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30564219082582417,
      "learning_rate": 8.561323281030379e-06,
      "loss": 0.2534,
      "step": 18272
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4381579633300493,
      "learning_rate": 8.56034172322423e-06,
      "loss": 0.2722,
      "step": 18273
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18799674240360464,
      "learning_rate": 8.559360179581914e-06,
      "loss": 0.0742,
      "step": 18274
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.29831256544277984,
      "learning_rate": 8.558378650113083e-06,
      "loss": 0.2463,
      "step": 18275
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7911099570002997,
      "learning_rate": 8.5573971348274e-06,
      "loss": 0.3315,
      "step": 18276
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3153295111228996,
      "learning_rate": 8.556415633734518e-06,
      "loss": 0.2895,
      "step": 18277
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.814625384647927,
      "learning_rate": 8.55543414684409e-06,
      "loss": 0.2813,
      "step": 18278
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34112566039684866,
      "learning_rate": 8.55445267416578e-06,
      "loss": 0.1773,
      "step": 18279
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.46257001956390714,
      "learning_rate": 8.55347121570924e-06,
      "loss": 0.2999,
      "step": 18280
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7538926916964234,
      "learning_rate": 8.552489771484127e-06,
      "loss": 0.6067,
      "step": 18281
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.668156487019058,
      "learning_rate": 8.551508341500096e-06,
      "loss": 0.9359,
      "step": 18282
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23632829023868707,
      "learning_rate": 8.5505269257668e-06,
      "loss": 0.1933,
      "step": 18283
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2606823881455657,
      "learning_rate": 8.549545524293899e-06,
      "loss": 0.2247,
      "step": 18284
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5562324438022668,
      "learning_rate": 8.548564137091047e-06,
      "loss": 0.2372,
      "step": 18285
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.082752516676632,
      "learning_rate": 8.547582764167897e-06,
      "loss": 0.5404,
      "step": 18286
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.122254716682912,
      "learning_rate": 8.546601405534106e-06,
      "loss": 0.1971,
      "step": 18287
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35104963719598753,
      "learning_rate": 8.54562006119933e-06,
      "loss": 0.2451,
      "step": 18288
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.45373341567801523,
      "learning_rate": 8.544638731173218e-06,
      "loss": 0.2361,
      "step": 18289
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.4662618755165329,
      "learning_rate": 8.543657415465434e-06,
      "loss": 0.7937,
      "step": 18290
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7783520193033815,
      "learning_rate": 8.542676114085626e-06,
      "loss": 0.5396,
      "step": 18291
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.21320569502780898,
      "learning_rate": 8.541694827043448e-06,
      "loss": 0.0905,
      "step": 18292
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3661495651294482,
      "learning_rate": 8.54071355434856e-06,
      "loss": 0.2769,
      "step": 18293
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6571091696147255,
      "learning_rate": 8.53973229601061e-06,
      "loss": 0.3456,
      "step": 18294
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3024109261136462,
      "learning_rate": 8.538751052039253e-06,
      "loss": 0.2911,
      "step": 18295
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.38517888027400904,
      "learning_rate": 8.537769822444147e-06,
      "loss": 0.1571,
      "step": 18296
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6582679899209455,
      "learning_rate": 8.53678860723494e-06,
      "loss": 0.3023,
      "step": 18297
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3128726831980333,
      "learning_rate": 8.535807406421287e-06,
      "loss": 0.1995,
      "step": 18298
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3271442676711618,
      "learning_rate": 8.534826220012845e-06,
      "loss": 0.7728,
      "step": 18299
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30656019176070426,
      "learning_rate": 8.533845048019266e-06,
      "loss": 0.2372,
      "step": 18300
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.834194171562382,
      "learning_rate": 8.532863890450197e-06,
      "loss": 0.4145,
      "step": 18301
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2792469451353538,
      "learning_rate": 8.531882747315302e-06,
      "loss": 0.1936,
      "step": 18302
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.552288113736498,
      "learning_rate": 8.530901618624226e-06,
      "loss": 0.4539,
      "step": 18303
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.1881747197522773,
      "learning_rate": 8.529920504386618e-06,
      "loss": 0.1189,
      "step": 18304
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1165157690245058,
      "learning_rate": 8.52893940461214e-06,
      "loss": 0.1687,
      "step": 18305
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6304139923022417,
      "learning_rate": 8.52795831931044e-06,
      "loss": 0.3191,
      "step": 18306
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24544369503569965,
      "learning_rate": 8.526977248491171e-06,
      "loss": 0.2274,
      "step": 18307
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2521342490409386,
      "learning_rate": 8.525996192163985e-06,
      "loss": 0.7908,
      "step": 18308
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.8863292555834397,
      "learning_rate": 8.525015150338528e-06,
      "loss": 0.2989,
      "step": 18309
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5389674753520487,
      "learning_rate": 8.524034123024463e-06,
      "loss": 0.4491,
      "step": 18310
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27117602446343336,
      "learning_rate": 8.523053110231436e-06,
      "loss": 0.1872,
      "step": 18311
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7824817641802893,
      "learning_rate": 8.522072111969095e-06,
      "loss": 0.4999,
      "step": 18312
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3112347391517643,
      "learning_rate": 8.5210911282471e-06,
      "loss": 0.2416,
      "step": 18313
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27420674736697725,
      "learning_rate": 8.520110159075094e-06,
      "loss": 0.1584,
      "step": 18314
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.32288810690402436,
      "learning_rate": 8.519129204462727e-06,
      "loss": 0.1628,
      "step": 18315
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5673810654551621,
      "learning_rate": 8.51814826441966e-06,
      "loss": 0.3656,
      "step": 18316
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9962656484938821,
      "learning_rate": 8.517167338955536e-06,
      "loss": 0.4627,
      "step": 18317
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.28493498112452703,
      "learning_rate": 8.516186428080008e-06,
      "loss": 0.2306,
      "step": 18318
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6007885004078526,
      "learning_rate": 8.515205531802727e-06,
      "loss": 0.3649,
      "step": 18319
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30355713787561694,
      "learning_rate": 8.514224650133343e-06,
      "loss": 0.2008,
      "step": 18320
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7306610650049477,
      "learning_rate": 8.5132437830815e-06,
      "loss": 0.4393,
      "step": 18321
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2123162336067149,
      "learning_rate": 8.512262930656859e-06,
      "loss": 0.0911,
      "step": 18322
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.833458458500546,
      "learning_rate": 8.511282092869067e-06,
      "loss": 0.8437,
      "step": 18323
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31323290238173074,
      "learning_rate": 8.510301269727764e-06,
      "loss": 0.1857,
      "step": 18324
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3782172719564944,
      "learning_rate": 8.509320461242614e-06,
      "loss": 0.2786,
      "step": 18325
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.43833400577068166,
      "learning_rate": 8.50833966742326e-06,
      "loss": 0.2915,
      "step": 18326
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8692735408003954,
      "learning_rate": 8.507358888279346e-06,
      "loss": 0.465,
      "step": 18327
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7470434857039533,
      "learning_rate": 8.50637812382053e-06,
      "loss": 0.3146,
      "step": 18328
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6309602037955594,
      "learning_rate": 8.505397374056455e-06,
      "loss": 0.3572,
      "step": 18329
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.34809690475562394,
      "learning_rate": 8.504416638996775e-06,
      "loss": 0.2171,
      "step": 18330
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2323513463164309,
      "learning_rate": 8.503435918651134e-06,
      "loss": 0.198,
      "step": 18331
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.40851837503924054,
      "learning_rate": 8.502455213029185e-06,
      "loss": 0.2196,
      "step": 18332
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.23072805997615065,
      "learning_rate": 8.501474522140574e-06,
      "loss": 0.0666,
      "step": 18333
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3532191987276839,
      "learning_rate": 8.500493845994951e-06,
      "loss": 0.2998,
      "step": 18334
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.224187026678175,
      "learning_rate": 8.499513184601958e-06,
      "loss": 0.3902,
      "step": 18335
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.0995628259033332,
      "learning_rate": 8.498532537971253e-06,
      "loss": 0.5645,
      "step": 18336
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3206252929436856,
      "learning_rate": 8.497551906112478e-06,
      "loss": 0.2582,
      "step": 18337
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36416107588584634,
      "learning_rate": 8.49657128903528e-06,
      "loss": 0.272,
      "step": 18338
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4283360706236224,
      "learning_rate": 8.49559068674931e-06,
      "loss": 0.2024,
      "step": 18339
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.854460648540313,
      "learning_rate": 8.494610099264215e-06,
      "loss": 0.8019,
      "step": 18340
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.18984091428361335,
      "learning_rate": 8.493629526589637e-06,
      "loss": 0.0892,
      "step": 18341
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.31153193156601916,
      "learning_rate": 8.49264896873523e-06,
      "loss": 0.2702,
      "step": 18342
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3365920882189288,
      "learning_rate": 8.491668425710639e-06,
      "loss": 0.2249,
      "step": 18343
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1835133914449874,
      "learning_rate": 8.490687897525509e-06,
      "loss": 0.4894,
      "step": 18344
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.219057335336078,
      "learning_rate": 8.48970738418949e-06,
      "loss": 0.4871,
      "step": 18345
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7428086985226042,
      "learning_rate": 8.488726885712226e-06,
      "loss": 0.2895,
      "step": 18346
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.32144470731809444,
      "learning_rate": 8.48774640210336e-06,
      "loss": 0.2561,
      "step": 18347
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.20516396209093,
      "learning_rate": 8.486765933372548e-06,
      "loss": 0.1595,
      "step": 18348
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3220852662153726,
      "learning_rate": 8.485785479529429e-06,
      "loss": 0.2867,
      "step": 18349
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.24351309140436686,
      "learning_rate": 8.48480504058365e-06,
      "loss": 0.1002,
      "step": 18350
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7134580834948828,
      "learning_rate": 8.483824616544858e-06,
      "loss": 0.4915,
      "step": 18351
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35567405610281055,
      "learning_rate": 8.482844207422699e-06,
      "loss": 0.2138,
      "step": 18352
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8742100068488324,
      "learning_rate": 8.481863813226813e-06,
      "loss": 0.4765,
      "step": 18353
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.39147839789159483,
      "learning_rate": 8.480883433966857e-06,
      "loss": 0.2725,
      "step": 18354
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.5412364974920623,
      "learning_rate": 8.479903069652466e-06,
      "loss": 0.8272,
      "step": 18355
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4723742063459923,
      "learning_rate": 8.47892272029329e-06,
      "loss": 0.1238,
      "step": 18356
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.317142914154847,
      "learning_rate": 8.477942385898975e-06,
      "loss": 0.2117,
      "step": 18357
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.480548522051543,
      "learning_rate": 8.476962066479157e-06,
      "loss": 0.6458,
      "step": 18358
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.17863397477309603,
      "learning_rate": 8.475981762043494e-06,
      "loss": 0.0727,
      "step": 18359
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3339108131854345,
      "learning_rate": 8.475001472601623e-06,
      "loss": 0.3054,
      "step": 18360
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2530330632041653,
      "learning_rate": 8.474021198163183e-06,
      "loss": 0.1749,
      "step": 18361
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.543424250044576,
      "learning_rate": 8.473040938737831e-06,
      "loss": 0.3905,
      "step": 18362
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6972269469395278,
      "learning_rate": 8.472060694335204e-06,
      "loss": 0.334,
      "step": 18363
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1944553304900174,
      "learning_rate": 8.471080464964945e-06,
      "loss": 0.631,
      "step": 18364
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3254375752822972,
      "learning_rate": 8.4701002506367e-06,
      "loss": 0.1666,
      "step": 18365
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35839921859903545,
      "learning_rate": 8.469120051360115e-06,
      "loss": 0.292,
      "step": 18366
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4161452366486556,
      "learning_rate": 8.468139867144824e-06,
      "loss": 0.246,
      "step": 18367
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36019016949966076,
      "learning_rate": 8.467159698000482e-06,
      "loss": 0.2032,
      "step": 18368
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3021942628843694,
      "learning_rate": 8.466179543936728e-06,
      "loss": 0.15,
      "step": 18369
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3315315874774126,
      "learning_rate": 8.465199404963202e-06,
      "loss": 0.2214,
      "step": 18370
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.821782967346051,
      "learning_rate": 8.46421928108955e-06,
      "loss": 0.5195,
      "step": 18371
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27644044073489576,
      "learning_rate": 8.463239172325416e-06,
      "loss": 0.2498,
      "step": 18372
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.216527133710873,
      "learning_rate": 8.462259078680435e-06,
      "loss": 0.8443,
      "step": 18373
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2896085034278296,
      "learning_rate": 8.461279000164259e-06,
      "loss": 0.1435,
      "step": 18374
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5450129778385788,
      "learning_rate": 8.460298936786529e-06,
      "loss": 0.3221,
      "step": 18375
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.30479327337652,
      "learning_rate": 8.45931888855688e-06,
      "loss": 0.2919,
      "step": 18376
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4125637400142327,
      "learning_rate": 8.458338855484963e-06,
      "loss": 0.2504,
      "step": 18377
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2597210307817243,
      "learning_rate": 8.457358837580413e-06,
      "loss": 0.1979,
      "step": 18378
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.552673310601732,
      "learning_rate": 8.456378834852872e-06,
      "loss": 0.351,
      "step": 18379
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37883231232636155,
      "learning_rate": 8.455398847311987e-06,
      "loss": 0.2775,
      "step": 18380
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4096861118390171,
      "learning_rate": 8.454418874967397e-06,
      "loss": 0.2636,
      "step": 18381
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1574954800306856,
      "learning_rate": 8.453438917828742e-06,
      "loss": 0.3731,
      "step": 18382
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.38036078951119806,
      "learning_rate": 8.452458975905664e-06,
      "loss": 0.0679,
      "step": 18383
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2709283105779863,
      "learning_rate": 8.451479049207801e-06,
      "loss": 0.2542,
      "step": 18384
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4074554777396767,
      "learning_rate": 8.450499137744798e-06,
      "loss": 0.239,
      "step": 18385
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8619976416110678,
      "learning_rate": 8.449519241526298e-06,
      "loss": 0.4945,
      "step": 18386
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5049845462020562,
      "learning_rate": 8.448539360561935e-06,
      "loss": 0.3243,
      "step": 18387
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3870064980682464,
      "learning_rate": 8.447559494861352e-06,
      "loss": 0.2713,
      "step": 18388
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3197905576326183,
      "learning_rate": 8.44657964443419e-06,
      "loss": 0.1712,
      "step": 18389
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3194440057787116,
      "learning_rate": 8.445599809290086e-06,
      "loss": 0.3048,
      "step": 18390
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8852542179304989,
      "learning_rate": 8.444619989438687e-06,
      "loss": 0.5645,
      "step": 18391
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30220724058524195,
      "learning_rate": 8.443640184889626e-06,
      "loss": 0.1529,
      "step": 18392
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.35843466342544433,
      "learning_rate": 8.442660395652542e-06,
      "loss": 0.2252,
      "step": 18393
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.685389856740792,
      "learning_rate": 8.441680621737082e-06,
      "loss": 0.8531,
      "step": 18394
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7156827726170295,
      "learning_rate": 8.440700863152882e-06,
      "loss": 0.3259,
      "step": 18395
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.30207439432471545,
      "learning_rate": 8.439721119909576e-06,
      "loss": 0.2977,
      "step": 18396
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.25973043687618375,
      "learning_rate": 8.438741392016809e-06,
      "loss": 0.1814,
      "step": 18397
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7109495305785437,
      "learning_rate": 8.43776167948422e-06,
      "loss": 0.3074,
      "step": 18398
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4314257327391582,
      "learning_rate": 8.43678198232144e-06,
      "loss": 0.299,
      "step": 18399
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2554007176909853,
      "learning_rate": 8.435802300538117e-06,
      "loss": 0.1339,
      "step": 18400
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3907424991938325,
      "learning_rate": 8.434822634143887e-06,
      "loss": 0.2842,
      "step": 18401
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3219692348528808,
      "learning_rate": 8.433842983148384e-06,
      "loss": 0.2134,
      "step": 18402
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.46304963412480615,
      "learning_rate": 8.432863347561251e-06,
      "loss": 0.3528,
      "step": 18403
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6552284008555599,
      "learning_rate": 8.431883727392123e-06,
      "loss": 0.318,
      "step": 18404
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.8891320995766281,
      "learning_rate": 8.430904122650637e-06,
      "loss": 0.4856,
      "step": 18405
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2719348745089196,
      "learning_rate": 8.429924533346436e-06,
      "loss": 0.0744,
      "step": 18406
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3598773646774953,
      "learning_rate": 8.42894495948915e-06,
      "loss": 0.2727,
      "step": 18407
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3270923685758919,
      "learning_rate": 8.427965401088424e-06,
      "loss": 0.2769,
      "step": 18408
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4605825954590173,
      "learning_rate": 8.426985858153892e-06,
      "loss": 0.2529,
      "step": 18409
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.27109589467898726,
      "learning_rate": 8.426006330695186e-06,
      "loss": 0.0717,
      "step": 18410
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2936213694017267,
      "learning_rate": 8.425026818721953e-06,
      "loss": 0.1974,
      "step": 18411
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6743018525753086,
      "learning_rate": 8.424047322243823e-06,
      "loss": 0.4416,
      "step": 18412
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.41049049439616453,
      "learning_rate": 8.423067841270432e-06,
      "loss": 0.2727,
      "step": 18413
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4761634121545937,
      "learning_rate": 8.422088375811421e-06,
      "loss": 0.3809,
      "step": 18414
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.28746163131175106,
      "learning_rate": 8.421108925876424e-06,
      "loss": 0.1586,
      "step": 18415
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.512826945330186,
      "learning_rate": 8.420129491475072e-06,
      "loss": 0.3589,
      "step": 18416
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9117673926814169,
      "learning_rate": 8.419150072617011e-06,
      "loss": 0.5581,
      "step": 18417
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4842508501677098,
      "learning_rate": 8.418170669311872e-06,
      "loss": 0.2547,
      "step": 18418
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.25260068098011595,
      "learning_rate": 8.417191281569286e-06,
      "loss": 0.1456,
      "step": 18419
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.3234653360672503,
      "learning_rate": 8.416211909398896e-06,
      "loss": 0.275,
      "step": 18420
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.38104175910907795,
      "learning_rate": 8.415232552810337e-06,
      "loss": 0.2212,
      "step": 18421
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6530786886958334,
      "learning_rate": 8.414253211813234e-06,
      "loss": 0.39,
      "step": 18422
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7301177480365065,
      "learning_rate": 8.413273886417237e-06,
      "loss": 0.3379,
      "step": 18423
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2797816764084812,
      "learning_rate": 8.412294576631973e-06,
      "loss": 0.1798,
      "step": 18424
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.2700501491295426,
      "learning_rate": 8.411315282467071e-06,
      "loss": 0.6107,
      "step": 18425
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2957852582206699,
      "learning_rate": 8.41033600393218e-06,
      "loss": 0.2612,
      "step": 18426
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.1109324336754358,
      "learning_rate": 8.409356741036922e-06,
      "loss": 0.7039,
      "step": 18427
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.17048919659226228,
      "learning_rate": 8.408377493790934e-06,
      "loss": 0.0689,
      "step": 18428
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5008929071812632,
      "learning_rate": 8.407398262203856e-06,
      "loss": 0.3287,
      "step": 18429
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5698539492318578,
      "learning_rate": 8.406419046285317e-06,
      "loss": 0.2925,
      "step": 18430
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.4593386088317388,
      "learning_rate": 8.405439846044949e-06,
      "loss": 0.3499,
      "step": 18431
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37426022815713716,
      "learning_rate": 8.40446066149239e-06,
      "loss": 0.2207,
      "step": 18432
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9417597296282996,
      "learning_rate": 8.403481492637272e-06,
      "loss": 0.4316,
      "step": 18433
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.296586767781574,
      "learning_rate": 8.40250233948923e-06,
      "loss": 0.1964,
      "step": 18434
    },
    {
      "epoch": 0.56,
      "grad_norm": 1.3460682855607322,
      "learning_rate": 8.401523202057896e-06,
      "loss": 0.8669,
      "step": 18435
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.7055024573467156,
      "learning_rate": 8.400544080352899e-06,
      "loss": 0.3318,
      "step": 18436
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2972628848336243,
      "learning_rate": 8.399564974383878e-06,
      "loss": 0.2208,
      "step": 18437
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.36671437520314104,
      "learning_rate": 8.398585884160465e-06,
      "loss": 0.2499,
      "step": 18438
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.37763582003495244,
      "learning_rate": 8.397606809692288e-06,
      "loss": 0.2784,
      "step": 18439
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.28216582660802536,
      "learning_rate": 8.396627750988987e-06,
      "loss": 0.1589,
      "step": 18440
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6408197685758386,
      "learning_rate": 8.395648708060187e-06,
      "loss": 0.0277,
      "step": 18441
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.59060245053284,
      "learning_rate": 8.394669680915519e-06,
      "loss": 0.2659,
      "step": 18442
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2643760527131214,
      "learning_rate": 8.393690669564622e-06,
      "loss": 0.2237,
      "step": 18443
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.5332398849628479,
      "learning_rate": 8.392711674017127e-06,
      "loss": 0.3603,
      "step": 18444
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.9007790592971808,
      "learning_rate": 8.39173269428266e-06,
      "loss": 0.464,
      "step": 18445
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6538327459157459,
      "learning_rate": 8.39075373037086e-06,
      "loss": 0.3983,
      "step": 18446
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.2800491792209357,
      "learning_rate": 8.389774782291352e-06,
      "loss": 0.1808,
      "step": 18447
    },
    {
      "epoch": 0.56,
      "grad_norm": 0.6710553639982515,
      "learning_rate": 8.388795850053766e-06,
      "loss": 0.3978,
      "step": 18448
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20268541883523267,
      "learning_rate": 8.387816933667741e-06,
      "loss": 0.1359,
      "step": 18449
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4471179139227992,
      "learning_rate": 8.386838033142905e-06,
      "loss": 0.2514,
      "step": 18450
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6030889643544318,
      "learning_rate": 8.385859148488884e-06,
      "loss": 0.2867,
      "step": 18451
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3423467234386755,
      "learning_rate": 8.384880279715313e-06,
      "loss": 0.2196,
      "step": 18452
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4695787913729614,
      "learning_rate": 8.38390142683182e-06,
      "loss": 0.7864,
      "step": 18453
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7772018489259771,
      "learning_rate": 8.382922589848034e-06,
      "loss": 0.3376,
      "step": 18454
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3025292602894253,
      "learning_rate": 8.381943768773592e-06,
      "loss": 0.2707,
      "step": 18455
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36001338770365726,
      "learning_rate": 8.380964963618121e-06,
      "loss": 0.182,
      "step": 18456
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4540013813076082,
      "learning_rate": 8.379986174391246e-06,
      "loss": 0.3159,
      "step": 18457
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1410127745523618,
      "learning_rate": 8.379007401102603e-06,
      "loss": 0.0704,
      "step": 18458
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0524091120597423,
      "learning_rate": 8.378028643761813e-06,
      "loss": 0.5347,
      "step": 18459
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37256320108732144,
      "learning_rate": 8.377049902378514e-06,
      "loss": 0.1485,
      "step": 18460
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34210790264623187,
      "learning_rate": 8.376071176962335e-06,
      "loss": 0.2835,
      "step": 18461
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3165884479065669,
      "learning_rate": 8.375092467522897e-06,
      "loss": 0.2543,
      "step": 18462
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7131620251933866,
      "learning_rate": 8.374113774069836e-06,
      "loss": 0.3308,
      "step": 18463
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8760971095217543,
      "learning_rate": 8.373135096612782e-06,
      "loss": 0.4834,
      "step": 18464
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31904199743769945,
      "learning_rate": 8.372156435161356e-06,
      "loss": 0.179,
      "step": 18465
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5152694581531041,
      "learning_rate": 8.371177789725193e-06,
      "loss": 0.3375,
      "step": 18466
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18793774801112711,
      "learning_rate": 8.370199160313918e-06,
      "loss": 0.1716,
      "step": 18467
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.9252355323498624,
      "learning_rate": 8.369220546937157e-06,
      "loss": 0.7741,
      "step": 18468
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.868954328836206,
      "learning_rate": 8.368241949604545e-06,
      "loss": 0.2029,
      "step": 18469
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33620255917436664,
      "learning_rate": 8.367263368325707e-06,
      "loss": 0.2724,
      "step": 18470
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8215383201947941,
      "learning_rate": 8.366284803110265e-06,
      "loss": 0.5538,
      "step": 18471
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.183615406416721,
      "learning_rate": 8.365306253967853e-06,
      "loss": 0.47,
      "step": 18472
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34917105681773103,
      "learning_rate": 8.364327720908096e-06,
      "loss": 0.2674,
      "step": 18473
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3516533870238066,
      "learning_rate": 8.363349203940618e-06,
      "loss": 0.224,
      "step": 18474
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4463992118635192,
      "learning_rate": 8.362370703075054e-06,
      "loss": 0.2115,
      "step": 18475
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.313019213450016,
      "learning_rate": 8.361392218321023e-06,
      "loss": 0.11,
      "step": 18476
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4267239616460634,
      "learning_rate": 8.360413749688154e-06,
      "loss": 0.2143,
      "step": 18477
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2997002109408969,
      "learning_rate": 8.359435297186076e-06,
      "loss": 0.1548,
      "step": 18478
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3968653797100202,
      "learning_rate": 8.358456860824413e-06,
      "loss": 0.2968,
      "step": 18479
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2753276270058609,
      "learning_rate": 8.357478440612787e-06,
      "loss": 0.2306,
      "step": 18480
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8213474468635984,
      "learning_rate": 8.356500036560834e-06,
      "loss": 0.4165,
      "step": 18481
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9457216991822118,
      "learning_rate": 8.355521648678172e-06,
      "loss": 0.3197,
      "step": 18482
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9138800013387762,
      "learning_rate": 8.354543276974431e-06,
      "loss": 0.416,
      "step": 18483
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.28609200056072154,
      "learning_rate": 8.353564921459236e-06,
      "loss": 0.1944,
      "step": 18484
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30510664218425987,
      "learning_rate": 8.352586582142204e-06,
      "loss": 0.2838,
      "step": 18485
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23164340302257788,
      "learning_rate": 8.351608259032974e-06,
      "loss": 0.0911,
      "step": 18486
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.590778676688068,
      "learning_rate": 8.350629952141163e-06,
      "loss": 0.7901,
      "step": 18487
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2883290791924199,
      "learning_rate": 8.349651661476394e-06,
      "loss": 0.1819,
      "step": 18488
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8148759969353795,
      "learning_rate": 8.3486733870483e-06,
      "loss": 0.3457,
      "step": 18489
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.43535140313637466,
      "learning_rate": 8.3476951288665e-06,
      "loss": 0.3738,
      "step": 18490
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34679853157035634,
      "learning_rate": 8.346716886940613e-06,
      "loss": 0.2186,
      "step": 18491
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5659931758203129,
      "learning_rate": 8.345738661280273e-06,
      "loss": 0.236,
      "step": 18492
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3276653563873606,
      "learning_rate": 8.344760451895104e-06,
      "loss": 0.2315,
      "step": 18493
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5671726084911742,
      "learning_rate": 8.343782258794719e-06,
      "loss": 0.2936,
      "step": 18494
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32598571361928375,
      "learning_rate": 8.342804081988754e-06,
      "loss": 0.0976,
      "step": 18495
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4636272474927571,
      "learning_rate": 8.341825921486829e-06,
      "loss": 0.3808,
      "step": 18496
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.262928950912933,
      "learning_rate": 8.340847777298561e-06,
      "loss": 0.1858,
      "step": 18497
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4401493030673904,
      "learning_rate": 8.339869649433582e-06,
      "loss": 0.3121,
      "step": 18498
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7855713367349351,
      "learning_rate": 8.338891537901513e-06,
      "loss": 0.32,
      "step": 18499
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1684556798269903,
      "learning_rate": 8.33791344271197e-06,
      "loss": 0.671,
      "step": 18500
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44754720235583556,
      "learning_rate": 8.336935363874587e-06,
      "loss": 0.1151,
      "step": 18501
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34960540271496987,
      "learning_rate": 8.335957301398982e-06,
      "loss": 0.213,
      "step": 18502
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4845952965664532,
      "learning_rate": 8.334979255294771e-06,
      "loss": 0.3481,
      "step": 18503
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22440536013491436,
      "learning_rate": 8.334001225571587e-06,
      "loss": 0.1423,
      "step": 18504
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6019136699638912,
      "learning_rate": 8.333023212239046e-06,
      "loss": 0.4045,
      "step": 18505
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2985641558296894,
      "learning_rate": 8.332045215306768e-06,
      "loss": 0.192,
      "step": 18506
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7987999548303318,
      "learning_rate": 8.33106723478438e-06,
      "loss": 0.4161,
      "step": 18507
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3501066210275693,
      "learning_rate": 8.330089270681501e-06,
      "loss": 0.2761,
      "step": 18508
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44749672506801336,
      "learning_rate": 8.329111323007756e-06,
      "loss": 0.3157,
      "step": 18509
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5642814148559627,
      "learning_rate": 8.328133391772763e-06,
      "loss": 0.1343,
      "step": 18510
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.342602967910325,
      "learning_rate": 8.327155476986139e-06,
      "loss": 0.2864,
      "step": 18511
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9555333707683166,
      "learning_rate": 8.326177578657514e-06,
      "loss": 0.1476,
      "step": 18512
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30818768689399945,
      "learning_rate": 8.325199696796506e-06,
      "loss": 0.2157,
      "step": 18513
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46159038215957143,
      "learning_rate": 8.324221831412731e-06,
      "loss": 0.2634,
      "step": 18514
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.25830534648499137,
      "learning_rate": 8.323243982515816e-06,
      "loss": 0.2016,
      "step": 18515
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5463109218365871,
      "learning_rate": 8.322266150115377e-06,
      "loss": 0.334,
      "step": 18516
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9645310146668894,
      "learning_rate": 8.321288334221034e-06,
      "loss": 0.6084,
      "step": 18517
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.3781967722452504,
      "learning_rate": 8.32031053484241e-06,
      "loss": 0.6275,
      "step": 18518
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.48466104336293087,
      "learning_rate": 8.319332751989126e-06,
      "loss": 0.1161,
      "step": 18519
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3920946829114333,
      "learning_rate": 8.318354985670796e-06,
      "loss": 0.3023,
      "step": 18520
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3192062397518094,
      "learning_rate": 8.317377235897046e-06,
      "loss": 0.2273,
      "step": 18521
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.8677881757361392,
      "learning_rate": 8.316399502677492e-06,
      "loss": 0.6498,
      "step": 18522
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3924819999886054,
      "learning_rate": 8.31542178602175e-06,
      "loss": 0.2096,
      "step": 18523
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37377296589014414,
      "learning_rate": 8.314444085939446e-06,
      "loss": 0.252,
      "step": 18524
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4290135030492876,
      "learning_rate": 8.313466402440196e-06,
      "loss": 0.2888,
      "step": 18525
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5254042083152874,
      "learning_rate": 8.312488735533615e-06,
      "loss": 0.827,
      "step": 18526
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23425612054813574,
      "learning_rate": 8.311511085229331e-06,
      "loss": 0.199,
      "step": 18527
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.28111145700585627,
      "learning_rate": 8.310533451536953e-06,
      "loss": 0.0697,
      "step": 18528
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4138299430588771,
      "learning_rate": 8.309555834466102e-06,
      "loss": 0.3064,
      "step": 18529
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1602319015288378,
      "learning_rate": 8.308578234026398e-06,
      "loss": 0.2324,
      "step": 18530
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6616188305833027,
      "learning_rate": 8.30760065022746e-06,
      "loss": 0.4274,
      "step": 18531
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29502066447044734,
      "learning_rate": 8.306623083078896e-06,
      "loss": 0.2404,
      "step": 18532
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4961286899691244,
      "learning_rate": 8.305645532590338e-06,
      "loss": 0.3382,
      "step": 18533
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2726310252194179,
      "learning_rate": 8.304667998771395e-06,
      "loss": 0.1873,
      "step": 18534
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.285958794822157,
      "learning_rate": 8.303690481631688e-06,
      "loss": 0.8012,
      "step": 18535
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2191368158779627,
      "learning_rate": 8.302712981180832e-06,
      "loss": 0.0964,
      "step": 18536
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9074134935062598,
      "learning_rate": 8.30173549742844e-06,
      "loss": 0.4252,
      "step": 18537
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2866729359565968,
      "learning_rate": 8.300758030384136e-06,
      "loss": 0.1893,
      "step": 18538
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34164685564128056,
      "learning_rate": 8.299780580057536e-06,
      "loss": 0.3143,
      "step": 18539
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5930750812560478,
      "learning_rate": 8.29880314645825e-06,
      "loss": 0.3002,
      "step": 18540
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9989908802935068,
      "learning_rate": 8.2978257295959e-06,
      "loss": 0.2893,
      "step": 18541
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6028809758916792,
      "learning_rate": 8.296848329480104e-06,
      "loss": 0.2966,
      "step": 18542
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3438657356123655,
      "learning_rate": 8.295870946120467e-06,
      "loss": 0.2396,
      "step": 18543
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.456358232210342,
      "learning_rate": 8.294893579526617e-06,
      "loss": 0.3493,
      "step": 18544
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2472547180436911,
      "learning_rate": 8.293916229708165e-06,
      "loss": 0.1681,
      "step": 18545
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4171744952762338,
      "learning_rate": 8.292938896674726e-06,
      "loss": 0.2125,
      "step": 18546
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.285510266780038,
      "learning_rate": 8.291961580435917e-06,
      "loss": 0.1858,
      "step": 18547
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9701132210771906,
      "learning_rate": 8.290984281001353e-06,
      "loss": 0.5092,
      "step": 18548
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9733021206392971,
      "learning_rate": 8.290006998380642e-06,
      "loss": 0.3422,
      "step": 18549
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32288187788172446,
      "learning_rate": 8.28902973258341e-06,
      "loss": 0.3226,
      "step": 18550
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3477268401987492,
      "learning_rate": 8.288052483619265e-06,
      "loss": 0.1767,
      "step": 18551
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.525524428459726,
      "learning_rate": 8.287075251497823e-06,
      "loss": 0.3549,
      "step": 18552
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8944325410156339,
      "learning_rate": 8.286098036228698e-06,
      "loss": 0.477,
      "step": 18553
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1661438489754375,
      "learning_rate": 8.285120837821507e-06,
      "loss": 0.0834,
      "step": 18554
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7980430622846034,
      "learning_rate": 8.284143656285858e-06,
      "loss": 0.3513,
      "step": 18555
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2585733732321875,
      "learning_rate": 8.28316649163137e-06,
      "loss": 0.214,
      "step": 18556
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4815006615654547,
      "learning_rate": 8.282189343867655e-06,
      "loss": 0.3549,
      "step": 18557
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1423067116682404,
      "learning_rate": 8.281212213004325e-06,
      "loss": 0.3633,
      "step": 18558
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.404628803116114,
      "learning_rate": 8.280235099050998e-06,
      "loss": 0.6493,
      "step": 18559
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3022199082004215,
      "learning_rate": 8.279258002017277e-06,
      "loss": 0.1183,
      "step": 18560
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38818564092797153,
      "learning_rate": 8.278280921912788e-06,
      "loss": 0.2899,
      "step": 18561
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3495997801056739,
      "learning_rate": 8.27730385874714e-06,
      "loss": 0.2537,
      "step": 18562
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30789771903341523,
      "learning_rate": 8.276326812529936e-06,
      "loss": 0.2408,
      "step": 18563
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.13350855668070408,
      "learning_rate": 8.275349783270802e-06,
      "loss": 0.063,
      "step": 18564
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40915185353044364,
      "learning_rate": 8.274372770979341e-06,
      "loss": 0.2516,
      "step": 18565
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5178401075004974,
      "learning_rate": 8.273395775665169e-06,
      "loss": 0.2907,
      "step": 18566
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4356470733206978,
      "learning_rate": 8.2724187973379e-06,
      "loss": 0.2483,
      "step": 18567
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46712600385402026,
      "learning_rate": 8.271441836007144e-06,
      "loss": 0.3472,
      "step": 18568
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27631854074614565,
      "learning_rate": 8.270464891682507e-06,
      "loss": 0.0716,
      "step": 18569
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36780158656419737,
      "learning_rate": 8.269487964373608e-06,
      "loss": 0.3185,
      "step": 18570
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1473928314981148,
      "learning_rate": 8.268511054090058e-06,
      "loss": 0.4539,
      "step": 18571
    },
    {
      "epoch": 0.57,
      "grad_norm": 2.0991561290494642,
      "learning_rate": 8.267534160841461e-06,
      "loss": 0.6953,
      "step": 18572
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.19283277126683637,
      "learning_rate": 8.266557284637438e-06,
      "loss": 0.1599,
      "step": 18573
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38157746627998085,
      "learning_rate": 8.265580425487593e-06,
      "loss": 0.2668,
      "step": 18574
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31065045271537167,
      "learning_rate": 8.264603583401536e-06,
      "loss": 0.2543,
      "step": 18575
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8287451175370789,
      "learning_rate": 8.263626758388884e-06,
      "loss": 0.5072,
      "step": 18576
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8801059224544104,
      "learning_rate": 8.262649950459243e-06,
      "loss": 0.1291,
      "step": 18577
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8998197703854747,
      "learning_rate": 8.261673159622219e-06,
      "loss": 0.4208,
      "step": 18578
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3014176682559601,
      "learning_rate": 8.260696385887432e-06,
      "loss": 0.2064,
      "step": 18579
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7581854373951364,
      "learning_rate": 8.259719629264485e-06,
      "loss": 0.8047,
      "step": 18580
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3091146953908739,
      "learning_rate": 8.258742889762984e-06,
      "loss": 0.2417,
      "step": 18581
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3121827468112836,
      "learning_rate": 8.257766167392548e-06,
      "loss": 0.1844,
      "step": 18582
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3970547080915299,
      "learning_rate": 8.256789462162782e-06,
      "loss": 0.285,
      "step": 18583
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7514600195983097,
      "learning_rate": 8.255812774083293e-06,
      "loss": 0.3132,
      "step": 18584
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46757824975473683,
      "learning_rate": 8.254836103163693e-06,
      "loss": 0.2471,
      "step": 18585
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2944861475035443,
      "learning_rate": 8.253859449413586e-06,
      "loss": 0.2371,
      "step": 18586
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5995274423523681,
      "learning_rate": 8.252882812842589e-06,
      "loss": 0.3045,
      "step": 18587
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33341820192420524,
      "learning_rate": 8.251906193460304e-06,
      "loss": 0.2113,
      "step": 18588
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.4101670282246868,
      "learning_rate": 8.250929591276341e-06,
      "loss": 0.7707,
      "step": 18589
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6929440109987874,
      "learning_rate": 8.24995300630031e-06,
      "loss": 0.321,
      "step": 18590
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3299680852220491,
      "learning_rate": 8.248976438541816e-06,
      "loss": 0.3127,
      "step": 18591
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2868090135255409,
      "learning_rate": 8.247999888010463e-06,
      "loss": 0.071,
      "step": 18592
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33395609165264833,
      "learning_rate": 8.24702335471587e-06,
      "loss": 0.2843,
      "step": 18593
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24785208109077494,
      "learning_rate": 8.246046838667637e-06,
      "loss": 0.1021,
      "step": 18594
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7957847647117978,
      "learning_rate": 8.24507033987537e-06,
      "loss": 0.2893,
      "step": 18595
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8621335388656934,
      "learning_rate": 8.24409385834868e-06,
      "loss": 0.445,
      "step": 18596
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31343275517976776,
      "learning_rate": 8.243117394097173e-06,
      "loss": 0.2039,
      "step": 18597
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32732203624215356,
      "learning_rate": 8.242140947130454e-06,
      "loss": 0.2962,
      "step": 18598
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6173283568174917,
      "learning_rate": 8.241164517458133e-06,
      "loss": 0.0224,
      "step": 18599
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6348662141295609,
      "learning_rate": 8.240188105089812e-06,
      "loss": 0.4192,
      "step": 18600
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31617461269174246,
      "learning_rate": 8.239211710035098e-06,
      "loss": 0.1942,
      "step": 18601
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5304469097136508,
      "learning_rate": 8.238235332303601e-06,
      "loss": 0.375,
      "step": 18602
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.1950742634263164,
      "learning_rate": 8.237258971904926e-06,
      "loss": 0.0708,
      "step": 18603
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3581867530310875,
      "learning_rate": 8.236282628848673e-06,
      "loss": 0.303,
      "step": 18604
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7241435812725334,
      "learning_rate": 8.235306303144456e-06,
      "loss": 0.228,
      "step": 18605
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3599218981020231,
      "learning_rate": 8.234329994801874e-06,
      "loss": 0.271,
      "step": 18606
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1473180970728325,
      "learning_rate": 8.233353703830533e-06,
      "loss": 0.4774,
      "step": 18607
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5952081323423511,
      "learning_rate": 8.232377430240043e-06,
      "loss": 0.3137,
      "step": 18608
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3128524947027005,
      "learning_rate": 8.231401174040002e-06,
      "loss": 0.2943,
      "step": 18609
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2818383349519491,
      "learning_rate": 8.230424935240022e-06,
      "loss": 0.1583,
      "step": 18610
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.511781555268212,
      "learning_rate": 8.229448713849702e-06,
      "loss": 0.3317,
      "step": 18611
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21341114777752895,
      "learning_rate": 8.228472509878646e-06,
      "loss": 0.0883,
      "step": 18612
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0652723131681465,
      "learning_rate": 8.227496323336462e-06,
      "loss": 0.6929,
      "step": 18613
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32357843454909346,
      "learning_rate": 8.226520154232754e-06,
      "loss": 0.1682,
      "step": 18614
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5401489499410046,
      "learning_rate": 8.225544002577123e-06,
      "loss": 0.3534,
      "step": 18615
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2599810634510545,
      "learning_rate": 8.224567868379174e-06,
      "loss": 0.2426,
      "step": 18616
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6274836807069551,
      "learning_rate": 8.223591751648512e-06,
      "loss": 0.4397,
      "step": 18617
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8255310223115155,
      "learning_rate": 8.222615652394735e-06,
      "loss": 0.2942,
      "step": 18618
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9168160644623838,
      "learning_rate": 8.221639570627453e-06,
      "loss": 0.4336,
      "step": 18619
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29914700416556333,
      "learning_rate": 8.220663506356267e-06,
      "loss": 0.1958,
      "step": 18620
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.18673234813558767,
      "learning_rate": 8.219687459590776e-06,
      "loss": 0.1333,
      "step": 18621
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4316821982422532,
      "learning_rate": 8.218711430340588e-06,
      "loss": 0.3308,
      "step": 18622
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8339778397247288,
      "learning_rate": 8.217735418615302e-06,
      "loss": 0.5436,
      "step": 18623
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4023743944552329,
      "learning_rate": 8.216759424424519e-06,
      "loss": 0.2936,
      "step": 18624
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5305718598483811,
      "learning_rate": 8.215783447777847e-06,
      "loss": 0.2887,
      "step": 18625
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8507260199862171,
      "learning_rate": 8.214807488684885e-06,
      "loss": 0.4595,
      "step": 18626
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3998435875931833,
      "learning_rate": 8.213831547155231e-06,
      "loss": 0.2502,
      "step": 18627
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3939990256282033,
      "learning_rate": 8.212855623198493e-06,
      "loss": 0.3013,
      "step": 18628
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3374331460580979,
      "learning_rate": 8.21187971682427e-06,
      "loss": 0.2085,
      "step": 18629
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4690588935777519,
      "learning_rate": 8.210903828042158e-06,
      "loss": 0.2333,
      "step": 18630
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.21273534657307266,
      "learning_rate": 8.209927956861766e-06,
      "loss": 0.0696,
      "step": 18631
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8500317064950257,
      "learning_rate": 8.208952103292694e-06,
      "loss": 0.5293,
      "step": 18632
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27964257175599777,
      "learning_rate": 8.207976267344534e-06,
      "loss": 0.1884,
      "step": 18633
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30203317050742207,
      "learning_rate": 8.207000449026898e-06,
      "loss": 0.2368,
      "step": 18634
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6005944502000843,
      "learning_rate": 8.20602464834938e-06,
      "loss": 0.433,
      "step": 18635
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8932146876974615,
      "learning_rate": 8.205048865321583e-06,
      "loss": 0.4397,
      "step": 18636
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6500264091667247,
      "learning_rate": 8.204073099953108e-06,
      "loss": 0.3396,
      "step": 18637
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2925306543149909,
      "learning_rate": 8.203097352253547e-06,
      "loss": 0.1885,
      "step": 18638
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.47278997032493136,
      "learning_rate": 8.20212162223251e-06,
      "loss": 0.2333,
      "step": 18639
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22667452802723706,
      "learning_rate": 8.20114590989959e-06,
      "loss": 0.2016,
      "step": 18640
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7711283533090358,
      "learning_rate": 8.200170215264389e-06,
      "loss": 0.4768,
      "step": 18641
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24556467208942778,
      "learning_rate": 8.199194538336507e-06,
      "loss": 0.073,
      "step": 18642
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.330336409449463,
      "learning_rate": 8.198218879125541e-06,
      "loss": 0.2884,
      "step": 18643
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6152443946632296,
      "learning_rate": 8.197243237641088e-06,
      "loss": 0.409,
      "step": 18644
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3229464956252919,
      "learning_rate": 8.196267613892752e-06,
      "loss": 0.303,
      "step": 18645
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.33116273070734253,
      "learning_rate": 8.19529200789013e-06,
      "loss": 0.063,
      "step": 18646
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3224632420872674,
      "learning_rate": 8.194316419642817e-06,
      "loss": 0.2094,
      "step": 18647
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6245037824469457,
      "learning_rate": 8.193340849160412e-06,
      "loss": 0.7139,
      "step": 18648
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.16637835743667656,
      "learning_rate": 8.192365296452518e-06,
      "loss": 0.0876,
      "step": 18649
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8364755230819721,
      "learning_rate": 8.191389761528723e-06,
      "loss": 0.4532,
      "step": 18650
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2408121700418751,
      "learning_rate": 8.190414244398635e-06,
      "loss": 0.1884,
      "step": 18651
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32537522214724357,
      "learning_rate": 8.189438745071846e-06,
      "loss": 0.288,
      "step": 18652
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7749786322819294,
      "learning_rate": 8.188463263557956e-06,
      "loss": 0.3834,
      "step": 18653
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.548177222973561,
      "learning_rate": 8.18748779986656e-06,
      "loss": 0.8178,
      "step": 18654
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.541718915566945,
      "learning_rate": 8.186512354007254e-06,
      "loss": 0.099,
      "step": 18655
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34691986934427266,
      "learning_rate": 8.185536925989635e-06,
      "loss": 0.2771,
      "step": 18656
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.46273452298058654,
      "learning_rate": 8.184561515823303e-06,
      "loss": 0.2507,
      "step": 18657
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.28914183602140653,
      "learning_rate": 8.183586123517852e-06,
      "loss": 0.2284,
      "step": 18658
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5057529336588938,
      "learning_rate": 8.182610749082878e-06,
      "loss": 0.2267,
      "step": 18659
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2668495786479623,
      "learning_rate": 8.181635392527979e-06,
      "loss": 0.1726,
      "step": 18660
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9174181719949858,
      "learning_rate": 8.180660053862744e-06,
      "loss": 0.522,
      "step": 18661
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0321318334233651,
      "learning_rate": 8.179684733096777e-06,
      "loss": 0.6113,
      "step": 18662
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3148358221454032,
      "learning_rate": 8.178709430239674e-06,
      "loss": 0.2893,
      "step": 18663
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.48610403812327146,
      "learning_rate": 8.17773414530102e-06,
      "loss": 0.1611,
      "step": 18664
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5735241292932662,
      "learning_rate": 8.176758878290423e-06,
      "loss": 0.4004,
      "step": 18665
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4378434072801061,
      "learning_rate": 8.175783629217472e-06,
      "loss": 0.2012,
      "step": 18666
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7593675619505318,
      "learning_rate": 8.174808398091758e-06,
      "loss": 0.4565,
      "step": 18667
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3032179928870372,
      "learning_rate": 8.173833184922884e-06,
      "loss": 0.2059,
      "step": 18668
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9136252440315705,
      "learning_rate": 8.17285798972044e-06,
      "loss": 0.4379,
      "step": 18669
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.23461630588900342,
      "learning_rate": 8.171882812494015e-06,
      "loss": 0.2127,
      "step": 18670
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2745702056023611,
      "learning_rate": 8.170907653253211e-06,
      "loss": 0.7838,
      "step": 18671
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2958469471848023,
      "learning_rate": 8.169932512007622e-06,
      "loss": 0.1233,
      "step": 18672
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6256967318659231,
      "learning_rate": 8.168957388766837e-06,
      "loss": 0.1397,
      "step": 18673
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34585833269576055,
      "learning_rate": 8.167982283540452e-06,
      "loss": 0.2616,
      "step": 18674
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3144004778071238,
      "learning_rate": 8.167007196338062e-06,
      "loss": 0.2455,
      "step": 18675
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7206037543350496,
      "learning_rate": 8.166032127169253e-06,
      "loss": 0.4315,
      "step": 18676
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6543647522524607,
      "learning_rate": 8.165057076043628e-06,
      "loss": 0.2984,
      "step": 18677
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31725433250997237,
      "learning_rate": 8.164082042970774e-06,
      "loss": 0.1859,
      "step": 18678
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3169766784972816,
      "learning_rate": 8.163107027960284e-06,
      "loss": 0.2364,
      "step": 18679
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40806874473416826,
      "learning_rate": 8.162132031021754e-06,
      "loss": 0.2689,
      "step": 18680
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.326787712760798,
      "learning_rate": 8.161157052164773e-06,
      "loss": 0.2341,
      "step": 18681
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.371680185225876,
      "learning_rate": 8.160182091398929e-06,
      "loss": 0.6719,
      "step": 18682
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.2840540000273715,
      "learning_rate": 8.159207148733826e-06,
      "loss": 0.191,
      "step": 18683
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.977680224348938,
      "learning_rate": 8.158232224179045e-06,
      "loss": 0.5051,
      "step": 18684
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9530841885685307,
      "learning_rate": 8.157257317744181e-06,
      "loss": 0.3473,
      "step": 18685
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36515110563599984,
      "learning_rate": 8.156282429438829e-06,
      "loss": 0.2664,
      "step": 18686
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37431060795401905,
      "learning_rate": 8.155307559272572e-06,
      "loss": 0.2783,
      "step": 18687
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36173177136781337,
      "learning_rate": 8.154332707255011e-06,
      "loss": 0.221,
      "step": 18688
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2478100911248269,
      "learning_rate": 8.153357873395731e-06,
      "loss": 0.7993,
      "step": 18689
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20307170310582567,
      "learning_rate": 8.152383057704323e-06,
      "loss": 0.083,
      "step": 18690
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8232426146211529,
      "learning_rate": 8.15140826019038e-06,
      "loss": 0.4465,
      "step": 18691
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.35424340393899656,
      "learning_rate": 8.150433480863492e-06,
      "loss": 0.171,
      "step": 18692
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3782285949303964,
      "learning_rate": 8.14945871973324e-06,
      "loss": 0.2891,
      "step": 18693
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6790139795950326,
      "learning_rate": 8.14848397680923e-06,
      "loss": 0.3111,
      "step": 18694
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0576844208288503,
      "learning_rate": 8.147509252101043e-06,
      "loss": 0.5604,
      "step": 18695
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44789600575885424,
      "learning_rate": 8.146534545618264e-06,
      "loss": 0.1561,
      "step": 18696
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3608960231601062,
      "learning_rate": 8.145559857370495e-06,
      "loss": 0.2504,
      "step": 18697
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8150806862684028,
      "learning_rate": 8.144585187367317e-06,
      "loss": 0.5525,
      "step": 18698
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.168802721363503,
      "learning_rate": 8.143610535618316e-06,
      "loss": 0.1622,
      "step": 18699
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.7398963633849729,
      "learning_rate": 8.142635902133089e-06,
      "loss": 0.7586,
      "step": 18700
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3076548874616857,
      "learning_rate": 8.141661286921222e-06,
      "loss": 0.1731,
      "step": 18701
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.42047814057046917,
      "learning_rate": 8.140686689992298e-06,
      "loss": 0.3195,
      "step": 18702
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6473277564974567,
      "learning_rate": 8.139712111355915e-06,
      "loss": 0.3108,
      "step": 18703
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.6295011101720047,
      "learning_rate": 8.138737551021656e-06,
      "loss": 0.6299,
      "step": 18704
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.26748823061628774,
      "learning_rate": 8.137763008999107e-06,
      "loss": 0.2109,
      "step": 18705
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39044966456496105,
      "learning_rate": 8.13678848529786e-06,
      "loss": 0.3196,
      "step": 18706
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.098665986059328,
      "learning_rate": 8.135813979927502e-06,
      "loss": 0.4501,
      "step": 18707
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.43091199333790714,
      "learning_rate": 8.134839492897614e-06,
      "loss": 0.2567,
      "step": 18708
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3496470205880606,
      "learning_rate": 8.133865024217793e-06,
      "loss": 0.2003,
      "step": 18709
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3464907871037031,
      "learning_rate": 8.132890573897623e-06,
      "loss": 0.2273,
      "step": 18710
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.28075421380793497,
      "learning_rate": 8.131916141946687e-06,
      "loss": 0.2422,
      "step": 18711
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.37566928979882697,
      "learning_rate": 8.130941728374578e-06,
      "loss": 0.2656,
      "step": 18712
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0969288070530911,
      "learning_rate": 8.129967333190874e-06,
      "loss": 0.5414,
      "step": 18713
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3040460770756911,
      "learning_rate": 8.128992956405171e-06,
      "loss": 0.1543,
      "step": 18714
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.36708295218226017,
      "learning_rate": 8.128018598027052e-06,
      "loss": 0.2955,
      "step": 18715
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9444210325722268,
      "learning_rate": 8.127044258066098e-06,
      "loss": 0.4662,
      "step": 18716
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3142536335129223,
      "learning_rate": 8.126069936531902e-06,
      "loss": 0.2742,
      "step": 18717
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.39458751606672016,
      "learning_rate": 8.125095633434046e-06,
      "loss": 0.2221,
      "step": 18718
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6483622510135605,
      "learning_rate": 8.124121348782113e-06,
      "loss": 0.3524,
      "step": 18719
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29198238135032895,
      "learning_rate": 8.123147082585694e-06,
      "loss": 0.2295,
      "step": 18720
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.44275297362342714,
      "learning_rate": 8.122172834854372e-06,
      "loss": 0.2516,
      "step": 18721
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.30040400602506495,
      "learning_rate": 8.121198605597731e-06,
      "loss": 0.2406,
      "step": 18722
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.631179196924451,
      "learning_rate": 8.120224394825358e-06,
      "loss": 0.8258,
      "step": 18723
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24830684376012646,
      "learning_rate": 8.119250202546835e-06,
      "loss": 0.1785,
      "step": 18724
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0036507933090921,
      "learning_rate": 8.118276028771743e-06,
      "loss": 0.4432,
      "step": 18725
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3651068486023361,
      "learning_rate": 8.117301873509675e-06,
      "loss": 0.2423,
      "step": 18726
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5399405237601451,
      "learning_rate": 8.11632773677021e-06,
      "loss": 0.3217,
      "step": 18727
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.47606599481873657,
      "learning_rate": 8.115353618562933e-06,
      "loss": 0.3367,
      "step": 18728
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20775324546192003,
      "learning_rate": 8.114379518897427e-06,
      "loss": 0.2183,
      "step": 18729
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.415730537098444,
      "learning_rate": 8.113405437783278e-06,
      "loss": 0.2312,
      "step": 18730
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1670222027114991,
      "learning_rate": 8.112431375230061e-06,
      "loss": 0.1864,
      "step": 18731
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.2131243033915473,
      "learning_rate": 8.111457331247368e-06,
      "loss": 0.7518,
      "step": 18732
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27006751465707746,
      "learning_rate": 8.110483305844781e-06,
      "loss": 0.183,
      "step": 18733
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4375297494715851,
      "learning_rate": 8.109509299031877e-06,
      "loss": 0.3262,
      "step": 18734
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3888766499760558,
      "learning_rate": 8.108535310818248e-06,
      "loss": 0.2189,
      "step": 18735
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.695586210575377,
      "learning_rate": 8.107561341213467e-06,
      "loss": 0.4098,
      "step": 18736
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29156866578630725,
      "learning_rate": 8.106587390227122e-06,
      "loss": 0.1938,
      "step": 18737
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24561399495501895,
      "learning_rate": 8.105613457868794e-06,
      "loss": 0.1458,
      "step": 18738
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4228642555178762,
      "learning_rate": 8.10463954414806e-06,
      "loss": 0.2602,
      "step": 18739
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3129481571627471,
      "learning_rate": 8.10366564907451e-06,
      "loss": 0.2289,
      "step": 18740
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8421613701308597,
      "learning_rate": 8.102691772657722e-06,
      "loss": 0.5116,
      "step": 18741
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.28246826439990785,
      "learning_rate": 8.101717914907273e-06,
      "loss": 0.1907,
      "step": 18742
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.407290128817867,
      "learning_rate": 8.100744075832751e-06,
      "loss": 0.7115,
      "step": 18743
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6950818193753259,
      "learning_rate": 8.099770255443734e-06,
      "loss": 0.3092,
      "step": 18744
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.858694571633427,
      "learning_rate": 8.098796453749797e-06,
      "loss": 0.4729,
      "step": 18745
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.29985596818212035,
      "learning_rate": 8.097822670760531e-06,
      "loss": 0.179,
      "step": 18746
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.40608019060617334,
      "learning_rate": 8.09684890648551e-06,
      "loss": 0.3022,
      "step": 18747
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.22147082563440837,
      "learning_rate": 8.095875160934317e-06,
      "loss": 0.1405,
      "step": 18748
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.5908491831126883,
      "learning_rate": 8.09490143411653e-06,
      "loss": 0.7286,
      "step": 18749
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.1031790934212158,
      "learning_rate": 8.09392772604173e-06,
      "loss": 0.4894,
      "step": 18750
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.31461208841921184,
      "learning_rate": 8.092954036719493e-06,
      "loss": 0.1856,
      "step": 18751
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.49660201805558657,
      "learning_rate": 8.091980366159404e-06,
      "loss": 0.309,
      "step": 18752
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.4609578419228898,
      "learning_rate": 8.09100671437104e-06,
      "loss": 0.2526,
      "step": 18753
    },
    {
      "epoch": 0.57,
      "grad_norm": 1.0263617568120618,
      "learning_rate": 8.090033081363982e-06,
      "loss": 0.5076,
      "step": 18754
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27802212908649865,
      "learning_rate": 8.089059467147805e-06,
      "loss": 0.0721,
      "step": 18755
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3821581208251685,
      "learning_rate": 8.088085871732091e-06,
      "loss": 0.2962,
      "step": 18756
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.24936438559402702,
      "learning_rate": 8.087112295126413e-06,
      "loss": 0.1016,
      "step": 18757
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5043659427500201,
      "learning_rate": 8.086138737340359e-06,
      "loss": 0.363,
      "step": 18758
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3726662248607489,
      "learning_rate": 8.0851651983835e-06,
      "loss": 0.2716,
      "step": 18759
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.38852645241976896,
      "learning_rate": 8.084191678265415e-06,
      "loss": 0.2606,
      "step": 18760
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.47830786049658286,
      "learning_rate": 8.083218176995685e-06,
      "loss": 0.2715,
      "step": 18761
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8674729381969538,
      "learning_rate": 8.08224469458388e-06,
      "loss": 0.5009,
      "step": 18762
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.7803047223075934,
      "learning_rate": 8.081271231039587e-06,
      "loss": 0.2788,
      "step": 18763
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.25736477625272974,
      "learning_rate": 8.08029778637238e-06,
      "loss": 0.2148,
      "step": 18764
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3842539749278271,
      "learning_rate": 8.079324360591829e-06,
      "loss": 0.311,
      "step": 18765
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.20604171822715275,
      "learning_rate": 8.078350953707523e-06,
      "loss": 0.072,
      "step": 18766
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.5296094933002861,
      "learning_rate": 8.077377565729032e-06,
      "loss": 0.3062,
      "step": 18767
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6365584841802149,
      "learning_rate": 8.07640419666593e-06,
      "loss": 0.2796,
      "step": 18768
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.34244748515005974,
      "learning_rate": 8.075430846527802e-06,
      "loss": 0.2583,
      "step": 18769
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.8629527967295215,
      "learning_rate": 8.074457515324214e-06,
      "loss": 0.3123,
      "step": 18770
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.3069383445273208,
      "learning_rate": 8.073484203064746e-06,
      "loss": 0.2891,
      "step": 18771
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.9405608975967964,
      "learning_rate": 8.072510909758976e-06,
      "loss": 0.2879,
      "step": 18772
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.6037133835124844,
      "learning_rate": 8.071537635416478e-06,
      "loss": 0.2939,
      "step": 18773
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.32800987900305184,
      "learning_rate": 8.070564380046828e-06,
      "loss": 0.2063,
      "step": 18774
    },
    {
      "epoch": 0.57,
      "grad_norm": 0.27513291351420144,
      "learning_rate": 8.069591143659598e-06,
      "loss": 0.1937,
      "step": 18775
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29368118807964494,
      "learning_rate": 8.06861792626437e-06,
      "loss": 0.2387,
      "step": 18776
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5513307077690477,
      "learning_rate": 8.067644727870706e-06,
      "loss": 0.797,
      "step": 18777
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5695032819192865,
      "learning_rate": 8.066671548488197e-06,
      "loss": 0.1963,
      "step": 18778
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4661675935409063,
      "learning_rate": 8.065698388126408e-06,
      "loss": 0.2638,
      "step": 18779
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9390233004714365,
      "learning_rate": 8.064725246794914e-06,
      "loss": 0.5681,
      "step": 18780
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3585946477535,
      "learning_rate": 8.063752124503289e-06,
      "loss": 0.2407,
      "step": 18781
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3013427106259892,
      "learning_rate": 8.062779021261109e-06,
      "loss": 0.2621,
      "step": 18782
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25688588451827554,
      "learning_rate": 8.061805937077943e-06,
      "loss": 0.1786,
      "step": 18783
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45533378345830766,
      "learning_rate": 8.060832871963372e-06,
      "loss": 0.2257,
      "step": 18784
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2131963336790541,
      "learning_rate": 8.059859825926965e-06,
      "loss": 0.1013,
      "step": 18785
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8011356533591505,
      "learning_rate": 8.058886798978293e-06,
      "loss": 0.4476,
      "step": 18786
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36061071435793796,
      "learning_rate": 8.057913791126933e-06,
      "loss": 0.2085,
      "step": 18787
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3551395406514816,
      "learning_rate": 8.056940802382453e-06,
      "loss": 0.2959,
      "step": 18788
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3962292840649992,
      "learning_rate": 8.055967832754432e-06,
      "loss": 0.233,
      "step": 18789
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.920468255740197,
      "learning_rate": 8.05499488225244e-06,
      "loss": 0.7474,
      "step": 18790
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6892868633912397,
      "learning_rate": 8.054021950886046e-06,
      "loss": 0.214,
      "step": 18791
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2736462155787086,
      "learning_rate": 8.053049038664826e-06,
      "loss": 0.1882,
      "step": 18792
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.228884981825042,
      "learning_rate": 8.052076145598351e-06,
      "loss": 0.6076,
      "step": 18793
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.21831209415789,
      "learning_rate": 8.051103271696189e-06,
      "loss": 0.1961,
      "step": 18794
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8164470491816054,
      "learning_rate": 8.050130416967914e-06,
      "loss": 0.4217,
      "step": 18795
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34199161602530503,
      "learning_rate": 8.049157581423102e-06,
      "loss": 0.2067,
      "step": 18796
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.538605870979444,
      "learning_rate": 8.048184765071313e-06,
      "loss": 0.3336,
      "step": 18797
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2552895187279729,
      "learning_rate": 8.047211967922131e-06,
      "loss": 0.1479,
      "step": 18798
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.48342163213195943,
      "learning_rate": 8.046239189985119e-06,
      "loss": 0.35,
      "step": 18799
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3437163574768451,
      "learning_rate": 8.045266431269843e-06,
      "loss": 0.1897,
      "step": 18800
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4410646029016311,
      "learning_rate": 8.044293691785885e-06,
      "loss": 0.2989,
      "step": 18801
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.46703168528790207,
      "learning_rate": 8.043320971542809e-06,
      "loss": 0.2217,
      "step": 18802
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41854498803363177,
      "learning_rate": 8.04234827055018e-06,
      "loss": 0.2671,
      "step": 18803
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5058112329023688,
      "learning_rate": 8.041375588817578e-06,
      "loss": 0.3132,
      "step": 18804
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.282501928158588,
      "learning_rate": 8.040402926354566e-06,
      "loss": 0.1194,
      "step": 18805
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29732340014294956,
      "learning_rate": 8.039430283170716e-06,
      "loss": 0.2587,
      "step": 18806
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25753909005202724,
      "learning_rate": 8.038457659275594e-06,
      "loss": 0.1841,
      "step": 18807
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3385538097495586,
      "learning_rate": 8.037485054678773e-06,
      "loss": 0.5495,
      "step": 18808
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1333115286737816,
      "learning_rate": 8.036512469389814e-06,
      "loss": 0.1702,
      "step": 18809
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4218300565046885,
      "learning_rate": 8.035539903418299e-06,
      "loss": 0.2894,
      "step": 18810
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.403111002871982,
      "learning_rate": 8.034567356773786e-06,
      "loss": 0.2293,
      "step": 18811
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.462366003715252,
      "learning_rate": 8.033594829465845e-06,
      "loss": 0.3412,
      "step": 18812
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.41022680992465227,
      "learning_rate": 8.032622321504047e-06,
      "loss": 0.2687,
      "step": 18813
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6130203795450899,
      "learning_rate": 8.031649832897953e-06,
      "loss": 0.3248,
      "step": 18814
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31423305484780645,
      "learning_rate": 8.03067736365714e-06,
      "loss": 0.2298,
      "step": 18815
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.8842224967489,
      "learning_rate": 8.029704913791173e-06,
      "loss": 0.7659,
      "step": 18816
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.195140404290552,
      "learning_rate": 8.028732483309613e-06,
      "loss": 0.1419,
      "step": 18817
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39644169821799796,
      "learning_rate": 8.027760072222034e-06,
      "loss": 0.2085,
      "step": 18818
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3261724148339992,
      "learning_rate": 8.026787680538e-06,
      "loss": 0.2615,
      "step": 18819
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.46303876539502753,
      "learning_rate": 8.025815308267075e-06,
      "loss": 0.0172,
      "step": 18820
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.623275239540783,
      "learning_rate": 8.024842955418832e-06,
      "loss": 0.3996,
      "step": 18821
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8400235071034989,
      "learning_rate": 8.023870622002835e-06,
      "loss": 0.3718,
      "step": 18822
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3844560790584591,
      "learning_rate": 8.022898308028646e-06,
      "loss": 0.2887,
      "step": 18823
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3627421019477515,
      "learning_rate": 8.021926013505837e-06,
      "loss": 0.2226,
      "step": 18824
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4556367355995448,
      "learning_rate": 8.02095373844397e-06,
      "loss": 0.3758,
      "step": 18825
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.14578292003733287,
      "learning_rate": 8.019981482852606e-06,
      "loss": 0.0708,
      "step": 18826
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1242072367254448,
      "learning_rate": 8.019009246741323e-06,
      "loss": 0.4817,
      "step": 18827
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3052653102963079,
      "learning_rate": 8.018037030119676e-06,
      "loss": 0.0675,
      "step": 18828
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35888718696482635,
      "learning_rate": 8.017064832997232e-06,
      "loss": 0.2927,
      "step": 18829
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31843397585062294,
      "learning_rate": 8.016092655383558e-06,
      "loss": 0.2514,
      "step": 18830
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9658926534512429,
      "learning_rate": 8.015120497288217e-06,
      "loss": 0.3232,
      "step": 18831
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1475470010612425,
      "learning_rate": 8.014148358720772e-06,
      "loss": 0.496,
      "step": 18832
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.304162707303805,
      "learning_rate": 8.013176239690789e-06,
      "loss": 0.2034,
      "step": 18833
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.373808798963114,
      "learning_rate": 8.012204140207836e-06,
      "loss": 0.2105,
      "step": 18834
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2231368284202454,
      "learning_rate": 8.011232060281466e-06,
      "loss": 0.1868,
      "step": 18835
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3686217490928116,
      "learning_rate": 8.010259999921256e-06,
      "loss": 0.5898,
      "step": 18836
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3249174432982401,
      "learning_rate": 8.00928795913676e-06,
      "loss": 0.1777,
      "step": 18837
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5051436947155145,
      "learning_rate": 8.008315937937543e-06,
      "loss": 0.3354,
      "step": 18838
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9449315514625451,
      "learning_rate": 8.007343936333172e-06,
      "loss": 0.3102,
      "step": 18839
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.356425741822291,
      "learning_rate": 8.006371954333201e-06,
      "loss": 0.6082,
      "step": 18840
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28145561628970733,
      "learning_rate": 8.005399991947205e-06,
      "loss": 0.2493,
      "step": 18841
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3384971798336163,
      "learning_rate": 8.004428049184737e-06,
      "loss": 0.2776,
      "step": 18842
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9805046653336162,
      "learning_rate": 8.003456126055364e-06,
      "loss": 0.4633,
      "step": 18843
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23132961764108864,
      "learning_rate": 8.002484222568647e-06,
      "loss": 0.0932,
      "step": 18844
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8943775981628528,
      "learning_rate": 8.001512338734148e-06,
      "loss": 0.4979,
      "step": 18845
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29209916924188917,
      "learning_rate": 8.000540474561425e-06,
      "loss": 0.1927,
      "step": 18846
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9096250640586764,
      "learning_rate": 7.999568630060045e-06,
      "loss": 0.4977,
      "step": 18847
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2859558267288249,
      "learning_rate": 7.998596805239568e-06,
      "loss": 0.2475,
      "step": 18848
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.606165810753197,
      "learning_rate": 7.997625000109551e-06,
      "loss": 0.7499,
      "step": 18849
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32810685324481037,
      "learning_rate": 7.996653214679561e-06,
      "loss": 0.1536,
      "step": 18850
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5510514965536132,
      "learning_rate": 7.995681448959156e-06,
      "loss": 0.3398,
      "step": 18851
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4576620645428888,
      "learning_rate": 7.994709702957892e-06,
      "loss": 0.2544,
      "step": 18852
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2253875188327185,
      "learning_rate": 7.993737976685339e-06,
      "loss": 0.1841,
      "step": 18853
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45902098319532175,
      "learning_rate": 7.992766270151051e-06,
      "loss": 0.2737,
      "step": 18854
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0248244027422921,
      "learning_rate": 7.991794583364587e-06,
      "loss": 0.4746,
      "step": 18855
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30914031419900434,
      "learning_rate": 7.99082291633551e-06,
      "loss": 0.2203,
      "step": 18856
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1683713975243224,
      "learning_rate": 7.989851269073379e-06,
      "loss": 0.2792,
      "step": 18857
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4637465445313832,
      "learning_rate": 7.988879641587749e-06,
      "loss": 0.3446,
      "step": 18858
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.387569447366909,
      "learning_rate": 7.987908033888186e-06,
      "loss": 0.2569,
      "step": 18859
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39795219177768537,
      "learning_rate": 7.986936445984246e-06,
      "loss": 0.2447,
      "step": 18860
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44185911873533085,
      "learning_rate": 7.985964877885486e-06,
      "loss": 0.2296,
      "step": 18861
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39217288420638186,
      "learning_rate": 7.984993329601468e-06,
      "loss": 0.2168,
      "step": 18862
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7951731610132732,
      "learning_rate": 7.984021801141745e-06,
      "loss": 0.3288,
      "step": 18863
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.535247406381392,
      "learning_rate": 7.98305029251588e-06,
      "loss": 0.3258,
      "step": 18864
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20721119333273327,
      "learning_rate": 7.982078803733435e-06,
      "loss": 0.1874,
      "step": 18865
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.47565895898293004,
      "learning_rate": 7.981107334803956e-06,
      "loss": 0.3427,
      "step": 18866
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1043338163489613,
      "learning_rate": 7.98013588573701e-06,
      "loss": 0.3098,
      "step": 18867
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3523258160597187,
      "learning_rate": 7.979164456542154e-06,
      "loss": 0.8324,
      "step": 18868
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2773211796455596,
      "learning_rate": 7.978193047228941e-06,
      "loss": 0.1892,
      "step": 18869
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.485975160853875,
      "learning_rate": 7.97722165780693e-06,
      "loss": 0.0179,
      "step": 18870
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.40011969488987004,
      "learning_rate": 7.97625028828568e-06,
      "loss": 0.3172,
      "step": 18871
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.46837515523770357,
      "learning_rate": 7.97527893867474e-06,
      "loss": 0.3052,
      "step": 18872
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2962091427251115,
      "learning_rate": 7.974307608983679e-06,
      "loss": 0.154,
      "step": 18873
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3493719332156102,
      "learning_rate": 7.973336299222043e-06,
      "loss": 0.2074,
      "step": 18874
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44930053828504,
      "learning_rate": 7.97236500939939e-06,
      "loss": 0.2888,
      "step": 18875
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3894486855915424,
      "learning_rate": 7.97139373952528e-06,
      "loss": 0.2647,
      "step": 18876
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4190225703779012,
      "learning_rate": 7.970422489609266e-06,
      "loss": 0.3166,
      "step": 18877
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3196872810887496,
      "learning_rate": 7.9694512596609e-06,
      "loss": 0.173,
      "step": 18878
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5287948251255893,
      "learning_rate": 7.968480049689741e-06,
      "loss": 0.3767,
      "step": 18879
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6047318101700416,
      "learning_rate": 7.967508859705348e-06,
      "loss": 0.3137,
      "step": 18880
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6337557952220205,
      "learning_rate": 7.966537689717267e-06,
      "loss": 0.4616,
      "step": 18881
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18941556027990575,
      "learning_rate": 7.965566539735061e-06,
      "loss": 0.0703,
      "step": 18882
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25544604527602555,
      "learning_rate": 7.96459540976828e-06,
      "loss": 0.2154,
      "step": 18883
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2953542254975715,
      "learning_rate": 7.963624299826475e-06,
      "loss": 0.232,
      "step": 18884
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8982591438080914,
      "learning_rate": 7.962653209919207e-06,
      "loss": 0.3523,
      "step": 18885
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6767231511958283,
      "learning_rate": 7.96168214005603e-06,
      "loss": 0.7841,
      "step": 18886
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2669243461458369,
      "learning_rate": 7.960711090246492e-06,
      "loss": 0.1647,
      "step": 18887
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44357353388085907,
      "learning_rate": 7.95974006050015e-06,
      "loss": 0.3383,
      "step": 18888
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3229754718872148,
      "learning_rate": 7.958769050826552e-06,
      "loss": 0.252,
      "step": 18889
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6717039069403713,
      "learning_rate": 7.95779806123526e-06,
      "loss": 0.4584,
      "step": 18890
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6199325588704768,
      "learning_rate": 7.956827091735825e-06,
      "loss": 0.1418,
      "step": 18891
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3461694030913435,
      "learning_rate": 7.955856142337794e-06,
      "loss": 0.2855,
      "step": 18892
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1541152452594124,
      "learning_rate": 7.954885213050724e-06,
      "loss": 0.0688,
      "step": 18893
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1803792420100505,
      "learning_rate": 7.953914303884168e-06,
      "loss": 0.6173,
      "step": 18894
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.27826048038983275,
      "learning_rate": 7.95294341484767e-06,
      "loss": 0.2394,
      "step": 18895
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26405141571084995,
      "learning_rate": 7.951972545950795e-06,
      "loss": 0.1822,
      "step": 18896
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8224630166569237,
      "learning_rate": 7.951001697203085e-06,
      "loss": 0.5509,
      "step": 18897
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.48148992103576765,
      "learning_rate": 7.950030868614095e-06,
      "loss": 0.319,
      "step": 18898
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.232749415969314,
      "learning_rate": 7.949060060193377e-06,
      "loss": 0.5926,
      "step": 18899
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2993501566621198,
      "learning_rate": 7.948089271950481e-06,
      "loss": 0.1916,
      "step": 18900
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5383829225028497,
      "learning_rate": 7.947118503894953e-06,
      "loss": 0.3401,
      "step": 18901
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18889034399154137,
      "learning_rate": 7.946147756036354e-06,
      "loss": 0.1576,
      "step": 18902
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2730308602744722,
      "learning_rate": 7.945177028384228e-06,
      "loss": 0.6564,
      "step": 18903
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7653762243691874,
      "learning_rate": 7.944206320948123e-06,
      "loss": 0.3663,
      "step": 18904
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9634245437412382,
      "learning_rate": 7.943235633737598e-06,
      "loss": 0.4972,
      "step": 18905
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.295103734556703,
      "learning_rate": 7.942264966762196e-06,
      "loss": 0.2213,
      "step": 18906
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31668422062292373,
      "learning_rate": 7.941294320031467e-06,
      "loss": 0.2861,
      "step": 18907
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7964817926975326,
      "learning_rate": 7.940323693554964e-06,
      "loss": 0.2928,
      "step": 18908
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35174538593747495,
      "learning_rate": 7.939353087342233e-06,
      "loss": 0.0693,
      "step": 18909
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3567935385124289,
      "learning_rate": 7.938382501402821e-06,
      "loss": 0.2757,
      "step": 18910
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18908850712115335,
      "learning_rate": 7.937411935746284e-06,
      "loss": 0.0811,
      "step": 18911
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4630284799906921,
      "learning_rate": 7.936441390382167e-06,
      "loss": 0.3714,
      "step": 18912
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4187882279294612,
      "learning_rate": 7.935470865320018e-06,
      "loss": 0.2617,
      "step": 18913
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5535770843544328,
      "learning_rate": 7.934500360569388e-06,
      "loss": 0.3708,
      "step": 18914
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35076397708560103,
      "learning_rate": 7.933529876139819e-06,
      "loss": 0.2442,
      "step": 18915
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8688237540225736,
      "learning_rate": 7.932559412040865e-06,
      "loss": 0.5947,
      "step": 18916
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.684125601657833,
      "learning_rate": 7.931588968282074e-06,
      "loss": 0.0248,
      "step": 18917
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34121037324530384,
      "learning_rate": 7.930618544872987e-06,
      "loss": 0.293,
      "step": 18918
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2564702221793337,
      "learning_rate": 7.92964814182316e-06,
      "loss": 0.1758,
      "step": 18919
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2774317676679254,
      "learning_rate": 7.928677759142135e-06,
      "loss": 0.174,
      "step": 18920
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0727995657066387,
      "learning_rate": 7.927707396839455e-06,
      "loss": 0.2823,
      "step": 18921
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7535841551748319,
      "learning_rate": 7.926737054924675e-06,
      "loss": 0.3344,
      "step": 18922
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5189485783168071,
      "learning_rate": 7.925766733407341e-06,
      "loss": 0.33,
      "step": 18923
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3142204590642501,
      "learning_rate": 7.924796432296993e-06,
      "loss": 0.2235,
      "step": 18924
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33356787016893436,
      "learning_rate": 7.923826151603181e-06,
      "loss": 0.3191,
      "step": 18925
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2828126953508263,
      "learning_rate": 7.922855891335452e-06,
      "loss": 0.3078,
      "step": 18926
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0166049007398634,
      "learning_rate": 7.921885651503345e-06,
      "loss": 0.4929,
      "step": 18927
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30981900032088405,
      "learning_rate": 7.920915432116416e-06,
      "loss": 0.1655,
      "step": 18928
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29053165224694627,
      "learning_rate": 7.919945233184205e-06,
      "loss": 0.2068,
      "step": 18929
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.24375696222621715,
      "learning_rate": 7.918975054716255e-06,
      "loss": 0.1773,
      "step": 18930
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.45260164681313503,
      "learning_rate": 7.918004896722114e-06,
      "loss": 0.3145,
      "step": 18931
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6833137406582833,
      "learning_rate": 7.917034759211328e-06,
      "loss": 0.2297,
      "step": 18932
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33524275414204496,
      "learning_rate": 7.916064642193433e-06,
      "loss": 0.2814,
      "step": 18933
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7188085469842387,
      "learning_rate": 7.915094545677986e-06,
      "loss": 0.3617,
      "step": 18934
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.109307462302392,
      "learning_rate": 7.914124469674524e-06,
      "loss": 0.2769,
      "step": 18935
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4597666186287545,
      "learning_rate": 7.913154414192587e-06,
      "loss": 0.3268,
      "step": 18936
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39519831534863786,
      "learning_rate": 7.91218437924173e-06,
      "loss": 0.1959,
      "step": 18937
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.438753738053119,
      "learning_rate": 7.911214364831488e-06,
      "loss": 0.3672,
      "step": 18938
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.46169907026908724,
      "learning_rate": 7.910244370971404e-06,
      "loss": 0.2206,
      "step": 18939
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.98542602056162,
      "learning_rate": 7.909274397671026e-06,
      "loss": 0.5144,
      "step": 18940
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6539665691551965,
      "learning_rate": 7.908304444939891e-06,
      "loss": 0.2061,
      "step": 18941
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.38821685194169236,
      "learning_rate": 7.907334512787549e-06,
      "loss": 0.3145,
      "step": 18942
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2270403473750566,
      "learning_rate": 7.906364601223538e-06,
      "loss": 0.1904,
      "step": 18943
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4815347028771226,
      "learning_rate": 7.905394710257398e-06,
      "loss": 0.7728,
      "step": 18944
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0937921685843406,
      "learning_rate": 7.90442483989868e-06,
      "loss": 0.1839,
      "step": 18945
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3718481006380109,
      "learning_rate": 7.903454990156916e-06,
      "loss": 0.2479,
      "step": 18946
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28577765974127173,
      "learning_rate": 7.90248516104165e-06,
      "loss": 0.1814,
      "step": 18947
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43257854502944276,
      "learning_rate": 7.901515352562428e-06,
      "loss": 0.2504,
      "step": 18948
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44625777723045124,
      "learning_rate": 7.900545564728789e-06,
      "loss": 0.3172,
      "step": 18949
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3881154037552574,
      "learning_rate": 7.899575797550271e-06,
      "loss": 0.16,
      "step": 18950
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.370678092994536,
      "learning_rate": 7.898606051036419e-06,
      "loss": 0.2717,
      "step": 18951
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20291852586859963,
      "learning_rate": 7.897636325196774e-06,
      "loss": 0.1163,
      "step": 18952
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3412490237880774,
      "learning_rate": 7.896666620040867e-06,
      "loss": 0.6318,
      "step": 18953
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2687220874305175,
      "learning_rate": 7.895696935578255e-06,
      "loss": 0.2093,
      "step": 18954
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.453514704755186,
      "learning_rate": 7.894727271818465e-06,
      "loss": 0.3154,
      "step": 18955
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.50181354976085,
      "learning_rate": 7.893757628771039e-06,
      "loss": 0.2211,
      "step": 18956
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7085013770452147,
      "learning_rate": 7.892788006445521e-06,
      "loss": 0.4429,
      "step": 18957
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6766040148702074,
      "learning_rate": 7.891818404851451e-06,
      "loss": 0.3454,
      "step": 18958
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9183993903061001,
      "learning_rate": 7.890848823998358e-06,
      "loss": 0.4741,
      "step": 18959
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2827495282744518,
      "learning_rate": 7.889879263895793e-06,
      "loss": 0.184,
      "step": 18960
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36217411768635027,
      "learning_rate": 7.88890972455329e-06,
      "loss": 0.2851,
      "step": 18961
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.1486446947523879,
      "learning_rate": 7.887940205980385e-06,
      "loss": 0.0717,
      "step": 18962
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6268751479909308,
      "learning_rate": 7.886970708186623e-06,
      "loss": 0.0467,
      "step": 18963
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6324151870212292,
      "learning_rate": 7.886001231181537e-06,
      "loss": 0.2915,
      "step": 18964
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3236795478049681,
      "learning_rate": 7.885031774974663e-06,
      "loss": 0.2526,
      "step": 18965
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30896457874301286,
      "learning_rate": 7.884062339575547e-06,
      "loss": 0.283,
      "step": 18966
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7260248542842677,
      "learning_rate": 7.883092924993716e-06,
      "loss": 0.3033,
      "step": 18967
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4634603381643838,
      "learning_rate": 7.882123531238719e-06,
      "loss": 0.8827,
      "step": 18968
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26707958626422224,
      "learning_rate": 7.881154158320088e-06,
      "loss": 0.1822,
      "step": 18969
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31063606879597006,
      "learning_rate": 7.880184806247356e-06,
      "loss": 0.1792,
      "step": 18970
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0773548290700208,
      "learning_rate": 7.879215475030066e-06,
      "loss": 0.2227,
      "step": 18971
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3227423133375077,
      "learning_rate": 7.878246164677753e-06,
      "loss": 0.2846,
      "step": 18972
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.336435221673856,
      "learning_rate": 7.877276875199948e-06,
      "loss": 0.1838,
      "step": 18973
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4538248500813264,
      "learning_rate": 7.876307606606195e-06,
      "loss": 0.3455,
      "step": 18974
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8772323048290199,
      "learning_rate": 7.875338358906026e-06,
      "loss": 0.3087,
      "step": 18975
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0336014310591486,
      "learning_rate": 7.874369132108976e-06,
      "loss": 0.4238,
      "step": 18976
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8594755396671062,
      "learning_rate": 7.873399926224584e-06,
      "loss": 0.4935,
      "step": 18977
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2767457927776296,
      "learning_rate": 7.872430741262383e-06,
      "loss": 0.185,
      "step": 18978
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34256982663194246,
      "learning_rate": 7.871461577231906e-06,
      "loss": 0.3178,
      "step": 18979
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23629831767391649,
      "learning_rate": 7.870492434142692e-06,
      "loss": 0.0968,
      "step": 18980
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8203920329817078,
      "learning_rate": 7.869523312004275e-06,
      "loss": 0.4996,
      "step": 18981
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.37026588351765016,
      "learning_rate": 7.868554210826188e-06,
      "loss": 0.1787,
      "step": 18982
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3725152014017955,
      "learning_rate": 7.867585130617966e-06,
      "loss": 0.3141,
      "step": 18983
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2823631370838138,
      "learning_rate": 7.866616071389144e-06,
      "loss": 0.2497,
      "step": 18984
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.0504057752792852,
      "learning_rate": 7.865647033149251e-06,
      "loss": 0.5847,
      "step": 18985
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4241509072538272,
      "learning_rate": 7.864678015907829e-06,
      "loss": 0.2096,
      "step": 18986
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5291767253151113,
      "learning_rate": 7.863709019674406e-06,
      "loss": 0.3175,
      "step": 18987
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.48341332943674326,
      "learning_rate": 7.862740044458516e-06,
      "loss": 0.2441,
      "step": 18988
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20455354723276792,
      "learning_rate": 7.861771090269694e-06,
      "loss": 0.1424,
      "step": 18989
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.46784343797935174,
      "learning_rate": 7.86080215711747e-06,
      "loss": 0.3274,
      "step": 18990
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.675922307975299,
      "learning_rate": 7.859833245011379e-06,
      "loss": 0.2287,
      "step": 18991
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3395329486916007,
      "learning_rate": 7.858864353960954e-06,
      "loss": 0.2838,
      "step": 18992
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7082777361977814,
      "learning_rate": 7.857895483975723e-06,
      "loss": 0.2923,
      "step": 18993
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6453190859530256,
      "learning_rate": 7.856926635065224e-06,
      "loss": 0.8865,
      "step": 18994
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3557387450652398,
      "learning_rate": 7.855957807238987e-06,
      "loss": 0.2515,
      "step": 18995
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4205981132202398,
      "learning_rate": 7.854989000506536e-06,
      "loss": 0.2723,
      "step": 18996
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3081043613902257,
      "learning_rate": 7.854020214877416e-06,
      "loss": 0.2211,
      "step": 18997
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4472832143827936,
      "learning_rate": 7.85305145036115e-06,
      "loss": 0.2482,
      "step": 18998
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5690807385836675,
      "learning_rate": 7.852082706967267e-06,
      "loss": 0.2317,
      "step": 18999
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8082421305621955,
      "learning_rate": 7.851113984705306e-06,
      "loss": 0.3991,
      "step": 19000
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2972780547451887,
      "learning_rate": 7.85014528358479e-06,
      "loss": 0.1907,
      "step": 19001
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29784909819336736,
      "learning_rate": 7.84917660361525e-06,
      "loss": 0.2266,
      "step": 19002
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1461512198536927,
      "learning_rate": 7.84820794480622e-06,
      "loss": 0.4788,
      "step": 19003
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.448897142243355,
      "learning_rate": 7.847239307167228e-06,
      "loss": 0.2633,
      "step": 19004
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32484666397441886,
      "learning_rate": 7.846270690707804e-06,
      "loss": 0.2553,
      "step": 19005
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.326859221239113,
      "learning_rate": 7.845302095437479e-06,
      "loss": 0.2393,
      "step": 19006
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.47047623792878335,
      "learning_rate": 7.844333521365781e-06,
      "loss": 0.2557,
      "step": 19007
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33630247146369574,
      "learning_rate": 7.843364968502237e-06,
      "loss": 0.25,
      "step": 19008
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.675654893445613,
      "learning_rate": 7.84239643685638e-06,
      "loss": 0.3132,
      "step": 19009
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.31288884305534725,
      "learning_rate": 7.841427926437737e-06,
      "loss": 0.2038,
      "step": 19010
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4681363666596971,
      "learning_rate": 7.840459437255833e-06,
      "loss": 0.2571,
      "step": 19011
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9970896373179758,
      "learning_rate": 7.839490969320203e-06,
      "loss": 0.5528,
      "step": 19012
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3388934552843922,
      "learning_rate": 7.838522522640372e-06,
      "loss": 0.2897,
      "step": 19013
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23861271225362043,
      "learning_rate": 7.837554097225866e-06,
      "loss": 0.0714,
      "step": 19014
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.39770715061420153,
      "learning_rate": 7.836585693086216e-06,
      "loss": 0.2212,
      "step": 19015
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8070090717134106,
      "learning_rate": 7.835617310230943e-06,
      "loss": 0.4588,
      "step": 19016
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6351925887354406,
      "learning_rate": 7.834648948669583e-06,
      "loss": 0.3143,
      "step": 19017
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.28048273818084274,
      "learning_rate": 7.83368060841166e-06,
      "loss": 0.1768,
      "step": 19018
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.29271814802135165,
      "learning_rate": 7.832712289466698e-06,
      "loss": 0.198,
      "step": 19019
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35171846876556273,
      "learning_rate": 7.831743991844228e-06,
      "loss": 0.311,
      "step": 19020
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9293530539484337,
      "learning_rate": 7.830775715553774e-06,
      "loss": 0.2446,
      "step": 19021
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2781571388620556,
      "learning_rate": 7.829807460604856e-06,
      "loss": 0.7713,
      "step": 19022
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33986982954431033,
      "learning_rate": 7.828839227007014e-06,
      "loss": 0.168,
      "step": 19023
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6549353245141509,
      "learning_rate": 7.827871014769763e-06,
      "loss": 0.3937,
      "step": 19024
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1247327722645204,
      "learning_rate": 7.82690282390263e-06,
      "loss": 0.3298,
      "step": 19025
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3085152049623268,
      "learning_rate": 7.825934654415147e-06,
      "loss": 0.3019,
      "step": 19026
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2872799029665258,
      "learning_rate": 7.824966506316832e-06,
      "loss": 0.1021,
      "step": 19027
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2220549318257129,
      "learning_rate": 7.823998379617208e-06,
      "loss": 0.1582,
      "step": 19028
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.6839551205877736,
      "learning_rate": 7.82303027432581e-06,
      "loss": 0.86,
      "step": 19029
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9848670922071225,
      "learning_rate": 7.822062190452153e-06,
      "loss": 0.5429,
      "step": 19030
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.36393349634514044,
      "learning_rate": 7.821094128005764e-06,
      "loss": 0.3155,
      "step": 19031
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3849187335193303,
      "learning_rate": 7.820126086996172e-06,
      "loss": 0.1796,
      "step": 19032
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5204899765807659,
      "learning_rate": 7.819158067432894e-06,
      "loss": 0.3246,
      "step": 19033
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.459311772067768,
      "learning_rate": 7.818190069325454e-06,
      "loss": 0.3445,
      "step": 19034
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.030405746671231,
      "learning_rate": 7.817222092683382e-06,
      "loss": 0.5417,
      "step": 19035
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.23375865091018427,
      "learning_rate": 7.816254137516198e-06,
      "loss": 0.0849,
      "step": 19036
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5393216183238767,
      "learning_rate": 7.815286203833418e-06,
      "loss": 0.3431,
      "step": 19037
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2436198222218669,
      "learning_rate": 7.814318291644579e-06,
      "loss": 0.1977,
      "step": 19038
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.7790124536120395,
      "learning_rate": 7.813350400959192e-06,
      "loss": 0.6323,
      "step": 19039
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.4796229404801309,
      "learning_rate": 7.812382531786784e-06,
      "loss": 0.4079,
      "step": 19040
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6478642654859609,
      "learning_rate": 7.811414684136876e-06,
      "loss": 0.2128,
      "step": 19041
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4205540258992338,
      "learning_rate": 7.810446858018988e-06,
      "loss": 0.2979,
      "step": 19042
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44662287835131076,
      "learning_rate": 7.809479053442647e-06,
      "loss": 0.2739,
      "step": 19043
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2363023024232624,
      "learning_rate": 7.808511270417374e-06,
      "loss": 0.5158,
      "step": 19044
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.18694865823922835,
      "learning_rate": 7.807543508952686e-06,
      "loss": 0.068,
      "step": 19045
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3578725298273948,
      "learning_rate": 7.806575769058107e-06,
      "loss": 0.2538,
      "step": 19046
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25637587493261904,
      "learning_rate": 7.805608050743158e-06,
      "loss": 0.16,
      "step": 19047
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3392959128785729,
      "learning_rate": 7.804640354017355e-06,
      "loss": 0.8292,
      "step": 19048
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30469922999764515,
      "learning_rate": 7.803672678890225e-06,
      "loss": 0.2472,
      "step": 19049
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8794841882357528,
      "learning_rate": 7.802705025371287e-06,
      "loss": 0.3264,
      "step": 19050
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34475638563798605,
      "learning_rate": 7.80173739347006e-06,
      "loss": 0.2535,
      "step": 19051
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8200057404557775,
      "learning_rate": 7.800769783196062e-06,
      "loss": 0.4394,
      "step": 19052
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.3689989373058988,
      "learning_rate": 7.799802194558818e-06,
      "loss": 0.3232,
      "step": 19053
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3176039873042749,
      "learning_rate": 7.798834627567839e-06,
      "loss": 0.1049,
      "step": 19054
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.32928054300407955,
      "learning_rate": 7.797867082232652e-06,
      "loss": 0.2738,
      "step": 19055
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2966276601738961,
      "learning_rate": 7.796899558562774e-06,
      "loss": 0.2396,
      "step": 19056
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.44324764607367834,
      "learning_rate": 7.795932056567722e-06,
      "loss": 0.3127,
      "step": 19057
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6702038848226739,
      "learning_rate": 7.794964576257016e-06,
      "loss": 0.3721,
      "step": 19058
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5526521269550939,
      "learning_rate": 7.793997117640175e-06,
      "loss": 0.3123,
      "step": 19059
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34579531064685964,
      "learning_rate": 7.793029680726711e-06,
      "loss": 0.2137,
      "step": 19060
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7740997434330169,
      "learning_rate": 7.792062265526152e-06,
      "loss": 0.4564,
      "step": 19061
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3028887008055993,
      "learning_rate": 7.791094872048013e-06,
      "loss": 0.2132,
      "step": 19062
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.43437021391991004,
      "learning_rate": 7.790127500301805e-06,
      "loss": 0.2016,
      "step": 19063
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2695286750666199,
      "learning_rate": 7.789160150297052e-06,
      "loss": 0.0653,
      "step": 19064
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2716446350160011,
      "learning_rate": 7.78819282204327e-06,
      "loss": 0.2307,
      "step": 19065
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.680913084006121,
      "learning_rate": 7.78722551554997e-06,
      "loss": 0.4059,
      "step": 19066
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2660955213127231,
      "learning_rate": 7.786258230826677e-06,
      "loss": 0.2145,
      "step": 19067
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.657252572234752,
      "learning_rate": 7.7852909678829e-06,
      "loss": 0.3737,
      "step": 19068
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.3172480731872456,
      "learning_rate": 7.784323726728164e-06,
      "loss": 0.2111,
      "step": 19069
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5549979077898985,
      "learning_rate": 7.783356507371982e-06,
      "loss": 0.8655,
      "step": 19070
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.346469389523999,
      "learning_rate": 7.782389309823863e-06,
      "loss": 0.1731,
      "step": 19071
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.228178054292337,
      "learning_rate": 7.78142213409333e-06,
      "loss": 0.4675,
      "step": 19072
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2431580791346491,
      "learning_rate": 7.780454980189896e-06,
      "loss": 0.1799,
      "step": 19073
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.25581978741734646,
      "learning_rate": 7.779487848123074e-06,
      "loss": 0.2141,
      "step": 19074
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7708657451824589,
      "learning_rate": 7.778520737902384e-06,
      "loss": 0.3672,
      "step": 19075
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6770328576374828,
      "learning_rate": 7.777553649537339e-06,
      "loss": 0.4234,
      "step": 19076
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.5682858452490709,
      "learning_rate": 7.77658658303745e-06,
      "loss": 0.2823,
      "step": 19077
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.33555539328121137,
      "learning_rate": 7.775619538412233e-06,
      "loss": 0.2583,
      "step": 19078
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.21588643130873048,
      "learning_rate": 7.774652515671206e-06,
      "loss": 0.0623,
      "step": 19079
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.30888047598282464,
      "learning_rate": 7.773685514823875e-06,
      "loss": 0.2204,
      "step": 19080
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.5208544863943478,
      "learning_rate": 7.772718535879763e-06,
      "loss": 0.6419,
      "step": 19081
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2697794307357884,
      "learning_rate": 7.771751578848377e-06,
      "loss": 0.1824,
      "step": 19082
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4233965096328721,
      "learning_rate": 7.770784643739232e-06,
      "loss": 0.2212,
      "step": 19083
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.8521082045671956,
      "learning_rate": 7.769817730561844e-06,
      "loss": 0.3633,
      "step": 19084
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.34919202369922625,
      "learning_rate": 7.768850839325721e-06,
      "loss": 0.2987,
      "step": 19085
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4140526808690369,
      "learning_rate": 7.767883970040374e-06,
      "loss": 0.195,
      "step": 19086
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.583383302935578,
      "learning_rate": 7.766917122715325e-06,
      "loss": 0.3012,
      "step": 19087
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.26361179390520284,
      "learning_rate": 7.765950297360077e-06,
      "loss": 0.167,
      "step": 19088
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.2259848920502847,
      "learning_rate": 7.764983493984146e-06,
      "loss": 0.6361,
      "step": 19089
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2930249867830626,
      "learning_rate": 7.764016712597042e-06,
      "loss": 0.2376,
      "step": 19090
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.9382557384663419,
      "learning_rate": 7.763049953208279e-06,
      "loss": 0.3749,
      "step": 19091
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2744507157649015,
      "learning_rate": 7.762083215827361e-06,
      "loss": 0.1908,
      "step": 19092
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7241272126672853,
      "learning_rate": 7.761116500463811e-06,
      "loss": 0.3462,
      "step": 19093
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.884040187479329,
      "learning_rate": 7.76014980712713e-06,
      "loss": 0.5532,
      "step": 19094
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.20065971347491557,
      "learning_rate": 7.759183135826836e-06,
      "loss": 0.0999,
      "step": 19095
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.4051530886258652,
      "learning_rate": 7.758216486572434e-06,
      "loss": 0.2978,
      "step": 19096
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.2058810391672634,
      "learning_rate": 7.757249859373431e-06,
      "loss": 0.1964,
      "step": 19097
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.1948069994940427,
      "learning_rate": 7.756283254239347e-06,
      "loss": 0.6245,
      "step": 19098
    },
    {
      "epoch": 0.58,
      "grad_norm": 1.404391924798978,
      "learning_rate": 7.755316671179687e-06,
      "loss": 0.2357,
      "step": 19099
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.6250279663291224,
      "learning_rate": 7.754350110203958e-06,
      "loss": 0.3111,
      "step": 19100
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.35639375071578444,
      "learning_rate": 7.753383571321673e-06,
      "loss": 0.259,
      "step": 19101
    },
    {
      "epoch": 0.58,
      "grad_norm": 0.7460046818571662,
      "learning_rate": 7.75241705454234e-06,
      "loss": 0.4437,
      "step": 19102
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29091705712838867,
      "learning_rate": 7.751450559875465e-06,
      "loss": 0.2457,
      "step": 19103
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2991862929568676,
      "learning_rate": 7.750484087330562e-06,
      "loss": 0.8466,
      "step": 19104
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2689725324488764,
      "learning_rate": 7.749517636917137e-06,
      "loss": 0.1744,
      "step": 19105
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2868523282008654,
      "learning_rate": 7.748551208644693e-06,
      "loss": 0.0966,
      "step": 19106
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.46419645768054046,
      "learning_rate": 7.747584802522748e-06,
      "loss": 0.2931,
      "step": 19107
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2974484683789173,
      "learning_rate": 7.746618418560805e-06,
      "loss": 0.2399,
      "step": 19108
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6422936798894642,
      "learning_rate": 7.745652056768366e-06,
      "loss": 0.374,
      "step": 19109
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3340016433280612,
      "learning_rate": 7.74468571715495e-06,
      "loss": 0.2162,
      "step": 19110
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6413991169260856,
      "learning_rate": 7.743719399730058e-06,
      "loss": 0.4004,
      "step": 19111
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6398490666035295,
      "learning_rate": 7.742753104503191e-06,
      "loss": 0.3978,
      "step": 19112
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6939976376923404,
      "learning_rate": 7.741786831483866e-06,
      "loss": 0.3915,
      "step": 19113
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3658735792503138,
      "learning_rate": 7.740820580681586e-06,
      "loss": 0.1872,
      "step": 19114
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24170133057916293,
      "learning_rate": 7.739854352105854e-06,
      "loss": 0.221,
      "step": 19115
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2617641176174921,
      "learning_rate": 7.738888145766181e-06,
      "loss": 0.1688,
      "step": 19116
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.876342742835906,
      "learning_rate": 7.737921961672065e-06,
      "loss": 0.51,
      "step": 19117
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5805976024519541,
      "learning_rate": 7.736955799833023e-06,
      "loss": 0.2333,
      "step": 19118
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2987370071138804,
      "learning_rate": 7.735989660258554e-06,
      "loss": 0.2107,
      "step": 19119
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7893946128331673,
      "learning_rate": 7.73502354295816e-06,
      "loss": 0.495,
      "step": 19120
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28629765328254303,
      "learning_rate": 7.734057447941355e-06,
      "loss": 0.2416,
      "step": 19121
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5581649685759889,
      "learning_rate": 7.733091375217636e-06,
      "loss": 0.7623,
      "step": 19122
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35877777925809184,
      "learning_rate": 7.732125324796507e-06,
      "loss": 0.1839,
      "step": 19123
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29460758134357506,
      "learning_rate": 7.731159296687479e-06,
      "loss": 0.2229,
      "step": 19124
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31577276549935746,
      "learning_rate": 7.73019329090005e-06,
      "loss": 0.1564,
      "step": 19125
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4422868737353379,
      "learning_rate": 7.729227307443728e-06,
      "loss": 0.3118,
      "step": 19126
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33386223294605794,
      "learning_rate": 7.728261346328015e-06,
      "loss": 0.2205,
      "step": 19127
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38577214256274844,
      "learning_rate": 7.727295407562414e-06,
      "loss": 0.3265,
      "step": 19128
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0610611432902106,
      "learning_rate": 7.726329491156425e-06,
      "loss": 0.2932,
      "step": 19129
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3824599636448662,
      "learning_rate": 7.72536359711956e-06,
      "loss": 0.8515,
      "step": 19130
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1146698818118825,
      "learning_rate": 7.724397725461314e-06,
      "loss": 0.3824,
      "step": 19131
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.26772835871999545,
      "learning_rate": 7.723431876191192e-06,
      "loss": 0.1827,
      "step": 19132
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.323018238007122,
      "learning_rate": 7.722466049318698e-06,
      "loss": 0.2953,
      "step": 19133
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2309022235713163,
      "learning_rate": 7.72150024485333e-06,
      "loss": 0.0887,
      "step": 19134
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8104713703997785,
      "learning_rate": 7.720534462804592e-06,
      "loss": 0.5131,
      "step": 19135
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5667644592981701,
      "learning_rate": 7.719568703181988e-06,
      "loss": 0.2209,
      "step": 19136
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39774091870765665,
      "learning_rate": 7.718602965995018e-06,
      "loss": 0.3133,
      "step": 19137
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1307725287931631,
      "learning_rate": 7.717637251253179e-06,
      "loss": 0.2566,
      "step": 19138
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3359847004995946,
      "learning_rate": 7.716671558965977e-06,
      "loss": 0.2753,
      "step": 19139
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.115843317631679,
      "learning_rate": 7.715705889142914e-06,
      "loss": 0.5374,
      "step": 19140
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6326452017059018,
      "learning_rate": 7.714740241793487e-06,
      "loss": 0.3447,
      "step": 19141
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37073666742446304,
      "learning_rate": 7.713774616927197e-06,
      "loss": 0.2223,
      "step": 19142
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29584180013041317,
      "learning_rate": 7.712809014553542e-06,
      "loss": 0.1935,
      "step": 19143
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3488692770341608,
      "learning_rate": 7.711843434682029e-06,
      "loss": 0.2583,
      "step": 19144
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.060285485506954,
      "learning_rate": 7.710877877322152e-06,
      "loss": 0.5418,
      "step": 19145
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2518056629679185,
      "learning_rate": 7.70991234248341e-06,
      "loss": 0.1874,
      "step": 19146
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0141476038973143,
      "learning_rate": 7.708946830175305e-06,
      "loss": 0.2201,
      "step": 19147
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3842785421500192,
      "learning_rate": 7.707981340407338e-06,
      "loss": 0.9038,
      "step": 19148
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6880206052278256,
      "learning_rate": 7.707015873188998e-06,
      "loss": 0.2667,
      "step": 19149
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3706755512412228,
      "learning_rate": 7.706050428529795e-06,
      "loss": 0.2866,
      "step": 19150
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28730876969236935,
      "learning_rate": 7.705085006439223e-06,
      "loss": 0.2315,
      "step": 19151
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7357702072388684,
      "learning_rate": 7.704119606926778e-06,
      "loss": 0.4275,
      "step": 19152
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7685849490045226,
      "learning_rate": 7.703154230001961e-06,
      "loss": 0.3009,
      "step": 19153
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30069419030858346,
      "learning_rate": 7.702188875674268e-06,
      "loss": 0.1871,
      "step": 19154
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2721369075543421,
      "learning_rate": 7.701223543953193e-06,
      "loss": 0.1776,
      "step": 19155
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4410063935459793,
      "learning_rate": 7.700258234848243e-06,
      "loss": 0.3205,
      "step": 19156
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3889172281958546,
      "learning_rate": 7.699292948368909e-06,
      "loss": 0.2504,
      "step": 19157
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3255780587624437,
      "learning_rate": 7.698327684524685e-06,
      "loss": 0.7655,
      "step": 19158
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30971499500081623,
      "learning_rate": 7.697362443325072e-06,
      "loss": 0.1638,
      "step": 19159
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43777594738750913,
      "learning_rate": 7.696397224779566e-06,
      "loss": 0.2117,
      "step": 19160
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9841918686603806,
      "learning_rate": 7.695432028897657e-06,
      "loss": 0.4435,
      "step": 19161
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3160805720154774,
      "learning_rate": 7.69446685568885e-06,
      "loss": 0.2557,
      "step": 19162
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24441152738876545,
      "learning_rate": 7.693501705162638e-06,
      "loss": 0.1364,
      "step": 19163
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2867740332641481,
      "learning_rate": 7.692536577328514e-06,
      "loss": 0.1939,
      "step": 19164
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.234975825154939,
      "learning_rate": 7.691571472195974e-06,
      "loss": 0.5494,
      "step": 19165
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9356551476952438,
      "learning_rate": 7.690606389774515e-06,
      "loss": 0.5226,
      "step": 19166
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4407051388886209,
      "learning_rate": 7.689641330073626e-06,
      "loss": 0.3309,
      "step": 19167
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3011459757727523,
      "learning_rate": 7.688676293102808e-06,
      "loss": 0.1989,
      "step": 19168
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3226606220063929,
      "learning_rate": 7.68771127887155e-06,
      "loss": 0.2583,
      "step": 19169
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6971135432092845,
      "learning_rate": 7.686746287389358e-06,
      "loss": 0.3153,
      "step": 19170
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8325257779784612,
      "learning_rate": 7.685781318665712e-06,
      "loss": 0.4505,
      "step": 19171
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.15832919049163158,
      "learning_rate": 7.684816372710108e-06,
      "loss": 0.0709,
      "step": 19172
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31791688905330484,
      "learning_rate": 7.683851449532046e-06,
      "loss": 0.1674,
      "step": 19173
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.34040588896730367,
      "learning_rate": 7.682886549141015e-06,
      "loss": 0.287,
      "step": 19174
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40907732022934995,
      "learning_rate": 7.681921671546507e-06,
      "loss": 0.2806,
      "step": 19175
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7270584011930301,
      "learning_rate": 7.680956816758017e-06,
      "loss": 0.7805,
      "step": 19176
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4572642190979213,
      "learning_rate": 7.67999198478504e-06,
      "loss": 0.1899,
      "step": 19177
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3813101070538763,
      "learning_rate": 7.679027175637062e-06,
      "loss": 0.2971,
      "step": 19178
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8121581133909167,
      "learning_rate": 7.67806238932358e-06,
      "loss": 0.3189,
      "step": 19179
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.49450571061053333,
      "learning_rate": 7.677097625854084e-06,
      "loss": 0.3549,
      "step": 19180
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19673101063822387,
      "learning_rate": 7.676132885238062e-06,
      "loss": 0.1395,
      "step": 19181
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4179847507085037,
      "learning_rate": 7.675168167485014e-06,
      "loss": 0.3118,
      "step": 19182
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4849092395008833,
      "learning_rate": 7.674203472604425e-06,
      "loss": 0.2325,
      "step": 19183
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6231597384200818,
      "learning_rate": 7.673238800605788e-06,
      "loss": 0.8285,
      "step": 19184
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9149853149499854,
      "learning_rate": 7.672274151498593e-06,
      "loss": 0.3656,
      "step": 19185
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.23848144858340004,
      "learning_rate": 7.671309525292333e-06,
      "loss": 0.1953,
      "step": 19186
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.36356965989320095,
      "learning_rate": 7.67034492199649e-06,
      "loss": 0.2949,
      "step": 19187
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8876145341153129,
      "learning_rate": 7.669380341620565e-06,
      "loss": 0.3161,
      "step": 19188
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7921637187195718,
      "learning_rate": 7.668415784174043e-06,
      "loss": 0.9097,
      "step": 19189
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1503733107724573,
      "learning_rate": 7.667451249666414e-06,
      "loss": 0.0707,
      "step": 19190
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35174531341697246,
      "learning_rate": 7.666486738107167e-06,
      "loss": 0.2754,
      "step": 19191
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4534198793602425,
      "learning_rate": 7.66552224950579e-06,
      "loss": 0.2365,
      "step": 19192
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4657559864476155,
      "learning_rate": 7.66455778387177e-06,
      "loss": 0.3967,
      "step": 19193
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9150578224978034,
      "learning_rate": 7.663593341214605e-06,
      "loss": 0.3651,
      "step": 19194
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6878120085323304,
      "learning_rate": 7.662628921543773e-06,
      "loss": 0.3763,
      "step": 19195
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3200503743605314,
      "learning_rate": 7.66166452486877e-06,
      "loss": 0.206,
      "step": 19196
    },
    {
      "epoch": 0.59,
      "grad_norm": 2.153559141123182,
      "learning_rate": 7.66070015119908e-06,
      "loss": 0.4639,
      "step": 19197
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28426133423269057,
      "learning_rate": 7.659735800544189e-06,
      "loss": 0.2328,
      "step": 19198
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4711193348926386,
      "learning_rate": 7.658771472913591e-06,
      "loss": 0.0282,
      "step": 19199
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32691592100791667,
      "learning_rate": 7.657807168316769e-06,
      "loss": 0.1693,
      "step": 19200
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.22313218914701088,
      "learning_rate": 7.65684288676321e-06,
      "loss": 0.17,
      "step": 19201
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8968972788020121,
      "learning_rate": 7.655878628262401e-06,
      "loss": 0.5254,
      "step": 19202
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9257861036167911,
      "learning_rate": 7.654914392823832e-06,
      "loss": 0.3321,
      "step": 19203
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3293698449533884,
      "learning_rate": 7.653950180456981e-06,
      "loss": 0.2909,
      "step": 19204
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2706243648542945,
      "learning_rate": 7.652985991171345e-06,
      "loss": 0.1842,
      "step": 19205
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.4650434196418873,
      "learning_rate": 7.652021824976406e-06,
      "loss": 0.6243,
      "step": 19206
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2525539085640987,
      "learning_rate": 7.651057681881642e-06,
      "loss": 0.285,
      "step": 19207
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3885626480356718,
      "learning_rate": 7.650093561896552e-06,
      "loss": 0.2001,
      "step": 19208
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19488175017094,
      "learning_rate": 7.649129465030614e-06,
      "loss": 0.1251,
      "step": 19209
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.283149849796421,
      "learning_rate": 7.648165391293312e-06,
      "loss": 0.3003,
      "step": 19210
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9167096504191857,
      "learning_rate": 7.647201340694135e-06,
      "loss": 0.463,
      "step": 19211
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8629415990745801,
      "learning_rate": 7.646237313242565e-06,
      "loss": 0.3366,
      "step": 19212
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7370977927840047,
      "learning_rate": 7.645273308948082e-06,
      "loss": 0.4011,
      "step": 19213
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30576413245371686,
      "learning_rate": 7.64430932782018e-06,
      "loss": 0.1998,
      "step": 19214
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.540793892758332,
      "learning_rate": 7.643345369868337e-06,
      "loss": 0.701,
      "step": 19215
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3197673376802121,
      "learning_rate": 7.642381435102037e-06,
      "loss": 0.243,
      "step": 19216
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0756097637071802,
      "learning_rate": 7.641417523530766e-06,
      "loss": 0.6339,
      "step": 19217
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.16779820389297967,
      "learning_rate": 7.640453635164005e-06,
      "loss": 0.0702,
      "step": 19218
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3527279265496034,
      "learning_rate": 7.639489770011235e-06,
      "loss": 0.2689,
      "step": 19219
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6496358241075009,
      "learning_rate": 7.638525928081943e-06,
      "loss": 0.3371,
      "step": 19220
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4117297753525889,
      "learning_rate": 7.637562109385608e-06,
      "loss": 0.3245,
      "step": 19221
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31127949637387825,
      "learning_rate": 7.636598313931718e-06,
      "loss": 0.2017,
      "step": 19222
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5374733672080773,
      "learning_rate": 7.635634541729752e-06,
      "loss": 0.2753,
      "step": 19223
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2720955922669156,
      "learning_rate": 7.634670792789186e-06,
      "loss": 0.1572,
      "step": 19224
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.764385925952616,
      "learning_rate": 7.63370706711951e-06,
      "loss": 0.0534,
      "step": 19225
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4619349267719891,
      "learning_rate": 7.632743364730203e-06,
      "loss": 0.257,
      "step": 19226
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28836882879421394,
      "learning_rate": 7.631779685630746e-06,
      "loss": 0.1161,
      "step": 19227
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29171833715488066,
      "learning_rate": 7.63081602983062e-06,
      "loss": 0.279,
      "step": 19228
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3647798672252481,
      "learning_rate": 7.629852397339305e-06,
      "loss": 0.2862,
      "step": 19229
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8097846173668268,
      "learning_rate": 7.628888788166279e-06,
      "loss": 0.4864,
      "step": 19230
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.23280602488307342,
      "learning_rate": 7.627925202321028e-06,
      "loss": 0.1436,
      "step": 19231
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.39576743774874135,
      "learning_rate": 7.626961639813031e-06,
      "loss": 0.2517,
      "step": 19232
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30592396688921025,
      "learning_rate": 7.625998100651763e-06,
      "loss": 0.2216,
      "step": 19233
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4637575244340384,
      "learning_rate": 7.6250345848467085e-06,
      "loss": 0.3487,
      "step": 19234
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9381987040946552,
      "learning_rate": 7.624071092407347e-06,
      "loss": 0.227,
      "step": 19235
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8604963120264457,
      "learning_rate": 7.62310762334315e-06,
      "loss": 0.4605,
      "step": 19236
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28671908638126425,
      "learning_rate": 7.622144177663607e-06,
      "loss": 0.215,
      "step": 19237
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6222655398423516,
      "learning_rate": 7.621180755378192e-06,
      "loss": 0.309,
      "step": 19238
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4008238893343117,
      "learning_rate": 7.62021735649638e-06,
      "loss": 0.3418,
      "step": 19239
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3307108132316925,
      "learning_rate": 7.619253981027656e-06,
      "loss": 0.2254,
      "step": 19240
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32372051719386946,
      "learning_rate": 7.618290628981495e-06,
      "loss": 0.2515,
      "step": 19241
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.15701126010462166,
      "learning_rate": 7.617327300367373e-06,
      "loss": 0.0704,
      "step": 19242
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5825777397829306,
      "learning_rate": 7.616363995194771e-06,
      "loss": 0.8082,
      "step": 19243
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9999387926549331,
      "learning_rate": 7.615400713473163e-06,
      "loss": 0.2936,
      "step": 19244
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4177278611732072,
      "learning_rate": 7.6144374552120245e-06,
      "loss": 0.2446,
      "step": 19245
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3461779496835774,
      "learning_rate": 7.613474220420839e-06,
      "loss": 0.2208,
      "step": 19246
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4101762259027956,
      "learning_rate": 7.612511009109077e-06,
      "loss": 0.3275,
      "step": 19247
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7711019297841648,
      "learning_rate": 7.6115478212862205e-06,
      "loss": 0.3897,
      "step": 19248
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41319939720936266,
      "learning_rate": 7.610584656961741e-06,
      "loss": 0.2298,
      "step": 19249
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3464326156482754,
      "learning_rate": 7.6096215161451125e-06,
      "loss": 0.0704,
      "step": 19250
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19705496280352294,
      "learning_rate": 7.608658398845818e-06,
      "loss": 0.1758,
      "step": 19251
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41646829317921447,
      "learning_rate": 7.60769530507333e-06,
      "loss": 0.3135,
      "step": 19252
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7301195563533319,
      "learning_rate": 7.6067322348371195e-06,
      "loss": 0.3222,
      "step": 19253
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6795432814178868,
      "learning_rate": 7.605769188146667e-06,
      "loss": 0.3646,
      "step": 19254
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2662909594383092,
      "learning_rate": 7.604806165011445e-06,
      "loss": 0.2068,
      "step": 19255
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8901510179679071,
      "learning_rate": 7.6038431654409236e-06,
      "loss": 0.5154,
      "step": 19256
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40061844713539957,
      "learning_rate": 7.602880189444587e-06,
      "loss": 0.261,
      "step": 19257
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.47491469308443385,
      "learning_rate": 7.601917237031902e-06,
      "loss": 0.3841,
      "step": 19258
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32914478359816696,
      "learning_rate": 7.600954308212344e-06,
      "loss": 0.1701,
      "step": 19259
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24562063941428455,
      "learning_rate": 7.599991402995388e-06,
      "loss": 0.184,
      "step": 19260
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3715989834337197,
      "learning_rate": 7.599028521390507e-06,
      "loss": 0.0159,
      "step": 19261
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7998790118950106,
      "learning_rate": 7.5980656634071685e-06,
      "loss": 0.4275,
      "step": 19262
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3003936947586905,
      "learning_rate": 7.597102829054855e-06,
      "loss": 0.2233,
      "step": 19263
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.29776384340342776,
      "learning_rate": 7.596140018343035e-06,
      "loss": 0.2176,
      "step": 19264
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8580524416479541,
      "learning_rate": 7.595177231281177e-06,
      "loss": 0.4738,
      "step": 19265
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0319237414862292,
      "learning_rate": 7.5942144678787595e-06,
      "loss": 0.4345,
      "step": 19266
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.091018313928724,
      "learning_rate": 7.593251728145253e-06,
      "loss": 0.508,
      "step": 19267
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28819131697309536,
      "learning_rate": 7.5922890120901214e-06,
      "loss": 0.188,
      "step": 19268
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.44058779395161235,
      "learning_rate": 7.5913263197228485e-06,
      "loss": 0.2255,
      "step": 19269
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24316759595586696,
      "learning_rate": 7.590363651052897e-06,
      "loss": 0.1913,
      "step": 19270
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7403934815310562,
      "learning_rate": 7.5894010060897425e-06,
      "loss": 0.458,
      "step": 19271
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4991334669914284,
      "learning_rate": 7.588438384842856e-06,
      "loss": 0.2156,
      "step": 19272
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5119489379838279,
      "learning_rate": 7.5874757873217e-06,
      "loss": 0.3164,
      "step": 19273
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4157674517473387,
      "learning_rate": 7.586513213535756e-06,
      "loss": 0.2265,
      "step": 19274
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30694563765672017,
      "learning_rate": 7.585550663494489e-06,
      "loss": 0.2974,
      "step": 19275
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0699829920723831,
      "learning_rate": 7.5845881372073645e-06,
      "loss": 0.4476,
      "step": 19276
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2910467283771874,
      "learning_rate": 7.583625634683861e-06,
      "loss": 0.0705,
      "step": 19277
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.26770677461033077,
      "learning_rate": 7.582663155933444e-06,
      "loss": 0.2108,
      "step": 19278
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5006288208478513,
      "learning_rate": 7.58170070096558e-06,
      "loss": 0.2057,
      "step": 19279
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4996572336225328,
      "learning_rate": 7.580738269789742e-06,
      "loss": 0.3409,
      "step": 19280
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41598930987832655,
      "learning_rate": 7.579775862415397e-06,
      "loss": 0.2298,
      "step": 19281
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3724879998484323,
      "learning_rate": 7.578813478852009e-06,
      "loss": 0.2549,
      "step": 19282
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4423945107952102,
      "learning_rate": 7.5778511191090565e-06,
      "loss": 0.2354,
      "step": 19283
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5494291860924403,
      "learning_rate": 7.576888783196002e-06,
      "loss": 0.7523,
      "step": 19284
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2416877343260524,
      "learning_rate": 7.575926471122308e-06,
      "loss": 0.3017,
      "step": 19285
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6403159468204908,
      "learning_rate": 7.574964182897451e-06,
      "loss": 0.3082,
      "step": 19286
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.21645973272190794,
      "learning_rate": 7.574001918530893e-06,
      "loss": 0.1854,
      "step": 19287
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3101660665161852,
      "learning_rate": 7.5730396780320995e-06,
      "loss": 0.2284,
      "step": 19288
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5246990152961191,
      "learning_rate": 7.572077461410544e-06,
      "loss": 0.3189,
      "step": 19289
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5365245014264558,
      "learning_rate": 7.571115268675689e-06,
      "loss": 0.2141,
      "step": 19290
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33884626934644274,
      "learning_rate": 7.570153099837e-06,
      "loss": 0.2642,
      "step": 19291
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2427120210544687,
      "learning_rate": 7.569190954903945e-06,
      "loss": 0.344,
      "step": 19292
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3167123086833345,
      "learning_rate": 7.568228833885991e-06,
      "loss": 0.3017,
      "step": 19293
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8677256038712163,
      "learning_rate": 7.567266736792596e-06,
      "loss": 0.4328,
      "step": 19294
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6665472453581279,
      "learning_rate": 7.566304663633237e-06,
      "loss": 0.3299,
      "step": 19295
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42362624107949853,
      "learning_rate": 7.56534261441737e-06,
      "loss": 0.2005,
      "step": 19296
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8100966531160779,
      "learning_rate": 7.564380589154466e-06,
      "loss": 0.4249,
      "step": 19297
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6044968723815052,
      "learning_rate": 7.563418587853988e-06,
      "loss": 0.2135,
      "step": 19298
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2610073256006523,
      "learning_rate": 7.562456610525395e-06,
      "loss": 0.2047,
      "step": 19299
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33326466136030136,
      "learning_rate": 7.56149465717816e-06,
      "loss": 0.16,
      "step": 19300
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5369877568198861,
      "learning_rate": 7.560532727821744e-06,
      "loss": 0.3516,
      "step": 19301
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.015513473154282,
      "learning_rate": 7.559570822465605e-06,
      "loss": 0.6168,
      "step": 19302
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7339313678666285,
      "learning_rate": 7.558608941119216e-06,
      "loss": 0.2883,
      "step": 19303
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8652337374883637,
      "learning_rate": 7.5576470837920335e-06,
      "loss": 0.475,
      "step": 19304
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.24203168802899525,
      "learning_rate": 7.556685250493519e-06,
      "loss": 0.204,
      "step": 19305
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.43230466665455236,
      "learning_rate": 7.5557234412331434e-06,
      "loss": 0.3379,
      "step": 19306
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3980126301898682,
      "learning_rate": 7.554761656020366e-06,
      "loss": 0.1624,
      "step": 19307
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5151716502156657,
      "learning_rate": 7.553799894864643e-06,
      "loss": 0.2784,
      "step": 19308
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3142212029891559,
      "learning_rate": 7.552838157775445e-06,
      "loss": 0.1724,
      "step": 19309
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4948663236310631,
      "learning_rate": 7.5518764447622316e-06,
      "loss": 0.346,
      "step": 19310
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.28519275421813917,
      "learning_rate": 7.55091475583446e-06,
      "loss": 0.2387,
      "step": 19311
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.6986099062667135,
      "learning_rate": 7.549953091001598e-06,
      "loss": 0.8739,
      "step": 19312
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5140476779212781,
      "learning_rate": 7.548991450273103e-06,
      "loss": 0.2454,
      "step": 19313
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32001285428500975,
      "learning_rate": 7.548029833658432e-06,
      "loss": 0.2646,
      "step": 19314
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5275140989548531,
      "learning_rate": 7.547068241167056e-06,
      "loss": 0.3099,
      "step": 19315
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3693218772759312,
      "learning_rate": 7.546106672808429e-06,
      "loss": 0.2126,
      "step": 19316
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.23953657694505617,
      "learning_rate": 7.545145128592009e-06,
      "loss": 0.1863,
      "step": 19317
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25744266498890944,
      "learning_rate": 7.544183608527262e-06,
      "loss": 0.1803,
      "step": 19318
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.8286039064217756,
      "learning_rate": 7.543222112623645e-06,
      "loss": 0.7679,
      "step": 19319
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7703990494217859,
      "learning_rate": 7.542260640890613e-06,
      "loss": 0.5893,
      "step": 19320
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8352921196714388,
      "learning_rate": 7.541299193337633e-06,
      "loss": 0.4669,
      "step": 19321
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3076707978719951,
      "learning_rate": 7.540337769974159e-06,
      "loss": 0.2016,
      "step": 19322
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5228942059216376,
      "learning_rate": 7.539376370809653e-06,
      "loss": 0.3339,
      "step": 19323
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5699068908390462,
      "learning_rate": 7.538414995853572e-06,
      "loss": 0.2572,
      "step": 19324
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.2321667244933137,
      "learning_rate": 7.53745364511537e-06,
      "loss": 0.5884,
      "step": 19325
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.16940053502867913,
      "learning_rate": 7.536492318604514e-06,
      "loss": 0.0857,
      "step": 19326
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8679164044817455,
      "learning_rate": 7.535531016330457e-06,
      "loss": 0.3698,
      "step": 19327
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.27869226565424426,
      "learning_rate": 7.534569738302655e-06,
      "loss": 0.1958,
      "step": 19328
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30681322571220004,
      "learning_rate": 7.533608484530569e-06,
      "loss": 0.2658,
      "step": 19329
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6751394070244751,
      "learning_rate": 7.532647255023654e-06,
      "loss": 0.4413,
      "step": 19330
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.766075938307698,
      "learning_rate": 7.5316860497913625e-06,
      "loss": 0.2236,
      "step": 19331
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3313166783268595,
      "learning_rate": 7.530724868843161e-06,
      "loss": 0.2622,
      "step": 19332
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7532291285498883,
      "learning_rate": 7.5297637121885e-06,
      "loss": 0.2919,
      "step": 19333
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3649412435324004,
      "learning_rate": 7.528802579836835e-06,
      "loss": 0.3084,
      "step": 19334
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33884786285133106,
      "learning_rate": 7.527841471797626e-06,
      "loss": 0.1011,
      "step": 19335
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32367615383461035,
      "learning_rate": 7.5268803880803244e-06,
      "loss": 0.1505,
      "step": 19336
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.34634308009754605,
      "learning_rate": 7.525919328694384e-06,
      "loss": 0.2463,
      "step": 19337
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.408162384041879,
      "learning_rate": 7.524958293649267e-06,
      "loss": 0.6596,
      "step": 19338
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9273112327682944,
      "learning_rate": 7.523997282954425e-06,
      "loss": 0.3075,
      "step": 19339
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3351628401866659,
      "learning_rate": 7.523036296619308e-06,
      "loss": 0.3165,
      "step": 19340
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.27945715925681747,
      "learning_rate": 7.52207533465338e-06,
      "loss": 0.1877,
      "step": 19341
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6992774021808564,
      "learning_rate": 7.521114397066092e-06,
      "loss": 0.4715,
      "step": 19342
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5024987458094596,
      "learning_rate": 7.52015348386689e-06,
      "loss": 0.0257,
      "step": 19343
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.30267055775722485,
      "learning_rate": 7.519192595065238e-06,
      "loss": 0.1222,
      "step": 19344
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6493142547413823,
      "learning_rate": 7.518231730670586e-06,
      "loss": 0.3797,
      "step": 19345
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3383685942729668,
      "learning_rate": 7.517270890692381e-06,
      "loss": 0.2235,
      "step": 19346
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.33429136012974914,
      "learning_rate": 7.516310075140087e-06,
      "loss": 0.2814,
      "step": 19347
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.999830537450412,
      "learning_rate": 7.51534928402315e-06,
      "loss": 0.3435,
      "step": 19348
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3592450064613523,
      "learning_rate": 7.514388517351027e-06,
      "loss": 0.2148,
      "step": 19349
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3511214470574946,
      "learning_rate": 7.513427775133167e-06,
      "loss": 0.1736,
      "step": 19350
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4951379453869973,
      "learning_rate": 7.512467057379018e-06,
      "loss": 0.328,
      "step": 19351
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.38907745311949765,
      "learning_rate": 7.511506364098041e-06,
      "loss": 0.2243,
      "step": 19352
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3291371478862059,
      "learning_rate": 7.510545695299683e-06,
      "loss": 0.2329,
      "step": 19353
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3416988884072457,
      "learning_rate": 7.509585050993394e-06,
      "loss": 0.1463,
      "step": 19354
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35177231292777444,
      "learning_rate": 7.508624431188629e-06,
      "loss": 0.274,
      "step": 19355
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1398117859419579,
      "learning_rate": 7.5076638358948364e-06,
      "loss": 0.3629,
      "step": 19356
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5273697010785119,
      "learning_rate": 7.506703265121464e-06,
      "loss": 0.2993,
      "step": 19357
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.48267985314688044,
      "learning_rate": 7.505742718877968e-06,
      "loss": 0.3572,
      "step": 19358
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.19956569554837666,
      "learning_rate": 7.504782197173795e-06,
      "loss": 0.119,
      "step": 19359
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5263138797403766,
      "learning_rate": 7.5038217000183954e-06,
      "loss": 0.3199,
      "step": 19360
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.1473667960099694,
      "learning_rate": 7.502861227421221e-06,
      "loss": 0.2882,
      "step": 19361
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5903085632667395,
      "learning_rate": 7.501900779391719e-06,
      "loss": 0.72,
      "step": 19362
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3008438014624946,
      "learning_rate": 7.500940355939335e-06,
      "loss": 0.1537,
      "step": 19363
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35776106443064026,
      "learning_rate": 7.499979957073527e-06,
      "loss": 0.2826,
      "step": 19364
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4548800862307429,
      "learning_rate": 7.499019582803738e-06,
      "loss": 0.2612,
      "step": 19365
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0782346480060185,
      "learning_rate": 7.498059233139417e-06,
      "loss": 0.5975,
      "step": 19366
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.25978505510054334,
      "learning_rate": 7.497098908090012e-06,
      "loss": 0.1322,
      "step": 19367
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4056100481721494,
      "learning_rate": 7.496138607664973e-06,
      "loss": 0.3168,
      "step": 19368
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.42921736123230975,
      "learning_rate": 7.49517833187374e-06,
      "loss": 0.1988,
      "step": 19369
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4305281787930915,
      "learning_rate": 7.494218080725774e-06,
      "loss": 0.2379,
      "step": 19370
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4729848105423598,
      "learning_rate": 7.493257854230514e-06,
      "loss": 0.3032,
      "step": 19371
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3297218390681759,
      "learning_rate": 7.492297652397405e-06,
      "loss": 0.1597,
      "step": 19372
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3915828348691479,
      "learning_rate": 7.491337475235901e-06,
      "loss": 0.3014,
      "step": 19373
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.764293359516464,
      "learning_rate": 7.490377322755437e-06,
      "loss": 0.319,
      "step": 19374
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7699563957163302,
      "learning_rate": 7.489417194965473e-06,
      "loss": 0.503,
      "step": 19375
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1794989368027822,
      "learning_rate": 7.488457091875447e-06,
      "loss": 0.1621,
      "step": 19376
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6442997016189408,
      "learning_rate": 7.487497013494804e-06,
      "loss": 0.3594,
      "step": 19377
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3062555017232311,
      "learning_rate": 7.486536959832997e-06,
      "loss": 0.2075,
      "step": 19378
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.7768237145053989,
      "learning_rate": 7.485576930899465e-06,
      "loss": 0.6798,
      "step": 19379
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.711503603600708,
      "learning_rate": 7.484616926703653e-06,
      "loss": 0.3378,
      "step": 19380
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6886318762268275,
      "learning_rate": 7.483656947255009e-06,
      "loss": 0.3735,
      "step": 19381
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3221474374113902,
      "learning_rate": 7.482696992562977e-06,
      "loss": 0.2038,
      "step": 19382
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3073792895643706,
      "learning_rate": 7.481737062636995e-06,
      "loss": 0.2723,
      "step": 19383
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.33633606241586,
      "learning_rate": 7.480777157486516e-06,
      "loss": 0.8049,
      "step": 19384
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35548155105784246,
      "learning_rate": 7.479817277120982e-06,
      "loss": 0.1007,
      "step": 19385
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.35015053452538464,
      "learning_rate": 7.478857421549832e-06,
      "loss": 0.2061,
      "step": 19386
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3226794826356818,
      "learning_rate": 7.477897590782514e-06,
      "loss": 0.2064,
      "step": 19387
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3661731762558412,
      "learning_rate": 7.4769377848284696e-06,
      "loss": 0.303,
      "step": 19388
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3587748277821311,
      "learning_rate": 7.475978003697138e-06,
      "loss": 0.021,
      "step": 19389
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8615507890295411,
      "learning_rate": 7.4750182473979685e-06,
      "loss": 0.4998,
      "step": 19390
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.26524740647849065,
      "learning_rate": 7.474058515940401e-06,
      "loss": 0.18,
      "step": 19391
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6419619191220434,
      "learning_rate": 7.473098809333874e-06,
      "loss": 0.4205,
      "step": 19392
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8411032604835785,
      "learning_rate": 7.472139127587834e-06,
      "loss": 0.4388,
      "step": 19393
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.327760311264238,
      "learning_rate": 7.471179470711723e-06,
      "loss": 0.3015,
      "step": 19394
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3224410622504581,
      "learning_rate": 7.470219838714975e-06,
      "loss": 0.1758,
      "step": 19395
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1846685586455564,
      "learning_rate": 7.469260231607041e-06,
      "loss": 0.1149,
      "step": 19396
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.3853274541247294,
      "learning_rate": 7.468300649397355e-06,
      "loss": 0.5853,
      "step": 19397
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6644343675950057,
      "learning_rate": 7.467341092095364e-06,
      "loss": 0.301,
      "step": 19398
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4185459455964874,
      "learning_rate": 7.466381559710503e-06,
      "loss": 0.324,
      "step": 19399
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.31511861336259417,
      "learning_rate": 7.465422052252211e-06,
      "loss": 0.2011,
      "step": 19400
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4380688189977495,
      "learning_rate": 7.4644625697299335e-06,
      "loss": 0.3469,
      "step": 19401
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.8685870109739465,
      "learning_rate": 7.463503112153109e-06,
      "loss": 0.5237,
      "step": 19402
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3997356781296775,
      "learning_rate": 7.462543679531173e-06,
      "loss": 0.2247,
      "step": 19403
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.1372219402386205,
      "learning_rate": 7.461584271873569e-06,
      "loss": 0.0685,
      "step": 19404
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.3631687187948913,
      "learning_rate": 7.460624889189734e-06,
      "loss": 0.2553,
      "step": 19405
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2956463971545918,
      "learning_rate": 7.4596655314891034e-06,
      "loss": 0.2184,
      "step": 19406
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7333051720882318,
      "learning_rate": 7.458706198781123e-06,
      "loss": 0.4197,
      "step": 19407
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.6994756669713017,
      "learning_rate": 7.457746891075227e-06,
      "loss": 0.2927,
      "step": 19408
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.273428623673131,
      "learning_rate": 7.456787608380849e-06,
      "loss": 0.1805,
      "step": 19409
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0326402194852686,
      "learning_rate": 7.455828350707437e-06,
      "loss": 0.5198,
      "step": 19410
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.40994255532469426,
      "learning_rate": 7.4548691180644215e-06,
      "loss": 0.2924,
      "step": 19411
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.45285138720841867,
      "learning_rate": 7.453909910461239e-06,
      "loss": 0.2959,
      "step": 19412
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.41637748148666565,
      "learning_rate": 7.452950727907331e-06,
      "loss": 0.1147,
      "step": 19413
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2571011398225779,
      "learning_rate": 7.451991570412133e-06,
      "loss": 0.2095,
      "step": 19414
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4399074127347337,
      "learning_rate": 7.451032437985075e-06,
      "loss": 0.1686,
      "step": 19415
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.0420173363300154,
      "learning_rate": 7.450073330635602e-06,
      "loss": 0.5634,
      "step": 19416
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.37662429272870185,
      "learning_rate": 7.449114248373148e-06,
      "loss": 0.2573,
      "step": 19417
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.406262278694963,
      "learning_rate": 7.448155191207145e-06,
      "loss": 0.2774,
      "step": 19418
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.4495163402400589,
      "learning_rate": 7.447196159147032e-06,
      "loss": 0.2359,
      "step": 19419
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.5095286982377072,
      "learning_rate": 7.446237152202244e-06,
      "loss": 0.7311,
      "step": 19420
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.9269282578193382,
      "learning_rate": 7.445278170382211e-06,
      "loss": 0.3594,
      "step": 19421
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.18905794347543456,
      "learning_rate": 7.444319213696376e-06,
      "loss": 0.0708,
      "step": 19422
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.390540224997358,
      "learning_rate": 7.443360282154169e-06,
      "loss": 0.2917,
      "step": 19423
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.2916949469928677,
      "learning_rate": 7.442401375765024e-06,
      "loss": 0.2279,
      "step": 19424
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.7423549633742684,
      "learning_rate": 7.441442494538378e-06,
      "loss": 0.4163,
      "step": 19425
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.5222801460489865,
      "learning_rate": 7.440483638483658e-06,
      "loss": 0.2241,
      "step": 19426
    },
    {
      "epoch": 0.59,
      "grad_norm": 0.32364887690865535,
      "learning_rate": 7.4395248076103054e-06,
      "loss": 0.2527,
      "step": 19427
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.192020205564408,
      "learning_rate": 7.438566001927749e-06,
      "loss": 0.3124,
      "step": 19428
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5334229172735037,
      "learning_rate": 7.4376072214454245e-06,
      "loss": 0.8109,
      "step": 19429
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36147152156461815,
      "learning_rate": 7.436648466172764e-06,
      "loss": 0.239,
      "step": 19430
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6305879827413998,
      "learning_rate": 7.4356897361191986e-06,
      "loss": 0.304,
      "step": 19431
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3075921871532129,
      "learning_rate": 7.434731031294158e-06,
      "loss": 0.1988,
      "step": 19432
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39820749055189736,
      "learning_rate": 7.43377235170708e-06,
      "loss": 0.2642,
      "step": 19433
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37599764497662214,
      "learning_rate": 7.432813697367396e-06,
      "loss": 0.1959,
      "step": 19434
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38907201193761326,
      "learning_rate": 7.431855068284533e-06,
      "loss": 0.2208,
      "step": 19435
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36433407142148844,
      "learning_rate": 7.4308964644679274e-06,
      "loss": 0.2562,
      "step": 19436
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3273590987495688,
      "learning_rate": 7.429937885927006e-06,
      "loss": 0.229,
      "step": 19437
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.2567068006057507,
      "learning_rate": 7.4289793326711986e-06,
      "loss": 0.7418,
      "step": 19438
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9673753759642422,
      "learning_rate": 7.4280208047099436e-06,
      "loss": 0.0456,
      "step": 19439
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8578705167905706,
      "learning_rate": 7.4270623020526645e-06,
      "loss": 0.4443,
      "step": 19440
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2681149424073192,
      "learning_rate": 7.426103824708789e-06,
      "loss": 0.2066,
      "step": 19441
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2960973633711969,
      "learning_rate": 7.425145372687758e-06,
      "loss": 0.3006,
      "step": 19442
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5505544288682044,
      "learning_rate": 7.424186945998993e-06,
      "loss": 0.2967,
      "step": 19443
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3169128135773203,
      "learning_rate": 7.423228544651917e-06,
      "loss": 0.1779,
      "step": 19444
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31471197554786656,
      "learning_rate": 7.422270168655973e-06,
      "loss": 0.1631,
      "step": 19445
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.516412787662453,
      "learning_rate": 7.421311818020582e-06,
      "loss": 0.3522,
      "step": 19446
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8102436450279041,
      "learning_rate": 7.42035349275517e-06,
      "loss": 0.4554,
      "step": 19447
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3123616418745904,
      "learning_rate": 7.419395192869174e-06,
      "loss": 0.2481,
      "step": 19448
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8296711939959907,
      "learning_rate": 7.418436918372016e-06,
      "loss": 0.4123,
      "step": 19449
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30629731912437325,
      "learning_rate": 7.417478669273126e-06,
      "loss": 0.2163,
      "step": 19450
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9504591312734864,
      "learning_rate": 7.41652044558193e-06,
      "loss": 0.482,
      "step": 19451
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23502026623701072,
      "learning_rate": 7.415562247307852e-06,
      "loss": 0.1179,
      "step": 19452
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3312056188752581,
      "learning_rate": 7.414604074460328e-06,
      "loss": 0.3011,
      "step": 19453
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3436515391972112,
      "learning_rate": 7.4136459270487805e-06,
      "loss": 0.1629,
      "step": 19454
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5444560458721809,
      "learning_rate": 7.412687805082631e-06,
      "loss": 0.3486,
      "step": 19455
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8961201224294366,
      "learning_rate": 7.411729708571314e-06,
      "loss": 0.5972,
      "step": 19456
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8285036650256978,
      "learning_rate": 7.410771637524251e-06,
      "loss": 0.4869,
      "step": 19457
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5424648868357722,
      "learning_rate": 7.409813591950865e-06,
      "loss": 0.26,
      "step": 19458
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4370798683229919,
      "learning_rate": 7.408855571860589e-06,
      "loss": 0.2995,
      "step": 19459
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.24054170817650372,
      "learning_rate": 7.4078975772628435e-06,
      "loss": 0.2457,
      "step": 19460
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34317235335366975,
      "learning_rate": 7.406939608167054e-06,
      "loss": 0.1004,
      "step": 19461
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.47699294935018477,
      "learning_rate": 7.405981664582648e-06,
      "loss": 0.2286,
      "step": 19462
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.29860652444423896,
      "learning_rate": 7.405023746519046e-06,
      "loss": 0.1172,
      "step": 19463
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3483931919249973,
      "learning_rate": 7.4040658539856724e-06,
      "loss": 0.2877,
      "step": 19464
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2546278997314263,
      "learning_rate": 7.403107986991956e-06,
      "loss": 0.236,
      "step": 19465
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4188342086346022,
      "learning_rate": 7.402150145547318e-06,
      "loss": 0.6292,
      "step": 19466
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7244840207690904,
      "learning_rate": 7.4011923296611795e-06,
      "loss": 0.3244,
      "step": 19467
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3956420394176945,
      "learning_rate": 7.400234539342969e-06,
      "loss": 0.2594,
      "step": 19468
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40735275836168583,
      "learning_rate": 7.399276774602105e-06,
      "loss": 0.2524,
      "step": 19469
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4924592917897777,
      "learning_rate": 7.39831903544801e-06,
      "loss": 0.3003,
      "step": 19470
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21028464508478872,
      "learning_rate": 7.39736132189011e-06,
      "loss": 0.1805,
      "step": 19471
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6069139059878309,
      "learning_rate": 7.396403633937826e-06,
      "loss": 0.2563,
      "step": 19472
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3044754360427843,
      "learning_rate": 7.395445971600579e-06,
      "loss": 0.2281,
      "step": 19473
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.037105580002787,
      "learning_rate": 7.3944883348877925e-06,
      "loss": 0.4824,
      "step": 19474
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7216426076810485,
      "learning_rate": 7.393530723808884e-06,
      "loss": 0.4511,
      "step": 19475
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6591572467153637,
      "learning_rate": 7.3925731383732806e-06,
      "loss": 0.246,
      "step": 19476
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3028087511123256,
      "learning_rate": 7.391615578590402e-06,
      "loss": 0.2465,
      "step": 19477
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2831285723217831,
      "learning_rate": 7.390658044469663e-06,
      "loss": 0.2492,
      "step": 19478
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9458160601214877,
      "learning_rate": 7.389700536020492e-06,
      "loss": 0.4844,
      "step": 19479
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.14841090343648725,
      "learning_rate": 7.388743053252308e-06,
      "loss": 0.0856,
      "step": 19480
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6334478720785067,
      "learning_rate": 7.387785596174525e-06,
      "loss": 0.3617,
      "step": 19481
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32038806573463174,
      "learning_rate": 7.386828164796569e-06,
      "loss": 0.2125,
      "step": 19482
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44244570372491304,
      "learning_rate": 7.385870759127859e-06,
      "loss": 0.3577,
      "step": 19483
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5504403187904513,
      "learning_rate": 7.384913379177807e-06,
      "loss": 0.2865,
      "step": 19484
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.023515082508266,
      "learning_rate": 7.3839560249558425e-06,
      "loss": 0.5552,
      "step": 19485
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3034480891965932,
      "learning_rate": 7.3829986964713796e-06,
      "loss": 0.0702,
      "step": 19486
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32673207569566615,
      "learning_rate": 7.382041393733834e-06,
      "loss": 0.225,
      "step": 19487
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1923853181775865,
      "learning_rate": 7.3810841167526285e-06,
      "loss": 0.653,
      "step": 19488
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23497471956945562,
      "learning_rate": 7.380126865537181e-06,
      "loss": 0.1884,
      "step": 19489
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9225334128430003,
      "learning_rate": 7.379169640096902e-06,
      "loss": 0.4854,
      "step": 19490
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2897329237281162,
      "learning_rate": 7.378212440441217e-06,
      "loss": 0.2002,
      "step": 19491
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8027516704828357,
      "learning_rate": 7.377255266579543e-06,
      "loss": 0.5183,
      "step": 19492
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.901355071868286,
      "learning_rate": 7.376298118521291e-06,
      "loss": 0.3263,
      "step": 19493
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.46350916022493505,
      "learning_rate": 7.375340996275885e-06,
      "loss": 0.2539,
      "step": 19494
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2598597897451074,
      "learning_rate": 7.374383899852737e-06,
      "loss": 0.1944,
      "step": 19495
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38407968295299133,
      "learning_rate": 7.37342682926126e-06,
      "loss": 0.3133,
      "step": 19496
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5397718366140074,
      "learning_rate": 7.372469784510878e-06,
      "loss": 0.2046,
      "step": 19497
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4411833656684072,
      "learning_rate": 7.371512765611004e-06,
      "loss": 0.2404,
      "step": 19498
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4380147765141409,
      "learning_rate": 7.370555772571049e-06,
      "loss": 0.2664,
      "step": 19499
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.370067984926375,
      "learning_rate": 7.369598805400434e-06,
      "loss": 0.2218,
      "step": 19500
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4814690025873183,
      "learning_rate": 7.368641864108568e-06,
      "loss": 0.367,
      "step": 19501
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6501710286488595,
      "learning_rate": 7.367684948704872e-06,
      "loss": 0.3038,
      "step": 19502
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41191045108480456,
      "learning_rate": 7.366728059198759e-06,
      "loss": 0.2094,
      "step": 19503
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2765499539102643,
      "learning_rate": 7.365771195599638e-06,
      "loss": 0.1792,
      "step": 19504
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6339419440673941,
      "learning_rate": 7.364814357916931e-06,
      "loss": 0.7541,
      "step": 19505
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3497392365446952,
      "learning_rate": 7.363857546160046e-06,
      "loss": 0.2184,
      "step": 19506
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3147102035841118,
      "learning_rate": 7.362900760338394e-06,
      "loss": 0.3073,
      "step": 19507
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27568889284732234,
      "learning_rate": 7.361944000461396e-06,
      "loss": 0.1607,
      "step": 19508
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5721335072515563,
      "learning_rate": 7.360987266538462e-06,
      "loss": 0.3361,
      "step": 19509
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3951449713999949,
      "learning_rate": 7.360030558578999e-06,
      "loss": 0.2803,
      "step": 19510
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8072576395749115,
      "learning_rate": 7.359073876592426e-06,
      "loss": 0.5446,
      "step": 19511
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19079261588637209,
      "learning_rate": 7.358117220588157e-06,
      "loss": 0.1599,
      "step": 19512
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2276682935160538,
      "learning_rate": 7.357160590575595e-06,
      "loss": 0.0688,
      "step": 19513
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.376838589147112,
      "learning_rate": 7.35620398656416e-06,
      "loss": 0.2874,
      "step": 19514
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9004076867077647,
      "learning_rate": 7.3552474085632595e-06,
      "loss": 0.2613,
      "step": 19515
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.367983160482865,
      "learning_rate": 7.354290856582302e-06,
      "loss": 0.7502,
      "step": 19516
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6724396060656394,
      "learning_rate": 7.353334330630706e-06,
      "loss": 0.2991,
      "step": 19517
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38447476442106554,
      "learning_rate": 7.352377830717876e-06,
      "loss": 0.2631,
      "step": 19518
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.25892184359354586,
      "learning_rate": 7.3514213568532235e-06,
      "loss": 0.2438,
      "step": 19519
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7550566560991364,
      "learning_rate": 7.35046490904616e-06,
      "loss": 0.486,
      "step": 19520
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1601095849629616,
      "learning_rate": 7.349508487306096e-06,
      "loss": 0.0752,
      "step": 19521
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5943562210055867,
      "learning_rate": 7.348552091642435e-06,
      "loss": 0.2628,
      "step": 19522
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.343823902102427,
      "learning_rate": 7.347595722064594e-06,
      "loss": 0.211,
      "step": 19523
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.230942750911563,
      "learning_rate": 7.346639378581979e-06,
      "loss": 0.5299,
      "step": 19524
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3097901127275064,
      "learning_rate": 7.345683061203999e-06,
      "loss": 0.233,
      "step": 19525
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8329786524236111,
      "learning_rate": 7.344726769940064e-06,
      "loss": 0.4231,
      "step": 19526
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2814659320823843,
      "learning_rate": 7.343770504799576e-06,
      "loss": 0.184,
      "step": 19527
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6383941195115517,
      "learning_rate": 7.342814265791951e-06,
      "loss": 0.2994,
      "step": 19528
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3137174905653206,
      "learning_rate": 7.341858052926595e-06,
      "loss": 0.8035,
      "step": 19529
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23042820025455849,
      "learning_rate": 7.34090186621291e-06,
      "loss": 0.1927,
      "step": 19530
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3856429265425719,
      "learning_rate": 7.339945705660312e-06,
      "loss": 0.1935,
      "step": 19531
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3203683878058755,
      "learning_rate": 7.338989571278202e-06,
      "loss": 0.2031,
      "step": 19532
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.556478134973488,
      "learning_rate": 7.338033463075985e-06,
      "loss": 0.8205,
      "step": 19533
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6778454194260349,
      "learning_rate": 7.337077381063074e-06,
      "loss": 0.3127,
      "step": 19534
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6069204680644195,
      "learning_rate": 7.336121325248872e-06,
      "loss": 0.3868,
      "step": 19535
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3296731517924709,
      "learning_rate": 7.335165295642783e-06,
      "loss": 0.1672,
      "step": 19536
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3565907485188026,
      "learning_rate": 7.334209292254216e-06,
      "loss": 0.3148,
      "step": 19537
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38202996353471824,
      "learning_rate": 7.333253315092576e-06,
      "loss": 0.2776,
      "step": 19538
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41984006362485343,
      "learning_rate": 7.332297364167264e-06,
      "loss": 0.2141,
      "step": 19539
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19596935848912272,
      "learning_rate": 7.331341439487692e-06,
      "loss": 0.0692,
      "step": 19540
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3052818847718371,
      "learning_rate": 7.33038554106326e-06,
      "loss": 0.1931,
      "step": 19541
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.47585215437573525,
      "learning_rate": 7.329429668903372e-06,
      "loss": 0.3443,
      "step": 19542
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.46935029203288864,
      "learning_rate": 7.328473823017434e-06,
      "loss": 0.2677,
      "step": 19543
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9819177210371614,
      "learning_rate": 7.327518003414853e-06,
      "loss": 0.5174,
      "step": 19544
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3221773105099834,
      "learning_rate": 7.326562210105023e-06,
      "loss": 0.2043,
      "step": 19545
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.47769769967309506,
      "learning_rate": 7.3256064430973575e-06,
      "loss": 0.3317,
      "step": 19546
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1529653944174383,
      "learning_rate": 7.324650702401257e-06,
      "loss": 0.4087,
      "step": 19547
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5005817323466129,
      "learning_rate": 7.323694988026117e-06,
      "loss": 0.3831,
      "step": 19548
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2173296016717171,
      "learning_rate": 7.322739299981354e-06,
      "loss": 0.1538,
      "step": 19549
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2641118701444675,
      "learning_rate": 7.321783638276358e-06,
      "loss": 0.212,
      "step": 19550
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3368076294489422,
      "learning_rate": 7.320828002920539e-06,
      "loss": 0.2519,
      "step": 19551
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8087650906567373,
      "learning_rate": 7.319872393923295e-06,
      "loss": 0.4164,
      "step": 19552
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6836690729778548,
      "learning_rate": 7.318916811294023e-06,
      "loss": 0.3325,
      "step": 19553
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23525804810269005,
      "learning_rate": 7.317961255042136e-06,
      "loss": 0.1675,
      "step": 19554
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3241350700004641,
      "learning_rate": 7.317005725177028e-06,
      "loss": 0.2958,
      "step": 19555
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8447926120797903,
      "learning_rate": 7.316050221708097e-06,
      "loss": 0.5455,
      "step": 19556
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.675639206308634,
      "learning_rate": 7.3150947446447505e-06,
      "loss": 0.8573,
      "step": 19557
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.14749499823091491,
      "learning_rate": 7.314139293996385e-06,
      "loss": 0.0725,
      "step": 19558
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39408786617854247,
      "learning_rate": 7.313183869772398e-06,
      "loss": 0.3191,
      "step": 19559
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9209425862360103,
      "learning_rate": 7.3122284719821945e-06,
      "loss": 0.3392,
      "step": 19560
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.33792649984221673,
      "learning_rate": 7.311273100635172e-06,
      "loss": 0.2946,
      "step": 19561
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6932385299895877,
      "learning_rate": 7.310317755740728e-06,
      "loss": 0.2914,
      "step": 19562
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0570795736963308,
      "learning_rate": 7.309362437308265e-06,
      "loss": 0.4427,
      "step": 19563
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28604990308518047,
      "learning_rate": 7.30840714534718e-06,
      "loss": 0.197,
      "step": 19564
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.177008572470102,
      "learning_rate": 7.307451879866867e-06,
      "loss": 0.6195,
      "step": 19565
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2948219515657795,
      "learning_rate": 7.3064966408767316e-06,
      "loss": 0.2273,
      "step": 19566
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.19095781986684704,
      "learning_rate": 7.305541428386169e-06,
      "loss": 0.0763,
      "step": 19567
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.35300009344325267,
      "learning_rate": 7.3045862424045745e-06,
      "loss": 0.2762,
      "step": 19568
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.49349828130639806,
      "learning_rate": 7.303631082941349e-06,
      "loss": 0.2035,
      "step": 19569
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.823081932900886,
      "learning_rate": 7.302675950005888e-06,
      "loss": 0.4792,
      "step": 19570
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0588776417827046,
      "learning_rate": 7.301720843607585e-06,
      "loss": 0.355,
      "step": 19571
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3739069955610148,
      "learning_rate": 7.300765763755844e-06,
      "loss": 0.2846,
      "step": 19572
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.25320608330838346,
      "learning_rate": 7.2998107104600575e-06,
      "loss": 0.2327,
      "step": 19573
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3350744235220213,
      "learning_rate": 7.29885568372962e-06,
      "loss": 0.8214,
      "step": 19574
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0135758015707774,
      "learning_rate": 7.297900683573931e-06,
      "loss": 0.2563,
      "step": 19575
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.427663454090828,
      "learning_rate": 7.29694571000238e-06,
      "loss": 0.2788,
      "step": 19576
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27853188870532347,
      "learning_rate": 7.2959907630243706e-06,
      "loss": 0.1695,
      "step": 19577
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.41088896742780934,
      "learning_rate": 7.295035842649294e-06,
      "loss": 0.3117,
      "step": 19578
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39132474819223495,
      "learning_rate": 7.29408094888654e-06,
      "loss": 0.2495,
      "step": 19579
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.22174508137213322,
      "learning_rate": 7.293126081745514e-06,
      "loss": 0.098,
      "step": 19580
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3887307725624815,
      "learning_rate": 7.292171241235603e-06,
      "loss": 0.27,
      "step": 19581
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4871659734042957,
      "learning_rate": 7.291216427366199e-06,
      "loss": 0.2607,
      "step": 19582
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3059859764112502,
      "learning_rate": 7.290261640146703e-06,
      "loss": 0.6261,
      "step": 19583
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2952200702333237,
      "learning_rate": 7.289306879586504e-06,
      "loss": 0.2349,
      "step": 19584
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6950286861340733,
      "learning_rate": 7.288352145694992e-06,
      "loss": 0.37,
      "step": 19585
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39279472473250643,
      "learning_rate": 7.287397438481567e-06,
      "loss": 0.1924,
      "step": 19586
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4175590205108874,
      "learning_rate": 7.286442757955621e-06,
      "loss": 0.3376,
      "step": 19587
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.22760139211017857,
      "learning_rate": 7.285488104126539e-06,
      "loss": 0.1142,
      "step": 19588
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.45517692233047735,
      "learning_rate": 7.2845334770037215e-06,
      "loss": 0.3105,
      "step": 19589
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31777079788236345,
      "learning_rate": 7.283578876596558e-06,
      "loss": 0.2021,
      "step": 19590
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4695283268326836,
      "learning_rate": 7.282624302914434e-06,
      "loss": 0.2702,
      "step": 19591
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9485246727248129,
      "learning_rate": 7.2816697559667515e-06,
      "loss": 0.53,
      "step": 19592
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6989847115012583,
      "learning_rate": 7.280715235762895e-06,
      "loss": 0.2991,
      "step": 19593
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7589876945122128,
      "learning_rate": 7.2797607423122564e-06,
      "loss": 0.5262,
      "step": 19594
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.38431752071616776,
      "learning_rate": 7.2788062756242265e-06,
      "loss": 0.2047,
      "step": 19595
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3499006363823945,
      "learning_rate": 7.277851835708198e-06,
      "loss": 0.3277,
      "step": 19596
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2053628731126291,
      "learning_rate": 7.276897422573553e-06,
      "loss": 0.1394,
      "step": 19597
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3358619836794234,
      "learning_rate": 7.275943036229692e-06,
      "loss": 0.6811,
      "step": 19598
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27895977988194554,
      "learning_rate": 7.274988676686e-06,
      "loss": 0.1561,
      "step": 19599
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.402888964763717,
      "learning_rate": 7.274034343951863e-06,
      "loss": 0.309,
      "step": 19600
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4378219424259424,
      "learning_rate": 7.273080038036677e-06,
      "loss": 0.2696,
      "step": 19601
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.49084276989406495,
      "learning_rate": 7.272125758949822e-06,
      "loss": 0.3478,
      "step": 19602
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6782666154731953,
      "learning_rate": 7.271171506700695e-06,
      "loss": 0.3253,
      "step": 19603
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40321944142729066,
      "learning_rate": 7.270217281298681e-06,
      "loss": 0.2442,
      "step": 19604
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4289422419350284,
      "learning_rate": 7.269263082753165e-06,
      "loss": 0.2347,
      "step": 19605
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21180196532934967,
      "learning_rate": 7.2683089110735404e-06,
      "loss": 0.0682,
      "step": 19606
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31169556086600114,
      "learning_rate": 7.26735476626919e-06,
      "loss": 0.2571,
      "step": 19607
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28477061740498427,
      "learning_rate": 7.266400648349501e-06,
      "loss": 0.1878,
      "step": 19608
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3424972368990246,
      "learning_rate": 7.265446557323863e-06,
      "loss": 0.2724,
      "step": 19609
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8724240853330146,
      "learning_rate": 7.2644924932016636e-06,
      "loss": 0.2836,
      "step": 19610
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7500541726169334,
      "learning_rate": 7.263538455992282e-06,
      "loss": 0.4152,
      "step": 19611
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4989489819738347,
      "learning_rate": 7.262584445705114e-06,
      "loss": 0.2579,
      "step": 19612
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.40928255918435025,
      "learning_rate": 7.2616304623495415e-06,
      "loss": 0.2485,
      "step": 19613
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31659168049885505,
      "learning_rate": 7.260676505934947e-06,
      "loss": 0.2225,
      "step": 19614
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.48228253419784467,
      "learning_rate": 7.25972257647072e-06,
      "loss": 0.3525,
      "step": 19615
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.15962635947943074,
      "learning_rate": 7.258768673966244e-06,
      "loss": 0.0951,
      "step": 19616
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8596198746694965,
      "learning_rate": 7.2578147984309e-06,
      "loss": 0.4165,
      "step": 19617
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.30208161525374316,
      "learning_rate": 7.256860949874081e-06,
      "loss": 0.1896,
      "step": 19618
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.943814150802805,
      "learning_rate": 7.255907128305166e-06,
      "loss": 0.3232,
      "step": 19619
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3131698664975687,
      "learning_rate": 7.254953333733537e-06,
      "loss": 0.2726,
      "step": 19620
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6459280889083096,
      "learning_rate": 7.253999566168584e-06,
      "loss": 0.2902,
      "step": 19621
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6866514956425189,
      "learning_rate": 7.2530458256196854e-06,
      "loss": 0.3325,
      "step": 19622
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.340540576719395,
      "learning_rate": 7.252092112096223e-06,
      "loss": 0.2119,
      "step": 19623
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.48533997343713836,
      "learning_rate": 7.251138425607584e-06,
      "loss": 0.2768,
      "step": 19624
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4138875084225591,
      "learning_rate": 7.2501847661631534e-06,
      "loss": 0.245,
      "step": 19625
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4408609090664915,
      "learning_rate": 7.249231133772305e-06,
      "loss": 0.3359,
      "step": 19626
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.28258091119455697,
      "learning_rate": 7.248277528444429e-06,
      "loss": 0.18,
      "step": 19627
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4922114541964376,
      "learning_rate": 7.2473239501889e-06,
      "loss": 0.2907,
      "step": 19628
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1291038506648845,
      "learning_rate": 7.246370399015108e-06,
      "loss": 0.3411,
      "step": 19629
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0216281458090768,
      "learning_rate": 7.245416874932429e-06,
      "loss": 0.5117,
      "step": 19630
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34376854408302626,
      "learning_rate": 7.244463377950244e-06,
      "loss": 0.1741,
      "step": 19631
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2625576307488079,
      "learning_rate": 7.243509908077937e-06,
      "loss": 0.2261,
      "step": 19632
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4154019584947473,
      "learning_rate": 7.242556465324886e-06,
      "loss": 0.5989,
      "step": 19633
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23186739057354014,
      "learning_rate": 7.241603049700467e-06,
      "loss": 0.1323,
      "step": 19634
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.883326085220532,
      "learning_rate": 7.240649661214071e-06,
      "loss": 0.5659,
      "step": 19635
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31924626819587826,
      "learning_rate": 7.239696299875071e-06,
      "loss": 0.2046,
      "step": 19636
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9765515051309803,
      "learning_rate": 7.238742965692844e-06,
      "loss": 0.4728,
      "step": 19637
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4030999172094517,
      "learning_rate": 7.237789658676775e-06,
      "loss": 0.2609,
      "step": 19638
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5162165165169439,
      "learning_rate": 7.23683637883624e-06,
      "loss": 0.3218,
      "step": 19639
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3141519510339476,
      "learning_rate": 7.235883126180614e-06,
      "loss": 0.0669,
      "step": 19640
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.354597635079025,
      "learning_rate": 7.234929900719283e-06,
      "loss": 0.2684,
      "step": 19641
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0164344086508776,
      "learning_rate": 7.233976702461623e-06,
      "loss": 0.2093,
      "step": 19642
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3074976522280462,
      "learning_rate": 7.233023531417007e-06,
      "loss": 0.2572,
      "step": 19643
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.37616251416812857,
      "learning_rate": 7.232070387594819e-06,
      "loss": 0.2088,
      "step": 19644
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2866075455983489,
      "learning_rate": 7.231117271004431e-06,
      "loss": 0.1836,
      "step": 19645
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6436634058683299,
      "learning_rate": 7.23016418165522e-06,
      "loss": 0.4674,
      "step": 19646
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21277176025864736,
      "learning_rate": 7.22921111955657e-06,
      "loss": 0.1669,
      "step": 19647
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1092156651624874,
      "learning_rate": 7.228258084717851e-06,
      "loss": 0.445,
      "step": 19648
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2863032352423141,
      "learning_rate": 7.227305077148437e-06,
      "loss": 0.1747,
      "step": 19649
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2806217345193171,
      "learning_rate": 7.226352096857714e-06,
      "loss": 0.2886,
      "step": 19650
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4500201919697466,
      "learning_rate": 7.225399143855048e-06,
      "loss": 0.1955,
      "step": 19651
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.8277419797810939,
      "learning_rate": 7.22444621814982e-06,
      "loss": 0.8479,
      "step": 19652
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5449483279284771,
      "learning_rate": 7.223493319751404e-06,
      "loss": 0.357,
      "step": 19653
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3885524598839338,
      "learning_rate": 7.2225404486691705e-06,
      "loss": 0.2719,
      "step": 19654
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.29667773076515247,
      "learning_rate": 7.221587604912501e-06,
      "loss": 0.2653,
      "step": 19655
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4574866159636639,
      "learning_rate": 7.220634788490768e-06,
      "loss": 0.3292,
      "step": 19656
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1681751413532616,
      "learning_rate": 7.219681999413341e-06,
      "loss": 0.0976,
      "step": 19657
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5388954420658707,
      "learning_rate": 7.218729237689601e-06,
      "loss": 0.1197,
      "step": 19658
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39066036944078464,
      "learning_rate": 7.217776503328918e-06,
      "loss": 0.2953,
      "step": 19659
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.5267873164230357,
      "learning_rate": 7.216823796340661e-06,
      "loss": 0.2191,
      "step": 19660
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42737488414717933,
      "learning_rate": 7.21587111673421e-06,
      "loss": 0.2988,
      "step": 19661
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3039683490426565,
      "learning_rate": 7.214918464518937e-06,
      "loss": 0.2394,
      "step": 19662
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.33681657011283744,
      "learning_rate": 7.213965839704209e-06,
      "loss": 0.2669,
      "step": 19663
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6053481200387639,
      "learning_rate": 7.213013242299405e-06,
      "loss": 0.3026,
      "step": 19664
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.377393018350954,
      "learning_rate": 7.212060672313892e-06,
      "loss": 0.812,
      "step": 19665
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.22399722755517326,
      "learning_rate": 7.21110812975704e-06,
      "loss": 0.1554,
      "step": 19666
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3568335225270321,
      "learning_rate": 7.210155614638229e-06,
      "loss": 0.2745,
      "step": 19667
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3485168437827255,
      "learning_rate": 7.209203126966824e-06,
      "loss": 0.2197,
      "step": 19668
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.3103497729623463,
      "learning_rate": 7.208250666752195e-06,
      "loss": 0.6395,
      "step": 19669
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6475812018019004,
      "learning_rate": 7.2072982340037165e-06,
      "loss": 0.3064,
      "step": 19670
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5287405405508849,
      "learning_rate": 7.206345828730759e-06,
      "loss": 0.3115,
      "step": 19671
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6200057355485071,
      "learning_rate": 7.2053934509426835e-06,
      "loss": 0.2617,
      "step": 19672
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2582746929414032,
      "learning_rate": 7.2044411006488726e-06,
      "loss": 0.2264,
      "step": 19673
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.46622030381478874,
      "learning_rate": 7.203488777858689e-06,
      "loss": 0.3649,
      "step": 19674
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.1524066996103642,
      "learning_rate": 7.202536482581502e-06,
      "loss": 0.0643,
      "step": 19675
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5807593073323214,
      "learning_rate": 7.2015842148266825e-06,
      "loss": 0.2519,
      "step": 19676
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.29748618241271235,
      "learning_rate": 7.200631974603594e-06,
      "loss": 0.1865,
      "step": 19677
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.1941298255113912,
      "learning_rate": 7.1996797619216155e-06,
      "loss": 0.5562,
      "step": 19678
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3751079440933585,
      "learning_rate": 7.1987275767901075e-06,
      "loss": 0.2642,
      "step": 19679
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.42025561447287624,
      "learning_rate": 7.197775419218434e-06,
      "loss": 0.3221,
      "step": 19680
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3033053180357689,
      "learning_rate": 7.196823289215974e-06,
      "loss": 0.193,
      "step": 19681
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5187990643628396,
      "learning_rate": 7.195871186792088e-06,
      "loss": 0.3448,
      "step": 19682
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8949964273736489,
      "learning_rate": 7.1949191119561425e-06,
      "loss": 0.4714,
      "step": 19683
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.20466234254641733,
      "learning_rate": 7.193967064717506e-06,
      "loss": 0.1388,
      "step": 19684
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4202625403774964,
      "learning_rate": 7.193015045085548e-06,
      "loss": 0.2908,
      "step": 19685
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2927963658639678,
      "learning_rate": 7.192063053069624e-06,
      "loss": 0.1958,
      "step": 19686
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6016700856632768,
      "learning_rate": 7.191111088679115e-06,
      "loss": 0.8025,
      "step": 19687
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5653159424058657,
      "learning_rate": 7.190159151923377e-06,
      "loss": 0.3078,
      "step": 19688
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8049684462532278,
      "learning_rate": 7.189207242811776e-06,
      "loss": 0.4946,
      "step": 19689
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.27453071440730437,
      "learning_rate": 7.1882553613536824e-06,
      "loss": 0.1799,
      "step": 19690
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6736821108574456,
      "learning_rate": 7.187303507558456e-06,
      "loss": 0.8162,
      "step": 19691
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2946817594753922,
      "learning_rate": 7.18635168143546e-06,
      "loss": 0.2496,
      "step": 19692
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4587563833506952,
      "learning_rate": 7.185399882994066e-06,
      "loss": 0.2348,
      "step": 19693
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2376024712464846,
      "learning_rate": 7.184448112243636e-06,
      "loss": 0.0977,
      "step": 19694
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4960079066788983,
      "learning_rate": 7.183496369193528e-06,
      "loss": 0.3332,
      "step": 19695
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39934444234464644,
      "learning_rate": 7.182544653853111e-06,
      "loss": 0.248,
      "step": 19696
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2734831095630275,
      "learning_rate": 7.1815929662317475e-06,
      "loss": 0.2601,
      "step": 19697
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8889292831844066,
      "learning_rate": 7.180641306338796e-06,
      "loss": 0.5199,
      "step": 19698
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.31283345753861413,
      "learning_rate": 7.179689674183625e-06,
      "loss": 0.0726,
      "step": 19699
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36960462387670967,
      "learning_rate": 7.178738069775597e-06,
      "loss": 0.2837,
      "step": 19700
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.128692882741308,
      "learning_rate": 7.17778649312407e-06,
      "loss": 0.3727,
      "step": 19701
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4511637713137884,
      "learning_rate": 7.17683494423841e-06,
      "loss": 0.2262,
      "step": 19702
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2466864796925411,
      "learning_rate": 7.175883423127973e-06,
      "loss": 0.1896,
      "step": 19703
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.36078077619360077,
      "learning_rate": 7.174931929802127e-06,
      "loss": 0.2389,
      "step": 19704
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3949402813828142,
      "learning_rate": 7.173980464270231e-06,
      "loss": 0.2702,
      "step": 19705
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7010243776716905,
      "learning_rate": 7.1730290265416435e-06,
      "loss": 0.3953,
      "step": 19706
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0342870565239004,
      "learning_rate": 7.172077616625727e-06,
      "loss": 0.2856,
      "step": 19707
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8243535764693427,
      "learning_rate": 7.171126234531844e-06,
      "loss": 0.3962,
      "step": 19708
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.22825917307123314,
      "learning_rate": 7.170174880269347e-06,
      "loss": 0.2237,
      "step": 19709
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4282061523214415,
      "learning_rate": 7.169223553847604e-06,
      "loss": 0.7959,
      "step": 19710
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6943979588825013,
      "learning_rate": 7.168272255275972e-06,
      "loss": 0.0141,
      "step": 19711
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21174369871286702,
      "learning_rate": 7.167320984563804e-06,
      "loss": 0.1185,
      "step": 19712
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3344648645486202,
      "learning_rate": 7.166369741720469e-06,
      "loss": 0.2495,
      "step": 19713
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7415453784459144,
      "learning_rate": 7.1654185267553235e-06,
      "loss": 0.3049,
      "step": 19714
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32814252723524234,
      "learning_rate": 7.164467339677716e-06,
      "loss": 0.3082,
      "step": 19715
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.451884126006741,
      "learning_rate": 7.163516180497017e-06,
      "loss": 0.271,
      "step": 19716
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6068189238703715,
      "learning_rate": 7.162565049222579e-06,
      "loss": 0.2924,
      "step": 19717
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.34778403234256666,
      "learning_rate": 7.161613945863754e-06,
      "loss": 0.2033,
      "step": 19718
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4143890070313874,
      "learning_rate": 7.16066287042991e-06,
      "loss": 0.7427,
      "step": 19719
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.427105394520895,
      "learning_rate": 7.1597118229304e-06,
      "loss": 0.213,
      "step": 19720
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3075910152413832,
      "learning_rate": 7.158760803374576e-06,
      "loss": 0.2462,
      "step": 19721
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.39357347298155554,
      "learning_rate": 7.157809811771801e-06,
      "loss": 0.1759,
      "step": 19722
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6557567976682909,
      "learning_rate": 7.156858848131427e-06,
      "loss": 0.3596,
      "step": 19723
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5869626366789754,
      "learning_rate": 7.15590791246281e-06,
      "loss": 0.2073,
      "step": 19724
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0030905977285043,
      "learning_rate": 7.154957004775308e-06,
      "loss": 0.4609,
      "step": 19725
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3720850226627342,
      "learning_rate": 7.1540061250782766e-06,
      "loss": 0.2699,
      "step": 19726
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.29238968169448143,
      "learning_rate": 7.153055273381067e-06,
      "loss": 0.2135,
      "step": 19727
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0446406188395507,
      "learning_rate": 7.152104449693038e-06,
      "loss": 0.5983,
      "step": 19728
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0190609735061245,
      "learning_rate": 7.151153654023539e-06,
      "loss": 0.2343,
      "step": 19729
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.7788552034776112,
      "learning_rate": 7.150202886381932e-06,
      "loss": 0.4071,
      "step": 19730
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4202831636173462,
      "learning_rate": 7.149252146777565e-06,
      "loss": 0.1971,
      "step": 19731
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4042254066554231,
      "learning_rate": 7.148301435219793e-06,
      "loss": 0.2336,
      "step": 19732
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.23532426688206234,
      "learning_rate": 7.1473507517179715e-06,
      "loss": 0.2073,
      "step": 19733
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.6543490684748365,
      "learning_rate": 7.146400096281453e-06,
      "loss": 0.8691,
      "step": 19734
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5039127287623706,
      "learning_rate": 7.145449468919584e-06,
      "loss": 0.2752,
      "step": 19735
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3813127328556534,
      "learning_rate": 7.144498869641727e-06,
      "loss": 0.3049,
      "step": 19736
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0274590409713336,
      "learning_rate": 7.14354829845723e-06,
      "loss": 0.3429,
      "step": 19737
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4540114316668217,
      "learning_rate": 7.142597755375442e-06,
      "loss": 0.2563,
      "step": 19738
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4080814561827057,
      "learning_rate": 7.14164724040572e-06,
      "loss": 0.3081,
      "step": 19739
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2827561146608422,
      "learning_rate": 7.1406967535574125e-06,
      "loss": 0.1728,
      "step": 19740
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0295035404732593,
      "learning_rate": 7.139746294839866e-06,
      "loss": 0.5117,
      "step": 19741
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21975142571673498,
      "learning_rate": 7.1387958642624425e-06,
      "loss": 0.0867,
      "step": 19742
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.4376494162956577,
      "learning_rate": 7.137845461834487e-06,
      "loss": 0.8076,
      "step": 19743
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.2813794503022017,
      "learning_rate": 7.136895087565346e-06,
      "loss": 0.1949,
      "step": 19744
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.6404094443047154,
      "learning_rate": 7.135944741464376e-06,
      "loss": 0.3219,
      "step": 19745
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.32367461365193584,
      "learning_rate": 7.134994423540924e-06,
      "loss": 0.242,
      "step": 19746
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.837976421879707,
      "learning_rate": 7.134044133804336e-06,
      "loss": 0.4648,
      "step": 19747
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5597439012441594,
      "learning_rate": 7.133093872263968e-06,
      "loss": 0.2994,
      "step": 19748
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.9327546651071263,
      "learning_rate": 7.132143638929166e-06,
      "loss": 0.5297,
      "step": 19749
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3173272935397082,
      "learning_rate": 7.131193433809276e-06,
      "loss": 0.1935,
      "step": 19750
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.21853544666694297,
      "learning_rate": 7.130243256913651e-06,
      "loss": 0.1909,
      "step": 19751
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.4063358583728475,
      "learning_rate": 7.129293108251638e-06,
      "loss": 0.2591,
      "step": 19752
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.3558301966028391,
      "learning_rate": 7.128342987832582e-06,
      "loss": 0.1499,
      "step": 19753
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.5577328784748897,
      "learning_rate": 7.127392895665835e-06,
      "loss": 0.252,
      "step": 19754
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.8377345662260205,
      "learning_rate": 7.126442831760736e-06,
      "loss": 0.292,
      "step": 19755
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30910926624600377,
      "learning_rate": 7.125492796126644e-06,
      "loss": 0.2943,
      "step": 19756
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7654968313485099,
      "learning_rate": 7.124542788772899e-06,
      "loss": 0.3179,
      "step": 19757
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6609266244848586,
      "learning_rate": 7.123592809708846e-06,
      "loss": 0.3783,
      "step": 19758
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3099033248520551,
      "learning_rate": 7.122642858943836e-06,
      "loss": 0.2008,
      "step": 19759
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24684894234959964,
      "learning_rate": 7.121692936487212e-06,
      "loss": 0.1518,
      "step": 19760
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8616438760983293,
      "learning_rate": 7.120743042348315e-06,
      "loss": 0.4708,
      "step": 19761
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33783446088331925,
      "learning_rate": 7.1197931765365e-06,
      "loss": 0.3036,
      "step": 19762
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2590132700330029,
      "learning_rate": 7.118843339061108e-06,
      "loss": 0.1701,
      "step": 19763
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9314189276496964,
      "learning_rate": 7.117893529931481e-06,
      "loss": 0.3166,
      "step": 19764
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.943804546551888,
      "learning_rate": 7.1169437491569685e-06,
      "loss": 0.4126,
      "step": 19765
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.909782307569593,
      "learning_rate": 7.11599399674691e-06,
      "loss": 0.2782,
      "step": 19766
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6627566274743995,
      "learning_rate": 7.11504427271065e-06,
      "loss": 0.3725,
      "step": 19767
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.31877311367683414,
      "learning_rate": 7.114094577057535e-06,
      "loss": 0.2054,
      "step": 19768
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3305429365858679,
      "learning_rate": 7.11314490979691e-06,
      "loss": 0.3052,
      "step": 19769
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19067693054697063,
      "learning_rate": 7.112195270938111e-06,
      "loss": 0.1153,
      "step": 19770
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8592121321357404,
      "learning_rate": 7.111245660490489e-06,
      "loss": 0.4534,
      "step": 19771
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4257023865317895,
      "learning_rate": 7.110296078463381e-06,
      "loss": 0.186,
      "step": 19772
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3675133992701405,
      "learning_rate": 7.109346524866129e-06,
      "loss": 0.2685,
      "step": 19773
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44116137573140285,
      "learning_rate": 7.108396999708079e-06,
      "loss": 0.2774,
      "step": 19774
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0051164024566468,
      "learning_rate": 7.107447502998573e-06,
      "loss": 0.4598,
      "step": 19775
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6883583802914311,
      "learning_rate": 7.106498034746946e-06,
      "loss": 0.1245,
      "step": 19776
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29762988485184366,
      "learning_rate": 7.105548594962546e-06,
      "loss": 0.1957,
      "step": 19777
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44066472136368295,
      "learning_rate": 7.104599183654708e-06,
      "loss": 0.2311,
      "step": 19778
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3672316091365868,
      "learning_rate": 7.103649800832779e-06,
      "loss": 0.2756,
      "step": 19779
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4809135974440553,
      "learning_rate": 7.102700446506097e-06,
      "loss": 0.344,
      "step": 19780
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29828168382858866,
      "learning_rate": 7.101751120683997e-06,
      "loss": 0.1776,
      "step": 19781
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.414163770877297,
      "learning_rate": 7.100801823375827e-06,
      "loss": 0.3369,
      "step": 19782
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3076359457746064,
      "learning_rate": 7.099852554590923e-06,
      "loss": 0.2011,
      "step": 19783
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6650605737289885,
      "learning_rate": 7.098903314338623e-06,
      "loss": 0.836,
      "step": 19784
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24476268589500977,
      "learning_rate": 7.0979541026282674e-06,
      "loss": 0.0749,
      "step": 19785
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35025502972400524,
      "learning_rate": 7.097004919469195e-06,
      "loss": 0.2762,
      "step": 19786
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3172646790596658,
      "learning_rate": 7.096055764870737e-06,
      "loss": 0.2414,
      "step": 19787
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.507185122536014,
      "learning_rate": 7.0951066388422444e-06,
      "loss": 0.806,
      "step": 19788
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5884172642635308,
      "learning_rate": 7.094157541393048e-06,
      "loss": 0.3413,
      "step": 19789
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2765690835481292,
      "learning_rate": 7.093208472532484e-06,
      "loss": 0.1871,
      "step": 19790
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5920318466718175,
      "learning_rate": 7.092259432269893e-06,
      "loss": 0.4024,
      "step": 19791
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32433879503847984,
      "learning_rate": 7.091310420614611e-06,
      "loss": 0.2502,
      "step": 19792
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.31523542388310044,
      "learning_rate": 7.09036143757597e-06,
      "loss": 0.1862,
      "step": 19793
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28134616224484416,
      "learning_rate": 7.089412483163314e-06,
      "loss": 0.0666,
      "step": 19794
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38948497962798656,
      "learning_rate": 7.088463557385976e-06,
      "loss": 0.3107,
      "step": 19795
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0508616551996426,
      "learning_rate": 7.08751466025329e-06,
      "loss": 0.2242,
      "step": 19796
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.279138911120418,
      "learning_rate": 7.086565791774596e-06,
      "loss": 0.7454,
      "step": 19797
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3105180305404135,
      "learning_rate": 7.085616951959225e-06,
      "loss": 0.2578,
      "step": 19798
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6278178783484131,
      "learning_rate": 7.084668140816509e-06,
      "loss": 0.3337,
      "step": 19799
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3365928282730262,
      "learning_rate": 7.083719358355792e-06,
      "loss": 0.2256,
      "step": 19800
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.826291003385204,
      "learning_rate": 7.082770604586402e-06,
      "loss": 0.5633,
      "step": 19801
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.19653428886536492,
      "learning_rate": 7.081821879517674e-06,
      "loss": 0.0833,
      "step": 19802
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33593631484627257,
      "learning_rate": 7.080873183158945e-06,
      "loss": 0.0617,
      "step": 19803
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.375729438660196,
      "learning_rate": 7.07992451551954e-06,
      "loss": 0.2848,
      "step": 19804
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2763145433149354,
      "learning_rate": 7.0789758766088046e-06,
      "loss": 0.2171,
      "step": 19805
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7686522815044187,
      "learning_rate": 7.078027266436066e-06,
      "loss": 0.4875,
      "step": 19806
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.751544149173503,
      "learning_rate": 7.077078685010654e-06,
      "loss": 0.3112,
      "step": 19807
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5230142326737459,
      "learning_rate": 7.076130132341905e-06,
      "loss": 0.331,
      "step": 19808
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35171702432648,
      "learning_rate": 7.07518160843915e-06,
      "loss": 0.2334,
      "step": 19809
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33395814717853506,
      "learning_rate": 7.074233113311717e-06,
      "loss": 0.2628,
      "step": 19810
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.15719169811152772,
      "learning_rate": 7.073284646968946e-06,
      "loss": 0.0712,
      "step": 19811
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.29629324046568,
      "learning_rate": 7.0723362094201635e-06,
      "loss": 0.6588,
      "step": 19812
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28165829283884297,
      "learning_rate": 7.071387800674695e-06,
      "loss": 0.1877,
      "step": 19813
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.692403332107986,
      "learning_rate": 7.0704394207418835e-06,
      "loss": 0.7007,
      "step": 19814
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7842352470603479,
      "learning_rate": 7.069491069631052e-06,
      "loss": 0.3121,
      "step": 19815
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30253251064227255,
      "learning_rate": 7.0685427473515275e-06,
      "loss": 0.2449,
      "step": 19816
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36351076256418335,
      "learning_rate": 7.0675944539126475e-06,
      "loss": 0.2309,
      "step": 19817
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40724786638065774,
      "learning_rate": 7.0666461893237385e-06,
      "loss": 0.2363,
      "step": 19818
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5112143717173214,
      "learning_rate": 7.065697953594125e-06,
      "loss": 0.7952,
      "step": 19819
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.22463319149949393,
      "learning_rate": 7.064749746733145e-06,
      "loss": 0.088,
      "step": 19820
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.42722697721440456,
      "learning_rate": 7.063801568750123e-06,
      "loss": 0.2667,
      "step": 19821
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2608937456635776,
      "learning_rate": 7.062853419654385e-06,
      "loss": 0.1779,
      "step": 19822
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.49428752666466796,
      "learning_rate": 7.061905299455263e-06,
      "loss": 0.3345,
      "step": 19823
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6892564478980215,
      "learning_rate": 7.0609572081620844e-06,
      "loss": 0.3109,
      "step": 19824
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9920333482140075,
      "learning_rate": 7.060009145784171e-06,
      "loss": 0.4487,
      "step": 19825
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3978843991528419,
      "learning_rate": 7.05906111233086e-06,
      "loss": 0.15,
      "step": 19826
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3330891384993815,
      "learning_rate": 7.058113107811473e-06,
      "loss": 0.2612,
      "step": 19827
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3169715187033088,
      "learning_rate": 7.0571651322353355e-06,
      "loss": 0.2517,
      "step": 19828
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1630913732026783,
      "learning_rate": 7.056217185611776e-06,
      "loss": 0.0669,
      "step": 19829
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3475154661214752,
      "learning_rate": 7.055269267950118e-06,
      "loss": 0.5743,
      "step": 19830
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29221155205279875,
      "learning_rate": 7.0543213792596945e-06,
      "loss": 0.186,
      "step": 19831
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8509719689126246,
      "learning_rate": 7.0533735195498255e-06,
      "loss": 0.4681,
      "step": 19832
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32230822515954244,
      "learning_rate": 7.052425688829836e-06,
      "loss": 0.2471,
      "step": 19833
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9999238001250198,
      "learning_rate": 7.051477887109053e-06,
      "loss": 0.5239,
      "step": 19834
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3097459887758646,
      "learning_rate": 7.050530114396803e-06,
      "loss": 0.1534,
      "step": 19835
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36425989226903693,
      "learning_rate": 7.049582370702402e-06,
      "loss": 0.2923,
      "step": 19836
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9234109042772893,
      "learning_rate": 7.048634656035186e-06,
      "loss": 0.2463,
      "step": 19837
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.48019694913011224,
      "learning_rate": 7.047686970404473e-06,
      "loss": 0.2544,
      "step": 19838
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.23167740219504016,
      "learning_rate": 7.046739313819584e-06,
      "loss": 0.1848,
      "step": 19839
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3982091017442549,
      "learning_rate": 7.045791686289849e-06,
      "loss": 0.2581,
      "step": 19840
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5028417856776527,
      "learning_rate": 7.044844087824587e-06,
      "loss": 0.2673,
      "step": 19841
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1209065403916005,
      "learning_rate": 7.043896518433117e-06,
      "loss": 0.3214,
      "step": 19842
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2176217635572266,
      "learning_rate": 7.042948978124769e-06,
      "loss": 0.6246,
      "step": 19843
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2942679260476159,
      "learning_rate": 7.0420014669088635e-06,
      "loss": 0.1586,
      "step": 19844
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3881784792632908,
      "learning_rate": 7.041053984794717e-06,
      "loss": 0.2963,
      "step": 19845
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2965656950866634,
      "learning_rate": 7.04010653179166e-06,
      "loss": 0.2252,
      "step": 19846
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6250098656406773,
      "learning_rate": 7.039159107909006e-06,
      "loss": 0.6965,
      "step": 19847
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.12626312102031612,
      "learning_rate": 7.038211713156075e-06,
      "loss": 0.0725,
      "step": 19848
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3370828844768843,
      "learning_rate": 7.037264347542197e-06,
      "loss": 0.2569,
      "step": 19849
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0024770911620324,
      "learning_rate": 7.036317011076688e-06,
      "loss": 0.3288,
      "step": 19850
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5869135645842566,
      "learning_rate": 7.0353697037688615e-06,
      "loss": 0.3351,
      "step": 19851
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36615984676683494,
      "learning_rate": 7.034422425628048e-06,
      "loss": 0.2417,
      "step": 19852
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8330822014748591,
      "learning_rate": 7.0334751766635625e-06,
      "loss": 0.3589,
      "step": 19853
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2820603909412581,
      "learning_rate": 7.032527956884723e-06,
      "loss": 0.1936,
      "step": 19854
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3426666228154274,
      "learning_rate": 7.0315807663008505e-06,
      "loss": 0.3131,
      "step": 19855
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.27595104683297655,
      "learning_rate": 7.03063360492126e-06,
      "loss": 0.151,
      "step": 19856
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29274117393110627,
      "learning_rate": 7.029686472755277e-06,
      "loss": 0.2203,
      "step": 19857
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5649092493332994,
      "learning_rate": 7.028739369812217e-06,
      "loss": 0.3584,
      "step": 19858
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3748915890072315,
      "learning_rate": 7.027792296101392e-06,
      "loss": 0.251,
      "step": 19859
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.765114179911374,
      "learning_rate": 7.026845251632128e-06,
      "loss": 0.4189,
      "step": 19860
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8198747226353844,
      "learning_rate": 7.02589823641374e-06,
      "loss": 0.4755,
      "step": 19861
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.032381506707072,
      "learning_rate": 7.024951250455538e-06,
      "loss": 0.3984,
      "step": 19862
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30441509716241666,
      "learning_rate": 7.02400429376685e-06,
      "loss": 0.211,
      "step": 19863
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3281727299903682,
      "learning_rate": 7.023057366356986e-06,
      "loss": 0.3123,
      "step": 19864
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.47023315169208785,
      "learning_rate": 7.022110468235261e-06,
      "loss": 0.0192,
      "step": 19865
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6798384991576505,
      "learning_rate": 7.021163599410995e-06,
      "loss": 0.4243,
      "step": 19866
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3241846904366613,
      "learning_rate": 7.0202167598935025e-06,
      "loss": 0.2062,
      "step": 19867
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21145021624732363,
      "learning_rate": 7.019269949692094e-06,
      "loss": 0.1405,
      "step": 19868
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.6861327197397327,
      "learning_rate": 7.01832316881609e-06,
      "loss": 0.7268,
      "step": 19869
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2901983894606929,
      "learning_rate": 7.017376417274806e-06,
      "loss": 0.2475,
      "step": 19870
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8553499770102052,
      "learning_rate": 7.016429695077554e-06,
      "loss": 0.5086,
      "step": 19871
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3029833939114912,
      "learning_rate": 7.015483002233647e-06,
      "loss": 0.193,
      "step": 19872
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3389071389486475,
      "learning_rate": 7.014536338752402e-06,
      "loss": 0.5679,
      "step": 19873
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1722756723173893,
      "learning_rate": 7.013589704643127e-06,
      "loss": 0.2868,
      "step": 19874
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.313232828670023,
      "learning_rate": 7.012643099915142e-06,
      "loss": 0.3117,
      "step": 19875
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41813757349388664,
      "learning_rate": 7.011696524577758e-06,
      "loss": 0.1816,
      "step": 19876
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.37414489133584017,
      "learning_rate": 7.010749978640284e-06,
      "loss": 0.2786,
      "step": 19877
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2312542480626902,
      "learning_rate": 7.009803462112038e-06,
      "loss": 0.1215,
      "step": 19878
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2162802842365958,
      "learning_rate": 7.00885697500233e-06,
      "loss": 0.8162,
      "step": 19879
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6065122976495192,
      "learning_rate": 7.007910517320467e-06,
      "loss": 0.1832,
      "step": 19880
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28963339336290583,
      "learning_rate": 7.006964089075768e-06,
      "loss": 0.1978,
      "step": 19881
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33645306257944396,
      "learning_rate": 7.006017690277538e-06,
      "loss": 0.2842,
      "step": 19882
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6787275960503645,
      "learning_rate": 7.005071320935094e-06,
      "loss": 0.2891,
      "step": 19883
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6668901761282835,
      "learning_rate": 7.004124981057745e-06,
      "loss": 0.468,
      "step": 19884
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4239025676134604,
      "learning_rate": 7.003178670654798e-06,
      "loss": 0.1181,
      "step": 19885
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38647504704906677,
      "learning_rate": 7.002232389735566e-06,
      "loss": 0.2901,
      "step": 19886
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18412519583141193,
      "learning_rate": 7.00128613830936e-06,
      "loss": 0.1586,
      "step": 19887
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4211621815574846,
      "learning_rate": 7.000339916385483e-06,
      "loss": 0.7657,
      "step": 19888
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9777744573213042,
      "learning_rate": 6.999393723973253e-06,
      "loss": 0.4582,
      "step": 19889
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36555929057370967,
      "learning_rate": 6.998447561081975e-06,
      "loss": 0.2418,
      "step": 19890
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.48182791472968817,
      "learning_rate": 6.997501427720954e-06,
      "loss": 0.247,
      "step": 19891
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0647087389650565,
      "learning_rate": 6.996555323899506e-06,
      "loss": 0.5219,
      "step": 19892
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3423277415098846,
      "learning_rate": 6.995609249626934e-06,
      "loss": 0.2505,
      "step": 19893
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6265571087930034,
      "learning_rate": 6.9946632049125425e-06,
      "loss": 0.302,
      "step": 19894
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3385901006166027,
      "learning_rate": 6.993717189765648e-06,
      "loss": 0.2142,
      "step": 19895
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.20896168385523212,
      "learning_rate": 6.992771204195553e-06,
      "loss": 0.0693,
      "step": 19896
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.294524611551259,
      "learning_rate": 6.991825248211562e-06,
      "loss": 0.7397,
      "step": 19897
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24303148944274494,
      "learning_rate": 6.990879321822987e-06,
      "loss": 0.168,
      "step": 19898
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3577807995793001,
      "learning_rate": 6.989933425039131e-06,
      "loss": 0.2657,
      "step": 19899
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.449563664768152,
      "learning_rate": 6.9889875578692955e-06,
      "loss": 0.2449,
      "step": 19900
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0391386582499618,
      "learning_rate": 6.988041720322795e-06,
      "loss": 0.4406,
      "step": 19901
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.03381971199747,
      "learning_rate": 6.987095912408932e-06,
      "loss": 0.3055,
      "step": 19902
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5933661625355064,
      "learning_rate": 6.986150134137008e-06,
      "loss": 0.296,
      "step": 19903
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30307481911589107,
      "learning_rate": 6.985204385516333e-06,
      "loss": 0.2043,
      "step": 19904
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30006748560279994,
      "learning_rate": 6.984258666556205e-06,
      "loss": 0.2267,
      "step": 19905
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3016461641780273,
      "learning_rate": 6.983312977265937e-06,
      "loss": 0.214,
      "step": 19906
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.1590663915010821,
      "learning_rate": 6.982367317654829e-06,
      "loss": 0.6204,
      "step": 19907
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.26609769948192225,
      "learning_rate": 6.981421687732181e-06,
      "loss": 0.0759,
      "step": 19908
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.626316173721487,
      "learning_rate": 6.980476087507304e-06,
      "loss": 0.2716,
      "step": 19909
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2592560408281144,
      "learning_rate": 6.9795305169894946e-06,
      "loss": 0.4623,
      "step": 19910
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2833520152318187,
      "learning_rate": 6.978584976188054e-06,
      "loss": 0.2196,
      "step": 19911
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8755733153511163,
      "learning_rate": 6.977639465112294e-06,
      "loss": 0.4291,
      "step": 19912
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2780792546682049,
      "learning_rate": 6.97669398377151e-06,
      "loss": 0.1946,
      "step": 19913
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38257379268525904,
      "learning_rate": 6.975748532175004e-06,
      "loss": 0.2189,
      "step": 19914
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.929801850438154,
      "learning_rate": 6.974803110332081e-06,
      "loss": 0.4534,
      "step": 19915
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4303483055351731,
      "learning_rate": 6.973857718252041e-06,
      "loss": 0.3282,
      "step": 19916
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.26435677512132183,
      "learning_rate": 6.972912355944179e-06,
      "loss": 0.1966,
      "step": 19917
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5535911973274404,
      "learning_rate": 6.971967023417807e-06,
      "loss": 0.3181,
      "step": 19918
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6886567723919552,
      "learning_rate": 6.971021720682218e-06,
      "loss": 0.2982,
      "step": 19919
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44119870217362883,
      "learning_rate": 6.970076447746712e-06,
      "loss": 0.1664,
      "step": 19920
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6678775976154693,
      "learning_rate": 6.969131204620594e-06,
      "loss": 0.1189,
      "step": 19921
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30495597515961487,
      "learning_rate": 6.968185991313159e-06,
      "loss": 0.191,
      "step": 19922
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.47804594515874305,
      "learning_rate": 6.967240807833707e-06,
      "loss": 0.356,
      "step": 19923
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.23544353019256806,
      "learning_rate": 6.966295654191541e-06,
      "loss": 0.1697,
      "step": 19924
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7754437111476562,
      "learning_rate": 6.965350530395956e-06,
      "loss": 0.5082,
      "step": 19925
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.26191795410784885,
      "learning_rate": 6.964405436456245e-06,
      "loss": 0.1841,
      "step": 19926
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0367925249772563,
      "learning_rate": 6.963460372381718e-06,
      "loss": 0.4946,
      "step": 19927
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.38078073728505263,
      "learning_rate": 6.962515338181667e-06,
      "loss": 0.1937,
      "step": 19928
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3281526136003713,
      "learning_rate": 6.9615703338653876e-06,
      "loss": 0.2974,
      "step": 19929
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3947688896015118,
      "learning_rate": 6.96062535944218e-06,
      "loss": 0.1022,
      "step": 19930
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5489498150453065,
      "learning_rate": 6.959680414921337e-06,
      "loss": 0.3558,
      "step": 19931
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4535439330238148,
      "learning_rate": 6.9587355003121626e-06,
      "loss": 0.205,
      "step": 19932
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3437345781188514,
      "learning_rate": 6.957790615623949e-06,
      "loss": 0.7825,
      "step": 19933
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3109717279540784,
      "learning_rate": 6.95684576086599e-06,
      "loss": 0.2488,
      "step": 19934
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2790527432803417,
      "learning_rate": 6.955900936047586e-06,
      "loss": 0.0671,
      "step": 19935
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3790043093084751,
      "learning_rate": 6.954956141178031e-06,
      "loss": 0.3019,
      "step": 19936
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.23273457803663092,
      "learning_rate": 6.954011376266614e-06,
      "loss": 0.1314,
      "step": 19937
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0964817490660503,
      "learning_rate": 6.953066641322639e-06,
      "loss": 0.5727,
      "step": 19938
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2663086822140148,
      "learning_rate": 6.952121936355398e-06,
      "loss": 0.2395,
      "step": 19939
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3979562053407399,
      "learning_rate": 6.951177261374183e-06,
      "loss": 0.2836,
      "step": 19940
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2358620086157694,
      "learning_rate": 6.950232616388291e-06,
      "loss": 0.2046,
      "step": 19941
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4963610318014406,
      "learning_rate": 6.949288001407013e-06,
      "loss": 0.8215,
      "step": 19942
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5332791653958285,
      "learning_rate": 6.9483434164396406e-06,
      "loss": 0.3113,
      "step": 19943
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5070076985822132,
      "learning_rate": 6.947398861495472e-06,
      "loss": 0.3247,
      "step": 19944
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3666304060871194,
      "learning_rate": 6.946454336583799e-06,
      "loss": 0.2271,
      "step": 19945
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5500755710948986,
      "learning_rate": 6.945509841713911e-06,
      "loss": 0.2843,
      "step": 19946
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2364059753300794,
      "learning_rate": 6.944565376895105e-06,
      "loss": 0.187,
      "step": 19947
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.44063929022740933,
      "learning_rate": 6.943620942136669e-06,
      "loss": 0.068,
      "step": 19948
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3645213541042298,
      "learning_rate": 6.942676537447892e-06,
      "loss": 0.2792,
      "step": 19949
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5502973701409651,
      "learning_rate": 6.9417321628380726e-06,
      "loss": 0.0435,
      "step": 19950
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.797462210503083,
      "learning_rate": 6.9407878183165e-06,
      "loss": 0.52,
      "step": 19951
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3112528130629463,
      "learning_rate": 6.9398435038924584e-06,
      "loss": 0.2627,
      "step": 19952
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.524888619652471,
      "learning_rate": 6.9388992195752484e-06,
      "loss": 0.3095,
      "step": 19953
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3704961345340674,
      "learning_rate": 6.937954965374154e-06,
      "loss": 0.229,
      "step": 19954
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4404914244859306,
      "learning_rate": 6.9370107412984645e-06,
      "loss": 0.86,
      "step": 19955
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24205763277670847,
      "learning_rate": 6.936066547357474e-06,
      "loss": 0.1006,
      "step": 19956
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9581190099057854,
      "learning_rate": 6.935122383560464e-06,
      "loss": 0.424,
      "step": 19957
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35701362219771193,
      "learning_rate": 6.934178249916732e-06,
      "loss": 0.1741,
      "step": 19958
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2897500276749459,
      "learning_rate": 6.933234146435565e-06,
      "loss": 0.2661,
      "step": 19959
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7577963172329675,
      "learning_rate": 6.9322900731262464e-06,
      "loss": 0.3158,
      "step": 19960
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7285737307151242,
      "learning_rate": 6.931346029998069e-06,
      "loss": 0.3513,
      "step": 19961
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.838074580311708,
      "learning_rate": 6.93040201706032e-06,
      "loss": 0.3393,
      "step": 19962
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30722313833314707,
      "learning_rate": 6.929458034322282e-06,
      "loss": 0.2063,
      "step": 19963
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.48153795902982427,
      "learning_rate": 6.928514081793249e-06,
      "loss": 0.3188,
      "step": 19964
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1748600432048073,
      "learning_rate": 6.927570159482507e-06,
      "loss": 0.1277,
      "step": 19965
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.4848705711154029,
      "learning_rate": 6.926626267399337e-06,
      "loss": 0.5865,
      "step": 19966
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3318811716744259,
      "learning_rate": 6.925682405553031e-06,
      "loss": 0.176,
      "step": 19967
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.49584768346395014,
      "learning_rate": 6.924738573952875e-06,
      "loss": 0.3292,
      "step": 19968
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.894056601435129,
      "learning_rate": 6.923794772608147e-06,
      "loss": 0.3381,
      "step": 19969
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4790361919198691,
      "learning_rate": 6.9228510015281435e-06,
      "loss": 0.3572,
      "step": 19970
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3180404014756259,
      "learning_rate": 6.921907260722142e-06,
      "loss": 0.2274,
      "step": 19971
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36989243328420873,
      "learning_rate": 6.920963550199431e-06,
      "loss": 0.243,
      "step": 19972
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7989226001237175,
      "learning_rate": 6.920019869969294e-06,
      "loss": 0.2964,
      "step": 19973
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.14293344749745876,
      "learning_rate": 6.919076220041014e-06,
      "loss": 0.073,
      "step": 19974
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.480492671150001,
      "learning_rate": 6.918132600423874e-06,
      "loss": 0.3203,
      "step": 19975
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28893663475808823,
      "learning_rate": 6.917189011127163e-06,
      "loss": 0.1882,
      "step": 19976
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4560500762113744,
      "learning_rate": 6.91624545216016e-06,
      "loss": 0.3133,
      "step": 19977
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8909630403379813,
      "learning_rate": 6.915301923532148e-06,
      "loss": 0.331,
      "step": 19978
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8624918493032244,
      "learning_rate": 6.914358425252412e-06,
      "loss": 0.5505,
      "step": 19979
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29093067440534076,
      "learning_rate": 6.913414957330233e-06,
      "loss": 0.1575,
      "step": 19980
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5863356658570814,
      "learning_rate": 6.9124715197748895e-06,
      "loss": 0.3731,
      "step": 19981
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3142467267478384,
      "learning_rate": 6.911528112595671e-06,
      "loss": 0.2186,
      "step": 19982
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3056311770721876,
      "learning_rate": 6.910584735801851e-06,
      "loss": 0.2179,
      "step": 19983
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3466564139531571,
      "learning_rate": 6.909641389402719e-06,
      "loss": 0.1581,
      "step": 19984
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.005660978952803,
      "learning_rate": 6.908698073407551e-06,
      "loss": 0.4955,
      "step": 19985
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33514977500785503,
      "learning_rate": 6.907754787825629e-06,
      "loss": 0.226,
      "step": 19986
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7907811906073624,
      "learning_rate": 6.906811532666233e-06,
      "loss": 0.3198,
      "step": 19987
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33517453663777674,
      "learning_rate": 6.9058683079386435e-06,
      "loss": 0.2986,
      "step": 19988
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39500892317942193,
      "learning_rate": 6.904925113652135e-06,
      "loss": 0.1451,
      "step": 19989
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3589164100370635,
      "learning_rate": 6.903981949815997e-06,
      "loss": 0.251,
      "step": 19990
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2555679167852483,
      "learning_rate": 6.9030388164395025e-06,
      "loss": 0.3056,
      "step": 19991
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28873729123269326,
      "learning_rate": 6.902095713531929e-06,
      "loss": 0.1917,
      "step": 19992
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.57592352728965,
      "learning_rate": 6.90115264110256e-06,
      "loss": 0.2882,
      "step": 19993
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3585136309231427,
      "learning_rate": 6.90020959916067e-06,
      "loss": 0.2815,
      "step": 19994
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25102812981630784,
      "learning_rate": 6.899266587715534e-06,
      "loss": 0.2375,
      "step": 19995
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8714919548909574,
      "learning_rate": 6.898323606776438e-06,
      "loss": 0.4245,
      "step": 19996
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9876098279092691,
      "learning_rate": 6.897380656352655e-06,
      "loss": 0.4745,
      "step": 19997
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.5854230400056655,
      "learning_rate": 6.896437736453459e-06,
      "loss": 0.8426,
      "step": 19998
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2746882769841562,
      "learning_rate": 6.895494847088134e-06,
      "loss": 0.1825,
      "step": 19999
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.27764887267003385,
      "learning_rate": 6.89455198826595e-06,
      "loss": 0.2275,
      "step": 20000
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.209185774180766,
      "learning_rate": 6.893609159996182e-06,
      "loss": 0.594,
      "step": 20001
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3569028939759213,
      "learning_rate": 6.892666362288114e-06,
      "loss": 0.1536,
      "step": 20002
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4127792808856874,
      "learning_rate": 6.891723595151016e-06,
      "loss": 0.2829,
      "step": 20003
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.47304405083269235,
      "learning_rate": 6.890780858594161e-06,
      "loss": 0.2437,
      "step": 20004
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8219065311406694,
      "learning_rate": 6.889838152626831e-06,
      "loss": 0.4452,
      "step": 20005
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29320166423618294,
      "learning_rate": 6.888895477258296e-06,
      "loss": 0.2428,
      "step": 20006
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9087086451251603,
      "learning_rate": 6.887952832497827e-06,
      "loss": 0.4859,
      "step": 20007
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35771417074527234,
      "learning_rate": 6.887010218354706e-06,
      "loss": 0.1722,
      "step": 20008
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5089773573897453,
      "learning_rate": 6.8860676348382e-06,
      "loss": 0.3056,
      "step": 20009
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.670190248500642,
      "learning_rate": 6.885125081957589e-06,
      "loss": 0.253,
      "step": 20010
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.40968450864660283,
      "learning_rate": 6.8841825597221405e-06,
      "loss": 0.3326,
      "step": 20011
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33020703746751223,
      "learning_rate": 6.883240068141126e-06,
      "loss": 0.2029,
      "step": 20012
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3078680034518624,
      "learning_rate": 6.882297607223825e-06,
      "loss": 0.2184,
      "step": 20013
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2885236037835709,
      "learning_rate": 6.881355176979506e-06,
      "loss": 0.1659,
      "step": 20014
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.822147752227366,
      "learning_rate": 6.880412777417439e-06,
      "loss": 0.6114,
      "step": 20015
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.8011780307746021,
      "learning_rate": 6.8794704085469e-06,
      "loss": 0.7668,
      "step": 20016
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.28147956978708527,
      "learning_rate": 6.878528070377157e-06,
      "loss": 0.1418,
      "step": 20017
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2669778578150441,
      "learning_rate": 6.877585762917478e-06,
      "loss": 0.2544,
      "step": 20018
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.7085379272515546,
      "learning_rate": 6.876643486177143e-06,
      "loss": 0.2987,
      "step": 20019
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9218984812352057,
      "learning_rate": 6.8757012401654166e-06,
      "loss": 0.5009,
      "step": 20020
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4379190843015482,
      "learning_rate": 6.874759024891564e-06,
      "loss": 0.2077,
      "step": 20021
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25287600324434495,
      "learning_rate": 6.8738168403648665e-06,
      "loss": 0.2038,
      "step": 20022
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.23176278941532782,
      "learning_rate": 6.872874686594586e-06,
      "loss": 0.1604,
      "step": 20023
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4900848691614716,
      "learning_rate": 6.871932563589992e-06,
      "loss": 0.3134,
      "step": 20024
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9521904591633665,
      "learning_rate": 6.870990471360357e-06,
      "loss": 0.5469,
      "step": 20025
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.26109352308325107,
      "learning_rate": 6.870048409914946e-06,
      "loss": 0.1803,
      "step": 20026
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.2232661343142797,
      "learning_rate": 6.869106379263026e-06,
      "loss": 0.5281,
      "step": 20027
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8938702928099712,
      "learning_rate": 6.868164379413873e-06,
      "loss": 0.3146,
      "step": 20028
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.43216902860425344,
      "learning_rate": 6.867222410376747e-06,
      "loss": 0.3489,
      "step": 20029
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30187188768632556,
      "learning_rate": 6.866280472160915e-06,
      "loss": 0.1857,
      "step": 20030
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.36103495925194085,
      "learning_rate": 6.8653385647756525e-06,
      "loss": 0.2734,
      "step": 20031
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.1698938384053415,
      "learning_rate": 6.864396688230217e-06,
      "loss": 0.0678,
      "step": 20032
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.252106761649925,
      "learning_rate": 6.863454842533876e-06,
      "loss": 0.7657,
      "step": 20033
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6000807865754574,
      "learning_rate": 6.862513027695903e-06,
      "loss": 0.1811,
      "step": 20034
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4433264761797396,
      "learning_rate": 6.861571243725555e-06,
      "loss": 0.2792,
      "step": 20035
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2989598066487902,
      "learning_rate": 6.860629490632107e-06,
      "loss": 0.2236,
      "step": 20036
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.500780724719218,
      "learning_rate": 6.859687768424816e-06,
      "loss": 0.3327,
      "step": 20037
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.0446945738637268,
      "learning_rate": 6.858746077112948e-06,
      "loss": 0.3187,
      "step": 20038
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.32413743288651603,
      "learning_rate": 6.857804416705773e-06,
      "loss": 0.0638,
      "step": 20039
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3494004706335789,
      "learning_rate": 6.856862787212553e-06,
      "loss": 0.2682,
      "step": 20040
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25650562302615915,
      "learning_rate": 6.8559211886425466e-06,
      "loss": 0.1693,
      "step": 20041
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.449603001194854,
      "learning_rate": 6.854979621005026e-06,
      "loss": 0.3585,
      "step": 20042
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.18212419186076537,
      "learning_rate": 6.854038084309251e-06,
      "loss": 0.0934,
      "step": 20043
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5738342049315764,
      "learning_rate": 6.8530965785644794e-06,
      "loss": 0.289,
      "step": 20044
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3391712879741754,
      "learning_rate": 6.852155103779985e-06,
      "loss": 0.2489,
      "step": 20045
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.704222497934892,
      "learning_rate": 6.851213659965024e-06,
      "loss": 0.4161,
      "step": 20046
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.41187027442497326,
      "learning_rate": 6.850272247128855e-06,
      "loss": 0.2234,
      "step": 20047
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.48818513795014673,
      "learning_rate": 6.849330865280748e-06,
      "loss": 0.3689,
      "step": 20048
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.25209725447891995,
      "learning_rate": 6.848389514429959e-06,
      "loss": 0.1801,
      "step": 20049
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.45936034604404313,
      "learning_rate": 6.847448194585749e-06,
      "loss": 0.2572,
      "step": 20050
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9162430638801645,
      "learning_rate": 6.846506905757384e-06,
      "loss": 0.4855,
      "step": 20051
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3284457959055673,
      "learning_rate": 6.845565647954121e-06,
      "loss": 0.1766,
      "step": 20052
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4095434738604456,
      "learning_rate": 6.844624421185217e-06,
      "loss": 0.2704,
      "step": 20053
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.24417394868349004,
      "learning_rate": 6.843683225459941e-06,
      "loss": 0.2112,
      "step": 20054
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6342789285801178,
      "learning_rate": 6.842742060787549e-06,
      "loss": 0.4148,
      "step": 20055
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.797290242660339,
      "learning_rate": 6.841800927177296e-06,
      "loss": 0.2766,
      "step": 20056
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8280303525477752,
      "learning_rate": 6.840859824638447e-06,
      "loss": 0.4204,
      "step": 20057
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2993178407143957,
      "learning_rate": 6.839918753180255e-06,
      "loss": 0.2008,
      "step": 20058
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.254576373971881,
      "learning_rate": 6.838977712811985e-06,
      "loss": 0.6596,
      "step": 20059
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.33433171585548865,
      "learning_rate": 6.838036703542894e-06,
      "loss": 0.2749,
      "step": 20060
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.4814480022815188,
      "learning_rate": 6.837095725382234e-06,
      "loss": 0.2581,
      "step": 20061
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.47554865116892103,
      "learning_rate": 6.836154778339271e-06,
      "loss": 0.188,
      "step": 20062
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.42221378521210656,
      "learning_rate": 6.835213862423258e-06,
      "loss": 0.2781,
      "step": 20063
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.3434058759141421,
      "learning_rate": 6.834272977643448e-06,
      "loss": 0.1859,
      "step": 20064
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2866230836943786,
      "learning_rate": 6.833332124009107e-06,
      "loss": 0.2261,
      "step": 20065
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9471827290663828,
      "learning_rate": 6.832391301529485e-06,
      "loss": 0.4645,
      "step": 20066
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.29081042842442234,
      "learning_rate": 6.8314505102138385e-06,
      "loss": 0.1977,
      "step": 20067
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3358405794045405,
      "learning_rate": 6.830509750071427e-06,
      "loss": 0.4673,
      "step": 20068
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6828740710404718,
      "learning_rate": 6.829569021111502e-06,
      "loss": 0.3921,
      "step": 20069
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.9213265472581329,
      "learning_rate": 6.828628323343318e-06,
      "loss": 0.5073,
      "step": 20070
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.30272441685578183,
      "learning_rate": 6.827687656776134e-06,
      "loss": 0.2144,
      "step": 20071
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2958988520524691,
      "learning_rate": 6.826747021419205e-06,
      "loss": 0.2684,
      "step": 20072
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.21629821878216743,
      "learning_rate": 6.82580641728178e-06,
      "loss": 0.1208,
      "step": 20073
    },
    {
      "epoch": 0.61,
      "grad_norm": 1.3292506469841132,
      "learning_rate": 6.824865844373116e-06,
      "loss": 0.6512,
      "step": 20074
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.6373101743005659,
      "learning_rate": 6.823925302702469e-06,
      "loss": 0.0094,
      "step": 20075
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.39932249361662636,
      "learning_rate": 6.822984792279085e-06,
      "loss": 0.2447,
      "step": 20076
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.2942468745524863,
      "learning_rate": 6.822044313112226e-06,
      "loss": 0.2393,
      "step": 20077
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.377880146330611,
      "learning_rate": 6.8211038652111405e-06,
      "loss": 0.2456,
      "step": 20078
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.8018844316318907,
      "learning_rate": 6.8201634485850794e-06,
      "loss": 0.4303,
      "step": 20079
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.5737756526943573,
      "learning_rate": 6.8192230632433e-06,
      "loss": 0.2262,
      "step": 20080
    },
    {
      "epoch": 0.61,
      "grad_norm": 0.35140797438613325,
      "learning_rate": 6.81828270919505e-06,
      "loss": 0.2974,
      "step": 20081
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18666005722912413,
      "learning_rate": 6.817342386449577e-06,
      "loss": 0.0746,
      "step": 20082
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3310136872070776,
      "learning_rate": 6.81640209501614e-06,
      "loss": 0.2959,
      "step": 20083
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6443715713626297,
      "learning_rate": 6.815461834903984e-06,
      "loss": 0.1549,
      "step": 20084
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3299300183634146,
      "learning_rate": 6.814521606122366e-06,
      "loss": 0.2793,
      "step": 20085
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1986402608460158,
      "learning_rate": 6.813581408680534e-06,
      "loss": 0.2804,
      "step": 20086
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8628788721635435,
      "learning_rate": 6.812641242587732e-06,
      "loss": 0.59,
      "step": 20087
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5801418193709241,
      "learning_rate": 6.8117011078532165e-06,
      "loss": 0.2735,
      "step": 20088
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.44480434921676176,
      "learning_rate": 6.810761004486236e-06,
      "loss": 0.2928,
      "step": 20089
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33387451492238357,
      "learning_rate": 6.809820932496034e-06,
      "loss": 0.2062,
      "step": 20090
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5826735915114802,
      "learning_rate": 6.808880891891867e-06,
      "loss": 0.6889,
      "step": 20091
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23539875591036474,
      "learning_rate": 6.80794088268298e-06,
      "loss": 0.1041,
      "step": 20092
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4119853002837455,
      "learning_rate": 6.807000904878618e-06,
      "loss": 0.188,
      "step": 20093
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5946962405301932,
      "learning_rate": 6.806060958488034e-06,
      "loss": 0.3427,
      "step": 20094
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2887840158399286,
      "learning_rate": 6.805121043520474e-06,
      "loss": 0.2453,
      "step": 20095
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5073669537985344,
      "learning_rate": 6.804181159985181e-06,
      "loss": 0.3379,
      "step": 20096
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9268644917309531,
      "learning_rate": 6.8032413078914085e-06,
      "loss": 0.2927,
      "step": 20097
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8203950439306725,
      "learning_rate": 6.802301487248399e-06,
      "loss": 0.5,
      "step": 20098
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2967771544437546,
      "learning_rate": 6.801361698065398e-06,
      "loss": 0.1878,
      "step": 20099
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2918756372980552,
      "learning_rate": 6.800421940351656e-06,
      "loss": 0.1777,
      "step": 20100
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3034529457014193,
      "learning_rate": 6.799482214116415e-06,
      "loss": 0.2231,
      "step": 20101
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4132435955192553,
      "learning_rate": 6.798542519368918e-06,
      "loss": 0.6008,
      "step": 20102
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3218444441169744,
      "learning_rate": 6.797602856118416e-06,
      "loss": 0.1561,
      "step": 20103
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5712463632738604,
      "learning_rate": 6.7966632243741515e-06,
      "loss": 0.3903,
      "step": 20104
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3641808306008212,
      "learning_rate": 6.795723624145366e-06,
      "loss": 0.3091,
      "step": 20105
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5148192427578158,
      "learning_rate": 6.794784055441308e-06,
      "loss": 0.2662,
      "step": 20106
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37210289238915367,
      "learning_rate": 6.79384451827122e-06,
      "loss": 0.28,
      "step": 20107
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2986060771674308,
      "learning_rate": 6.7929050126443386e-06,
      "loss": 0.1961,
      "step": 20108
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8238837164410304,
      "learning_rate": 6.791965538569918e-06,
      "loss": 0.6246,
      "step": 20109
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2626089727060768,
      "learning_rate": 6.791026096057194e-06,
      "loss": 0.0893,
      "step": 20110
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5096765169641855,
      "learning_rate": 6.790086685115414e-06,
      "loss": 0.7833,
      "step": 20111
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47504223122702455,
      "learning_rate": 6.789147305753816e-06,
      "loss": 0.2097,
      "step": 20112
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5706460333292145,
      "learning_rate": 6.78820795798164e-06,
      "loss": 0.3543,
      "step": 20113
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5503422019295002,
      "learning_rate": 6.787268641808135e-06,
      "loss": 0.2753,
      "step": 20114
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.03626379543933,
      "learning_rate": 6.786329357242538e-06,
      "loss": 0.5508,
      "step": 20115
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.255356261532026,
      "learning_rate": 6.785390104294087e-06,
      "loss": 0.1193,
      "step": 20116
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.49301408930194385,
      "learning_rate": 6.78445088297203e-06,
      "loss": 0.2965,
      "step": 20117
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28112864425856215,
      "learning_rate": 6.783511693285603e-06,
      "loss": 0.1568,
      "step": 20118
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2265422992512629,
      "learning_rate": 6.7825725352440404e-06,
      "loss": 0.1805,
      "step": 20119
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.210070670869131,
      "learning_rate": 6.781633408856594e-06,
      "loss": 0.5075,
      "step": 20120
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4261503059942978,
      "learning_rate": 6.780694314132497e-06,
      "loss": 0.1978,
      "step": 20121
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3412786038689464,
      "learning_rate": 6.779755251080984e-06,
      "loss": 0.2896,
      "step": 20122
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7730098384503405,
      "learning_rate": 6.778816219711302e-06,
      "loss": 0.3526,
      "step": 20123
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4960273272125673,
      "learning_rate": 6.777877220032686e-06,
      "loss": 0.3406,
      "step": 20124
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3895708125562716,
      "learning_rate": 6.7769382520543724e-06,
      "loss": 0.2314,
      "step": 20125
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47069318194497467,
      "learning_rate": 6.775999315785603e-06,
      "loss": 0.2429,
      "step": 20126
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47115230824426807,
      "learning_rate": 6.775060411235613e-06,
      "loss": 0.2177,
      "step": 20127
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.25075556655945014,
      "learning_rate": 6.774121538413636e-06,
      "loss": 0.1397,
      "step": 20128
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.944830181247802,
      "learning_rate": 6.7731826973289165e-06,
      "loss": 0.3095,
      "step": 20129
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5408479750065424,
      "learning_rate": 6.772243887990688e-06,
      "loss": 0.3171,
      "step": 20130
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22342305228517373,
      "learning_rate": 6.771305110408183e-06,
      "loss": 0.2369,
      "step": 20131
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8884975685815489,
      "learning_rate": 6.770366364590644e-06,
      "loss": 0.3003,
      "step": 20132
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2405927052535681,
      "learning_rate": 6.769427650547302e-06,
      "loss": 0.8651,
      "step": 20133
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6337970597088928,
      "learning_rate": 6.76848896828739e-06,
      "loss": 0.198,
      "step": 20134
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3398349315454585,
      "learning_rate": 6.767550317820153e-06,
      "loss": 0.2746,
      "step": 20135
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5789060896113968,
      "learning_rate": 6.766611699154815e-06,
      "loss": 0.0278,
      "step": 20136
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35967660866269663,
      "learning_rate": 6.765673112300618e-06,
      "loss": 0.2966,
      "step": 20137
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3366095422657939,
      "learning_rate": 6.764734557266793e-06,
      "loss": 0.1626,
      "step": 20138
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5581233410203096,
      "learning_rate": 6.7637960340625705e-06,
      "loss": 0.2928,
      "step": 20139
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3753902639776391,
      "learning_rate": 6.762857542697191e-06,
      "loss": 0.2057,
      "step": 20140
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.838976131727466,
      "learning_rate": 6.761919083179884e-06,
      "loss": 0.4791,
      "step": 20141
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2796058589854069,
      "learning_rate": 6.760980655519882e-06,
      "loss": 0.2414,
      "step": 20142
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8142164399744661,
      "learning_rate": 6.760042259726418e-06,
      "loss": 0.4556,
      "step": 20143
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.345411121841696,
      "learning_rate": 6.7591038958087255e-06,
      "loss": 0.1736,
      "step": 20144
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4267579689426288,
      "learning_rate": 6.758165563776031e-06,
      "loss": 0.2048,
      "step": 20145
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47650738482690885,
      "learning_rate": 6.757227263637573e-06,
      "loss": 0.2903,
      "step": 20146
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.586142959537507,
      "learning_rate": 6.756288995402582e-06,
      "loss": 0.3028,
      "step": 20147
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3930260114697735,
      "learning_rate": 6.7553507590802855e-06,
      "loss": 0.2579,
      "step": 20148
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24266767541318918,
      "learning_rate": 6.754412554679916e-06,
      "loss": 0.2085,
      "step": 20149
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42353369397373275,
      "learning_rate": 6.7534743822107065e-06,
      "loss": 0.2661,
      "step": 20150
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9246808770325246,
      "learning_rate": 6.752536241681878e-06,
      "loss": 0.5569,
      "step": 20151
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3352687849382145,
      "learning_rate": 6.751598133102671e-06,
      "loss": 0.6827,
      "step": 20152
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29714533414180233,
      "learning_rate": 6.750660056482311e-06,
      "loss": 0.1577,
      "step": 20153
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7819822201371739,
      "learning_rate": 6.749722011830022e-06,
      "loss": 0.3309,
      "step": 20154
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.291587644112363,
      "learning_rate": 6.7487839991550415e-06,
      "loss": 0.2255,
      "step": 20155
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8319455128489023,
      "learning_rate": 6.747846018466594e-06,
      "loss": 0.4319,
      "step": 20156
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31649060414234503,
      "learning_rate": 6.746908069773905e-06,
      "loss": 0.1769,
      "step": 20157
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2204496692296681,
      "learning_rate": 6.7459701530862075e-06,
      "loss": 0.1568,
      "step": 20158
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3820488636599728,
      "learning_rate": 6.745032268412726e-06,
      "loss": 0.6271,
      "step": 20159
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3932530339844929,
      "learning_rate": 6.744094415762683e-06,
      "loss": 0.2803,
      "step": 20160
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42547672559095656,
      "learning_rate": 6.7431565951453156e-06,
      "loss": 0.2924,
      "step": 20161
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2950945370118293,
      "learning_rate": 6.742218806569843e-06,
      "loss": 0.1947,
      "step": 20162
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8203553691557237,
      "learning_rate": 6.7412810500454965e-06,
      "loss": 0.8029,
      "step": 20163
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6653295705296624,
      "learning_rate": 6.7403433255814974e-06,
      "loss": 0.2959,
      "step": 20164
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6476261891191288,
      "learning_rate": 6.73940563318707e-06,
      "loss": 0.4421,
      "step": 20165
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.25816287595500803,
      "learning_rate": 6.738467972871449e-06,
      "loss": 0.2116,
      "step": 20166
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5956146932014725,
      "learning_rate": 6.737530344643851e-06,
      "loss": 0.3551,
      "step": 20167
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.20928230836361283,
      "learning_rate": 6.736592748513502e-06,
      "loss": 0.1279,
      "step": 20168
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3856108752534053,
      "learning_rate": 6.735655184489629e-06,
      "loss": 0.7542,
      "step": 20169
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9781076338892125,
      "learning_rate": 6.7347176525814536e-06,
      "loss": 0.2769,
      "step": 20170
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22961373179822178,
      "learning_rate": 6.733780152798197e-06,
      "loss": 0.0722,
      "step": 20171
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37370635091448934,
      "learning_rate": 6.73284268514909e-06,
      "loss": 0.3033,
      "step": 20172
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.372965259533201,
      "learning_rate": 6.731905249643353e-06,
      "loss": 0.2595,
      "step": 20173
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8718983303002711,
      "learning_rate": 6.730967846290204e-06,
      "loss": 0.4454,
      "step": 20174
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3635680832584753,
      "learning_rate": 6.7300304750988696e-06,
      "loss": 0.0637,
      "step": 20175
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36684053572263525,
      "learning_rate": 6.7290931360785735e-06,
      "loss": 0.2818,
      "step": 20176
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1870496479969092,
      "learning_rate": 6.728155829238532e-06,
      "loss": 0.1054,
      "step": 20177
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3298641150783754,
      "learning_rate": 6.727218554587972e-06,
      "loss": 0.2904,
      "step": 20178
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9098954870316853,
      "learning_rate": 6.726281312136113e-06,
      "loss": 0.3332,
      "step": 20179
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6280716039161104,
      "learning_rate": 6.725344101892175e-06,
      "loss": 0.3158,
      "step": 20180
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3456454340350878,
      "learning_rate": 6.724406923865381e-06,
      "loss": 0.2268,
      "step": 20181
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8721733344041271,
      "learning_rate": 6.723469778064948e-06,
      "loss": 0.4188,
      "step": 20182
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8223203735926158,
      "learning_rate": 6.7225326645000945e-06,
      "loss": 0.274,
      "step": 20183
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2994216970014017,
      "learning_rate": 6.721595583180047e-06,
      "loss": 0.2236,
      "step": 20184
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33371295623592573,
      "learning_rate": 6.720658534114017e-06,
      "loss": 0.2629,
      "step": 20185
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1598099963694404,
      "learning_rate": 6.719721517311234e-06,
      "loss": 0.072,
      "step": 20186
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2806603566674046,
      "learning_rate": 6.718784532780906e-06,
      "loss": 0.7901,
      "step": 20187
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0020794041067742,
      "learning_rate": 6.717847580532254e-06,
      "loss": 0.3141,
      "step": 20188
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6064035587748359,
      "learning_rate": 6.7169106605745e-06,
      "loss": 0.3128,
      "step": 20189
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35123749766865786,
      "learning_rate": 6.71597377291686e-06,
      "loss": 0.2091,
      "step": 20190
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29898905072736437,
      "learning_rate": 6.7150369175685455e-06,
      "loss": 0.276,
      "step": 20191
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6702168780554714,
      "learning_rate": 6.714100094538784e-06,
      "loss": 0.2768,
      "step": 20192
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9155097632215398,
      "learning_rate": 6.713163303836786e-06,
      "loss": 0.4793,
      "step": 20193
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3684882526789919,
      "learning_rate": 6.712226545471768e-06,
      "loss": 0.182,
      "step": 20194
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32640590753162047,
      "learning_rate": 6.7112898194529475e-06,
      "loss": 0.2255,
      "step": 20195
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3039360457417942,
      "learning_rate": 6.710353125789541e-06,
      "loss": 0.2307,
      "step": 20196
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47297944732150055,
      "learning_rate": 6.7094164644907574e-06,
      "loss": 0.201,
      "step": 20197
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9337797833253157,
      "learning_rate": 6.708479835565823e-06,
      "loss": 0.5052,
      "step": 20198
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3570815014667208,
      "learning_rate": 6.707543239023946e-06,
      "loss": 0.2173,
      "step": 20199
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9508240510618632,
      "learning_rate": 6.706606674874341e-06,
      "loss": 0.521,
      "step": 20200
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42119722693115613,
      "learning_rate": 6.705670143126225e-06,
      "loss": 0.2165,
      "step": 20201
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.41388530112203586,
      "learning_rate": 6.70473364378881e-06,
      "loss": 0.2888,
      "step": 20202
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31495013117929715,
      "learning_rate": 6.703797176871305e-06,
      "loss": 0.1561,
      "step": 20203
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.574326606073024,
      "learning_rate": 6.7028607423829325e-06,
      "loss": 0.3568,
      "step": 20204
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23447823233729,
      "learning_rate": 6.701924340332902e-06,
      "loss": 0.1485,
      "step": 20205
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8618207168746056,
      "learning_rate": 6.700987970730421e-06,
      "loss": 0.4709,
      "step": 20206
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5332106456125484,
      "learning_rate": 6.700051633584709e-06,
      "loss": 0.221,
      "step": 20207
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.28051941218123827,
      "learning_rate": 6.699115328904975e-06,
      "loss": 0.2729,
      "step": 20208
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2751319034854635,
      "learning_rate": 6.698179056700426e-06,
      "loss": 0.1998,
      "step": 20209
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5546211929669201,
      "learning_rate": 6.697242816980282e-06,
      "loss": 0.017,
      "step": 20210
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7029884795339754,
      "learning_rate": 6.696306609753748e-06,
      "loss": 0.7862,
      "step": 20211
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26418860110515696,
      "learning_rate": 6.695370435030037e-06,
      "loss": 0.1684,
      "step": 20212
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3892841151163602,
      "learning_rate": 6.694434292818362e-06,
      "loss": 0.8187,
      "step": 20213
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2966504757332308,
      "learning_rate": 6.693498183127924e-06,
      "loss": 0.2364,
      "step": 20214
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.770586155194724,
      "learning_rate": 6.692562105967944e-06,
      "loss": 0.4618,
      "step": 20215
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6100751325209075,
      "learning_rate": 6.6916260613476266e-06,
      "loss": 0.2345,
      "step": 20216
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3751348043253867,
      "learning_rate": 6.690690049276178e-06,
      "loss": 0.296,
      "step": 20217
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2156673537231678,
      "learning_rate": 6.689754069762811e-06,
      "loss": 0.0854,
      "step": 20218
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.424330203332018,
      "learning_rate": 6.688818122816733e-06,
      "loss": 0.5912,
      "step": 20219
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3524732759238774,
      "learning_rate": 6.6878822084471485e-06,
      "loss": 0.2342,
      "step": 20220
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9090683341682297,
      "learning_rate": 6.686946326663272e-06,
      "loss": 0.4493,
      "step": 20221
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3305489619472809,
      "learning_rate": 6.686010477474307e-06,
      "loss": 0.2127,
      "step": 20222
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8261662368678582,
      "learning_rate": 6.685074660889459e-06,
      "loss": 0.3531,
      "step": 20223
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7956122316080234,
      "learning_rate": 6.684138876917941e-06,
      "loss": 0.4194,
      "step": 20224
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3522869022090762,
      "learning_rate": 6.683203125568955e-06,
      "loss": 0.1913,
      "step": 20225
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3228051211510193,
      "learning_rate": 6.682267406851706e-06,
      "loss": 0.2635,
      "step": 20226
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.19276487005720436,
      "learning_rate": 6.681331720775403e-06,
      "loss": 0.1324,
      "step": 20227
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5467895298422174,
      "learning_rate": 6.680396067349253e-06,
      "loss": 0.5329,
      "step": 20228
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4900043184212383,
      "learning_rate": 6.679460446582453e-06,
      "loss": 0.0222,
      "step": 20229
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.49274136008710573,
      "learning_rate": 6.678524858484219e-06,
      "loss": 0.2619,
      "step": 20230
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.326608975448095,
      "learning_rate": 6.6775893030637495e-06,
      "loss": 0.234,
      "step": 20231
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35442263812386116,
      "learning_rate": 6.6766537803302476e-06,
      "loss": 0.2927,
      "step": 20232
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0322319527987622,
      "learning_rate": 6.6757182902929205e-06,
      "loss": 0.3203,
      "step": 20233
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.916339547124325,
      "learning_rate": 6.674782832960973e-06,
      "loss": 0.4614,
      "step": 20234
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26808056590747925,
      "learning_rate": 6.673847408343601e-06,
      "loss": 0.1811,
      "step": 20235
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.14733933936393825,
      "learning_rate": 6.672912016450016e-06,
      "loss": 0.0726,
      "step": 20236
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0567068803550828,
      "learning_rate": 6.671976657289416e-06,
      "loss": 0.5313,
      "step": 20237
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31971523506697225,
      "learning_rate": 6.6710413308710065e-06,
      "loss": 0.2361,
      "step": 20238
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35655779858114456,
      "learning_rate": 6.670106037203987e-06,
      "loss": 0.2405,
      "step": 20239
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40121436558840984,
      "learning_rate": 6.669170776297558e-06,
      "loss": 0.2365,
      "step": 20240
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9633879728108563,
      "learning_rate": 6.668235548160924e-06,
      "loss": 0.4568,
      "step": 20241
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9628247433367257,
      "learning_rate": 6.667300352803287e-06,
      "loss": 0.3654,
      "step": 20242
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3921438946392328,
      "learning_rate": 6.666365190233843e-06,
      "loss": 0.2771,
      "step": 20243
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3806411539479186,
      "learning_rate": 6.665430060461798e-06,
      "loss": 0.1771,
      "step": 20244
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35615756015256916,
      "learning_rate": 6.6644949634963485e-06,
      "loss": 0.2759,
      "step": 20245
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.14361261106562823,
      "learning_rate": 6.663559899346692e-06,
      "loss": 0.0652,
      "step": 20246
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.055350676547686,
      "learning_rate": 6.662624868022034e-06,
      "loss": 0.603,
      "step": 20247
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6183491120244231,
      "learning_rate": 6.661689869531571e-06,
      "loss": 0.2337,
      "step": 20248
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35255162849177885,
      "learning_rate": 6.6607549038845e-06,
      "loss": 0.2287,
      "step": 20249
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3236437719085419,
      "learning_rate": 6.6598199710900225e-06,
      "loss": 0.2881,
      "step": 20250
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7790768482554957,
      "learning_rate": 6.658885071157335e-06,
      "loss": 0.4126,
      "step": 20251
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8047790556465177,
      "learning_rate": 6.657950204095632e-06,
      "loss": 0.7695,
      "step": 20252
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31662262684551534,
      "learning_rate": 6.657015369914119e-06,
      "loss": 0.1627,
      "step": 20253
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3431150141971942,
      "learning_rate": 6.656080568621988e-06,
      "loss": 0.2297,
      "step": 20254
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22975912293194306,
      "learning_rate": 6.655145800228433e-06,
      "loss": 0.1875,
      "step": 20255
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3362962951622495,
      "learning_rate": 6.654211064742657e-06,
      "loss": 0.6546,
      "step": 20256
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4968515354415596,
      "learning_rate": 6.653276362173855e-06,
      "loss": 0.1396,
      "step": 20257
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3371296749459188,
      "learning_rate": 6.652341692531219e-06,
      "loss": 0.3222,
      "step": 20258
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5404846940220116,
      "learning_rate": 6.651407055823948e-06,
      "loss": 0.3352,
      "step": 20259
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.7420512975589788,
      "learning_rate": 6.6504724520612385e-06,
      "loss": 0.8394,
      "step": 20260
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2988389974002585,
      "learning_rate": 6.649537881252277e-06,
      "loss": 0.2468,
      "step": 20261
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26239091791888797,
      "learning_rate": 6.64860334340627e-06,
      "loss": 0.1767,
      "step": 20262
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.42244195668297624,
      "learning_rate": 6.647668838532403e-06,
      "loss": 0.2159,
      "step": 20263
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2195888615057875,
      "learning_rate": 6.6467343666398756e-06,
      "loss": 0.0655,
      "step": 20264
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7430882573707255,
      "learning_rate": 6.645799927737877e-06,
      "loss": 0.466,
      "step": 20265
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36114269701237633,
      "learning_rate": 6.644865521835599e-06,
      "loss": 0.2252,
      "step": 20266
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2826813153562842,
      "learning_rate": 6.6439311489422445e-06,
      "loss": 0.2564,
      "step": 20267
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.37162822293368236,
      "learning_rate": 6.642996809066997e-06,
      "loss": 0.2438,
      "step": 20268
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2735223586416633,
      "learning_rate": 6.642062502219051e-06,
      "loss": 0.6658,
      "step": 20269
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7323947544075063,
      "learning_rate": 6.641128228407601e-06,
      "loss": 0.2327,
      "step": 20270
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4138610012774609,
      "learning_rate": 6.640193987641836e-06,
      "loss": 0.254,
      "step": 20271
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.45763420089813506,
      "learning_rate": 6.6392597799309434e-06,
      "loss": 0.1987,
      "step": 20272
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23997701721018916,
      "learning_rate": 6.638325605284124e-06,
      "loss": 0.1854,
      "step": 20273
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.346742580947579,
      "learning_rate": 6.637391463710563e-06,
      "loss": 0.2501,
      "step": 20274
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7056142096467717,
      "learning_rate": 6.63645735521945e-06,
      "loss": 0.3729,
      "step": 20275
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.30389836933860154,
      "learning_rate": 6.6355232798199775e-06,
      "loss": 0.1972,
      "step": 20276
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6142883592239861,
      "learning_rate": 6.634589237521334e-06,
      "loss": 0.3589,
      "step": 20277
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.336249205383674,
      "learning_rate": 6.633655228332705e-06,
      "loss": 0.8575,
      "step": 20278
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31510595252833884,
      "learning_rate": 6.632721252263288e-06,
      "loss": 0.2355,
      "step": 20279
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6198157476778042,
      "learning_rate": 6.631787309322267e-06,
      "loss": 0.2925,
      "step": 20280
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38144901534027637,
      "learning_rate": 6.630853399518828e-06,
      "loss": 0.2164,
      "step": 20281
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2686534483361336,
      "learning_rate": 6.6299195228621645e-06,
      "loss": 0.1534,
      "step": 20282
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8118556728640375,
      "learning_rate": 6.628985679361462e-06,
      "loss": 0.3046,
      "step": 20283
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6593610394688882,
      "learning_rate": 6.628051869025902e-06,
      "loss": 0.39,
      "step": 20284
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31058850204497546,
      "learning_rate": 6.6271180918646835e-06,
      "loss": 0.1984,
      "step": 20285
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.328396473855326,
      "learning_rate": 6.626184347886986e-06,
      "loss": 0.2837,
      "step": 20286
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9274638318386705,
      "learning_rate": 6.625250637101995e-06,
      "loss": 0.4564,
      "step": 20287
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9916133770334828,
      "learning_rate": 6.624316959518901e-06,
      "loss": 0.5257,
      "step": 20288
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32229395932684346,
      "learning_rate": 6.623383315146883e-06,
      "loss": 0.1662,
      "step": 20289
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24359969629009312,
      "learning_rate": 6.622449703995135e-06,
      "loss": 0.1411,
      "step": 20290
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4714650791514797,
      "learning_rate": 6.621516126072839e-06,
      "loss": 0.3253,
      "step": 20291
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5022447344060239,
      "learning_rate": 6.620582581389174e-06,
      "loss": 0.2802,
      "step": 20292
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.389706456441914,
      "learning_rate": 6.619649069953333e-06,
      "loss": 0.212,
      "step": 20293
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29167265057233993,
      "learning_rate": 6.618715591774498e-06,
      "loss": 0.1959,
      "step": 20294
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4839014207506314,
      "learning_rate": 6.617782146861851e-06,
      "loss": 0.7085,
      "step": 20295
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0026005594756369,
      "learning_rate": 6.616848735224576e-06,
      "loss": 0.5512,
      "step": 20296
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3367597761077586,
      "learning_rate": 6.615915356871856e-06,
      "loss": 0.3124,
      "step": 20297
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4199451047267864,
      "learning_rate": 6.6149820118128715e-06,
      "loss": 0.1028,
      "step": 20298
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34977275598732965,
      "learning_rate": 6.614048700056811e-06,
      "loss": 0.2842,
      "step": 20299
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6696006493965533,
      "learning_rate": 6.613115421612855e-06,
      "loss": 0.2944,
      "step": 20300
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6759490011191692,
      "learning_rate": 6.612182176490179e-06,
      "loss": 0.48,
      "step": 20301
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.17348603138377852,
      "learning_rate": 6.6112489646979734e-06,
      "loss": 0.0719,
      "step": 20302
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.29331760861572026,
      "learning_rate": 6.610315786245416e-06,
      "loss": 0.1961,
      "step": 20303
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33783685655053447,
      "learning_rate": 6.609382641141681e-06,
      "loss": 0.2781,
      "step": 20304
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8376854548745871,
      "learning_rate": 6.608449529395961e-06,
      "loss": 0.5346,
      "step": 20305
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.223618393209214,
      "learning_rate": 6.607516451017431e-06,
      "loss": 0.5518,
      "step": 20306
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5100522868530578,
      "learning_rate": 6.6065834060152675e-06,
      "loss": 0.1177,
      "step": 20307
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4028319465521841,
      "learning_rate": 6.6056503943986564e-06,
      "loss": 0.3079,
      "step": 20308
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3394092129408815,
      "learning_rate": 6.6047174161767725e-06,
      "loss": 0.2581,
      "step": 20309
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9180793447369935,
      "learning_rate": 6.603784471358792e-06,
      "loss": 0.533,
      "step": 20310
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2531595632943837,
      "learning_rate": 6.6028515599539e-06,
      "loss": 0.131,
      "step": 20311
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38354310857520946,
      "learning_rate": 6.601918681971272e-06,
      "loss": 0.2556,
      "step": 20312
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4396664299027006,
      "learning_rate": 6.600985837420088e-06,
      "loss": 0.2779,
      "step": 20313
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4977994638862007,
      "learning_rate": 6.600053026309523e-06,
      "loss": 0.8249,
      "step": 20314
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34050486747622916,
      "learning_rate": 6.599120248648752e-06,
      "loss": 0.2573,
      "step": 20315
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.30596359781038346,
      "learning_rate": 6.598187504446958e-06,
      "loss": 0.1548,
      "step": 20316
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47661252537845467,
      "learning_rate": 6.5972547937133145e-06,
      "loss": 0.3254,
      "step": 20317
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6744944601115314,
      "learning_rate": 6.596322116456997e-06,
      "loss": 0.3064,
      "step": 20318
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9022397907274636,
      "learning_rate": 6.595389472687184e-06,
      "loss": 0.4295,
      "step": 20319
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24876676101693307,
      "learning_rate": 6.59445686241305e-06,
      "loss": 0.1596,
      "step": 20320
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32838835981839704,
      "learning_rate": 6.593524285643764e-06,
      "loss": 0.2471,
      "step": 20321
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3847488520024002,
      "learning_rate": 6.592591742388513e-06,
      "loss": 0.2268,
      "step": 20322
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.296932708510234,
      "learning_rate": 6.591659232656465e-06,
      "loss": 0.8099,
      "step": 20323
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4535289825255963,
      "learning_rate": 6.590726756456791e-06,
      "loss": 0.1627,
      "step": 20324
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.574812354413758,
      "learning_rate": 6.589794313798674e-06,
      "loss": 0.3338,
      "step": 20325
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3828946553690941,
      "learning_rate": 6.588861904691282e-06,
      "loss": 0.221,
      "step": 20326
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.49053241479437437,
      "learning_rate": 6.587929529143786e-06,
      "loss": 0.3346,
      "step": 20327
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.38090347467991237,
      "learning_rate": 6.586997187165365e-06,
      "loss": 0.2015,
      "step": 20328
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3226794536996721,
      "learning_rate": 6.586064878765189e-06,
      "loss": 0.1115,
      "step": 20329
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6281300436733043,
      "learning_rate": 6.585132603952425e-06,
      "loss": 0.3285,
      "step": 20330
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32953982573880225,
      "learning_rate": 6.584200362736255e-06,
      "loss": 0.2299,
      "step": 20331
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.30116616068361535,
      "learning_rate": 6.5832681551258445e-06,
      "loss": 0.2446,
      "step": 20332
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.44653778000600874,
      "learning_rate": 6.582335981130366e-06,
      "loss": 0.239,
      "step": 20333
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6371014931649811,
      "learning_rate": 6.581403840758991e-06,
      "loss": 0.3693,
      "step": 20334
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3310446601775136,
      "learning_rate": 6.5804717340208905e-06,
      "loss": 0.2227,
      "step": 20335
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.4480958930376187,
      "learning_rate": 6.579539660925231e-06,
      "loss": 0.4182,
      "step": 20336
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3730507285282467,
      "learning_rate": 6.578607621481189e-06,
      "loss": 0.0173,
      "step": 20337
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4803295869528921,
      "learning_rate": 6.57767561569793e-06,
      "loss": 0.3592,
      "step": 20338
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2548188944968282,
      "learning_rate": 6.576743643584625e-06,
      "loss": 0.1856,
      "step": 20339
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5342113737635732,
      "learning_rate": 6.575811705150443e-06,
      "loss": 0.3953,
      "step": 20340
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9941799662383259,
      "learning_rate": 6.574879800404548e-06,
      "loss": 0.4624,
      "step": 20341
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.33309678880468696,
      "learning_rate": 6.573947929356116e-06,
      "loss": 0.1586,
      "step": 20342
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7134071127388232,
      "learning_rate": 6.573016092014313e-06,
      "loss": 0.3698,
      "step": 20343
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31985346848800456,
      "learning_rate": 6.572084288388302e-06,
      "loss": 0.1955,
      "step": 20344
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.48255547823627926,
      "learning_rate": 6.571152518487256e-06,
      "loss": 0.3417,
      "step": 20345
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3837307233824594,
      "learning_rate": 6.57022078232034e-06,
      "loss": 0.2066,
      "step": 20346
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3477953403976277,
      "learning_rate": 6.569289079896717e-06,
      "loss": 0.6619,
      "step": 20347
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2710537969850274,
      "learning_rate": 6.56835741122556e-06,
      "loss": 0.1745,
      "step": 20348
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.5114620977748154,
      "learning_rate": 6.567425776316032e-06,
      "loss": 0.776,
      "step": 20349
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3700063228332796,
      "learning_rate": 6.566494175177297e-06,
      "loss": 0.2449,
      "step": 20350
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.47706403376845924,
      "learning_rate": 6.565562607818524e-06,
      "loss": 0.351,
      "step": 20351
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6229766422897923,
      "learning_rate": 6.564631074248877e-06,
      "loss": 0.2177,
      "step": 20352
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5400418539295311,
      "learning_rate": 6.563699574477515e-06,
      "loss": 0.3364,
      "step": 20353
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.21170788439187613,
      "learning_rate": 6.562768108513611e-06,
      "loss": 0.1233,
      "step": 20354
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0867991549371667,
      "learning_rate": 6.5618366763663266e-06,
      "loss": 0.2317,
      "step": 20355
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32913965132257383,
      "learning_rate": 6.560905278044818e-06,
      "loss": 0.3174,
      "step": 20356
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.24854888688168916,
      "learning_rate": 6.55997391355826e-06,
      "loss": 0.0675,
      "step": 20357
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.412157898528608,
      "learning_rate": 6.559042582915814e-06,
      "loss": 0.3269,
      "step": 20358
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7870769733665778,
      "learning_rate": 6.558111286126632e-06,
      "loss": 0.3526,
      "step": 20359
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.7393028190287703,
      "learning_rate": 6.557180023199888e-06,
      "loss": 0.4194,
      "step": 20360
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.1963043515712688,
      "learning_rate": 6.556248794144739e-06,
      "loss": 0.1175,
      "step": 20361
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.40337100797668257,
      "learning_rate": 6.555317598970344e-06,
      "loss": 0.2576,
      "step": 20362
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.26338022080469936,
      "learning_rate": 6.554386437685873e-06,
      "loss": 0.2301,
      "step": 20363
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.364674246902576,
      "learning_rate": 6.553455310300478e-06,
      "loss": 0.5941,
      "step": 20364
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8391808137048898,
      "learning_rate": 6.552524216823327e-06,
      "loss": 0.2024,
      "step": 20365
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.651547624092878,
      "learning_rate": 6.551593157263577e-06,
      "loss": 0.3768,
      "step": 20366
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32655592142424955,
      "learning_rate": 6.5506621316303846e-06,
      "loss": 0.2294,
      "step": 20367
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.1970821960051563,
      "learning_rate": 6.549731139932916e-06,
      "loss": 0.3251,
      "step": 20368
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3293012614610856,
      "learning_rate": 6.548800182180328e-06,
      "loss": 0.292,
      "step": 20369
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5199911290621659,
      "learning_rate": 6.547869258381779e-06,
      "loss": 0.2187,
      "step": 20370
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3087103256027574,
      "learning_rate": 6.546938368546428e-06,
      "loss": 0.239,
      "step": 20371
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.18052893368890255,
      "learning_rate": 6.546007512683434e-06,
      "loss": 0.0672,
      "step": 20372
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.427618519032754,
      "learning_rate": 6.5450766908019515e-06,
      "loss": 0.5977,
      "step": 20373
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2928940654218132,
      "learning_rate": 6.544145902911145e-06,
      "loss": 0.2272,
      "step": 20374
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5509048200594501,
      "learning_rate": 6.543215149020168e-06,
      "loss": 0.2833,
      "step": 20375
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3433685669769734,
      "learning_rate": 6.542284429138176e-06,
      "loss": 0.2147,
      "step": 20376
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.9544391511989576,
      "learning_rate": 6.5413537432743304e-06,
      "loss": 0.4692,
      "step": 20377
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0435221758727005,
      "learning_rate": 6.540423091437782e-06,
      "loss": 0.345,
      "step": 20378
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.3415530478548011,
      "learning_rate": 6.539492473637688e-06,
      "loss": 0.2117,
      "step": 20379
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.35914255369045056,
      "learning_rate": 6.5385618898832085e-06,
      "loss": 0.1723,
      "step": 20380
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.21628748765960354,
      "learning_rate": 6.537631340183497e-06,
      "loss": 0.1822,
      "step": 20381
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.153589759827638,
      "learning_rate": 6.536700824547706e-06,
      "loss": 0.5063,
      "step": 20382
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.0372686656213295,
      "learning_rate": 6.535770342984993e-06,
      "loss": 0.4578,
      "step": 20383
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.8855096758190305,
      "learning_rate": 6.5348398955045115e-06,
      "loss": 0.4758,
      "step": 20384
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.34107710412331915,
      "learning_rate": 6.5339094821154105e-06,
      "loss": 0.2346,
      "step": 20385
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5100213841957584,
      "learning_rate": 6.532979102826853e-06,
      "loss": 0.342,
      "step": 20386
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.46019695562749335,
      "learning_rate": 6.532048757647989e-06,
      "loss": 0.2868,
      "step": 20387
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.6326744548756653,
      "learning_rate": 6.531118446587967e-06,
      "loss": 0.7098,
      "step": 20388
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.32176800809070216,
      "learning_rate": 6.530188169655945e-06,
      "loss": 0.1684,
      "step": 20389
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.22208554731782879,
      "learning_rate": 6.52925792686107e-06,
      "loss": 0.1863,
      "step": 20390
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.3658273563591494,
      "learning_rate": 6.528327718212502e-06,
      "loss": 0.1783,
      "step": 20391
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.31150135672406015,
      "learning_rate": 6.527397543719387e-06,
      "loss": 0.2837,
      "step": 20392
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.4966694541582685,
      "learning_rate": 6.526467403390875e-06,
      "loss": 0.2704,
      "step": 20393
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2827662519370385,
      "learning_rate": 6.5255372972361245e-06,
      "loss": 0.2118,
      "step": 20394
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6273417632866125,
      "learning_rate": 6.524607225264279e-06,
      "loss": 0.4582,
      "step": 20395
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.127751710870591,
      "learning_rate": 6.5236771874844896e-06,
      "loss": 0.3848,
      "step": 20396
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.46917904031919344,
      "learning_rate": 6.522747183905911e-06,
      "loss": 0.3206,
      "step": 20397
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23918823261769412,
      "learning_rate": 6.521817214537689e-06,
      "loss": 0.1772,
      "step": 20398
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.439703851579976,
      "learning_rate": 6.52088727938897e-06,
      "loss": 0.3114,
      "step": 20399
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.23440097657117479,
      "learning_rate": 6.519957378468911e-06,
      "loss": 0.0962,
      "step": 20400
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.6904768356663734,
      "learning_rate": 6.519027511786655e-06,
      "loss": 0.4018,
      "step": 20401
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.36865984311238764,
      "learning_rate": 6.518097679351352e-06,
      "loss": 0.2231,
      "step": 20402
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.5287992859522594,
      "learning_rate": 6.517167881172149e-06,
      "loss": 0.3497,
      "step": 20403
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2963369599422095,
      "learning_rate": 6.5162381172581965e-06,
      "loss": 0.2583,
      "step": 20404
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.48679825849355507,
      "learning_rate": 6.515308387618635e-06,
      "loss": 0.3442,
      "step": 20405
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.2197853541779031,
      "learning_rate": 6.514378692262621e-06,
      "loss": 0.194,
      "step": 20406
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.2245639804172004,
      "learning_rate": 6.513449031199295e-06,
      "loss": 0.0713,
      "step": 20407
    },
    {
      "epoch": 0.62,
      "grad_norm": 0.27296862592766835,
      "learning_rate": 6.512519404437803e-06,
      "loss": 0.2276,
      "step": 20408
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2931235025414657,
      "learning_rate": 6.511589811987296e-06,
      "loss": 0.092,
      "step": 20409
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30185254242143134,
      "learning_rate": 6.510660253856915e-06,
      "loss": 0.2902,
      "step": 20410
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5017372911274761,
      "learning_rate": 6.5097307300558025e-06,
      "loss": 0.2188,
      "step": 20411
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33162564485704515,
      "learning_rate": 6.5088012405931125e-06,
      "loss": 0.2577,
      "step": 20412
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5452273305478715,
      "learning_rate": 6.507871785477983e-06,
      "loss": 0.3265,
      "step": 20413
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1718288815888447,
      "learning_rate": 6.50694236471956e-06,
      "loss": 0.7422,
      "step": 20414
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4341055804217136,
      "learning_rate": 6.506012978326987e-06,
      "loss": 0.2362,
      "step": 20415
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3865889790141019,
      "learning_rate": 6.505083626309405e-06,
      "loss": 0.2658,
      "step": 20416
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3408083439908754,
      "learning_rate": 6.5041543086759655e-06,
      "loss": 0.2132,
      "step": 20417
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.25449356767271797,
      "learning_rate": 6.503225025435804e-06,
      "loss": 0.1496,
      "step": 20418
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6989772416510383,
      "learning_rate": 6.502295776598064e-06,
      "loss": 0.2917,
      "step": 20419
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5315491933494849,
      "learning_rate": 6.501366562171892e-06,
      "loss": 0.2465,
      "step": 20420
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3320443384285328,
      "learning_rate": 6.500437382166427e-06,
      "loss": 0.287,
      "step": 20421
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3986740317420287,
      "learning_rate": 6.4995082365908055e-06,
      "loss": 0.2571,
      "step": 20422
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4935797581682571,
      "learning_rate": 6.4985791254541785e-06,
      "loss": 0.3523,
      "step": 20423
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3355577932148055,
      "learning_rate": 6.497650048765683e-06,
      "loss": 0.326,
      "step": 20424
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37756019974286764,
      "learning_rate": 6.4967210065344525e-06,
      "loss": 0.259,
      "step": 20425
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4450510248261956,
      "learning_rate": 6.495791998769639e-06,
      "loss": 0.1956,
      "step": 20426
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32948458244390216,
      "learning_rate": 6.494863025480377e-06,
      "loss": 0.1919,
      "step": 20427
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3280889083435784,
      "learning_rate": 6.4939340866758035e-06,
      "loss": 0.2506,
      "step": 20428
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6799157790062814,
      "learning_rate": 6.493005182365062e-06,
      "loss": 0.3447,
      "step": 20429
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.34409738919171223,
      "learning_rate": 6.49207631255729e-06,
      "loss": 0.1708,
      "step": 20430
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4868272038828369,
      "learning_rate": 6.49114747726162e-06,
      "loss": 0.3513,
      "step": 20431
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0422457375075245,
      "learning_rate": 6.490218676487201e-06,
      "loss": 0.4263,
      "step": 20432
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.287162840791448,
      "learning_rate": 6.489289910243166e-06,
      "loss": 0.2404,
      "step": 20433
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8526846358848055,
      "learning_rate": 6.488361178538649e-06,
      "loss": 0.3724,
      "step": 20434
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30435246722069004,
      "learning_rate": 6.4874324813827915e-06,
      "loss": 0.193,
      "step": 20435
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45252055723319556,
      "learning_rate": 6.486503818784731e-06,
      "loss": 0.2634,
      "step": 20436
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32404930468238935,
      "learning_rate": 6.485575190753597e-06,
      "loss": 0.2092,
      "step": 20437
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.8532621113256158,
      "learning_rate": 6.484646597298534e-06,
      "loss": 0.8373,
      "step": 20438
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29632921896225556,
      "learning_rate": 6.483718038428674e-06,
      "loss": 0.1754,
      "step": 20439
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2526411977036693,
      "learning_rate": 6.482789514153154e-06,
      "loss": 0.2543,
      "step": 20440
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8555628723247352,
      "learning_rate": 6.481861024481109e-06,
      "loss": 0.5321,
      "step": 20441
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.7072457409809503,
      "learning_rate": 6.480932569421667e-06,
      "loss": 0.7466,
      "step": 20442
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23676637981455298,
      "learning_rate": 6.480004148983974e-06,
      "loss": 0.0688,
      "step": 20443
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3669070129900583,
      "learning_rate": 6.479075763177158e-06,
      "loss": 0.2792,
      "step": 20444
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36879197013507636,
      "learning_rate": 6.47814741201035e-06,
      "loss": 0.2751,
      "step": 20445
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32023017197606296,
      "learning_rate": 6.47721909549269e-06,
      "loss": 0.2351,
      "step": 20446
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2579393102843535,
      "learning_rate": 6.476290813633308e-06,
      "loss": 0.1698,
      "step": 20447
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27328350549066344,
      "learning_rate": 6.475362566441331e-06,
      "loss": 0.1831,
      "step": 20448
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3676950745421292,
      "learning_rate": 6.474434353925902e-06,
      "loss": 0.8265,
      "step": 20449
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0183478588530714,
      "learning_rate": 6.473506176096147e-06,
      "loss": 0.3142,
      "step": 20450
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3107180743598413,
      "learning_rate": 6.4725780329611965e-06,
      "loss": 0.2794,
      "step": 20451
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4706174803391088,
      "learning_rate": 6.471649924530188e-06,
      "loss": 0.1459,
      "step": 20452
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4967771846011373,
      "learning_rate": 6.470721850812246e-06,
      "loss": 0.308,
      "step": 20453
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5117420857510561,
      "learning_rate": 6.469793811816501e-06,
      "loss": 0.3057,
      "step": 20454
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7656125897159012,
      "learning_rate": 6.4688658075520895e-06,
      "loss": 0.4877,
      "step": 20455
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.19811935641447648,
      "learning_rate": 6.467937838028139e-06,
      "loss": 0.0867,
      "step": 20456
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4622390730835871,
      "learning_rate": 6.467009903253774e-06,
      "loss": 0.3052,
      "step": 20457
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27944212029155086,
      "learning_rate": 6.466082003238133e-06,
      "loss": 0.2326,
      "step": 20458
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3423717668738948,
      "learning_rate": 6.465154137990339e-06,
      "loss": 0.9136,
      "step": 20459
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3099322845507504,
      "learning_rate": 6.464226307519518e-06,
      "loss": 0.1663,
      "step": 20460
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5918325558080679,
      "learning_rate": 6.463298511834806e-06,
      "loss": 0.2,
      "step": 20461
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3437815628641298,
      "learning_rate": 6.462370750945326e-06,
      "loss": 0.2741,
      "step": 20462
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36383005644078303,
      "learning_rate": 6.461443024860203e-06,
      "loss": 0.2671,
      "step": 20463
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.42415120693579184,
      "learning_rate": 6.460515333588572e-06,
      "loss": 0.2966,
      "step": 20464
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32783910362107377,
      "learning_rate": 6.459587677139554e-06,
      "loss": 0.094,
      "step": 20465
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6004247861906684,
      "learning_rate": 6.458660055522279e-06,
      "loss": 0.3469,
      "step": 20466
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3351648903303348,
      "learning_rate": 6.457732468745872e-06,
      "loss": 0.2341,
      "step": 20467
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5067670302733458,
      "learning_rate": 6.456804916819454e-06,
      "loss": 0.3085,
      "step": 20468
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.41149144460259035,
      "learning_rate": 6.4558773997521595e-06,
      "loss": 0.223,
      "step": 20469
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3901399401285082,
      "learning_rate": 6.454949917553109e-06,
      "loss": 0.3262,
      "step": 20470
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.26020294062085336,
      "learning_rate": 6.4540224702314254e-06,
      "loss": 0.1792,
      "step": 20471
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6340210655424063,
      "learning_rate": 6.453095057796238e-06,
      "loss": 0.3858,
      "step": 20472
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7497977067218378,
      "learning_rate": 6.452167680256669e-06,
      "loss": 0.0546,
      "step": 20473
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2586916323508314,
      "learning_rate": 6.451240337621837e-06,
      "loss": 0.1845,
      "step": 20474
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40874020668460914,
      "learning_rate": 6.450313029900874e-06,
      "loss": 0.2826,
      "step": 20475
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2863311576370179,
      "learning_rate": 6.4493857571029005e-06,
      "loss": 0.2311,
      "step": 20476
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.440231441249659,
      "learning_rate": 6.4484585192370356e-06,
      "loss": 0.8081,
      "step": 20477
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37309566825488694,
      "learning_rate": 6.447531316312406e-06,
      "loss": 0.1904,
      "step": 20478
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5990432455698764,
      "learning_rate": 6.4466041483381335e-06,
      "loss": 0.3657,
      "step": 20479
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3009656439229873,
      "learning_rate": 6.445677015323333e-06,
      "loss": 0.198,
      "step": 20480
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2978392336772262,
      "learning_rate": 6.4447499172771365e-06,
      "loss": 0.5465,
      "step": 20481
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30954741764359817,
      "learning_rate": 6.44382285420866e-06,
      "loss": 0.231,
      "step": 20482
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0529688894199902,
      "learning_rate": 6.442895826127022e-06,
      "loss": 0.5894,
      "step": 20483
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2988127458426382,
      "learning_rate": 6.441968833041347e-06,
      "loss": 0.1175,
      "step": 20484
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3739095649163239,
      "learning_rate": 6.441041874960756e-06,
      "loss": 0.297,
      "step": 20485
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.16245770977356117,
      "learning_rate": 6.440114951894359e-06,
      "loss": 0.1002,
      "step": 20486
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32045635977035486,
      "learning_rate": 6.439188063851288e-06,
      "loss": 0.2509,
      "step": 20487
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8664378392952092,
      "learning_rate": 6.438261210840656e-06,
      "loss": 0.515,
      "step": 20488
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27776977168146777,
      "learning_rate": 6.43733439287158e-06,
      "loss": 0.1835,
      "step": 20489
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7142134193907225,
      "learning_rate": 6.436407609953184e-06,
      "loss": 0.4887,
      "step": 20490
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8439502778075015,
      "learning_rate": 6.435480862094576e-06,
      "loss": 0.0432,
      "step": 20491
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4520452243496567,
      "learning_rate": 6.434554149304886e-06,
      "loss": 0.7022,
      "step": 20492
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.25857087528317807,
      "learning_rate": 6.4336274715932245e-06,
      "loss": 0.2122,
      "step": 20493
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.37662774982598646,
      "learning_rate": 6.432700828968706e-06,
      "loss": 0.2729,
      "step": 20494
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7890799371969889,
      "learning_rate": 6.431774221440454e-06,
      "loss": 0.3484,
      "step": 20495
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9520024670167994,
      "learning_rate": 6.430847649017582e-06,
      "loss": 0.4429,
      "step": 20496
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23074610094546072,
      "learning_rate": 6.429921111709203e-06,
      "loss": 0.1506,
      "step": 20497
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3868379283862098,
      "learning_rate": 6.428994609524437e-06,
      "loss": 0.2576,
      "step": 20498
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24106988461865728,
      "learning_rate": 6.428068142472396e-06,
      "loss": 0.2135,
      "step": 20499
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1604089673464386,
      "learning_rate": 6.427141710562193e-06,
      "loss": 0.1962,
      "step": 20500
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2516053241925635,
      "learning_rate": 6.42621531380295e-06,
      "loss": 0.658,
      "step": 20501
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3909989137708067,
      "learning_rate": 6.4252889522037765e-06,
      "loss": 0.1655,
      "step": 20502
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.35743020545878174,
      "learning_rate": 6.424362625773783e-06,
      "loss": 0.2842,
      "step": 20503
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0440513781882965,
      "learning_rate": 6.42343633452209e-06,
      "loss": 0.3134,
      "step": 20504
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3252160008267346,
      "learning_rate": 6.422510078457806e-06,
      "loss": 0.2855,
      "step": 20505
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2553900161473333,
      "learning_rate": 6.421583857590041e-06,
      "loss": 0.1419,
      "step": 20506
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31249240684157253,
      "learning_rate": 6.420657671927917e-06,
      "loss": 0.2436,
      "step": 20507
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32484219274152715,
      "learning_rate": 6.419731521480539e-06,
      "loss": 0.1016,
      "step": 20508
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4144713544945742,
      "learning_rate": 6.41880540625702e-06,
      "loss": 0.6636,
      "step": 20509
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38695500168459573,
      "learning_rate": 6.417879326266472e-06,
      "loss": 0.2497,
      "step": 20510
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39834066809022883,
      "learning_rate": 6.416953281518008e-06,
      "loss": 0.3099,
      "step": 20511
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29689996241853683,
      "learning_rate": 6.416027272020731e-06,
      "loss": 0.1844,
      "step": 20512
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0437959155886145,
      "learning_rate": 6.415101297783761e-06,
      "loss": 0.3575,
      "step": 20513
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9544061223499678,
      "learning_rate": 6.414175358816204e-06,
      "loss": 0.4439,
      "step": 20514
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.21105041667423521,
      "learning_rate": 6.413249455127169e-06,
      "loss": 0.0709,
      "step": 20515
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3309915193593273,
      "learning_rate": 6.412323586725766e-06,
      "loss": 0.2022,
      "step": 20516
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2493351579368031,
      "learning_rate": 6.4113977536211e-06,
      "loss": 0.2363,
      "step": 20517
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2531784241090131,
      "learning_rate": 6.41047195582229e-06,
      "loss": 0.4663,
      "step": 20518
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3202142400549768,
      "learning_rate": 6.409546193338436e-06,
      "loss": 0.2873,
      "step": 20519
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.072595658223321,
      "learning_rate": 6.408620466178648e-06,
      "loss": 0.5053,
      "step": 20520
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3983667285328831,
      "learning_rate": 6.407694774352033e-06,
      "loss": 0.2184,
      "step": 20521
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1065094062769347,
      "learning_rate": 6.406769117867701e-06,
      "loss": 0.4336,
      "step": 20522
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2852033341761363,
      "learning_rate": 6.405843496734752e-06,
      "loss": 0.2206,
      "step": 20523
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5913173948726367,
      "learning_rate": 6.404917910962301e-06,
      "loss": 0.7614,
      "step": 20524
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.20909933337185474,
      "learning_rate": 6.40399236055945e-06,
      "loss": 0.1276,
      "step": 20525
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.25852978714266334,
      "learning_rate": 6.403066845535303e-06,
      "loss": 0.1399,
      "step": 20526
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6136241204198971,
      "learning_rate": 6.4021413658989706e-06,
      "loss": 0.6942,
      "step": 20527
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4083444313354199,
      "learning_rate": 6.401215921659556e-06,
      "loss": 0.2511,
      "step": 20528
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3805805974457887,
      "learning_rate": 6.400290512826162e-06,
      "loss": 0.2967,
      "step": 20529
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30190300952639354,
      "learning_rate": 6.399365139407895e-06,
      "loss": 0.2147,
      "step": 20530
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9425101990495297,
      "learning_rate": 6.398439801413858e-06,
      "loss": 0.4544,
      "step": 20531
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7630076211743293,
      "learning_rate": 6.397514498853152e-06,
      "loss": 0.3515,
      "step": 20532
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5153367351640143,
      "learning_rate": 6.396589231734889e-06,
      "loss": 0.2524,
      "step": 20533
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.26392313717350113,
      "learning_rate": 6.395664000068167e-06,
      "loss": 0.1965,
      "step": 20534
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5421001613387899,
      "learning_rate": 6.394738803862084e-06,
      "loss": 0.3235,
      "step": 20535
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2714158654926647,
      "learning_rate": 6.393813643125751e-06,
      "loss": 0.1653,
      "step": 20536
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3987819510661221,
      "learning_rate": 6.3928885178682655e-06,
      "loss": 0.8344,
      "step": 20537
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5467439461900838,
      "learning_rate": 6.391963428098725e-06,
      "loss": 0.262,
      "step": 20538
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27772221954121923,
      "learning_rate": 6.3910383738262405e-06,
      "loss": 0.1814,
      "step": 20539
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7278333396334851,
      "learning_rate": 6.390113355059907e-06,
      "loss": 0.4411,
      "step": 20540
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2798436783000421,
      "learning_rate": 6.389188371808826e-06,
      "loss": 0.2333,
      "step": 20541
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.9499871010925347,
      "learning_rate": 6.388263424082098e-06,
      "loss": 0.7291,
      "step": 20542
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.310233021135389,
      "learning_rate": 6.387338511888819e-06,
      "loss": 0.0706,
      "step": 20543
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2619530458783077,
      "learning_rate": 6.386413635238096e-06,
      "loss": 0.2172,
      "step": 20544
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29404123593702464,
      "learning_rate": 6.385488794139025e-06,
      "loss": 0.1185,
      "step": 20545
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40304676807042145,
      "learning_rate": 6.384563988600704e-06,
      "loss": 0.3215,
      "step": 20546
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.390978021972701,
      "learning_rate": 6.383639218632234e-06,
      "loss": 0.2814,
      "step": 20547
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3922253980798345,
      "learning_rate": 6.382714484242709e-06,
      "loss": 0.308,
      "step": 20548
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40739363793508243,
      "learning_rate": 6.381789785441228e-06,
      "loss": 0.2709,
      "step": 20549
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4456442681011046,
      "learning_rate": 6.380865122236892e-06,
      "loss": 0.7427,
      "step": 20550
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.389850830620926,
      "learning_rate": 6.379940494638795e-06,
      "loss": 0.2085,
      "step": 20551
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2087392766485853,
      "learning_rate": 6.3790159026560335e-06,
      "loss": 0.1262,
      "step": 20552
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.25630475027296595,
      "learning_rate": 6.378091346297708e-06,
      "loss": 0.2508,
      "step": 20553
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3013771288730929,
      "learning_rate": 6.377166825572911e-06,
      "loss": 0.1007,
      "step": 20554
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6858978582185662,
      "learning_rate": 6.376242340490734e-06,
      "loss": 0.4129,
      "step": 20555
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.48048302205131843,
      "learning_rate": 6.375317891060282e-06,
      "loss": 0.2648,
      "step": 20556
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3656272892971717,
      "learning_rate": 6.374393477290646e-06,
      "loss": 0.232,
      "step": 20557
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0739529051904937,
      "learning_rate": 6.373469099190917e-06,
      "loss": 0.4675,
      "step": 20558
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33461951552285013,
      "learning_rate": 6.372544756770194e-06,
      "loss": 0.2899,
      "step": 20559
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9156052382142572,
      "learning_rate": 6.371620450037569e-06,
      "loss": 0.2324,
      "step": 20560
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6375628234525378,
      "learning_rate": 6.3706961790021325e-06,
      "loss": 0.3051,
      "step": 20561
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3632350460237349,
      "learning_rate": 6.3697719436729846e-06,
      "loss": 0.2031,
      "step": 20562
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2873353698795273,
      "learning_rate": 6.368847744059216e-06,
      "loss": 0.1728,
      "step": 20563
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2860399591058989,
      "learning_rate": 6.367923580169913e-06,
      "loss": 0.2432,
      "step": 20564
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7844926807804918,
      "learning_rate": 6.366999452014178e-06,
      "loss": 0.262,
      "step": 20565
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.49209036648878335,
      "learning_rate": 6.366075359601095e-06,
      "loss": 0.3471,
      "step": 20566
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4072689937278888,
      "learning_rate": 6.36515130293976e-06,
      "loss": 0.2622,
      "step": 20567
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.2049754571611786,
      "learning_rate": 6.364227282039264e-06,
      "loss": 0.4838,
      "step": 20568
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1799529726356255,
      "learning_rate": 6.363303296908692e-06,
      "loss": 0.2248,
      "step": 20569
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.40369136735642513,
      "learning_rate": 6.362379347557142e-06,
      "loss": 0.2608,
      "step": 20570
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24095707114373416,
      "learning_rate": 6.3614554339937e-06,
      "loss": 0.2024,
      "step": 20571
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2918398751211057,
      "learning_rate": 6.360531556227456e-06,
      "loss": 0.2271,
      "step": 20572
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7185309424264644,
      "learning_rate": 6.359607714267501e-06,
      "loss": 0.3057,
      "step": 20573
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5528667415133937,
      "learning_rate": 6.358683908122924e-06,
      "loss": 0.8196,
      "step": 20574
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3110599284073564,
      "learning_rate": 6.357760137802809e-06,
      "loss": 0.1674,
      "step": 20575
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3675948416094179,
      "learning_rate": 6.356836403316253e-06,
      "loss": 0.2949,
      "step": 20576
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3978393049303874,
      "learning_rate": 6.355912704672338e-06,
      "loss": 0.2282,
      "step": 20577
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5377862588979663,
      "learning_rate": 6.354989041880152e-06,
      "loss": 0.0076,
      "step": 20578
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8143631853215955,
      "learning_rate": 6.3540654149487844e-06,
      "loss": 0.387,
      "step": 20579
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29482071259805653,
      "learning_rate": 6.353141823887321e-06,
      "loss": 0.211,
      "step": 20580
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7689677371316103,
      "learning_rate": 6.352218268704844e-06,
      "loss": 0.4577,
      "step": 20581
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31694490572179496,
      "learning_rate": 6.3512947494104484e-06,
      "loss": 0.2649,
      "step": 20582
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2918229755866768,
      "learning_rate": 6.350371266013215e-06,
      "loss": 0.2123,
      "step": 20583
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2702111580361796,
      "learning_rate": 6.349447818522228e-06,
      "loss": 0.176,
      "step": 20584
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5165668505341374,
      "learning_rate": 6.348524406946577e-06,
      "loss": 0.8583,
      "step": 20585
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1004917888681918,
      "learning_rate": 6.347601031295345e-06,
      "loss": 0.2964,
      "step": 20586
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.6251738385599312,
      "learning_rate": 6.346677691577611e-06,
      "loss": 0.7811,
      "step": 20587
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2779281101001885,
      "learning_rate": 6.345754387802467e-06,
      "loss": 0.208,
      "step": 20588
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5345922036666041,
      "learning_rate": 6.344831119978996e-06,
      "loss": 0.3248,
      "step": 20589
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3836019109619769,
      "learning_rate": 6.343907888116277e-06,
      "loss": 0.2457,
      "step": 20590
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38101735681338034,
      "learning_rate": 6.342984692223395e-06,
      "loss": 0.1841,
      "step": 20591
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4822664891608999,
      "learning_rate": 6.34206153230943e-06,
      "loss": 0.7702,
      "step": 20592
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2715497751247081,
      "learning_rate": 6.341138408383472e-06,
      "loss": 0.0702,
      "step": 20593
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3964028195927644,
      "learning_rate": 6.340215320454597e-06,
      "loss": 0.2643,
      "step": 20594
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23457139385281453,
      "learning_rate": 6.339292268531885e-06,
      "loss": 0.1971,
      "step": 20595
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.522087907438933,
      "learning_rate": 6.338369252624423e-06,
      "loss": 0.6927,
      "step": 20596
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5393038954554104,
      "learning_rate": 6.33744627274129e-06,
      "loss": 0.2054,
      "step": 20597
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.34279294495746954,
      "learning_rate": 6.336523328891564e-06,
      "loss": 0.2641,
      "step": 20598
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6595617897188997,
      "learning_rate": 6.3356004210843294e-06,
      "loss": 0.2958,
      "step": 20599
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3311394154472009,
      "learning_rate": 6.334677549328663e-06,
      "loss": 0.2663,
      "step": 20600
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.32338158487868174,
      "learning_rate": 6.33375471363364e-06,
      "loss": 0.1122,
      "step": 20601
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6282842466331848,
      "learning_rate": 6.3328319140083485e-06,
      "loss": 0.3446,
      "step": 20602
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3565812757067608,
      "learning_rate": 6.331909150461865e-06,
      "loss": 0.237,
      "step": 20603
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2654543494049247,
      "learning_rate": 6.330986423003262e-06,
      "loss": 0.0805,
      "step": 20604
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8554567149568292,
      "learning_rate": 6.330063731641626e-06,
      "loss": 0.4076,
      "step": 20605
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2519178093394958,
      "learning_rate": 6.32914107638603e-06,
      "loss": 0.2173,
      "step": 20606
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3429578741860789,
      "learning_rate": 6.328218457245547e-06,
      "loss": 0.2617,
      "step": 20607
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6994943523046194,
      "learning_rate": 6.3272958742292635e-06,
      "loss": 0.2886,
      "step": 20608
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3496202948228206,
      "learning_rate": 6.326373327346251e-06,
      "loss": 0.5477,
      "step": 20609
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.90157977028825,
      "learning_rate": 6.325450816605585e-06,
      "loss": 0.5333,
      "step": 20610
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6107162299172378,
      "learning_rate": 6.324528342016346e-06,
      "loss": 0.3739,
      "step": 20611
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3483001926260649,
      "learning_rate": 6.323605903587606e-06,
      "loss": 0.2138,
      "step": 20612
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24221929662765002,
      "learning_rate": 6.322683501328436e-06,
      "loss": 0.2235,
      "step": 20613
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.39724707562663936,
      "learning_rate": 6.32176113524792e-06,
      "loss": 0.1805,
      "step": 20614
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.598558446065084,
      "learning_rate": 6.320838805355129e-06,
      "loss": 0.3101,
      "step": 20615
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3780775900705086,
      "learning_rate": 6.3199165116591345e-06,
      "loss": 0.1855,
      "step": 20616
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4129950963106271,
      "learning_rate": 6.318994254169013e-06,
      "loss": 0.2302,
      "step": 20617
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.34610289153975154,
      "learning_rate": 6.318072032893834e-06,
      "loss": 0.2929,
      "step": 20618
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8501407758181797,
      "learning_rate": 6.3171498478426785e-06,
      "loss": 0.5352,
      "step": 20619
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8264523183110725,
      "learning_rate": 6.316227699024613e-06,
      "loss": 0.3666,
      "step": 20620
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2693643511132081,
      "learning_rate": 6.31530558644871e-06,
      "loss": 0.2006,
      "step": 20621
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2874433582908361,
      "learning_rate": 6.314383510124046e-06,
      "loss": 0.1751,
      "step": 20622
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3142289607760295,
      "learning_rate": 6.3134614700596895e-06,
      "loss": 0.2438,
      "step": 20623
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8780955525799375,
      "learning_rate": 6.312539466264708e-06,
      "loss": 0.487,
      "step": 20624
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3254291524972516,
      "learning_rate": 6.3116174987481795e-06,
      "loss": 0.1932,
      "step": 20625
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5367562908823554,
      "learning_rate": 6.310695567519174e-06,
      "loss": 0.3681,
      "step": 20626
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.042134069834725,
      "learning_rate": 6.309773672586753e-06,
      "loss": 0.175,
      "step": 20627
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5055206980056475,
      "learning_rate": 6.308851813959998e-06,
      "loss": 0.8254,
      "step": 20628
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3273236364962747,
      "learning_rate": 6.3079299916479745e-06,
      "loss": 0.154,
      "step": 20629
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.23633923374373286,
      "learning_rate": 6.307008205659746e-06,
      "loss": 0.2343,
      "step": 20630
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4384189846693015,
      "learning_rate": 6.30608645600439e-06,
      "loss": 0.2053,
      "step": 20631
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0378390284667722,
      "learning_rate": 6.30516474269097e-06,
      "loss": 0.3215,
      "step": 20632
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.38709707882850997,
      "learning_rate": 6.30424306572855e-06,
      "loss": 0.2217,
      "step": 20633
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.26896422912264456,
      "learning_rate": 6.30332142512621e-06,
      "loss": 0.1812,
      "step": 20634
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.640953477971037,
      "learning_rate": 6.302399820893007e-06,
      "loss": 0.8388,
      "step": 20635
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33103666434316337,
      "learning_rate": 6.301478253038012e-06,
      "loss": 0.2339,
      "step": 20636
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.439967237246359,
      "learning_rate": 6.30055672157029e-06,
      "loss": 0.8182,
      "step": 20637
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2988692213963256,
      "learning_rate": 6.29963522649891e-06,
      "loss": 0.1625,
      "step": 20638
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5008058962115187,
      "learning_rate": 6.2987137678329315e-06,
      "loss": 0.3108,
      "step": 20639
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8443867985906647,
      "learning_rate": 6.297792345581429e-06,
      "loss": 0.2541,
      "step": 20640
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5620790376360447,
      "learning_rate": 6.296870959753464e-06,
      "loss": 0.3225,
      "step": 20641
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24624493862572266,
      "learning_rate": 6.295949610358099e-06,
      "loss": 0.1564,
      "step": 20642
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9504168540281739,
      "learning_rate": 6.295028297404402e-06,
      "loss": 0.4225,
      "step": 20643
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30394048716637573,
      "learning_rate": 6.294107020901432e-06,
      "loss": 0.1894,
      "step": 20644
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.399050725830084,
      "learning_rate": 6.293185780858261e-06,
      "loss": 0.2979,
      "step": 20645
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.322258772091794,
      "learning_rate": 6.292264577283946e-06,
      "loss": 0.8097,
      "step": 20646
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3421543626297946,
      "learning_rate": 6.2913434101875525e-06,
      "loss": 0.1989,
      "step": 20647
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3642612068731874,
      "learning_rate": 6.290422279578143e-06,
      "loss": 0.2733,
      "step": 20648
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.43554118575459416,
      "learning_rate": 6.2895011854647814e-06,
      "loss": 0.2829,
      "step": 20649
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7368905713062532,
      "learning_rate": 6.288580127856524e-06,
      "loss": 0.4783,
      "step": 20650
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.13933851969857422,
      "learning_rate": 6.287659106762439e-06,
      "loss": 0.0705,
      "step": 20651
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6082758152995448,
      "learning_rate": 6.2867381221915865e-06,
      "loss": 0.3681,
      "step": 20652
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3702381381358343,
      "learning_rate": 6.285817174153024e-06,
      "loss": 0.2145,
      "step": 20653
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3220940707383158,
      "learning_rate": 6.284896262655815e-06,
      "loss": 0.2825,
      "step": 20654
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9880038259035123,
      "learning_rate": 6.28397538770902e-06,
      "loss": 0.4797,
      "step": 20655
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8193900257573057,
      "learning_rate": 6.283054549321694e-06,
      "loss": 0.5701,
      "step": 20656
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27390910674811875,
      "learning_rate": 6.2821337475029056e-06,
      "loss": 0.1891,
      "step": 20657
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6254372860955343,
      "learning_rate": 6.281212982261706e-06,
      "loss": 0.2944,
      "step": 20658
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6691593243982504,
      "learning_rate": 6.280292253607156e-06,
      "loss": 0.3392,
      "step": 20659
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27606154127405036,
      "learning_rate": 6.2793715615483175e-06,
      "loss": 0.1776,
      "step": 20660
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3763978179406121,
      "learning_rate": 6.278450906094246e-06,
      "loss": 0.2718,
      "step": 20661
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.24361779400588943,
      "learning_rate": 6.277530287253995e-06,
      "loss": 0.1384,
      "step": 20662
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.0785811952664228,
      "learning_rate": 6.276609705036628e-06,
      "loss": 0.5042,
      "step": 20663
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8268667176350267,
      "learning_rate": 6.275689159451202e-06,
      "loss": 0.599,
      "step": 20664
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.27684057870979145,
      "learning_rate": 6.274768650506766e-06,
      "loss": 0.2492,
      "step": 20665
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3358415893532185,
      "learning_rate": 6.273848178212385e-06,
      "loss": 0.1676,
      "step": 20666
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.41912812945175765,
      "learning_rate": 6.272927742577114e-06,
      "loss": 0.3038,
      "step": 20667
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.8420680104431629,
      "learning_rate": 6.272007343610003e-06,
      "loss": 0.2976,
      "step": 20668
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.25157292790929126,
      "learning_rate": 6.2710869813201125e-06,
      "loss": 0.1316,
      "step": 20669
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3545645077023701,
      "learning_rate": 6.270166655716492e-06,
      "loss": 0.0666,
      "step": 20670
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2961459953199734,
      "learning_rate": 6.2692463668082015e-06,
      "loss": 0.1934,
      "step": 20671
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3379203784854807,
      "learning_rate": 6.268326114604295e-06,
      "loss": 0.2762,
      "step": 20672
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6213783349214306,
      "learning_rate": 6.267405899113821e-06,
      "loss": 0.3291,
      "step": 20673
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9275237956059802,
      "learning_rate": 6.266485720345838e-06,
      "loss": 0.5376,
      "step": 20674
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3207166299674538,
      "learning_rate": 6.265565578309396e-06,
      "loss": 0.203,
      "step": 20675
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4549102117521005,
      "learning_rate": 6.264645473013547e-06,
      "loss": 0.3762,
      "step": 20676
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.29269310631404194,
      "learning_rate": 6.263725404467347e-06,
      "loss": 0.2302,
      "step": 20677
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3612010172675393,
      "learning_rate": 6.262805372679846e-06,
      "loss": 0.5917,
      "step": 20678
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.18869098386749666,
      "learning_rate": 6.261885377660094e-06,
      "loss": 0.0722,
      "step": 20679
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2705517228415291,
      "learning_rate": 6.260965419417146e-06,
      "loss": 0.2245,
      "step": 20680
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7356986176667726,
      "learning_rate": 6.26004549796005e-06,
      "loss": 0.2916,
      "step": 20681
    },
    {
      "epoch": 0.63,
      "grad_norm": 2.264808588968482,
      "learning_rate": 6.259125613297852e-06,
      "loss": 0.5217,
      "step": 20682
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.36312678102093193,
      "learning_rate": 6.258205765439612e-06,
      "loss": 0.2605,
      "step": 20683
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2774410110929129,
      "learning_rate": 6.257285954394375e-06,
      "loss": 0.1828,
      "step": 20684
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.824432882342336,
      "learning_rate": 6.256366180171188e-06,
      "loss": 0.523,
      "step": 20685
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4448968950162977,
      "learning_rate": 6.255446442779105e-06,
      "loss": 0.3038,
      "step": 20686
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.427898780169013,
      "learning_rate": 6.2545267422271715e-06,
      "loss": 0.2067,
      "step": 20687
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22309598782146542,
      "learning_rate": 6.253607078524433e-06,
      "loss": 0.0696,
      "step": 20688
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.33027765955243743,
      "learning_rate": 6.2526874516799445e-06,
      "loss": 0.2469,
      "step": 20689
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3131855853664974,
      "learning_rate": 6.251767861702749e-06,
      "loss": 0.2277,
      "step": 20690
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7564992566624308,
      "learning_rate": 6.250848308601893e-06,
      "loss": 0.4635,
      "step": 20691
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5973556739293729,
      "learning_rate": 6.249928792386429e-06,
      "loss": 0.2687,
      "step": 20692
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4335371188114109,
      "learning_rate": 6.249009313065397e-06,
      "loss": 0.2842,
      "step": 20693
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3928373825290742,
      "learning_rate": 6.248089870647842e-06,
      "loss": 0.2413,
      "step": 20694
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4576176382414131,
      "learning_rate": 6.247170465142817e-06,
      "loss": 0.3313,
      "step": 20695
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4080210127979049,
      "learning_rate": 6.246251096559361e-06,
      "loss": 0.2388,
      "step": 20696
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.22274061015565325,
      "learning_rate": 6.245331764906526e-06,
      "loss": 0.074,
      "step": 20697
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3222546630509167,
      "learning_rate": 6.244412470193351e-06,
      "loss": 0.2691,
      "step": 20698
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4285479526661388,
      "learning_rate": 6.2434932124288815e-06,
      "loss": 0.1529,
      "step": 20699
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4586344966487179,
      "learning_rate": 6.242573991622163e-06,
      "loss": 0.3475,
      "step": 20700
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4060998964766305,
      "learning_rate": 6.241654807782238e-06,
      "loss": 0.2409,
      "step": 20701
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6181074990685665,
      "learning_rate": 6.240735660918144e-06,
      "loss": 0.3373,
      "step": 20702
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3340376571877855,
      "learning_rate": 6.239816551038938e-06,
      "loss": 0.2331,
      "step": 20703
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.4443775705001431,
      "learning_rate": 6.238897478153649e-06,
      "loss": 0.8655,
      "step": 20704
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45227522825736766,
      "learning_rate": 6.2379784422713264e-06,
      "loss": 0.0212,
      "step": 20705
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.43622015608896614,
      "learning_rate": 6.237059443401009e-06,
      "loss": 0.225,
      "step": 20706
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.2381756082350269,
      "learning_rate": 6.23614048155174e-06,
      "loss": 0.1926,
      "step": 20707
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31263946174920515,
      "learning_rate": 6.235221556732555e-06,
      "loss": 0.2556,
      "step": 20708
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.9594354643310635,
      "learning_rate": 6.2343026689525035e-06,
      "loss": 0.2989,
      "step": 20709
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.7337214376354483,
      "learning_rate": 6.233383818220621e-06,
      "loss": 0.3178,
      "step": 20710
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.371588001438821,
      "learning_rate": 6.232465004545946e-06,
      "loss": 0.229,
      "step": 20711
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5074128610862485,
      "learning_rate": 6.231546227937522e-06,
      "loss": 0.2459,
      "step": 20712
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.45976709892483375,
      "learning_rate": 6.230627488404386e-06,
      "loss": 0.3124,
      "step": 20713
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4381731143935066,
      "learning_rate": 6.229708785955574e-06,
      "loss": 0.2359,
      "step": 20714
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3215180519721603,
      "learning_rate": 6.22879012060013e-06,
      "loss": 0.1478,
      "step": 20715
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.30258894750141224,
      "learning_rate": 6.227871492347091e-06,
      "loss": 0.1945,
      "step": 20716
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5960763930694948,
      "learning_rate": 6.226952901205491e-06,
      "loss": 0.2437,
      "step": 20717
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5583174728578826,
      "learning_rate": 6.226034347184372e-06,
      "loss": 0.3421,
      "step": 20718
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.34925176709108025,
      "learning_rate": 6.225115830292765e-06,
      "loss": 0.3047,
      "step": 20719
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4073160456871149,
      "learning_rate": 6.224197350539714e-06,
      "loss": 0.1151,
      "step": 20720
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3785072979039252,
      "learning_rate": 6.223278907934251e-06,
      "loss": 0.2822,
      "step": 20721
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.1359113945397035,
      "learning_rate": 6.222360502485414e-06,
      "loss": 0.3381,
      "step": 20722
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.3119699305291528,
      "learning_rate": 6.221442134202236e-06,
      "loss": 0.1881,
      "step": 20723
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.417673267309729,
      "learning_rate": 6.220523803093756e-06,
      "loss": 0.3036,
      "step": 20724
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31348985501983484,
      "learning_rate": 6.219605509169e-06,
      "loss": 0.2008,
      "step": 20725
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.601963864050546,
      "learning_rate": 6.218687252437014e-06,
      "loss": 0.3376,
      "step": 20726
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.5179677240413962,
      "learning_rate": 6.217769032906829e-06,
      "loss": 0.1723,
      "step": 20727
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5871545757466723,
      "learning_rate": 6.216850850587471e-06,
      "loss": 0.6735,
      "step": 20728
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.3165335783601394,
      "learning_rate": 6.215932705487984e-06,
      "loss": 0.1533,
      "step": 20729
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.4088308963849328,
      "learning_rate": 6.215014597617397e-06,
      "loss": 0.3057,
      "step": 20730
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.31125688775590493,
      "learning_rate": 6.214096526984735e-06,
      "loss": 0.2147,
      "step": 20731
    },
    {
      "epoch": 0.63,
      "grad_norm": 1.5886308573040784,
      "learning_rate": 6.213178493599044e-06,
      "loss": 0.584,
      "step": 20732
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.6227396615910884,
      "learning_rate": 6.212260497469347e-06,
      "loss": 0.2898,
      "step": 20733
    },
    {
      "epoch": 0.63,
      "grad_norm": 0.41313177216184216,
      "learning_rate": 6.211342538604673e-06,
      "loss": 0.2957,
      "step": 20734
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3643420654243935,
      "learning_rate": 6.210424617014064e-06,
      "loss": 0.2563,
      "step": 20735
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.15259749215453886,
      "learning_rate": 6.209506732706543e-06,
      "loss": 0.067,
      "step": 20736
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30483312874085294,
      "learning_rate": 6.2085888856911395e-06,
      "loss": 0.2761,
      "step": 20737
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2924511828702785,
      "learning_rate": 6.207671075976889e-06,
      "loss": 0.1737,
      "step": 20738
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.47594980266317743,
      "learning_rate": 6.206753303572818e-06,
      "loss": 0.3244,
      "step": 20739
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1594457374098777,
      "learning_rate": 6.205835568487952e-06,
      "loss": 0.137,
      "step": 20740
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5894483823075434,
      "learning_rate": 6.2049178707313286e-06,
      "loss": 0.6455,
      "step": 20741
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2731397233306701,
      "learning_rate": 6.2040002103119715e-06,
      "loss": 0.2271,
      "step": 20742
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36236649351781164,
      "learning_rate": 6.203082587238907e-06,
      "loss": 0.2775,
      "step": 20743
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6895940998041553,
      "learning_rate": 6.202165001521168e-06,
      "loss": 0.3181,
      "step": 20744
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8085205079784188,
      "learning_rate": 6.201247453167776e-06,
      "loss": 0.4324,
      "step": 20745
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24245356724799644,
      "learning_rate": 6.200329942187764e-06,
      "loss": 0.1441,
      "step": 20746
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.68420474325803,
      "learning_rate": 6.199412468590158e-06,
      "loss": 0.4311,
      "step": 20747
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35953467913155873,
      "learning_rate": 6.19849503238398e-06,
      "loss": 0.2194,
      "step": 20748
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2529928373544666,
      "learning_rate": 6.19757763357826e-06,
      "loss": 0.1915,
      "step": 20749
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8467971655451506,
      "learning_rate": 6.1966602721820225e-06,
      "loss": 0.5524,
      "step": 20750
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7048914149106946,
      "learning_rate": 6.1957429482042884e-06,
      "loss": 0.3162,
      "step": 20751
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5147607110521397,
      "learning_rate": 6.194825661654093e-06,
      "loss": 0.3363,
      "step": 20752
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30960226888501596,
      "learning_rate": 6.1939084125404525e-06,
      "loss": 0.2037,
      "step": 20753
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.48374985965895245,
      "learning_rate": 6.192991200872392e-06,
      "loss": 0.3634,
      "step": 20754
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.41057593208155546,
      "learning_rate": 6.19207402665894e-06,
      "loss": 0.2454,
      "step": 20755
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7997954329146982,
      "learning_rate": 6.191156889909115e-06,
      "loss": 0.5127,
      "step": 20756
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2978129705804799,
      "learning_rate": 6.19023979063194e-06,
      "loss": 0.2076,
      "step": 20757
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29060526138431336,
      "learning_rate": 6.189322728836442e-06,
      "loss": 0.1653,
      "step": 20758
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.091139377500302,
      "learning_rate": 6.1884057045316405e-06,
      "loss": 0.304,
      "step": 20759
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4912996007615832,
      "learning_rate": 6.187488717726556e-06,
      "loss": 0.3315,
      "step": 20760
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25668254420303327,
      "learning_rate": 6.186571768430214e-06,
      "loss": 0.1842,
      "step": 20761
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5280134896864526,
      "learning_rate": 6.185654856651635e-06,
      "loss": 0.2043,
      "step": 20762
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5961586454827026,
      "learning_rate": 6.184737982399834e-06,
      "loss": 0.6605,
      "step": 20763
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9483979170603116,
      "learning_rate": 6.183821145683839e-06,
      "loss": 0.5299,
      "step": 20764
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7477954382075305,
      "learning_rate": 6.182904346512668e-06,
      "loss": 0.2195,
      "step": 20765
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24995346121288678,
      "learning_rate": 6.181987584895336e-06,
      "loss": 0.1895,
      "step": 20766
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.45883944446813996,
      "learning_rate": 6.18107086084087e-06,
      "loss": 0.3432,
      "step": 20767
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6227597498642082,
      "learning_rate": 6.180154174358286e-06,
      "loss": 0.2113,
      "step": 20768
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0001211426564371,
      "learning_rate": 6.1792375254566004e-06,
      "loss": 0.4536,
      "step": 20769
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2763187768384752,
      "learning_rate": 6.178320914144835e-06,
      "loss": 0.1855,
      "step": 20770
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.8056481367784267,
      "learning_rate": 6.1774043404320025e-06,
      "loss": 0.6645,
      "step": 20771
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3996299543798343,
      "learning_rate": 6.176487804327127e-06,
      "loss": 0.2105,
      "step": 20772
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.45279395285846674,
      "learning_rate": 6.175571305839223e-06,
      "loss": 0.3421,
      "step": 20773
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3445874609972862,
      "learning_rate": 6.174654844977304e-06,
      "loss": 0.1554,
      "step": 20774
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31600508067217015,
      "learning_rate": 6.173738421750393e-06,
      "loss": 0.201,
      "step": 20775
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3984376480249961,
      "learning_rate": 6.1728220361675e-06,
      "loss": 0.2517,
      "step": 20776
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5527669142878824,
      "learning_rate": 6.1719056882376415e-06,
      "loss": 0.2169,
      "step": 20777
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3075398477543841,
      "learning_rate": 6.170989377969838e-06,
      "loss": 0.2742,
      "step": 20778
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2262751121279457,
      "learning_rate": 6.1700731053731e-06,
      "loss": 0.0695,
      "step": 20779
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3838538060445028,
      "learning_rate": 6.169156870456441e-06,
      "loss": 0.2791,
      "step": 20780
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.212260147570771,
      "learning_rate": 6.16824067322888e-06,
      "loss": 0.2606,
      "step": 20781
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3822437028698702,
      "learning_rate": 6.167324513699429e-06,
      "loss": 0.7737,
      "step": 20782
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39370493878486856,
      "learning_rate": 6.1664083918770946e-06,
      "loss": 0.2479,
      "step": 20783
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3745521258068565,
      "learning_rate": 6.1654923077709e-06,
      "loss": 0.2631,
      "step": 20784
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2787357068437573,
      "learning_rate": 6.164576261389854e-06,
      "loss": 0.2655,
      "step": 20785
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5901796861448741,
      "learning_rate": 6.1636602527429675e-06,
      "loss": 0.4389,
      "step": 20786
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.13793181448674655,
      "learning_rate": 6.162744281839256e-06,
      "loss": 0.0677,
      "step": 20787
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3015484111998239,
      "learning_rate": 6.161828348687728e-06,
      "loss": 0.1548,
      "step": 20788
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5136621634028995,
      "learning_rate": 6.1609124532973915e-06,
      "loss": 0.3037,
      "step": 20789
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4482096850532322,
      "learning_rate": 6.159996595677266e-06,
      "loss": 0.2511,
      "step": 20790
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46633688859489697,
      "learning_rate": 6.159080775836357e-06,
      "loss": 0.3751,
      "step": 20791
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6818502444021575,
      "learning_rate": 6.158164993783674e-06,
      "loss": 0.2208,
      "step": 20792
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3167554501938912,
      "learning_rate": 6.1572492495282296e-06,
      "loss": 0.255,
      "step": 20793
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.770844961090757,
      "learning_rate": 6.156333543079032e-06,
      "loss": 0.3008,
      "step": 20794
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6364323852461575,
      "learning_rate": 6.155417874445084e-06,
      "loss": 0.4043,
      "step": 20795
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17713403132916944,
      "learning_rate": 6.154502243635405e-06,
      "loss": 0.1616,
      "step": 20796
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5864775951787005,
      "learning_rate": 6.1535866506589945e-06,
      "loss": 0.2462,
      "step": 20797
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33354832054036726,
      "learning_rate": 6.152671095524869e-06,
      "loss": 0.2029,
      "step": 20798
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3015892263173654,
      "learning_rate": 6.151755578242029e-06,
      "loss": 0.5297,
      "step": 20799
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0259712932146616,
      "learning_rate": 6.150840098819484e-06,
      "loss": 0.5987,
      "step": 20800
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5586134027331482,
      "learning_rate": 6.149924657266242e-06,
      "loss": 0.2382,
      "step": 20801
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5263805812353315,
      "learning_rate": 6.149009253591307e-06,
      "loss": 0.291,
      "step": 20802
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2642929131840165,
      "learning_rate": 6.1480938878036835e-06,
      "loss": 0.2365,
      "step": 20803
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8902344754296762,
      "learning_rate": 6.147178559912383e-06,
      "loss": 0.527,
      "step": 20804
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.1688857189945558,
      "learning_rate": 6.146263269926408e-06,
      "loss": 0.0701,
      "step": 20805
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.902167353811049,
      "learning_rate": 6.145348017854761e-06,
      "loss": 0.4274,
      "step": 20806
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3245013997159389,
      "learning_rate": 6.1444328037064505e-06,
      "loss": 0.2065,
      "step": 20807
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32971926087029424,
      "learning_rate": 6.143517627490478e-06,
      "loss": 0.2661,
      "step": 20808
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9513513101640456,
      "learning_rate": 6.1426024892158455e-06,
      "loss": 0.5574,
      "step": 20809
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8207179163594455,
      "learning_rate": 6.141687388891562e-06,
      "loss": 0.4397,
      "step": 20810
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3486402815689641,
      "learning_rate": 6.140772326526627e-06,
      "loss": 0.1946,
      "step": 20811
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4632945270036634,
      "learning_rate": 6.139857302130042e-06,
      "loss": 0.3693,
      "step": 20812
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3840045388069868,
      "learning_rate": 6.138942315710814e-06,
      "loss": 0.0123,
      "step": 20813
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21316647495582466,
      "learning_rate": 6.138027367277941e-06,
      "loss": 0.1806,
      "step": 20814
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3866077048569837,
      "learning_rate": 6.137112456840421e-06,
      "loss": 0.2439,
      "step": 20815
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.282462598118429,
      "learning_rate": 6.136197584407265e-06,
      "loss": 0.1826,
      "step": 20816
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.7051050632562819,
      "learning_rate": 6.135282749987468e-06,
      "loss": 0.6901,
      "step": 20817
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9786792597305253,
      "learning_rate": 6.134367953590029e-06,
      "loss": 0.4017,
      "step": 20818
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43349069438412424,
      "learning_rate": 6.1334531952239525e-06,
      "loss": 0.3246,
      "step": 20819
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26646524375688474,
      "learning_rate": 6.132538474898235e-06,
      "loss": 0.1875,
      "step": 20820
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43123723778875417,
      "learning_rate": 6.131623792621873e-06,
      "loss": 0.304,
      "step": 20821
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7261626486422367,
      "learning_rate": 6.130709148403873e-06,
      "loss": 0.309,
      "step": 20822
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43836253765154165,
      "learning_rate": 6.1297945422532265e-06,
      "loss": 0.2218,
      "step": 20823
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23309525348600166,
      "learning_rate": 6.1288799741789385e-06,
      "loss": 0.0732,
      "step": 20824
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8108270680617263,
      "learning_rate": 6.127965444190001e-06,
      "loss": 0.3376,
      "step": 20825
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26304142673148767,
      "learning_rate": 6.127050952295411e-06,
      "loss": 0.2274,
      "step": 20826
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8187053237629969,
      "learning_rate": 6.1261364985041715e-06,
      "loss": 0.329,
      "step": 20827
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.737897063615463,
      "learning_rate": 6.125222082825277e-06,
      "loss": 0.4546,
      "step": 20828
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.28117487588039963,
      "learning_rate": 6.124307705267718e-06,
      "loss": 0.0718,
      "step": 20829
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3606462882288582,
      "learning_rate": 6.1233933658405e-06,
      "loss": 0.2931,
      "step": 20830
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.43587807559809527,
      "learning_rate": 6.122479064552613e-06,
      "loss": 0.2285,
      "step": 20831
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31203315302232354,
      "learning_rate": 6.121564801413048e-06,
      "loss": 0.2372,
      "step": 20832
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2289763902844221,
      "learning_rate": 6.1206505764308086e-06,
      "loss": 0.0671,
      "step": 20833
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34690755041212035,
      "learning_rate": 6.119736389614886e-06,
      "loss": 0.2782,
      "step": 20834
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.041781872801704,
      "learning_rate": 6.118822240974269e-06,
      "loss": 0.3281,
      "step": 20835
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.783985779316722,
      "learning_rate": 6.11790813051796e-06,
      "loss": 0.3897,
      "step": 20836
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3969549496442615,
      "learning_rate": 6.116994058254948e-06,
      "loss": 0.2517,
      "step": 20837
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36863051703415095,
      "learning_rate": 6.116080024194224e-06,
      "loss": 0.2848,
      "step": 20838
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.334853357424721,
      "learning_rate": 6.115166028344786e-06,
      "loss": 0.2383,
      "step": 20839
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6118829250907534,
      "learning_rate": 6.114252070715622e-06,
      "loss": 0.7346,
      "step": 20840
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2782291392256884,
      "learning_rate": 6.11333815131572e-06,
      "loss": 0.2272,
      "step": 20841
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.17792743141614337,
      "learning_rate": 6.1124242701540815e-06,
      "loss": 0.0783,
      "step": 20842
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33528508498806064,
      "learning_rate": 6.1115104272396905e-06,
      "loss": 0.2653,
      "step": 20843
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3663194581014632,
      "learning_rate": 6.110596622581539e-06,
      "loss": 0.2208,
      "step": 20844
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6506633903044315,
      "learning_rate": 6.109682856188619e-06,
      "loss": 0.3313,
      "step": 20845
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1920674687963095,
      "learning_rate": 6.1087691280699155e-06,
      "loss": 0.3732,
      "step": 20846
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3492480821724163,
      "learning_rate": 6.107855438234426e-06,
      "loss": 0.231,
      "step": 20847
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44103331272011176,
      "learning_rate": 6.106941786691134e-06,
      "loss": 0.2251,
      "step": 20848
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3647021696647976,
      "learning_rate": 6.10602817344903e-06,
      "loss": 0.6359,
      "step": 20849
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2953694489362076,
      "learning_rate": 6.105114598517103e-06,
      "loss": 0.2313,
      "step": 20850
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3235114243427561,
      "learning_rate": 6.104201061904339e-06,
      "loss": 0.1775,
      "step": 20851
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36177632522038955,
      "learning_rate": 6.103287563619724e-06,
      "loss": 0.1898,
      "step": 20852
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3490055450998585,
      "learning_rate": 6.102374103672252e-06,
      "loss": 0.2466,
      "step": 20853
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8020339995088274,
      "learning_rate": 6.101460682070905e-06,
      "loss": 0.3445,
      "step": 20854
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32218015603647093,
      "learning_rate": 6.100547298824671e-06,
      "loss": 0.2674,
      "step": 20855
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.893839996236883,
      "learning_rate": 6.099633953942536e-06,
      "loss": 0.4689,
      "step": 20856
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26884613637010935,
      "learning_rate": 6.098720647433485e-06,
      "loss": 0.1972,
      "step": 20857
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3692521122352295,
      "learning_rate": 6.097807379306501e-06,
      "loss": 0.6576,
      "step": 20858
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29629420275181106,
      "learning_rate": 6.096894149570576e-06,
      "loss": 0.084,
      "step": 20859
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7808270514725992,
      "learning_rate": 6.09598095823469e-06,
      "loss": 0.4511,
      "step": 20860
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2762676187218354,
      "learning_rate": 6.095067805307826e-06,
      "loss": 0.1861,
      "step": 20861
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2686926822989849,
      "learning_rate": 6.094154690798972e-06,
      "loss": 0.293,
      "step": 20862
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32421367096210246,
      "learning_rate": 6.093241614717109e-06,
      "loss": 0.2048,
      "step": 20863
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.436022630386409,
      "learning_rate": 6.0923285770712184e-06,
      "loss": 0.7887,
      "step": 20864
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2847928919724393,
      "learning_rate": 6.091415577870287e-06,
      "loss": 0.155,
      "step": 20865
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38214571232662686,
      "learning_rate": 6.090502617123296e-06,
      "loss": 0.2754,
      "step": 20866
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38834029258846403,
      "learning_rate": 6.089589694839222e-06,
      "loss": 0.2091,
      "step": 20867
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38191552439312876,
      "learning_rate": 6.0886768110270564e-06,
      "loss": 0.2131,
      "step": 20868
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7329027315903137,
      "learning_rate": 6.087763965695774e-06,
      "loss": 0.4448,
      "step": 20869
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2630141835430191,
      "learning_rate": 6.086851158854357e-06,
      "loss": 0.1713,
      "step": 20870
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6705106344610806,
      "learning_rate": 6.085938390511786e-06,
      "loss": 0.4366,
      "step": 20871
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2439504181689579,
      "learning_rate": 6.085025660677037e-06,
      "loss": 0.156,
      "step": 20872
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3350472957944137,
      "learning_rate": 6.0841129693590995e-06,
      "loss": 0.2967,
      "step": 20873
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31859685195754023,
      "learning_rate": 6.083200316566947e-06,
      "loss": 0.1561,
      "step": 20874
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40290253920961705,
      "learning_rate": 6.082287702309555e-06,
      "loss": 0.3042,
      "step": 20875
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3811983413913204,
      "learning_rate": 6.081375126595908e-06,
      "loss": 0.1265,
      "step": 20876
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1781274863713525,
      "learning_rate": 6.080462589434981e-06,
      "loss": 0.4747,
      "step": 20877
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5401004015358047,
      "learning_rate": 6.079550090835751e-06,
      "loss": 0.2677,
      "step": 20878
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8986911004714153,
      "learning_rate": 6.0786376308072e-06,
      "loss": 0.4433,
      "step": 20879
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24563728916986288,
      "learning_rate": 6.077725209358303e-06,
      "loss": 0.22,
      "step": 20880
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3458026878331302,
      "learning_rate": 6.076812826498033e-06,
      "loss": 0.1776,
      "step": 20881
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4960069394816884,
      "learning_rate": 6.0759004822353705e-06,
      "loss": 0.7904,
      "step": 20882
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2068987718736958,
      "learning_rate": 6.074988176579293e-06,
      "loss": 0.07,
      "step": 20883
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7893742683644867,
      "learning_rate": 6.074075909538767e-06,
      "loss": 0.2633,
      "step": 20884
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3805362196907255,
      "learning_rate": 6.073163681122778e-06,
      "loss": 0.2016,
      "step": 20885
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4224204635406952,
      "learning_rate": 6.072251491340299e-06,
      "loss": 0.3042,
      "step": 20886
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6166822247645125,
      "learning_rate": 6.071339340200298e-06,
      "loss": 0.2836,
      "step": 20887
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5253198605376826,
      "learning_rate": 6.070427227711756e-06,
      "loss": 0.3337,
      "step": 20888
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3354999838943555,
      "learning_rate": 6.069515153883645e-06,
      "loss": 0.2025,
      "step": 20889
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1720758030970724,
      "learning_rate": 6.068603118724932e-06,
      "loss": 0.6853,
      "step": 20890
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30700125746734974,
      "learning_rate": 6.067691122244599e-06,
      "loss": 0.2444,
      "step": 20891
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37560572627047584,
      "learning_rate": 6.0667791644516146e-06,
      "loss": 0.2197,
      "step": 20892
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2779129022540271,
      "learning_rate": 6.0658672453549495e-06,
      "loss": 0.1778,
      "step": 20893
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36564976406404043,
      "learning_rate": 6.0649553649635786e-06,
      "loss": 0.1008,
      "step": 20894
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9068987574946896,
      "learning_rate": 6.064043523286471e-06,
      "loss": 0.4567,
      "step": 20895
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33481776826857135,
      "learning_rate": 6.063131720332595e-06,
      "loss": 0.257,
      "step": 20896
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36719724764808903,
      "learning_rate": 6.062219956110931e-06,
      "loss": 0.235,
      "step": 20897
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3131468327724271,
      "learning_rate": 6.061308230630436e-06,
      "loss": 0.2264,
      "step": 20898
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.13127115679529,
      "learning_rate": 6.060396543900091e-06,
      "loss": 0.479,
      "step": 20899
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8556756187834466,
      "learning_rate": 6.05948489592886e-06,
      "loss": 0.5999,
      "step": 20900
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38775958468679156,
      "learning_rate": 6.058573286725713e-06,
      "loss": 0.1794,
      "step": 20901
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29789541155276344,
      "learning_rate": 6.057661716299621e-06,
      "loss": 0.186,
      "step": 20902
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.489061836317966,
      "learning_rate": 6.056750184659549e-06,
      "loss": 0.2632,
      "step": 20903
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30006164531501645,
      "learning_rate": 6.0558386918144615e-06,
      "loss": 0.2579,
      "step": 20904
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8328456971873054,
      "learning_rate": 6.054927237773337e-06,
      "loss": 0.3826,
      "step": 20905
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3621682525232487,
      "learning_rate": 6.054015822545136e-06,
      "loss": 0.0633,
      "step": 20906
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30392286971149707,
      "learning_rate": 6.053104446138822e-06,
      "loss": 0.1973,
      "step": 20907
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.29051563081965,
      "learning_rate": 6.052193108563369e-06,
      "loss": 0.7964,
      "step": 20908
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32416943617530375,
      "learning_rate": 6.051281809827738e-06,
      "loss": 0.257,
      "step": 20909
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.698735133071477,
      "learning_rate": 6.050370549940893e-06,
      "loss": 0.7805,
      "step": 20910
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27354395561598355,
      "learning_rate": 6.049459328911806e-06,
      "loss": 0.1803,
      "step": 20911
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30088371100907363,
      "learning_rate": 6.048548146749437e-06,
      "loss": 0.2055,
      "step": 20912
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6221365072897633,
      "learning_rate": 6.047637003462749e-06,
      "loss": 0.2943,
      "step": 20913
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.298551615365507,
      "learning_rate": 6.046725899060713e-06,
      "loss": 0.6511,
      "step": 20914
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2282903739418153,
      "learning_rate": 6.045814833552286e-06,
      "loss": 0.203,
      "step": 20915
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37511671936603186,
      "learning_rate": 6.044903806946431e-06,
      "loss": 0.2886,
      "step": 20916
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9102082514108631,
      "learning_rate": 6.043992819252118e-06,
      "loss": 0.4304,
      "step": 20917
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.6595148650701317,
      "learning_rate": 6.0430818704783046e-06,
      "loss": 0.7949,
      "step": 20918
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2119280120087158,
      "learning_rate": 6.0421709606339525e-06,
      "loss": 0.0929,
      "step": 20919
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2752147745948554,
      "learning_rate": 6.041260089728027e-06,
      "loss": 0.1858,
      "step": 20920
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.324242674921646,
      "learning_rate": 6.040349257769487e-06,
      "loss": 0.3031,
      "step": 20921
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8806876133594161,
      "learning_rate": 6.039438464767289e-06,
      "loss": 0.2975,
      "step": 20922
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.440264152679746,
      "learning_rate": 6.038527710730403e-06,
      "loss": 0.2699,
      "step": 20923
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5468064046732317,
      "learning_rate": 6.037616995667783e-06,
      "loss": 0.1403,
      "step": 20924
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40050296402047486,
      "learning_rate": 6.036706319588392e-06,
      "loss": 0.322,
      "step": 20925
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1734122221537766,
      "learning_rate": 6.035795682501191e-06,
      "loss": 0.2555,
      "step": 20926
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3445897699677588,
      "learning_rate": 6.034885084415129e-06,
      "loss": 0.3157,
      "step": 20927
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38579453782585643,
      "learning_rate": 6.033974525339179e-06,
      "loss": 0.2121,
      "step": 20928
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.44453274916900926,
      "learning_rate": 6.033064005282292e-06,
      "loss": 0.2903,
      "step": 20929
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5207120955392355,
      "learning_rate": 6.032153524253424e-06,
      "loss": 0.2565,
      "step": 20930
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.45199280002044406,
      "learning_rate": 6.031243082261538e-06,
      "loss": 0.2733,
      "step": 20931
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2246629203569805,
      "learning_rate": 6.030332679315587e-06,
      "loss": 0.1841,
      "step": 20932
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5377945380159992,
      "learning_rate": 6.029422315424526e-06,
      "loss": 0.0919,
      "step": 20933
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3407556029077668,
      "learning_rate": 6.028511990597319e-06,
      "loss": 0.2489,
      "step": 20934
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9512192261068252,
      "learning_rate": 6.027601704842917e-06,
      "loss": 0.3137,
      "step": 20935
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.260546746648627,
      "learning_rate": 6.026691458170273e-06,
      "loss": 0.7593,
      "step": 20936
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9224008001165588,
      "learning_rate": 6.02578125058835e-06,
      "loss": 0.2771,
      "step": 20937
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6081623044781814,
      "learning_rate": 6.024871082106098e-06,
      "loss": 0.3033,
      "step": 20938
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22490558285696366,
      "learning_rate": 6.023960952732473e-06,
      "loss": 0.2028,
      "step": 20939
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7877984994548853,
      "learning_rate": 6.023050862476427e-06,
      "loss": 0.4398,
      "step": 20940
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.16113392085408393,
      "learning_rate": 6.022140811346917e-06,
      "loss": 0.0729,
      "step": 20941
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0536436789884311,
      "learning_rate": 6.0212307993528905e-06,
      "loss": 0.4497,
      "step": 20942
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32472507002188145,
      "learning_rate": 6.020320826503308e-06,
      "loss": 0.1916,
      "step": 20943
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2567840851348544,
      "learning_rate": 6.01941089280712e-06,
      "loss": 0.6388,
      "step": 20944
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.314880149914829,
      "learning_rate": 6.018500998273274e-06,
      "loss": 0.2383,
      "step": 20945
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.024597448921463,
      "learning_rate": 6.017591142910728e-06,
      "loss": 0.3349,
      "step": 20946
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3834481823522024,
      "learning_rate": 6.016681326728431e-06,
      "loss": 0.2541,
      "step": 20947
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5056922918812246,
      "learning_rate": 6.015771549735329e-06,
      "loss": 0.245,
      "step": 20948
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5052628509149504,
      "learning_rate": 6.014861811940381e-06,
      "loss": 0.5419,
      "step": 20949
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.21822849823878215,
      "learning_rate": 6.013952113352532e-06,
      "loss": 0.1798,
      "step": 20950
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3844312104207948,
      "learning_rate": 6.0130424539807375e-06,
      "loss": 0.2164,
      "step": 20951
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3099667150324561,
      "learning_rate": 6.012132833833942e-06,
      "loss": 0.1907,
      "step": 20952
    },
    {
      "epoch": 0.64,
      "grad_norm": 3.2763077159068072,
      "learning_rate": 6.011223252921091e-06,
      "loss": 0.8302,
      "step": 20953
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0244525293896476,
      "learning_rate": 6.010313711251144e-06,
      "loss": 0.4095,
      "step": 20954
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9384749124986238,
      "learning_rate": 6.009404208833042e-06,
      "loss": 0.4163,
      "step": 20955
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2817258366737883,
      "learning_rate": 6.0084947456757324e-06,
      "loss": 0.1812,
      "step": 20956
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.36055052172044333,
      "learning_rate": 6.007585321788168e-06,
      "loss": 0.2875,
      "step": 20957
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4046959860914907,
      "learning_rate": 6.006675937179291e-06,
      "loss": 0.2158,
      "step": 20958
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2121174436117336,
      "learning_rate": 6.005766591858048e-06,
      "loss": 0.0888,
      "step": 20959
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.892327750735175,
      "learning_rate": 6.0048572858333905e-06,
      "loss": 0.4249,
      "step": 20960
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3055999778645244,
      "learning_rate": 6.003948019114263e-06,
      "loss": 0.1951,
      "step": 20961
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3188283451263707,
      "learning_rate": 6.003038791709606e-06,
      "loss": 0.6696,
      "step": 20962
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5384511093579636,
      "learning_rate": 6.002129603628372e-06,
      "loss": 0.267,
      "step": 20963
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5444554523147782,
      "learning_rate": 6.0012204548795015e-06,
      "loss": 0.4527,
      "step": 20964
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31116705375547904,
      "learning_rate": 6.000311345471936e-06,
      "loss": 0.1167,
      "step": 20965
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37287331524222744,
      "learning_rate": 5.999402275414627e-06,
      "loss": 0.3137,
      "step": 20966
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7685968767325332,
      "learning_rate": 5.998493244716517e-06,
      "loss": 0.0478,
      "step": 20967
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5068900676013747,
      "learning_rate": 5.997584253386541e-06,
      "loss": 0.3315,
      "step": 20968
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23653405464226587,
      "learning_rate": 5.996675301433652e-06,
      "loss": 0.1899,
      "step": 20969
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.39152779590059544,
      "learning_rate": 5.995766388866791e-06,
      "loss": 0.2537,
      "step": 20970
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5125204451564552,
      "learning_rate": 5.994857515694893e-06,
      "loss": 0.2522,
      "step": 20971
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0266997528182866,
      "learning_rate": 5.9939486819269075e-06,
      "loss": 0.2998,
      "step": 20972
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.873614531629548,
      "learning_rate": 5.99303988757177e-06,
      "loss": 0.4208,
      "step": 20973
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.25087061709980907,
      "learning_rate": 5.992131132638428e-06,
      "loss": 0.1893,
      "step": 20974
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3623646169111488,
      "learning_rate": 5.991222417135818e-06,
      "loss": 0.3034,
      "step": 20975
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2432107789748812,
      "learning_rate": 5.990313741072879e-06,
      "loss": 0.2506,
      "step": 20976
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5474173451957718,
      "learning_rate": 5.989405104458556e-06,
      "loss": 0.2762,
      "step": 20977
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31021579819274714,
      "learning_rate": 5.988496507301784e-06,
      "loss": 0.0945,
      "step": 20978
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3676570496581621,
      "learning_rate": 5.9875879496115e-06,
      "loss": 0.281,
      "step": 20979
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0480076730221979,
      "learning_rate": 5.9866794313966505e-06,
      "loss": 0.3336,
      "step": 20980
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3441209667727584,
      "learning_rate": 5.985770952666168e-06,
      "loss": 0.3019,
      "step": 20981
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.846402573596172,
      "learning_rate": 5.984862513428991e-06,
      "loss": 0.334,
      "step": 20982
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.594521430218786,
      "learning_rate": 5.983954113694059e-06,
      "loss": 0.3336,
      "step": 20983
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33088651473268144,
      "learning_rate": 5.983045753470308e-06,
      "loss": 0.2142,
      "step": 20984
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.230525226407223,
      "learning_rate": 5.98213743276667e-06,
      "loss": 0.1715,
      "step": 20985
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.33117872460691494,
      "learning_rate": 5.981229151592092e-06,
      "loss": 0.2772,
      "step": 20986
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.18033287783724788,
      "learning_rate": 5.980320909955501e-06,
      "loss": 0.0714,
      "step": 20987
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7427487331274154,
      "learning_rate": 5.979412707865836e-06,
      "loss": 0.4398,
      "step": 20988
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.42293826071588914,
      "learning_rate": 5.978504545332033e-06,
      "loss": 0.248,
      "step": 20989
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0029159553780687,
      "learning_rate": 5.977596422363026e-06,
      "loss": 0.4444,
      "step": 20990
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.40133805165204317,
      "learning_rate": 5.976688338967744e-06,
      "loss": 0.2581,
      "step": 20991
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3900115062555019,
      "learning_rate": 5.9757802951551305e-06,
      "loss": 0.2608,
      "step": 20992
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30326761477896946,
      "learning_rate": 5.974872290934114e-06,
      "loss": 0.2017,
      "step": 20993
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3522229065350786,
      "learning_rate": 5.973964326313626e-06,
      "loss": 0.5111,
      "step": 20994
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2579982777075163,
      "learning_rate": 5.973056401302606e-06,
      "loss": 0.2464,
      "step": 20995
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2761987581311992,
      "learning_rate": 5.972148515909979e-06,
      "loss": 0.1534,
      "step": 20996
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2918737969765396,
      "learning_rate": 5.9712406701446775e-06,
      "loss": 0.1942,
      "step": 20997
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.24745938631469871,
      "learning_rate": 5.970332864015642e-06,
      "loss": 0.2249,
      "step": 20998
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.0440901498818558,
      "learning_rate": 5.969425097531792e-06,
      "loss": 0.4403,
      "step": 20999
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.758317177320233,
      "learning_rate": 5.968517370702067e-06,
      "loss": 0.3851,
      "step": 21000
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.145285291581605,
      "learning_rate": 5.9676096835353984e-06,
      "loss": 0.4122,
      "step": 21001
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.27498532891236827,
      "learning_rate": 5.966702036040706e-06,
      "loss": 0.1776,
      "step": 21002
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.567425498530572,
      "learning_rate": 5.96579442822693e-06,
      "loss": 0.7429,
      "step": 21003
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30933722186072543,
      "learning_rate": 5.964886860102996e-06,
      "loss": 0.2288,
      "step": 21004
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.416228135631844,
      "learning_rate": 5.9639793316778285e-06,
      "loss": 0.2365,
      "step": 21005
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2546633926841336,
      "learning_rate": 5.963071842960364e-06,
      "loss": 0.1693,
      "step": 21006
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9135956347321338,
      "learning_rate": 5.962164393959528e-06,
      "loss": 0.4665,
      "step": 21007
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.37645969916152866,
      "learning_rate": 5.961256984684245e-06,
      "loss": 0.2031,
      "step": 21008
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.368927413316555,
      "learning_rate": 5.960349615143446e-06,
      "loss": 0.8076,
      "step": 21009
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30691987938827536,
      "learning_rate": 5.959442285346058e-06,
      "loss": 0.2089,
      "step": 21010
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35603727900474186,
      "learning_rate": 5.958534995301002e-06,
      "loss": 0.2151,
      "step": 21011
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.4219003699640216,
      "learning_rate": 5.957627745017211e-06,
      "loss": 0.5726,
      "step": 21012
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7479070605971341,
      "learning_rate": 5.9567205345036105e-06,
      "loss": 0.2687,
      "step": 21013
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.7062977584941397,
      "learning_rate": 5.95581336376912e-06,
      "loss": 0.386,
      "step": 21014
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3007271454382585,
      "learning_rate": 5.9549062328226715e-06,
      "loss": 0.0703,
      "step": 21015
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.288078149707924,
      "learning_rate": 5.953999141673186e-06,
      "loss": 0.2607,
      "step": 21016
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.29611634539242504,
      "learning_rate": 5.9530920903295845e-06,
      "loss": 0.1818,
      "step": 21017
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.5528498326308033,
      "learning_rate": 5.952185078800798e-06,
      "loss": 0.7458,
      "step": 21018
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2030257366402757,
      "learning_rate": 5.951278107095748e-06,
      "loss": 0.0935,
      "step": 21019
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.315388213351756,
      "learning_rate": 5.950371175223353e-06,
      "loss": 0.2498,
      "step": 21020
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2020966139104396,
      "learning_rate": 5.949464283192541e-06,
      "loss": 0.2621,
      "step": 21021
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4177514202812858,
      "learning_rate": 5.9485574310122315e-06,
      "loss": 0.3172,
      "step": 21022
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6971180825798419,
      "learning_rate": 5.947650618691345e-06,
      "loss": 0.2909,
      "step": 21023
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4560479017733366,
      "learning_rate": 5.946743846238808e-06,
      "loss": 0.2826,
      "step": 21024
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3203782337417977,
      "learning_rate": 5.945837113663537e-06,
      "loss": 0.2282,
      "step": 21025
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.22619742850666597,
      "learning_rate": 5.944930420974455e-06,
      "loss": 0.101,
      "step": 21026
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.3449548260966755,
      "learning_rate": 5.944023768180485e-06,
      "loss": 0.7494,
      "step": 21027
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2149433214183424,
      "learning_rate": 5.943117155290536e-06,
      "loss": 0.1771,
      "step": 21028
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32298499650093065,
      "learning_rate": 5.942210582313542e-06,
      "loss": 0.2493,
      "step": 21029
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2486415521967629,
      "learning_rate": 5.941304049258414e-06,
      "loss": 0.2484,
      "step": 21030
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8219289821605469,
      "learning_rate": 5.940397556134072e-06,
      "loss": 0.4629,
      "step": 21031
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6024802116214807,
      "learning_rate": 5.939491102949435e-06,
      "loss": 0.3004,
      "step": 21032
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.34584445405316266,
      "learning_rate": 5.938584689713421e-06,
      "loss": 0.2659,
      "step": 21033
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3319528235080125,
      "learning_rate": 5.937678316434944e-06,
      "loss": 0.2097,
      "step": 21034
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.4562488450414829,
      "learning_rate": 5.9367719831229275e-06,
      "loss": 0.2927,
      "step": 21035
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8647054263202935,
      "learning_rate": 5.935865689786285e-06,
      "loss": 0.5812,
      "step": 21036
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.20463793706353517,
      "learning_rate": 5.934959436433929e-06,
      "loss": 0.1117,
      "step": 21037
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3406714162334193,
      "learning_rate": 5.934053223074785e-06,
      "loss": 0.1681,
      "step": 21038
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.470382429330591,
      "learning_rate": 5.933147049717762e-06,
      "loss": 0.2316,
      "step": 21039
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.31956116389105377,
      "learning_rate": 5.932240916371777e-06,
      "loss": 0.2842,
      "step": 21040
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.6375922436455636,
      "learning_rate": 5.931334823045742e-06,
      "loss": 0.276,
      "step": 21041
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9991658011321412,
      "learning_rate": 5.930428769748578e-06,
      "loss": 0.5265,
      "step": 21042
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.3216758540343945,
      "learning_rate": 5.9295227564891875e-06,
      "loss": 0.1954,
      "step": 21043
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2996332960861596,
      "learning_rate": 5.928616783276498e-06,
      "loss": 0.677,
      "step": 21044
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.35767094139746114,
      "learning_rate": 5.927710850119415e-06,
      "loss": 0.2373,
      "step": 21045
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.46468178234815904,
      "learning_rate": 5.926804957026851e-06,
      "loss": 0.3116,
      "step": 21046
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.23774838789108135,
      "learning_rate": 5.925899104007722e-06,
      "loss": 0.1348,
      "step": 21047
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38076085210670657,
      "learning_rate": 5.924993291070938e-06,
      "loss": 0.2371,
      "step": 21048
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8290320085188604,
      "learning_rate": 5.924087518225405e-06,
      "loss": 0.299,
      "step": 21049
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9082070598573995,
      "learning_rate": 5.923181785480048e-06,
      "loss": 0.5113,
      "step": 21050
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.26145056573142905,
      "learning_rate": 5.922276092843765e-06,
      "loss": 0.0736,
      "step": 21051
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.229954429722911,
      "learning_rate": 5.921370440325476e-06,
      "loss": 0.2257,
      "step": 21052
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.2941575834403862,
      "learning_rate": 5.920464827934085e-06,
      "loss": 0.6781,
      "step": 21053
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.9091947379655037,
      "learning_rate": 5.919559255678499e-06,
      "loss": 0.5249,
      "step": 21054
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.30499551317408646,
      "learning_rate": 5.9186537235676355e-06,
      "loss": 0.185,
      "step": 21055
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.2838292807946733,
      "learning_rate": 5.917748231610401e-06,
      "loss": 0.1898,
      "step": 21056
    },
    {
      "epoch": 0.64,
      "grad_norm": 1.1488299515696205,
      "learning_rate": 5.9168427798157005e-06,
      "loss": 0.5089,
      "step": 21057
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.38149096061436055,
      "learning_rate": 5.915937368192446e-06,
      "loss": 0.2579,
      "step": 21058
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.8339781853868734,
      "learning_rate": 5.9150319967495426e-06,
      "loss": 0.3992,
      "step": 21059
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.32615621030365605,
      "learning_rate": 5.914126665495897e-06,
      "loss": 0.068,
      "step": 21060
    },
    {
      "epoch": 0.64,
      "grad_norm": 0.5270535607875124,
      "learning_rate": 5.913221374440418e-06,
      "loss": 0.3433,
      "step": 21061
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4529141981184872,
      "learning_rate": 5.9123161235920145e-06,
      "loss": 0.2542,
      "step": 21062
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4614702423689232,
      "learning_rate": 5.911410912959588e-06,
      "loss": 0.3718,
      "step": 21063
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4289936408309332,
      "learning_rate": 5.910505742552047e-06,
      "loss": 0.2643,
      "step": 21064
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2227657367976054,
      "learning_rate": 5.909600612378297e-06,
      "loss": 0.0656,
      "step": 21065
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36788070894803765,
      "learning_rate": 5.908695522447237e-06,
      "loss": 0.2825,
      "step": 21066
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3495593423266555,
      "learning_rate": 5.907790472767783e-06,
      "loss": 0.1535,
      "step": 21067
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0690523552887365,
      "learning_rate": 5.906885463348831e-06,
      "loss": 0.5332,
      "step": 21068
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.34598419580613654,
      "learning_rate": 5.905980494199283e-06,
      "loss": 0.1964,
      "step": 21069
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.30776239609688033,
      "learning_rate": 5.905075565328049e-06,
      "loss": 0.2445,
      "step": 21070
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3823755773140527,
      "learning_rate": 5.90417067674403e-06,
      "loss": 0.2533,
      "step": 21071
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4529432980653048,
      "learning_rate": 5.903265828456126e-06,
      "loss": 0.7314,
      "step": 21072
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1207420291586985,
      "learning_rate": 5.902361020473242e-06,
      "loss": 0.2829,
      "step": 21073
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5234683233216465,
      "learning_rate": 5.901456252804278e-06,
      "loss": 0.3786,
      "step": 21074
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2568066405280095,
      "learning_rate": 5.900551525458134e-06,
      "loss": 0.2229,
      "step": 21075
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.29550458020903336,
      "learning_rate": 5.899646838443716e-06,
      "loss": 0.2447,
      "step": 21076
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.342540835857004,
      "learning_rate": 5.898742191769919e-06,
      "loss": 0.1008,
      "step": 21077
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6410034799035187,
      "learning_rate": 5.897837585445649e-06,
      "loss": 0.1343,
      "step": 21078
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3500725047895569,
      "learning_rate": 5.8969330194798025e-06,
      "loss": 0.2661,
      "step": 21079
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.942599803566011,
      "learning_rate": 5.896028493881274e-06,
      "loss": 0.546,
      "step": 21080
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46091757721706916,
      "learning_rate": 5.8951240086589725e-06,
      "loss": 0.3682,
      "step": 21081
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4353159174221417,
      "learning_rate": 5.894219563821791e-06,
      "loss": 0.2829,
      "step": 21082
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40502791832685714,
      "learning_rate": 5.893315159378627e-06,
      "loss": 0.2788,
      "step": 21083
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4405578750339171,
      "learning_rate": 5.892410795338381e-06,
      "loss": 0.2361,
      "step": 21084
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8836684769979017,
      "learning_rate": 5.891506471709949e-06,
      "loss": 0.4812,
      "step": 21085
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18793471086250557,
      "learning_rate": 5.890602188502224e-06,
      "loss": 0.0776,
      "step": 21086
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2936621022506325,
      "learning_rate": 5.889697945724113e-06,
      "loss": 0.2669,
      "step": 21087
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.347793577668865,
      "learning_rate": 5.888793743384505e-06,
      "loss": 0.1739,
      "step": 21088
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5735286702658622,
      "learning_rate": 5.887889581492294e-06,
      "loss": 0.3415,
      "step": 21089
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7564350526335533,
      "learning_rate": 5.88698546005638e-06,
      "loss": 0.3734,
      "step": 21090
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6222364629886872,
      "learning_rate": 5.886081379085658e-06,
      "loss": 0.2954,
      "step": 21091
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0034613448188976,
      "learning_rate": 5.885177338589016e-06,
      "loss": 0.5507,
      "step": 21092
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2985426219235356,
      "learning_rate": 5.884273338575358e-06,
      "loss": 0.1893,
      "step": 21093
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3355379220002568,
      "learning_rate": 5.8833693790535726e-06,
      "loss": 0.2783,
      "step": 21094
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.14736010018814596,
      "learning_rate": 5.882465460032553e-06,
      "loss": 0.0698,
      "step": 21095
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.6604998916960017,
      "learning_rate": 5.881561581521194e-06,
      "loss": 0.8564,
      "step": 21096
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2769453814047196,
      "learning_rate": 5.880657743528386e-06,
      "loss": 0.1603,
      "step": 21097
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.49568293192504725,
      "learning_rate": 5.8797539460630206e-06,
      "loss": 0.3507,
      "step": 21098
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3052442076209488,
      "learning_rate": 5.878850189133996e-06,
      "loss": 0.2392,
      "step": 21099
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0104865053022845,
      "learning_rate": 5.8779464727501935e-06,
      "loss": 0.5017,
      "step": 21100
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5063469352121652,
      "learning_rate": 5.877042796920514e-06,
      "loss": 0.2178,
      "step": 21101
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37708719256326934,
      "learning_rate": 5.876139161653848e-06,
      "loss": 0.2819,
      "step": 21102
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3507557486005432,
      "learning_rate": 5.875235566959074e-06,
      "loss": 0.1036,
      "step": 21103
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.434191745016662,
      "learning_rate": 5.874332012845094e-06,
      "loss": 0.0851,
      "step": 21104
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3495495838068774,
      "learning_rate": 5.873428499320795e-06,
      "loss": 0.3028,
      "step": 21105
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.29514209365998545,
      "learning_rate": 5.872525026395057e-06,
      "loss": 0.1667,
      "step": 21106
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4117595222238728,
      "learning_rate": 5.871621594076783e-06,
      "loss": 0.7766,
      "step": 21107
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8714084397825149,
      "learning_rate": 5.870718202374852e-06,
      "loss": 0.3237,
      "step": 21108
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7525159414715382,
      "learning_rate": 5.869814851298154e-06,
      "loss": 0.4465,
      "step": 21109
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2576402119094228,
      "learning_rate": 5.868911540855578e-06,
      "loss": 0.0648,
      "step": 21110
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37535121248026465,
      "learning_rate": 5.868008271056007e-06,
      "loss": 0.2713,
      "step": 21111
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.30717700386691676,
      "learning_rate": 5.867105041908329e-06,
      "loss": 0.21,
      "step": 21112
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4522456929315977,
      "learning_rate": 5.866201853421435e-06,
      "loss": 0.2441,
      "step": 21113
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2252920590076413,
      "learning_rate": 5.8652987056042075e-06,
      "loss": 0.0939,
      "step": 21114
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5942557244266052,
      "learning_rate": 5.864395598465529e-06,
      "loss": 0.2925,
      "step": 21115
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7889158451428084,
      "learning_rate": 5.86349253201429e-06,
      "loss": 0.2609,
      "step": 21116
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4157930052913058,
      "learning_rate": 5.862589506259372e-06,
      "loss": 0.2553,
      "step": 21117
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9919755630701623,
      "learning_rate": 5.861686521209657e-06,
      "loss": 0.5792,
      "step": 21118
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3698473561624567,
      "learning_rate": 5.860783576874034e-06,
      "loss": 0.0659,
      "step": 21119
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35115443891337184,
      "learning_rate": 5.859880673261384e-06,
      "loss": 0.2667,
      "step": 21120
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.170186857847083,
      "learning_rate": 5.8589778103805904e-06,
      "loss": 0.1266,
      "step": 21121
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23987225690952288,
      "learning_rate": 5.858074988240536e-06,
      "loss": 0.1853,
      "step": 21122
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.47526545732566466,
      "learning_rate": 5.857172206850103e-06,
      "loss": 0.2319,
      "step": 21123
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5685512137543633,
      "learning_rate": 5.8562694662181696e-06,
      "loss": 0.2954,
      "step": 21124
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3901657455485679,
      "learning_rate": 5.855366766353625e-06,
      "loss": 0.2431,
      "step": 21125
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7948511728557593,
      "learning_rate": 5.854464107265343e-06,
      "loss": 0.4213,
      "step": 21126
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9525358062034714,
      "learning_rate": 5.853561488962209e-06,
      "loss": 0.466,
      "step": 21127
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.47691846733890275,
      "learning_rate": 5.852658911453102e-06,
      "loss": 0.3467,
      "step": 21128
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24862712705946505,
      "learning_rate": 5.8517563747468975e-06,
      "loss": 0.1685,
      "step": 21129
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40507490898461473,
      "learning_rate": 5.850853878852483e-06,
      "loss": 0.2279,
      "step": 21130
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7291059003925562,
      "learning_rate": 5.849951423778733e-06,
      "loss": 0.6883,
      "step": 21131
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3011235076272595,
      "learning_rate": 5.849049009534523e-06,
      "loss": 0.1386,
      "step": 21132
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40358667840562,
      "learning_rate": 5.848146636128737e-06,
      "loss": 0.2648,
      "step": 21133
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.544294960731746,
      "learning_rate": 5.847244303570252e-06,
      "loss": 0.263,
      "step": 21134
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4253602140608788,
      "learning_rate": 5.8463420118679405e-06,
      "loss": 0.3163,
      "step": 21135
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37676748693013906,
      "learning_rate": 5.845439761030684e-06,
      "loss": 0.2621,
      "step": 21136
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8563113927295546,
      "learning_rate": 5.844537551067362e-06,
      "loss": 0.4214,
      "step": 21137
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2893618281252168,
      "learning_rate": 5.8436353819868406e-06,
      "loss": 0.1917,
      "step": 21138
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2897714107934777,
      "learning_rate": 5.842733253798005e-06,
      "loss": 0.6516,
      "step": 21139
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.42056810686792584,
      "learning_rate": 5.8418311665097295e-06,
      "loss": 0.2067,
      "step": 21140
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2821337122655131,
      "learning_rate": 5.840929120130885e-06,
      "loss": 0.2262,
      "step": 21141
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3774439658289192,
      "learning_rate": 5.840027114670349e-06,
      "loss": 0.1763,
      "step": 21142
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.34371836079603557,
      "learning_rate": 5.839125150136996e-06,
      "loss": 0.2343,
      "step": 21143
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7782340470388216,
      "learning_rate": 5.838223226539697e-06,
      "loss": 0.5528,
      "step": 21144
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26502193114553496,
      "learning_rate": 5.837321343887328e-06,
      "loss": 0.1217,
      "step": 21145
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5134532225336851,
      "learning_rate": 5.836419502188763e-06,
      "loss": 0.3563,
      "step": 21146
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2830782941220241,
      "learning_rate": 5.835517701452866e-06,
      "loss": 0.168,
      "step": 21147
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4822941666990578,
      "learning_rate": 5.834615941688525e-06,
      "loss": 0.2904,
      "step": 21148
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3162826153724141,
      "learning_rate": 5.833714222904598e-06,
      "loss": 0.0966,
      "step": 21149
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8138872387562045,
      "learning_rate": 5.832812545109963e-06,
      "loss": 0.5766,
      "step": 21150
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4492466946090776,
      "learning_rate": 5.831910908313491e-06,
      "loss": 0.2209,
      "step": 21151
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38445804764793867,
      "learning_rate": 5.831009312524045e-06,
      "loss": 0.3094,
      "step": 21152
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.291728633263842,
      "learning_rate": 5.830107757750508e-06,
      "loss": 0.2433,
      "step": 21153
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5825398601377756,
      "learning_rate": 5.829206244001739e-06,
      "loss": 0.7989,
      "step": 21154
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.276628525537828,
      "learning_rate": 5.8283047712866125e-06,
      "loss": 0.0962,
      "step": 21155
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.27316407162235323,
      "learning_rate": 5.827403339614e-06,
      "loss": 0.1772,
      "step": 21156
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.088424958968443,
      "learning_rate": 5.826501948992761e-06,
      "loss": 0.4361,
      "step": 21157
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.419378935497932,
      "learning_rate": 5.82560059943177e-06,
      "loss": 0.228,
      "step": 21158
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38260523067470287,
      "learning_rate": 5.824699290939893e-06,
      "loss": 0.3062,
      "step": 21159
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4542941561176533,
      "learning_rate": 5.823798023526004e-06,
      "loss": 0.1975,
      "step": 21160
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.49159741733536616,
      "learning_rate": 5.822896797198958e-06,
      "loss": 0.3616,
      "step": 21161
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0163424193097184,
      "learning_rate": 5.82199561196763e-06,
      "loss": 0.4661,
      "step": 21162
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4729054674129522,
      "learning_rate": 5.8210944678408864e-06,
      "loss": 0.2776,
      "step": 21163
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21372031663786129,
      "learning_rate": 5.820193364827583e-06,
      "loss": 0.1768,
      "step": 21164
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3212595934783338,
      "learning_rate": 5.819292302936601e-06,
      "loss": 0.2573,
      "step": 21165
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8941464150689599,
      "learning_rate": 5.8183912821767914e-06,
      "loss": 0.065,
      "step": 21166
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9397732326127664,
      "learning_rate": 5.817490302557026e-06,
      "loss": 0.5309,
      "step": 21167
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8087190606262806,
      "learning_rate": 5.8165893640861715e-06,
      "loss": 0.3013,
      "step": 21168
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33613437475425895,
      "learning_rate": 5.815688466773081e-06,
      "loss": 0.0657,
      "step": 21169
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3284660433604403,
      "learning_rate": 5.814787610626627e-06,
      "loss": 0.2697,
      "step": 21170
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.30362314284838,
      "learning_rate": 5.813886795655668e-06,
      "loss": 0.2423,
      "step": 21171
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.398439476009075,
      "learning_rate": 5.812986021869073e-06,
      "loss": 0.2638,
      "step": 21172
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3851370969555216,
      "learning_rate": 5.812085289275691e-06,
      "loss": 0.1056,
      "step": 21173
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.615359718203934,
      "learning_rate": 5.811184597884398e-06,
      "loss": 0.2348,
      "step": 21174
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.326337080840224,
      "learning_rate": 5.810283947704047e-06,
      "loss": 0.1935,
      "step": 21175
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4719940918630821,
      "learning_rate": 5.809383338743502e-06,
      "loss": 0.3695,
      "step": 21176
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43513137448404976,
      "learning_rate": 5.808482771011625e-06,
      "loss": 0.2298,
      "step": 21177
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7131964452882104,
      "learning_rate": 5.8075822445172666e-06,
      "loss": 0.4099,
      "step": 21178
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3057572770826636,
      "learning_rate": 5.806681759269302e-06,
      "loss": 0.198,
      "step": 21179
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5604102141533271,
      "learning_rate": 5.805781315276576e-06,
      "loss": 0.7075,
      "step": 21180
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9760479989587041,
      "learning_rate": 5.804880912547954e-06,
      "loss": 0.4757,
      "step": 21181
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.20461915492632968,
      "learning_rate": 5.8039805510923e-06,
      "loss": 0.1692,
      "step": 21182
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38136770543383647,
      "learning_rate": 5.803080230918461e-06,
      "loss": 0.2543,
      "step": 21183
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4252873265797549,
      "learning_rate": 5.802179952035299e-06,
      "loss": 0.2387,
      "step": 21184
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8387964534871163,
      "learning_rate": 5.801279714451672e-06,
      "loss": 0.531,
      "step": 21185
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2834506378808088,
      "learning_rate": 5.800379518176441e-06,
      "loss": 0.2895,
      "step": 21186
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7237274812322401,
      "learning_rate": 5.799479363218454e-06,
      "loss": 0.3252,
      "step": 21187
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24632559062024365,
      "learning_rate": 5.7985792495865714e-06,
      "loss": 0.2008,
      "step": 21188
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.502897664085838,
      "learning_rate": 5.797679177289652e-06,
      "loss": 0.36,
      "step": 21189
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.94219624523495,
      "learning_rate": 5.796779146336541e-06,
      "loss": 0.5889,
      "step": 21190
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.25816167557658526,
      "learning_rate": 5.795879156736106e-06,
      "loss": 0.1615,
      "step": 21191
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2582614528515901,
      "learning_rate": 5.7949792084971935e-06,
      "loss": 0.1749,
      "step": 21192
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9844144748341954,
      "learning_rate": 5.794079301628656e-06,
      "loss": 0.4856,
      "step": 21193
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5162121542776928,
      "learning_rate": 5.7931794361393575e-06,
      "loss": 0.2692,
      "step": 21194
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43129809166062816,
      "learning_rate": 5.792279612038138e-06,
      "loss": 0.2216,
      "step": 21195
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9350703560204817,
      "learning_rate": 5.791379829333856e-06,
      "loss": 0.4929,
      "step": 21196
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32430474341135257,
      "learning_rate": 5.790480088035366e-06,
      "loss": 0.1996,
      "step": 21197
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3553281357993452,
      "learning_rate": 5.7895803881515215e-06,
      "loss": 0.7464,
      "step": 21198
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.44281683837632496,
      "learning_rate": 5.788680729691165e-06,
      "loss": 0.258,
      "step": 21199
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.28086692063927815,
      "learning_rate": 5.787781112663153e-06,
      "loss": 0.2153,
      "step": 21200
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.14891523562942177,
      "learning_rate": 5.786881537076342e-06,
      "loss": 0.0691,
      "step": 21201
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3756489653094567,
      "learning_rate": 5.785982002939568e-06,
      "loss": 0.2849,
      "step": 21202
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.076911285420237,
      "learning_rate": 5.785082510261698e-06,
      "loss": 0.3013,
      "step": 21203
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.947021627991,
      "learning_rate": 5.784183059051565e-06,
      "loss": 0.7386,
      "step": 21204
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4578584404094403,
      "learning_rate": 5.783283649318034e-06,
      "loss": 0.2386,
      "step": 21205
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38347968299986307,
      "learning_rate": 5.782384281069941e-06,
      "loss": 0.2424,
      "step": 21206
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35659810896489996,
      "learning_rate": 5.781484954316141e-06,
      "loss": 0.2572,
      "step": 21207
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5920992082442949,
      "learning_rate": 5.780585669065484e-06,
      "loss": 0.7598,
      "step": 21208
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.941685113585205,
      "learning_rate": 5.779686425326809e-06,
      "loss": 0.2988,
      "step": 21209
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37347113369476936,
      "learning_rate": 5.778787223108968e-06,
      "loss": 0.1872,
      "step": 21210
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5384746154661997,
      "learning_rate": 5.777888062420806e-06,
      "loss": 0.3564,
      "step": 21211
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26276394988150914,
      "learning_rate": 5.776988943271176e-06,
      "loss": 0.1675,
      "step": 21212
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3206896007418086,
      "learning_rate": 5.776089865668914e-06,
      "loss": 0.2296,
      "step": 21213
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3720775412387693,
      "learning_rate": 5.77519082962287e-06,
      "loss": 0.0612,
      "step": 21214
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3609182207838268,
      "learning_rate": 5.774291835141892e-06,
      "loss": 0.2773,
      "step": 21215
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0055055985886776,
      "learning_rate": 5.773392882234814e-06,
      "loss": 0.4723,
      "step": 21216
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2603966516325782,
      "learning_rate": 5.7724939709104945e-06,
      "loss": 0.5972,
      "step": 21217
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4423442649702948,
      "learning_rate": 5.7715951011777675e-06,
      "loss": 0.263,
      "step": 21218
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5878295119735307,
      "learning_rate": 5.770696273045479e-06,
      "loss": 0.2982,
      "step": 21219
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36688800721263415,
      "learning_rate": 5.769797486522475e-06,
      "loss": 0.2122,
      "step": 21220
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6485137033459845,
      "learning_rate": 5.768898741617592e-06,
      "loss": 0.2527,
      "step": 21221
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3432107418751967,
      "learning_rate": 5.768000038339676e-06,
      "loss": 0.0844,
      "step": 21222
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36232633821807453,
      "learning_rate": 5.767101376697568e-06,
      "loss": 0.1866,
      "step": 21223
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.45234754722126275,
      "learning_rate": 5.766202756700112e-06,
      "loss": 0.2824,
      "step": 21224
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33909823354468893,
      "learning_rate": 5.7653041783561445e-06,
      "loss": 0.246,
      "step": 21225
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2659850418694216,
      "learning_rate": 5.764405641674507e-06,
      "loss": 0.7122,
      "step": 21226
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7781238608424561,
      "learning_rate": 5.76350714666404e-06,
      "loss": 0.3028,
      "step": 21227
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.742665785411662,
      "learning_rate": 5.762608693333584e-06,
      "loss": 0.3551,
      "step": 21228
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.29750458170830296,
      "learning_rate": 5.761710281691983e-06,
      "loss": 0.1994,
      "step": 21229
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35145151660890256,
      "learning_rate": 5.760811911748065e-06,
      "loss": 0.2991,
      "step": 21230
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1707634414970534,
      "learning_rate": 5.759913583510676e-06,
      "loss": 0.0722,
      "step": 21231
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.049390825638423,
      "learning_rate": 5.759015296988655e-06,
      "loss": 0.4961,
      "step": 21232
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2839984002551263,
      "learning_rate": 5.75811705219083e-06,
      "loss": 0.1597,
      "step": 21233
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5558886609832305,
      "learning_rate": 5.757218849126053e-06,
      "loss": 0.3988,
      "step": 21234
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.1443282541479105,
      "learning_rate": 5.756320687803149e-06,
      "loss": 0.3341,
      "step": 21235
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.38327597752731196,
      "learning_rate": 5.755422568230958e-06,
      "loss": 0.2505,
      "step": 21236
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6485567839954464,
      "learning_rate": 5.754524490418321e-06,
      "loss": 0.3595,
      "step": 21237
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.28418336506604636,
      "learning_rate": 5.753626454374064e-06,
      "loss": 0.1962,
      "step": 21238
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7288562357594737,
      "learning_rate": 5.752728460107028e-06,
      "loss": 0.8287,
      "step": 21239
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.21923370918293034,
      "learning_rate": 5.751830507626046e-06,
      "loss": 0.082,
      "step": 21240
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39597291645072574,
      "learning_rate": 5.750932596939958e-06,
      "loss": 0.2514,
      "step": 21241
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2734428471604835,
      "learning_rate": 5.750034728057585e-06,
      "loss": 0.1741,
      "step": 21242
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.667722045663053,
      "learning_rate": 5.749136900987777e-06,
      "loss": 0.7156,
      "step": 21243
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2563013603684492,
      "learning_rate": 5.7482391157393556e-06,
      "loss": 0.364,
      "step": 21244
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9002217792495137,
      "learning_rate": 5.747341372321156e-06,
      "loss": 0.4504,
      "step": 21245
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3304923236654476,
      "learning_rate": 5.746443670742016e-06,
      "loss": 0.0666,
      "step": 21246
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5660367799006647,
      "learning_rate": 5.745546011010759e-06,
      "loss": 0.3777,
      "step": 21247
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2599609082491102,
      "learning_rate": 5.744648393136219e-06,
      "loss": 0.2255,
      "step": 21248
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.23253023478839316,
      "learning_rate": 5.743750817127228e-06,
      "loss": 0.0809,
      "step": 21249
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2569062210426118,
      "learning_rate": 5.742853282992622e-06,
      "loss": 0.785,
      "step": 21250
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2500896820599945,
      "learning_rate": 5.741955790741222e-06,
      "loss": 0.0682,
      "step": 21251
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.37295932106526347,
      "learning_rate": 5.741058340381862e-06,
      "loss": 0.2668,
      "step": 21252
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8588551681955703,
      "learning_rate": 5.7401609319233705e-06,
      "loss": 0.3043,
      "step": 21253
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5598410319927772,
      "learning_rate": 5.739263565374578e-06,
      "loss": 0.3396,
      "step": 21254
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6449785230789258,
      "learning_rate": 5.738366240744318e-06,
      "loss": 0.109,
      "step": 21255
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3360966487135582,
      "learning_rate": 5.737468958041408e-06,
      "loss": 0.27,
      "step": 21256
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.272671746356958,
      "learning_rate": 5.73657171727468e-06,
      "loss": 0.2483,
      "step": 21257
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2846785806033968,
      "learning_rate": 5.735674518452967e-06,
      "loss": 0.1741,
      "step": 21258
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.29611205617574154,
      "learning_rate": 5.734777361585085e-06,
      "loss": 0.2324,
      "step": 21259
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6453888050695338,
      "learning_rate": 5.733880246679873e-06,
      "loss": 0.337,
      "step": 21260
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3965437658616677,
      "learning_rate": 5.732983173746146e-06,
      "loss": 0.2391,
      "step": 21261
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6450025539946711,
      "learning_rate": 5.732086142792737e-06,
      "loss": 0.3025,
      "step": 21262
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8130503806409509,
      "learning_rate": 5.7311891538284704e-06,
      "loss": 0.5624,
      "step": 21263
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3549058346218998,
      "learning_rate": 5.730292206862167e-06,
      "loss": 0.0628,
      "step": 21264
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.33850246863067346,
      "learning_rate": 5.729395301902653e-06,
      "loss": 0.257,
      "step": 21265
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31740954029881885,
      "learning_rate": 5.728498438958754e-06,
      "loss": 0.2413,
      "step": 21266
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.554036683853026,
      "learning_rate": 5.727601618039295e-06,
      "loss": 0.2357,
      "step": 21267
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3658614301811624,
      "learning_rate": 5.726704839153095e-06,
      "loss": 0.1657,
      "step": 21268
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.505830460058215,
      "learning_rate": 5.725808102308978e-06,
      "loss": 0.3028,
      "step": 21269
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36686349550918546,
      "learning_rate": 5.724911407515773e-06,
      "loss": 0.2126,
      "step": 21270
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7137833448979976,
      "learning_rate": 5.724014754782288e-06,
      "loss": 0.4186,
      "step": 21271
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32680588238197666,
      "learning_rate": 5.723118144117359e-06,
      "loss": 0.2542,
      "step": 21272
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8504920180117688,
      "learning_rate": 5.722221575529798e-06,
      "loss": 0.3868,
      "step": 21273
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3725624043521182,
      "learning_rate": 5.721325049028428e-06,
      "loss": 0.1606,
      "step": 21274
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.45213924341322503,
      "learning_rate": 5.720428564622071e-06,
      "loss": 0.1942,
      "step": 21275
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.28709661601020897,
      "learning_rate": 5.7195321223195486e-06,
      "loss": 0.1591,
      "step": 21276
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2668487235686889,
      "learning_rate": 5.718635722129675e-06,
      "loss": 0.2379,
      "step": 21277
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.887575560502963,
      "learning_rate": 5.717739364061272e-06,
      "loss": 0.4523,
      "step": 21278
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2908669401608682,
      "learning_rate": 5.716843048123157e-06,
      "loss": 0.2091,
      "step": 21279
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7633707453742731,
      "learning_rate": 5.7159467743241504e-06,
      "loss": 0.4174,
      "step": 21280
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8216031179633826,
      "learning_rate": 5.715050542673073e-06,
      "loss": 0.4421,
      "step": 21281
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.334932091267474,
      "learning_rate": 5.7141543531787335e-06,
      "loss": 0.5989,
      "step": 21282
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26093935510396576,
      "learning_rate": 5.713258205849954e-06,
      "loss": 0.189,
      "step": 21283
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3602697765168542,
      "learning_rate": 5.712362100695556e-06,
      "loss": 0.2954,
      "step": 21284
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18909394338968086,
      "learning_rate": 5.711466037724342e-06,
      "loss": 0.0635,
      "step": 21285
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.44178574963075096,
      "learning_rate": 5.710570016945143e-06,
      "loss": 0.2584,
      "step": 21286
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4105018118000279,
      "learning_rate": 5.709674038366765e-06,
      "loss": 0.2109,
      "step": 21287
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.29875756637833145,
      "learning_rate": 5.7087781019980246e-06,
      "loss": 0.2417,
      "step": 21288
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4696614799600883,
      "learning_rate": 5.7078822078477414e-06,
      "loss": 0.3567,
      "step": 21289
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3908545486938711,
      "learning_rate": 5.7069863559247214e-06,
      "loss": 0.2519,
      "step": 21290
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.737702570515397,
      "learning_rate": 5.706090546237782e-06,
      "loss": 0.552,
      "step": 21291
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2858880675022833,
      "learning_rate": 5.705194778795737e-06,
      "loss": 0.1806,
      "step": 21292
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.182862353685575,
      "learning_rate": 5.704299053607403e-06,
      "loss": 0.5384,
      "step": 21293
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8082439344776688,
      "learning_rate": 5.703403370681585e-06,
      "loss": 0.2765,
      "step": 21294
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3270687266393195,
      "learning_rate": 5.702507730027098e-06,
      "loss": 0.275,
      "step": 21295
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.46279588688414924,
      "learning_rate": 5.701612131652757e-06,
      "loss": 0.2098,
      "step": 21296
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5597387283847614,
      "learning_rate": 5.700716575567363e-06,
      "loss": 0.3134,
      "step": 21297
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2661861549405014,
      "learning_rate": 5.699821061779742e-06,
      "loss": 0.157,
      "step": 21298
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4150134382863379,
      "learning_rate": 5.698925590298691e-06,
      "loss": 0.2554,
      "step": 21299
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35915502798125654,
      "learning_rate": 5.698030161133028e-06,
      "loss": 0.1919,
      "step": 21300
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36508515074430836,
      "learning_rate": 5.697134774291562e-06,
      "loss": 0.1761,
      "step": 21301
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3606099221623998,
      "learning_rate": 5.696239429783097e-06,
      "loss": 0.3038,
      "step": 21302
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7884469041552372,
      "learning_rate": 5.6953441276164435e-06,
      "loss": 0.2517,
      "step": 21303
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6120665986812316,
      "learning_rate": 5.694448867800411e-06,
      "loss": 0.3939,
      "step": 21304
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6781562062640618,
      "learning_rate": 5.693553650343808e-06,
      "loss": 0.1285,
      "step": 21305
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3338912421524312,
      "learning_rate": 5.6926584752554405e-06,
      "loss": 0.2384,
      "step": 21306
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3180689564191909,
      "learning_rate": 5.691763342544121e-06,
      "loss": 0.2401,
      "step": 21307
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4236341688034889,
      "learning_rate": 5.690868252218647e-06,
      "loss": 0.2496,
      "step": 21308
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.24187755649482146,
      "learning_rate": 5.689973204287827e-06,
      "loss": 0.0661,
      "step": 21309
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6436585196914889,
      "learning_rate": 5.689078198760476e-06,
      "loss": 0.3282,
      "step": 21310
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.36931454849380063,
      "learning_rate": 5.688183235645382e-06,
      "loss": 0.2049,
      "step": 21311
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9472918945613751,
      "learning_rate": 5.687288314951369e-06,
      "loss": 0.408,
      "step": 21312
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35861930418071336,
      "learning_rate": 5.686393436687229e-06,
      "loss": 0.2393,
      "step": 21313
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6991259317332584,
      "learning_rate": 5.685498600861769e-06,
      "loss": 0.2231,
      "step": 21314
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32235943924136495,
      "learning_rate": 5.684603807483797e-06,
      "loss": 0.2562,
      "step": 21315
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0106454488344487,
      "learning_rate": 5.68370905656211e-06,
      "loss": 0.5538,
      "step": 21316
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4839758432261646,
      "learning_rate": 5.682814348105512e-06,
      "loss": 0.2413,
      "step": 21317
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31714818469170825,
      "learning_rate": 5.6819196821228086e-06,
      "loss": 0.1818,
      "step": 21318
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3999403344609002,
      "learning_rate": 5.681025058622803e-06,
      "loss": 0.2457,
      "step": 21319
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3240042665133821,
      "learning_rate": 5.680130477614291e-06,
      "loss": 0.2288,
      "step": 21320
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.823013518876695,
      "learning_rate": 5.6792359391060754e-06,
      "loss": 0.4301,
      "step": 21321
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7420234014205839,
      "learning_rate": 5.678341443106964e-06,
      "loss": 0.2737,
      "step": 21322
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9379880601865579,
      "learning_rate": 5.677446989625743e-06,
      "loss": 0.3952,
      "step": 21323
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3278879176419484,
      "learning_rate": 5.6765525786712284e-06,
      "loss": 0.1579,
      "step": 21324
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3534520588746159,
      "learning_rate": 5.675658210252209e-06,
      "loss": 0.3017,
      "step": 21325
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43297311403132915,
      "learning_rate": 5.6747638843774855e-06,
      "loss": 0.2573,
      "step": 21326
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.16248757288086907,
      "learning_rate": 5.673869601055863e-06,
      "loss": 0.0695,
      "step": 21327
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8961834691526633,
      "learning_rate": 5.672975360296131e-06,
      "loss": 0.4982,
      "step": 21328
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.43467071315102884,
      "learning_rate": 5.672081162107091e-06,
      "loss": 0.2069,
      "step": 21329
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0048878479201864,
      "learning_rate": 5.671187006497541e-06,
      "loss": 0.4058,
      "step": 21330
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.28868853111056525,
      "learning_rate": 5.670292893476276e-06,
      "loss": 0.2071,
      "step": 21331
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.7206192651117544,
      "learning_rate": 5.669398823052099e-06,
      "loss": 0.8247,
      "step": 21332
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.27546688632480765,
      "learning_rate": 5.668504795233798e-06,
      "loss": 0.1752,
      "step": 21333
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.255622554425532,
      "learning_rate": 5.667610810030172e-06,
      "loss": 0.7627,
      "step": 21334
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.15116613244520907,
      "learning_rate": 5.666716867450015e-06,
      "loss": 0.0868,
      "step": 21335
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4735756411612615,
      "learning_rate": 5.66582296750213e-06,
      "loss": 0.6439,
      "step": 21336
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.26519252609256627,
      "learning_rate": 5.664929110195294e-06,
      "loss": 0.2022,
      "step": 21337
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4583712301398966,
      "learning_rate": 5.664035295538322e-06,
      "loss": 0.2913,
      "step": 21338
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.7633782152978628,
      "learning_rate": 5.663141523539997e-06,
      "loss": 0.2951,
      "step": 21339
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.301399153637073,
      "learning_rate": 5.662247794209104e-06,
      "loss": 0.2737,
      "step": 21340
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2936709496099466,
      "learning_rate": 5.661354107554452e-06,
      "loss": 0.6479,
      "step": 21341
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.27513476286326954,
      "learning_rate": 5.6604604635848226e-06,
      "loss": 0.176,
      "step": 21342
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31952858740647816,
      "learning_rate": 5.6595668623090114e-06,
      "loss": 0.2912,
      "step": 21343
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0136466806611093,
      "learning_rate": 5.658673303735809e-06,
      "loss": 0.3315,
      "step": 21344
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.41968599714845245,
      "learning_rate": 5.657779787874012e-06,
      "loss": 0.2727,
      "step": 21345
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.44098834311987967,
      "learning_rate": 5.6568863147324016e-06,
      "loss": 0.2123,
      "step": 21346
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3770498898587419,
      "learning_rate": 5.655992884319772e-06,
      "loss": 0.2781,
      "step": 21347
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5283227706573932,
      "learning_rate": 5.655099496644918e-06,
      "loss": 0.1493,
      "step": 21348
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3149811885826827,
      "learning_rate": 5.654206151716618e-06,
      "loss": 0.2762,
      "step": 21349
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6160594796784967,
      "learning_rate": 5.653312849543675e-06,
      "loss": 0.1761,
      "step": 21350
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4272293290377262,
      "learning_rate": 5.652419590134867e-06,
      "loss": 0.271,
      "step": 21351
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4066054909991709,
      "learning_rate": 5.6515263734989844e-06,
      "loss": 0.2371,
      "step": 21352
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0779079645937055,
      "learning_rate": 5.650633199644821e-06,
      "loss": 0.2331,
      "step": 21353
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.35544389043566543,
      "learning_rate": 5.649740068581151e-06,
      "loss": 0.2864,
      "step": 21354
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.5843101653544635,
      "learning_rate": 5.648846980316777e-06,
      "loss": 0.2037,
      "step": 21355
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.39344215961258694,
      "learning_rate": 5.647953934860477e-06,
      "loss": 0.2618,
      "step": 21356
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.27311090497987867,
      "learning_rate": 5.647060932221036e-06,
      "loss": 0.1456,
      "step": 21357
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3535843772587135,
      "learning_rate": 5.646167972407246e-06,
      "loss": 0.566,
      "step": 21358
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8071300040707267,
      "learning_rate": 5.6452750554278836e-06,
      "loss": 0.4536,
      "step": 21359
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6781680967054874,
      "learning_rate": 5.644382181291739e-06,
      "loss": 0.37,
      "step": 21360
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2660227584813762,
      "learning_rate": 5.643489350007596e-06,
      "loss": 0.2338,
      "step": 21361
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5290335301136184,
      "learning_rate": 5.6425965615842415e-06,
      "loss": 0.808,
      "step": 21362
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6299370233583877,
      "learning_rate": 5.641703816030452e-06,
      "loss": 0.2809,
      "step": 21363
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.8543675602725205,
      "learning_rate": 5.640811113355014e-06,
      "loss": 0.3391,
      "step": 21364
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.32645152902358476,
      "learning_rate": 5.639918453566715e-06,
      "loss": 0.1932,
      "step": 21365
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.18968215709526146,
      "learning_rate": 5.639025836674325e-06,
      "loss": 0.1612,
      "step": 21366
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.4129436157236968,
      "learning_rate": 5.638133262686641e-06,
      "loss": 0.6227,
      "step": 21367
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0651109040133162,
      "learning_rate": 5.637240731612434e-06,
      "loss": 0.3761,
      "step": 21368
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.40321204362479346,
      "learning_rate": 5.636348243460488e-06,
      "loss": 0.2535,
      "step": 21369
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.4509497718463792,
      "learning_rate": 5.635455798239589e-06,
      "loss": 0.2473,
      "step": 21370
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.0678390751191797,
      "learning_rate": 5.6345633959585056e-06,
      "loss": 0.5174,
      "step": 21371
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3551436578286951,
      "learning_rate": 5.633671036626026e-06,
      "loss": 0.2542,
      "step": 21372
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.6764942192773848,
      "learning_rate": 5.632778720250926e-06,
      "loss": 0.3764,
      "step": 21373
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3186192603406736,
      "learning_rate": 5.631886446841991e-06,
      "loss": 0.1923,
      "step": 21374
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.3435150471579662,
      "learning_rate": 5.6309942164079854e-06,
      "loss": 0.5431,
      "step": 21375
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.1636857914104891,
      "learning_rate": 5.630102028957706e-06,
      "loss": 0.0724,
      "step": 21376
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.2452718320787182,
      "learning_rate": 5.6292098844999156e-06,
      "loss": 0.7955,
      "step": 21377
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.30510130763109045,
      "learning_rate": 5.628317783043397e-06,
      "loss": 0.1697,
      "step": 21378
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2803572425269732,
      "learning_rate": 5.627425724596929e-06,
      "loss": 0.2206,
      "step": 21379
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9499818523236584,
      "learning_rate": 5.626533709169279e-06,
      "loss": 0.4694,
      "step": 21380
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.9201688864068623,
      "learning_rate": 5.625641736769237e-06,
      "loss": 0.3006,
      "step": 21381
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.858379466449414,
      "learning_rate": 5.6247498074055675e-06,
      "loss": 0.3976,
      "step": 21382
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.31666740865473253,
      "learning_rate": 5.6238579210870485e-06,
      "loss": 0.1998,
      "step": 21383
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.3118676357009991,
      "learning_rate": 5.6229660778224595e-06,
      "loss": 0.2196,
      "step": 21384
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2858003433115755,
      "learning_rate": 5.6220742776205665e-06,
      "loss": 0.1618,
      "step": 21385
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.5392013799109185,
      "learning_rate": 5.621182520490147e-06,
      "loss": 0.7886,
      "step": 21386
    },
    {
      "epoch": 0.65,
      "grad_norm": 0.2812661147749116,
      "learning_rate": 5.6202908064399744e-06,
      "loss": 0.1527,
      "step": 21387
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3683376078271896,
      "learning_rate": 5.619399135478826e-06,
      "loss": 0.3104,
      "step": 21388
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1147608013929249,
      "learning_rate": 5.618507507615467e-06,
      "loss": 0.3299,
      "step": 21389
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5005208835276953,
      "learning_rate": 5.617615922858672e-06,
      "loss": 0.3562,
      "step": 21390
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43726719698067457,
      "learning_rate": 5.6167243812172176e-06,
      "loss": 0.2362,
      "step": 21391
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26462013355851033,
      "learning_rate": 5.615832882699863e-06,
      "loss": 0.166,
      "step": 21392
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4798964936707222,
      "learning_rate": 5.614941427315397e-06,
      "loss": 0.3122,
      "step": 21393
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1940180964688258,
      "learning_rate": 5.6140500150725744e-06,
      "loss": 0.0744,
      "step": 21394
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2145713329068886,
      "learning_rate": 5.61315864598017e-06,
      "loss": 0.8005,
      "step": 21395
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24864145456454692,
      "learning_rate": 5.61226732004696e-06,
      "loss": 0.175,
      "step": 21396
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33729453325726855,
      "learning_rate": 5.611376037281702e-06,
      "loss": 0.3062,
      "step": 21397
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8262264530080096,
      "learning_rate": 5.610484797693173e-06,
      "loss": 0.3094,
      "step": 21398
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8233049803808508,
      "learning_rate": 5.609593601290137e-06,
      "loss": 0.4647,
      "step": 21399
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7472769639990517,
      "learning_rate": 5.608702448081369e-06,
      "loss": 0.146,
      "step": 21400
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3633457694373261,
      "learning_rate": 5.6078113380756275e-06,
      "loss": 0.2702,
      "step": 21401
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4958163185535278,
      "learning_rate": 5.606920271281682e-06,
      "loss": 0.0178,
      "step": 21402
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2713189415541302,
      "learning_rate": 5.606029247708304e-06,
      "loss": 0.2505,
      "step": 21403
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2302364120588646,
      "learning_rate": 5.6051382673642495e-06,
      "loss": 0.144,
      "step": 21404
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5601592806531996,
      "learning_rate": 5.604247330258299e-06,
      "loss": 0.2968,
      "step": 21405
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36090336090107766,
      "learning_rate": 5.603356436399201e-06,
      "loss": 0.1982,
      "step": 21406
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8966346933606385,
      "learning_rate": 5.602465585795737e-06,
      "loss": 0.3164,
      "step": 21407
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3562063459641142,
      "learning_rate": 5.60157477845666e-06,
      "loss": 0.2815,
      "step": 21408
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1687663980214793,
      "learning_rate": 5.600684014390736e-06,
      "loss": 0.2008,
      "step": 21409
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6551524777625799,
      "learning_rate": 5.599793293606737e-06,
      "loss": 0.2506,
      "step": 21410
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3495016165092818,
      "learning_rate": 5.598902616113413e-06,
      "loss": 0.2075,
      "step": 21411
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2657760937460827,
      "learning_rate": 5.598011981919534e-06,
      "loss": 0.1733,
      "step": 21412
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8759315767494282,
      "learning_rate": 5.597121391033861e-06,
      "loss": 0.3013,
      "step": 21413
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3098322855457353,
      "learning_rate": 5.596230843465163e-06,
      "loss": 0.2682,
      "step": 21414
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.34306001295093314,
      "learning_rate": 5.595340339222188e-06,
      "loss": 0.2134,
      "step": 21415
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7204379955950262,
      "learning_rate": 5.5944498783137055e-06,
      "loss": 0.4562,
      "step": 21416
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3656512614537795,
      "learning_rate": 5.593559460748479e-06,
      "loss": 0.285,
      "step": 21417
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.369775557727709,
      "learning_rate": 5.592669086535255e-06,
      "loss": 0.5459,
      "step": 21418
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3545297091833184,
      "learning_rate": 5.5917787556828116e-06,
      "loss": 0.1718,
      "step": 21419
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24416788479685683,
      "learning_rate": 5.590888468199896e-06,
      "loss": 0.2221,
      "step": 21420
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3816753685904057,
      "learning_rate": 5.58999822409527e-06,
      "loss": 0.5938,
      "step": 21421
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4322443629762303,
      "learning_rate": 5.589108023377697e-06,
      "loss": 0.2114,
      "step": 21422
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7349697132421272,
      "learning_rate": 5.588217866055925e-06,
      "loss": 0.3489,
      "step": 21423
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3239435523743258,
      "learning_rate": 5.587327752138718e-06,
      "loss": 0.2029,
      "step": 21424
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5735165768722885,
      "learning_rate": 5.586437681634833e-06,
      "loss": 0.7767,
      "step": 21425
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2412467416195281,
      "learning_rate": 5.58554765455303e-06,
      "loss": 0.2128,
      "step": 21426
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4743197430823627,
      "learning_rate": 5.584657670902057e-06,
      "loss": 0.6442,
      "step": 21427
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2606585862829846,
      "learning_rate": 5.583767730690676e-06,
      "loss": 0.1782,
      "step": 21428
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.5978247273714907,
      "learning_rate": 5.582877833927645e-06,
      "loss": 0.6902,
      "step": 21429
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1479015085462107,
      "learning_rate": 5.581987980621706e-06,
      "loss": 0.0699,
      "step": 21430
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4405242753324309,
      "learning_rate": 5.581098170781632e-06,
      "loss": 0.3106,
      "step": 21431
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.329780117590321,
      "learning_rate": 5.580208404416165e-06,
      "loss": 0.2006,
      "step": 21432
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37504294739732397,
      "learning_rate": 5.579318681534061e-06,
      "loss": 0.2218,
      "step": 21433
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8643148261377621,
      "learning_rate": 5.578429002144079e-06,
      "loss": 0.4876,
      "step": 21434
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.321603177021937,
      "learning_rate": 5.577539366254958e-06,
      "loss": 0.2736,
      "step": 21435
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5090812728578433,
      "learning_rate": 5.576649773875469e-06,
      "loss": 0.6347,
      "step": 21436
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3693218920260463,
      "learning_rate": 5.5757602250143505e-06,
      "loss": 0.0659,
      "step": 21437
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2960768590212823,
      "learning_rate": 5.57487071968036e-06,
      "loss": 0.2608,
      "step": 21438
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4093684245801906,
      "learning_rate": 5.5739812578822505e-06,
      "loss": 0.2414,
      "step": 21439
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6961886431796316,
      "learning_rate": 5.573091839628766e-06,
      "loss": 0.4065,
      "step": 21440
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.28792552981557723,
      "learning_rate": 5.57220246492866e-06,
      "loss": 0.1297,
      "step": 21441
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3482827763446832,
      "learning_rate": 5.571313133790684e-06,
      "loss": 0.2652,
      "step": 21442
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8601505014640133,
      "learning_rate": 5.570423846223589e-06,
      "loss": 0.6121,
      "step": 21443
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.311829837738094,
      "learning_rate": 5.569534602236115e-06,
      "loss": 0.2711,
      "step": 21444
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2503961975610925,
      "learning_rate": 5.5686454018370254e-06,
      "loss": 0.072,
      "step": 21445
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3454190291665341,
      "learning_rate": 5.567756245035056e-06,
      "loss": 0.1698,
      "step": 21446
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5046072369857226,
      "learning_rate": 5.5668671318389585e-06,
      "loss": 0.3319,
      "step": 21447
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7770995631679328,
      "learning_rate": 5.565978062257487e-06,
      "loss": 0.2737,
      "step": 21448
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40728978556035716,
      "learning_rate": 5.565089036299376e-06,
      "loss": 0.3109,
      "step": 21449
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3205728952048962,
      "learning_rate": 5.564200053973379e-06,
      "loss": 0.1808,
      "step": 21450
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3604878176377471,
      "learning_rate": 5.5633111152882414e-06,
      "loss": 0.2691,
      "step": 21451
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0586182236419042,
      "learning_rate": 5.562422220252715e-06,
      "loss": 0.4762,
      "step": 21452
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4233888557660555,
      "learning_rate": 5.561533368875531e-06,
      "loss": 0.2648,
      "step": 21453
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35021488337769585,
      "learning_rate": 5.560644561165445e-06,
      "loss": 0.0963,
      "step": 21454
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3529819924951595,
      "learning_rate": 5.559755797131202e-06,
      "loss": 0.2499,
      "step": 21455
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3330162626962551,
      "learning_rate": 5.558867076781534e-06,
      "loss": 0.2048,
      "step": 21456
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4603901297703607,
      "learning_rate": 5.557978400125201e-06,
      "loss": 0.3323,
      "step": 21457
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9400208181137008,
      "learning_rate": 5.557089767170936e-06,
      "loss": 0.3037,
      "step": 21458
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2745393677271987,
      "learning_rate": 5.556201177927481e-06,
      "loss": 0.1223,
      "step": 21459
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5302802068158406,
      "learning_rate": 5.555312632403587e-06,
      "loss": 0.3746,
      "step": 21460
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42434976097681637,
      "learning_rate": 5.5544241306079806e-06,
      "loss": 0.259,
      "step": 21461
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26106848062555965,
      "learning_rate": 5.553535672549421e-06,
      "loss": 0.2391,
      "step": 21462
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23466830295831836,
      "learning_rate": 5.5526472582366366e-06,
      "loss": 0.0659,
      "step": 21463
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9216328878501729,
      "learning_rate": 5.551758887678372e-06,
      "loss": 0.3543,
      "step": 21464
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33342727700697533,
      "learning_rate": 5.55087056088337e-06,
      "loss": 0.2176,
      "step": 21465
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9400578629448368,
      "learning_rate": 5.5499822778603654e-06,
      "loss": 0.3842,
      "step": 21466
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.31186640204475563,
      "learning_rate": 5.549094038618098e-06,
      "loss": 0.2273,
      "step": 21467
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.157369419824378,
      "learning_rate": 5.5482058431653084e-06,
      "loss": 0.5557,
      "step": 21468
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36984264993042765,
      "learning_rate": 5.547317691510738e-06,
      "loss": 0.169,
      "step": 21469
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5292897537672028,
      "learning_rate": 5.546429583663118e-06,
      "loss": 0.3662,
      "step": 21470
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30633541389187685,
      "learning_rate": 5.545541519631188e-06,
      "loss": 0.1119,
      "step": 21471
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3548690462295203,
      "learning_rate": 5.544653499423691e-06,
      "loss": 0.1494,
      "step": 21472
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3624137509728101,
      "learning_rate": 5.543765523049351e-06,
      "loss": 0.2742,
      "step": 21473
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.29171293779786456,
      "learning_rate": 5.54287759051692e-06,
      "loss": 0.2359,
      "step": 21474
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9854973824142914,
      "learning_rate": 5.541989701835122e-06,
      "loss": 0.5108,
      "step": 21475
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6963287863435226,
      "learning_rate": 5.541101857012696e-06,
      "loss": 0.2684,
      "step": 21476
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4199953562494338,
      "learning_rate": 5.540214056058378e-06,
      "loss": 0.8066,
      "step": 21477
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.27547539707184293,
      "learning_rate": 5.539326298980905e-06,
      "loss": 0.1857,
      "step": 21478
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5380174059286549,
      "learning_rate": 5.5384385857890034e-06,
      "loss": 0.6883,
      "step": 21479
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.29990016226111266,
      "learning_rate": 5.53755091649141e-06,
      "loss": 0.2334,
      "step": 21480
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2524370003953334,
      "learning_rate": 5.536663291096864e-06,
      "loss": 0.1388,
      "step": 21481
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6308503346405712,
      "learning_rate": 5.535775709614085e-06,
      "loss": 0.2118,
      "step": 21482
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5447421753404209,
      "learning_rate": 5.534888172051822e-06,
      "loss": 0.3681,
      "step": 21483
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5406104699560119,
      "learning_rate": 5.534000678418794e-06,
      "loss": 0.2521,
      "step": 21484
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36582090581008436,
      "learning_rate": 5.533113228723737e-06,
      "loss": 0.2364,
      "step": 21485
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.6148212600626146,
      "learning_rate": 5.5322258229753856e-06,
      "loss": 0.9045,
      "step": 21486
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3100727941209542,
      "learning_rate": 5.531338461182458e-06,
      "loss": 0.0704,
      "step": 21487
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35053563325131515,
      "learning_rate": 5.530451143353703e-06,
      "loss": 0.2611,
      "step": 21488
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3275826493608006,
      "learning_rate": 5.529563869497835e-06,
      "loss": 0.1238,
      "step": 21489
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.46932166623097854,
      "learning_rate": 5.528676639623589e-06,
      "loss": 0.3323,
      "step": 21490
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3127277557220166,
      "learning_rate": 5.527789453739697e-06,
      "loss": 0.197,
      "step": 21491
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3639922300913885,
      "learning_rate": 5.526902311854881e-06,
      "loss": 0.2854,
      "step": 21492
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.49257660921966007,
      "learning_rate": 5.52601521397787e-06,
      "loss": 0.1918,
      "step": 21493
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1630980194581428,
      "learning_rate": 5.525128160117395e-06,
      "loss": 0.5685,
      "step": 21494
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0604365926993238,
      "learning_rate": 5.524241150282185e-06,
      "loss": 0.4687,
      "step": 21495
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6065148257566958,
      "learning_rate": 5.523354184480959e-06,
      "loss": 0.3535,
      "step": 21496
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2617701792151086,
      "learning_rate": 5.522467262722449e-06,
      "loss": 0.2254,
      "step": 21497
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.40685433867762943,
      "learning_rate": 5.521580385015381e-06,
      "loss": 0.215,
      "step": 21498
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7956278462413132,
      "learning_rate": 5.520693551368472e-06,
      "loss": 0.4487,
      "step": 21499
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6104679499008531,
      "learning_rate": 5.519806761790462e-06,
      "loss": 0.2449,
      "step": 21500
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3391469628873626,
      "learning_rate": 5.5189200162900634e-06,
      "loss": 0.2615,
      "step": 21501
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1253816385238585,
      "learning_rate": 5.518033314876004e-06,
      "loss": 0.0661,
      "step": 21502
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.45819728234963486,
      "learning_rate": 5.517146657557011e-06,
      "loss": 0.3046,
      "step": 21503
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35735590118883875,
      "learning_rate": 5.5162600443418e-06,
      "loss": 0.2299,
      "step": 21504
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3896413564857884,
      "learning_rate": 5.515373475239099e-06,
      "loss": 0.2509,
      "step": 21505
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.44721831091416003,
      "learning_rate": 5.514486950257629e-06,
      "loss": 0.2205,
      "step": 21506
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.683131647015245,
      "learning_rate": 5.513600469406113e-06,
      "loss": 0.7358,
      "step": 21507
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5445491934166143,
      "learning_rate": 5.51271403269327e-06,
      "loss": 0.2699,
      "step": 21508
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.49308414613445417,
      "learning_rate": 5.511827640127829e-06,
      "loss": 0.3006,
      "step": 21509
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35847927877625535,
      "learning_rate": 5.510941291718499e-06,
      "loss": 0.1718,
      "step": 21510
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.18836191044879916,
      "learning_rate": 5.510054987474006e-06,
      "loss": 0.1177,
      "step": 21511
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.5456923811156404,
      "learning_rate": 5.509168727403074e-06,
      "loss": 0.6201,
      "step": 21512
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8926234806998737,
      "learning_rate": 5.508282511514409e-06,
      "loss": 0.5906,
      "step": 21513
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7910056748608169,
      "learning_rate": 5.507396339816746e-06,
      "loss": 0.3784,
      "step": 21514
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24588448120166556,
      "learning_rate": 5.506510212318794e-06,
      "loss": 0.2077,
      "step": 21515
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.43286160972834103,
      "learning_rate": 5.505624129029273e-06,
      "loss": 0.337,
      "step": 21516
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5473087625193014,
      "learning_rate": 5.504738089956903e-06,
      "loss": 0.3005,
      "step": 21517
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.752870633994216,
      "learning_rate": 5.503852095110396e-06,
      "loss": 0.5967,
      "step": 21518
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30820919946961123,
      "learning_rate": 5.502966144498471e-06,
      "loss": 0.1494,
      "step": 21519
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25768388095217554,
      "learning_rate": 5.502080238129844e-06,
      "loss": 0.1855,
      "step": 21520
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3069321532795988,
      "learning_rate": 5.501194376013237e-06,
      "loss": 0.2318,
      "step": 21521
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.544374529725893,
      "learning_rate": 5.5003085581573554e-06,
      "loss": 0.6237,
      "step": 21522
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30899699919684864,
      "learning_rate": 5.4994227845709196e-06,
      "loss": 0.1622,
      "step": 21523
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3066915153701007,
      "learning_rate": 5.498537055262646e-06,
      "loss": 0.2389,
      "step": 21524
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6329686297201794,
      "learning_rate": 5.4976513702412395e-06,
      "loss": 0.4103,
      "step": 21525
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8174127552732959,
      "learning_rate": 5.496765729515427e-06,
      "loss": 0.2666,
      "step": 21526
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32338596267685316,
      "learning_rate": 5.495880133093911e-06,
      "loss": 0.273,
      "step": 21527
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2660435453386954,
      "learning_rate": 5.494994580985409e-06,
      "loss": 0.1776,
      "step": 21528
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4551689812891737,
      "learning_rate": 5.494109073198637e-06,
      "loss": 0.2326,
      "step": 21529
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.31378818895809324,
      "learning_rate": 5.493223609742298e-06,
      "loss": 0.0803,
      "step": 21530
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.310666371300116,
      "learning_rate": 5.492338190625108e-06,
      "loss": 0.7717,
      "step": 21531
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2791300135827619,
      "learning_rate": 5.491452815855778e-06,
      "loss": 0.1821,
      "step": 21532
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37613193038631554,
      "learning_rate": 5.4905674854430195e-06,
      "loss": 0.2726,
      "step": 21533
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.550289669313405,
      "learning_rate": 5.489682199395545e-06,
      "loss": 0.2617,
      "step": 21534
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8718165161830521,
      "learning_rate": 5.488796957722057e-06,
      "loss": 0.4369,
      "step": 21535
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.651456094247287,
      "learning_rate": 5.487911760431269e-06,
      "loss": 0.101,
      "step": 21536
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.34813323090941545,
      "learning_rate": 5.48702660753189e-06,
      "loss": 0.166,
      "step": 21537
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3564195829302837,
      "learning_rate": 5.4861414990326325e-06,
      "loss": 0.2267,
      "step": 21538
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2252761842414007,
      "learning_rate": 5.485256434942196e-06,
      "loss": 0.1749,
      "step": 21539
    },
    {
      "epoch": 0.66,
      "grad_norm": 2.2863013248867277,
      "learning_rate": 5.484371415269292e-06,
      "loss": 0.7627,
      "step": 21540
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7383496122532164,
      "learning_rate": 5.483486440022632e-06,
      "loss": 0.2116,
      "step": 21541
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35797532282626926,
      "learning_rate": 5.4826015092109105e-06,
      "loss": 0.2972,
      "step": 21542
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8570889604950013,
      "learning_rate": 5.481716622842849e-06,
      "loss": 0.3137,
      "step": 21543
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.47080518239105124,
      "learning_rate": 5.480831780927144e-06,
      "loss": 0.3299,
      "step": 21544
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37956953666476684,
      "learning_rate": 5.479946983472502e-06,
      "loss": 0.1855,
      "step": 21545
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6285158690052898,
      "learning_rate": 5.479062230487629e-06,
      "loss": 0.3667,
      "step": 21546
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.38070520488618576,
      "learning_rate": 5.478177521981233e-06,
      "loss": 0.2066,
      "step": 21547
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2821766422283463,
      "learning_rate": 5.47729285796201e-06,
      "loss": 0.1585,
      "step": 21548
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0682008188780003,
      "learning_rate": 5.4764082384386684e-06,
      "loss": 0.3216,
      "step": 21549
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3094223966272712,
      "learning_rate": 5.475523663419915e-06,
      "loss": 0.2294,
      "step": 21550
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3553642116638466,
      "learning_rate": 5.474639132914441e-06,
      "loss": 0.2663,
      "step": 21551
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6428132941938529,
      "learning_rate": 5.473754646930965e-06,
      "loss": 0.3028,
      "step": 21552
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4599059758189619,
      "learning_rate": 5.472870205478174e-06,
      "loss": 0.6736,
      "step": 21553
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1021941657435141,
      "learning_rate": 5.471985808564777e-06,
      "loss": 0.2033,
      "step": 21554
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.38327409265729306,
      "learning_rate": 5.4711014561994765e-06,
      "loss": 0.253,
      "step": 21555
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25532577749833835,
      "learning_rate": 5.470217148390967e-06,
      "loss": 0.134,
      "step": 21556
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.34595420580168285,
      "learning_rate": 5.469332885147951e-06,
      "loss": 0.3138,
      "step": 21557
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3857079314309691,
      "learning_rate": 5.4684486664791294e-06,
      "loss": 0.1886,
      "step": 21558
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6690377055920138,
      "learning_rate": 5.467564492393201e-06,
      "loss": 0.3906,
      "step": 21559
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2963308522405076,
      "learning_rate": 5.466680362898868e-06,
      "loss": 0.1874,
      "step": 21560
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0245869323394514,
      "learning_rate": 5.465796278004823e-06,
      "loss": 0.4828,
      "step": 21561
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3892853036009954,
      "learning_rate": 5.464912237719764e-06,
      "loss": 0.2519,
      "step": 21562
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4709237894543325,
      "learning_rate": 5.46402824205239e-06,
      "loss": 0.225,
      "step": 21563
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.021029310859858,
      "learning_rate": 5.463144291011404e-06,
      "loss": 0.5301,
      "step": 21564
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3151269670077694,
      "learning_rate": 5.4622603846054935e-06,
      "loss": 0.1886,
      "step": 21565
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4509599961209459,
      "learning_rate": 5.4613765228433595e-06,
      "loss": 0.2556,
      "step": 21566
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.72913013874337,
      "learning_rate": 5.460492705733699e-06,
      "loss": 0.3128,
      "step": 21567
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4946664548290353,
      "learning_rate": 5.459608933285196e-06,
      "loss": 0.3464,
      "step": 21568
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24952217975009097,
      "learning_rate": 5.458725205506565e-06,
      "loss": 0.1725,
      "step": 21569
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4661711718111653,
      "learning_rate": 5.457841522406484e-06,
      "loss": 0.3294,
      "step": 21570
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.201369350057821,
      "learning_rate": 5.456957883993653e-06,
      "loss": 0.0931,
      "step": 21571
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4220207159273819,
      "learning_rate": 5.4560742902767696e-06,
      "loss": 0.4994,
      "step": 21572
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.46549673588117685,
      "learning_rate": 5.4551907412645174e-06,
      "loss": 0.1012,
      "step": 21573
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3659371035190862,
      "learning_rate": 5.454307236965595e-06,
      "loss": 0.282,
      "step": 21574
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24204839159818534,
      "learning_rate": 5.453423777388694e-06,
      "loss": 0.1984,
      "step": 21575
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6766425244039745,
      "learning_rate": 5.452540362542509e-06,
      "loss": 0.2779,
      "step": 21576
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7754388589270418,
      "learning_rate": 5.451656992435721e-06,
      "loss": 0.5314,
      "step": 21577
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2654365148108166,
      "learning_rate": 5.450773667077036e-06,
      "loss": 0.1759,
      "step": 21578
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.44593201346092,
      "learning_rate": 5.449890386475132e-06,
      "loss": 0.2683,
      "step": 21579
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4343428251926389,
      "learning_rate": 5.449007150638704e-06,
      "loss": 0.2335,
      "step": 21580
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4784828560282039,
      "learning_rate": 5.4481239595764455e-06,
      "loss": 0.3189,
      "step": 21581
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3084418608778658,
      "learning_rate": 5.4472408132970365e-06,
      "loss": 0.1649,
      "step": 21582
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.517596430632847,
      "learning_rate": 5.446357711809171e-06,
      "loss": 0.3066,
      "step": 21583
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9069543738401095,
      "learning_rate": 5.445474655121536e-06,
      "loss": 0.26,
      "step": 21584
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7820839496723639,
      "learning_rate": 5.44459164324282e-06,
      "loss": 0.3617,
      "step": 21585
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25796486220986714,
      "learning_rate": 5.443708676181716e-06,
      "loss": 0.2162,
      "step": 21586
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3445793616969427,
      "learning_rate": 5.4428257539469e-06,
      "loss": 0.2577,
      "step": 21587
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8697063176083248,
      "learning_rate": 5.4419428765470635e-06,
      "loss": 0.4348,
      "step": 21588
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23874518421564522,
      "learning_rate": 5.441060043990893e-06,
      "loss": 0.149,
      "step": 21589
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.784565176731024,
      "learning_rate": 5.440177256287078e-06,
      "loss": 0.1843,
      "step": 21590
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3553105571596318,
      "learning_rate": 5.4392945134442955e-06,
      "loss": 0.0669,
      "step": 21591
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36020868476683704,
      "learning_rate": 5.438411815471234e-06,
      "loss": 0.2742,
      "step": 21592
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4697762391102484,
      "learning_rate": 5.437529162376582e-06,
      "loss": 0.2595,
      "step": 21593
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1199741723440895,
      "learning_rate": 5.436646554169012e-06,
      "loss": 0.4564,
      "step": 21594
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0585014560813104,
      "learning_rate": 5.435763990857222e-06,
      "loss": 0.4677,
      "step": 21595
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6577014854420608,
      "learning_rate": 5.4348814724498846e-06,
      "loss": 0.3607,
      "step": 21596
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35550158558511824,
      "learning_rate": 5.433998998955684e-06,
      "loss": 0.2302,
      "step": 21597
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2856365644758661,
      "learning_rate": 5.433116570383309e-06,
      "loss": 0.2429,
      "step": 21598
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.31422654703761094,
      "learning_rate": 5.43223418674143e-06,
      "loss": 0.0786,
      "step": 21599
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.838167896024288,
      "learning_rate": 5.4313518480387365e-06,
      "loss": 0.3484,
      "step": 21600
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.37829806878582956,
      "learning_rate": 5.430469554283906e-06,
      "loss": 0.2007,
      "step": 21601
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.0426350973999308,
      "learning_rate": 5.429587305485623e-06,
      "loss": 0.3921,
      "step": 21602
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.342847006647542,
      "learning_rate": 5.42870510165256e-06,
      "loss": 0.2166,
      "step": 21603
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3022894064245922,
      "learning_rate": 5.427822942793402e-06,
      "loss": 0.2371,
      "step": 21604
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4346393810441749,
      "learning_rate": 5.426940828916827e-06,
      "loss": 0.2974,
      "step": 21605
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4202895000559468,
      "learning_rate": 5.426058760031507e-06,
      "loss": 0.2425,
      "step": 21606
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4135366339412508,
      "learning_rate": 5.425176736146134e-06,
      "loss": 0.2008,
      "step": 21607
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1885713712458989,
      "learning_rate": 5.424294757269374e-06,
      "loss": 0.0722,
      "step": 21608
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7010931285950779,
      "learning_rate": 5.4234128234099056e-06,
      "loss": 0.3513,
      "step": 21609
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.391555874368261,
      "learning_rate": 5.422530934576414e-06,
      "loss": 0.2005,
      "step": 21610
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.45286803107571166,
      "learning_rate": 5.421649090777561e-06,
      "loss": 0.2796,
      "step": 21611
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9458450050442585,
      "learning_rate": 5.4207672920220375e-06,
      "loss": 0.2839,
      "step": 21612
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2440114695278977,
      "learning_rate": 5.41988553831851e-06,
      "loss": 0.7631,
      "step": 21613
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30943084244552066,
      "learning_rate": 5.419003829675654e-06,
      "loss": 0.1607,
      "step": 21614
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5186236577166853,
      "learning_rate": 5.418122166102146e-06,
      "loss": 0.3149,
      "step": 21615
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3186101229255252,
      "learning_rate": 5.417240547606664e-06,
      "loss": 0.2242,
      "step": 21616
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.24135115221087133,
      "learning_rate": 5.416358974197873e-06,
      "loss": 0.1222,
      "step": 21617
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7998388476976341,
      "learning_rate": 5.415477445884452e-06,
      "loss": 0.3543,
      "step": 21618
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30962294274485425,
      "learning_rate": 5.414595962675075e-06,
      "loss": 0.1937,
      "step": 21619
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9876056495690981,
      "learning_rate": 5.413714524578404e-06,
      "loss": 0.4282,
      "step": 21620
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4673157507483008,
      "learning_rate": 5.4128331316031265e-06,
      "loss": 0.2519,
      "step": 21621
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5138772831987765,
      "learning_rate": 5.411951783757902e-06,
      "loss": 0.3527,
      "step": 21622
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.27449688165610603,
      "learning_rate": 5.411070481051406e-06,
      "loss": 0.1146,
      "step": 21623
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3898378961074791,
      "learning_rate": 5.410189223492312e-06,
      "loss": 0.2866,
      "step": 21624
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.1876911272512959,
      "learning_rate": 5.409308011089284e-06,
      "loss": 0.0683,
      "step": 21625
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9202685754795913,
      "learning_rate": 5.4084268438509935e-06,
      "loss": 0.4858,
      "step": 21626
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.41709481319369723,
      "learning_rate": 5.40754572178611e-06,
      "loss": 0.2646,
      "step": 21627
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3783197233805021,
      "learning_rate": 5.406664644903307e-06,
      "loss": 0.2754,
      "step": 21628
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3129486483720174,
      "learning_rate": 5.405783613211244e-06,
      "loss": 0.2053,
      "step": 21629
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.44549943617782056,
      "learning_rate": 5.404902626718595e-06,
      "loss": 0.0182,
      "step": 21630
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3424174417799999,
      "learning_rate": 5.404021685434029e-06,
      "loss": 0.7618,
      "step": 21631
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2964962693851959,
      "learning_rate": 5.4031407893662025e-06,
      "loss": 0.1702,
      "step": 21632
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5653300250355926,
      "learning_rate": 5.402259938523798e-06,
      "loss": 0.3385,
      "step": 21633
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2942013583045931,
      "learning_rate": 5.401379132915467e-06,
      "loss": 0.224,
      "step": 21634
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4276199784992439,
      "learning_rate": 5.400498372549883e-06,
      "loss": 0.2655,
      "step": 21635
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5400351938938658,
      "learning_rate": 5.399617657435713e-06,
      "loss": 0.2566,
      "step": 21636
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3291151044653083,
      "learning_rate": 5.39873698758161e-06,
      "loss": 0.2585,
      "step": 21637
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19695487340230353,
      "learning_rate": 5.397856362996254e-06,
      "loss": 0.0615,
      "step": 21638
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.662267146958705,
      "learning_rate": 5.396975783688298e-06,
      "loss": 0.7708,
      "step": 21639
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2928887497597068,
      "learning_rate": 5.39609524966641e-06,
      "loss": 0.2442,
      "step": 21640
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.600418558233363,
      "learning_rate": 5.3952147609392534e-06,
      "loss": 0.3773,
      "step": 21641
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3215261937348487,
      "learning_rate": 5.394334317515487e-06,
      "loss": 0.196,
      "step": 21642
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7572727327456104,
      "learning_rate": 5.393453919403775e-06,
      "loss": 0.2656,
      "step": 21643
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6097681348704046,
      "learning_rate": 5.392573566612779e-06,
      "loss": 0.3746,
      "step": 21644
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3421053558956819,
      "learning_rate": 5.391693259151163e-06,
      "loss": 0.1919,
      "step": 21645
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4186613950448722,
      "learning_rate": 5.390812997027579e-06,
      "loss": 0.2592,
      "step": 21646
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2010456870011657,
      "learning_rate": 5.3899327802506995e-06,
      "loss": 0.1525,
      "step": 21647
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.010118433741099,
      "learning_rate": 5.389052608829175e-06,
      "loss": 0.4989,
      "step": 21648
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.014742270899946,
      "learning_rate": 5.3881724827716685e-06,
      "loss": 0.5487,
      "step": 21649
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7612803105823763,
      "learning_rate": 5.387292402086842e-06,
      "loss": 0.287,
      "step": 21650
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2965929657015164,
      "learning_rate": 5.386412366783347e-06,
      "loss": 0.1919,
      "step": 21651
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3187103207825278,
      "learning_rate": 5.385532376869845e-06,
      "loss": 0.2843,
      "step": 21652
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9067261767268732,
      "learning_rate": 5.384652432354995e-06,
      "loss": 0.2881,
      "step": 21653
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9089791539452753,
      "learning_rate": 5.383772533247456e-06,
      "loss": 0.4755,
      "step": 21654
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.19767455462997177,
      "learning_rate": 5.382892679555878e-06,
      "loss": 0.1163,
      "step": 21655
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26315505318069704,
      "learning_rate": 5.38201287128892e-06,
      "loss": 0.1437,
      "step": 21656
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.4812201710441404,
      "learning_rate": 5.381133108455243e-06,
      "loss": 0.8662,
      "step": 21657
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32939005226559637,
      "learning_rate": 5.380253391063491e-06,
      "loss": 0.2685,
      "step": 21658
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8259026226395562,
      "learning_rate": 5.379373719122335e-06,
      "loss": 0.4905,
      "step": 21659
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32833107779549825,
      "learning_rate": 5.378494092640416e-06,
      "loss": 0.1959,
      "step": 21660
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.1230159747025277,
      "learning_rate": 5.3776145116263925e-06,
      "loss": 0.5072,
      "step": 21661
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8877809028951409,
      "learning_rate": 5.376734976088922e-06,
      "loss": 0.3006,
      "step": 21662
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33104779992554595,
      "learning_rate": 5.375855486036647e-06,
      "loss": 0.2954,
      "step": 21663
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2789845122404183,
      "learning_rate": 5.374976041478235e-06,
      "loss": 0.1821,
      "step": 21664
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.7617083764086607,
      "learning_rate": 5.374096642422326e-06,
      "loss": 0.6479,
      "step": 21665
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.15342362391273429,
      "learning_rate": 5.373217288877577e-06,
      "loss": 0.0786,
      "step": 21666
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3460557272062936,
      "learning_rate": 5.372337980852643e-06,
      "loss": 0.7891,
      "step": 21667
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5019239495987666,
      "learning_rate": 5.371458718356166e-06,
      "loss": 0.1667,
      "step": 21668
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3333019965153957,
      "learning_rate": 5.370579501396802e-06,
      "loss": 0.2059,
      "step": 21669
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.33491819349832025,
      "learning_rate": 5.3697003299832e-06,
      "loss": 0.2772,
      "step": 21670
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9299622174159436,
      "learning_rate": 5.368821204124013e-06,
      "loss": 0.29,
      "step": 21671
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.45953956783916866,
      "learning_rate": 5.367942123827883e-06,
      "loss": 0.2691,
      "step": 21672
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.23901915482730673,
      "learning_rate": 5.367063089103462e-06,
      "loss": 0.0688,
      "step": 21673
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.28289523617588075,
      "learning_rate": 5.366184099959404e-06,
      "loss": 0.2125,
      "step": 21674
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30691266951441404,
      "learning_rate": 5.365305156404342e-06,
      "loss": 0.2476,
      "step": 21675
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3858265958833351,
      "learning_rate": 5.364426258446942e-06,
      "loss": 0.765,
      "step": 21676
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5569532479409561,
      "learning_rate": 5.363547406095838e-06,
      "loss": 0.2198,
      "step": 21677
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.35002429670661583,
      "learning_rate": 5.3626685993596795e-06,
      "loss": 0.2854,
      "step": 21678
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.7462620398582998,
      "learning_rate": 5.361789838247113e-06,
      "loss": 0.301,
      "step": 21679
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.329473981552498,
      "learning_rate": 5.36091112276679e-06,
      "loss": 0.6264,
      "step": 21680
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.29545754817488684,
      "learning_rate": 5.360032452927344e-06,
      "loss": 0.2174,
      "step": 21681
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.36286863882683834,
      "learning_rate": 5.359153828737427e-06,
      "loss": 0.1625,
      "step": 21682
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3670449863771443,
      "learning_rate": 5.358275250205685e-06,
      "loss": 0.2327,
      "step": 21683
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2583870680574275,
      "learning_rate": 5.357396717340749e-06,
      "loss": 0.1028,
      "step": 21684
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.2677092996856076,
      "learning_rate": 5.356518230151281e-06,
      "loss": 0.842,
      "step": 21685
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5236509457073613,
      "learning_rate": 5.35563978864591e-06,
      "loss": 0.2715,
      "step": 21686
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3533972190760922,
      "learning_rate": 5.354761392833283e-06,
      "loss": 0.2356,
      "step": 21687
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.32252326199497117,
      "learning_rate": 5.353883042722048e-06,
      "loss": 0.2407,
      "step": 21688
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.8667237075337841,
      "learning_rate": 5.353004738320831e-06,
      "loss": 0.4317,
      "step": 21689
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.565183871032969,
      "learning_rate": 5.3521264796382884e-06,
      "loss": 0.036,
      "step": 21690
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6160371406978757,
      "learning_rate": 5.351248266683053e-06,
      "loss": 0.2779,
      "step": 21691
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3507401340332094,
      "learning_rate": 5.350370099463766e-06,
      "loss": 0.1886,
      "step": 21692
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2735708677496132,
      "learning_rate": 5.349491977989072e-06,
      "loss": 0.1669,
      "step": 21693
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3113781404655217,
      "learning_rate": 5.348613902267603e-06,
      "loss": 0.2507,
      "step": 21694
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4723337869934932,
      "learning_rate": 5.347735872308e-06,
      "loss": 0.225,
      "step": 21695
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5051203751831288,
      "learning_rate": 5.346857888118904e-06,
      "loss": 0.285,
      "step": 21696
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.42558120373545016,
      "learning_rate": 5.345979949708952e-06,
      "loss": 0.2533,
      "step": 21697
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.242864287881476,
      "learning_rate": 5.345102057086779e-06,
      "loss": 0.4479,
      "step": 21698
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.30780038007494775,
      "learning_rate": 5.344224210261024e-06,
      "loss": 0.233,
      "step": 21699
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.9019943485405323,
      "learning_rate": 5.343346409240326e-06,
      "loss": 0.4234,
      "step": 21700
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3175060621115266,
      "learning_rate": 5.3424686540333105e-06,
      "loss": 0.1971,
      "step": 21701
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26690038232875823,
      "learning_rate": 5.341590944648629e-06,
      "loss": 0.1853,
      "step": 21702
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.5388928315812231,
      "learning_rate": 5.340713281094905e-06,
      "loss": 0.2874,
      "step": 21703
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.857576337408842,
      "learning_rate": 5.339835663380776e-06,
      "loss": 0.5062,
      "step": 21704
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.2617905917137441,
      "learning_rate": 5.338958091514881e-06,
      "loss": 0.1843,
      "step": 21705
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.3632282016376881,
      "learning_rate": 5.338080565505847e-06,
      "loss": 0.2864,
      "step": 21706
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.103244352689092,
      "learning_rate": 5.33720308536231e-06,
      "loss": 0.229,
      "step": 21707
    },
    {
      "epoch": 0.66,
      "grad_norm": 1.3296174185632472,
      "learning_rate": 5.336325651092903e-06,
      "loss": 0.2282,
      "step": 21708
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6498991183445932,
      "learning_rate": 5.335448262706262e-06,
      "loss": 0.3331,
      "step": 21709
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.25248109471079117,
      "learning_rate": 5.334570920211013e-06,
      "loss": 0.1626,
      "step": 21710
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4962879375980748,
      "learning_rate": 5.33369362361579e-06,
      "loss": 0.3273,
      "step": 21711
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.4155881039870201,
      "learning_rate": 5.332816372929226e-06,
      "loss": 0.2525,
      "step": 21712
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.6697657903848047,
      "learning_rate": 5.331939168159949e-06,
      "loss": 0.39,
      "step": 21713
    },
    {
      "epoch": 0.66,
      "grad_norm": 0.26439147832188287,
      "learning_rate": 5.331062009316594e-06,
      "loss": 0.1765,
      "step": 21714
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5492467258267142,
      "learning_rate": 5.33018489640778e-06,
      "loss": 0.7668,
      "step": 21715
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2653141648059883,
      "learning_rate": 5.329307829442151e-06,
      "loss": 0.0996,
      "step": 21716
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35272321713475724,
      "learning_rate": 5.328430808428325e-06,
      "loss": 0.2948,
      "step": 21717
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5063869366978514,
      "learning_rate": 5.327553833374933e-06,
      "loss": 0.1057,
      "step": 21718
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5215609373064599,
      "learning_rate": 5.326676904290609e-06,
      "loss": 0.2997,
      "step": 21719
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3093438949953456,
      "learning_rate": 5.325800021183971e-06,
      "loss": 0.1828,
      "step": 21720
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8237278785926262,
      "learning_rate": 5.324923184063649e-06,
      "loss": 0.3088,
      "step": 21721
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41634580213090255,
      "learning_rate": 5.324046392938273e-06,
      "loss": 0.3041,
      "step": 21722
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3889322678996433,
      "learning_rate": 5.323169647816469e-06,
      "loss": 0.1872,
      "step": 21723
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.32989568588174933,
      "learning_rate": 5.32229294870686e-06,
      "loss": 0.2741,
      "step": 21724
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20500977532394435,
      "learning_rate": 5.32141629561807e-06,
      "loss": 0.0894,
      "step": 21725
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.274032160967333,
      "learning_rate": 5.32053968855873e-06,
      "loss": 0.5414,
      "step": 21726
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6549029566434015,
      "learning_rate": 5.319663127537452e-06,
      "loss": 0.1042,
      "step": 21727
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3540008820636247,
      "learning_rate": 5.318786612562877e-06,
      "loss": 0.2539,
      "step": 21728
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37649788617586316,
      "learning_rate": 5.317910143643615e-06,
      "loss": 0.2579,
      "step": 21729
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9235935170698362,
      "learning_rate": 5.317033720788294e-06,
      "loss": 0.3699,
      "step": 21730
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0503177536257846,
      "learning_rate": 5.316157344005542e-06,
      "loss": 0.4585,
      "step": 21731
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6232500738224594,
      "learning_rate": 5.315281013303968e-06,
      "loss": 0.3848,
      "step": 21732
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.25090070075568094,
      "learning_rate": 5.314404728692202e-06,
      "loss": 0.1816,
      "step": 21733
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36690921868409593,
      "learning_rate": 5.313528490178865e-06,
      "loss": 0.0987,
      "step": 21734
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37119066429132574,
      "learning_rate": 5.3126522977725805e-06,
      "loss": 0.2855,
      "step": 21735
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3742180699447716,
      "learning_rate": 5.31177615148196e-06,
      "loss": 0.0619,
      "step": 21736
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.352004301165734,
      "learning_rate": 5.310900051315629e-06,
      "loss": 0.2622,
      "step": 21737
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9578727462790532,
      "learning_rate": 5.310023997282206e-06,
      "loss": 0.3146,
      "step": 21738
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8917352110099931,
      "learning_rate": 5.309147989390311e-06,
      "loss": 0.4171,
      "step": 21739
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2972628975503107,
      "learning_rate": 5.308272027648565e-06,
      "loss": 0.255,
      "step": 21740
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3740676910509222,
      "learning_rate": 5.307396112065577e-06,
      "loss": 0.235,
      "step": 21741
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4251905225246538,
      "learning_rate": 5.306520242649973e-06,
      "loss": 0.2264,
      "step": 21742
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4497345141231121,
      "learning_rate": 5.30564441941037e-06,
      "loss": 0.2612,
      "step": 21743
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3829957857708987,
      "learning_rate": 5.3047686423553735e-06,
      "loss": 0.1081,
      "step": 21744
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0516248706451743,
      "learning_rate": 5.303892911493616e-06,
      "loss": 0.3761,
      "step": 21745
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36680576693588013,
      "learning_rate": 5.3030172268337e-06,
      "loss": 0.1799,
      "step": 21746
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29268186634801957,
      "learning_rate": 5.302141588384249e-06,
      "loss": 0.2375,
      "step": 21747
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1146333935235808,
      "learning_rate": 5.301265996153872e-06,
      "loss": 0.549,
      "step": 21748
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7801697715005543,
      "learning_rate": 5.300390450151192e-06,
      "loss": 0.5995,
      "step": 21749
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.860564995108643,
      "learning_rate": 5.299514950384812e-06,
      "loss": 0.4072,
      "step": 21750
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2951930230417143,
      "learning_rate": 5.2986394968633516e-06,
      "loss": 0.1915,
      "step": 21751
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3106364752427346,
      "learning_rate": 5.297764089595428e-06,
      "loss": 0.1873,
      "step": 21752
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.30973640671535335,
      "learning_rate": 5.296888728589637e-06,
      "loss": 0.2252,
      "step": 21753
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8783903577599491,
      "learning_rate": 5.296013413854612e-06,
      "loss": 0.4528,
      "step": 21754
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3133912986140401,
      "learning_rate": 5.295138145398953e-06,
      "loss": 0.1849,
      "step": 21755
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9423928108746304,
      "learning_rate": 5.29426292323127e-06,
      "loss": 0.4076,
      "step": 21756
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.281658409418863,
      "learning_rate": 5.293387747360181e-06,
      "loss": 0.3902,
      "step": 21757
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6127367807399786,
      "learning_rate": 5.29251261779429e-06,
      "loss": 0.3572,
      "step": 21758
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28510671302739743,
      "learning_rate": 5.291637534542206e-06,
      "loss": 0.1735,
      "step": 21759
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3214291785692666,
      "learning_rate": 5.290762497612544e-06,
      "loss": 0.2158,
      "step": 21760
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2710930468712451,
      "learning_rate": 5.289887507013909e-06,
      "loss": 0.161,
      "step": 21761
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8770385587749296,
      "learning_rate": 5.289012562754914e-06,
      "loss": 0.2755,
      "step": 21762
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7197424610891645,
      "learning_rate": 5.28813766484416e-06,
      "loss": 0.37,
      "step": 21763
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26083748096359743,
      "learning_rate": 5.287262813290259e-06,
      "loss": 0.1702,
      "step": 21764
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3316849064915445,
      "learning_rate": 5.286388008101816e-06,
      "loss": 0.2926,
      "step": 21765
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3050689966523124,
      "learning_rate": 5.285513249287443e-06,
      "loss": 0.1796,
      "step": 21766
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.254114557818378,
      "learning_rate": 5.284638536855737e-06,
      "loss": 0.7342,
      "step": 21767
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.455052889739596,
      "learning_rate": 5.2837638708153104e-06,
      "loss": 0.1369,
      "step": 21768
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37633050916362826,
      "learning_rate": 5.28288925117477e-06,
      "loss": 0.2947,
      "step": 21769
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19060013274389714,
      "learning_rate": 5.282014677942708e-06,
      "loss": 0.0699,
      "step": 21770
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3178510024258533,
      "learning_rate": 5.2811401511277484e-06,
      "loss": 0.3,
      "step": 21771
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6556818940809336,
      "learning_rate": 5.280265670738478e-06,
      "loss": 0.273,
      "step": 21772
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4289824338062051,
      "learning_rate": 5.279391236783509e-06,
      "loss": 0.2616,
      "step": 21773
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5028063614257401,
      "learning_rate": 5.278516849271445e-06,
      "loss": 0.2173,
      "step": 21774
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1094973755511206,
      "learning_rate": 5.277642508210883e-06,
      "loss": 0.323,
      "step": 21775
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31636653332268605,
      "learning_rate": 5.2767682136104285e-06,
      "loss": 0.2925,
      "step": 21776
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.386980819916311,
      "learning_rate": 5.27589396547868e-06,
      "loss": 0.144,
      "step": 21777
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33668215755236747,
      "learning_rate": 5.275019763824247e-06,
      "loss": 0.2554,
      "step": 21778
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8687573571683962,
      "learning_rate": 5.274145608655716e-06,
      "loss": 0.252,
      "step": 21779
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6119744454288666,
      "learning_rate": 5.273271499981703e-06,
      "loss": 0.3614,
      "step": 21780
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1513768759941834,
      "learning_rate": 5.2723974378108e-06,
      "loss": 0.071,
      "step": 21781
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3598206003503133,
      "learning_rate": 5.2715234221515985e-06,
      "loss": 0.2612,
      "step": 21782
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28981804792407034,
      "learning_rate": 5.270649453012715e-06,
      "loss": 0.2235,
      "step": 21783
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.6145286120620823,
      "learning_rate": 5.269775530402732e-06,
      "loss": 0.7339,
      "step": 21784
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1118589507021988,
      "learning_rate": 5.268901654330256e-06,
      "loss": 0.4788,
      "step": 21785
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.90346095813227,
      "learning_rate": 5.26802782480388e-06,
      "loss": 0.4219,
      "step": 21786
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3109368282910859,
      "learning_rate": 5.267154041832205e-06,
      "loss": 0.1851,
      "step": 21787
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2993669216580666,
      "learning_rate": 5.26628030542383e-06,
      "loss": 0.243,
      "step": 21788
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9216434611293509,
      "learning_rate": 5.265406615587343e-06,
      "loss": 0.5213,
      "step": 21789
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24688949611764607,
      "learning_rate": 5.264532972331343e-06,
      "loss": 0.1386,
      "step": 21790
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3947859315517502,
      "learning_rate": 5.263659375664426e-06,
      "loss": 0.2482,
      "step": 21791
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.46480593137422765,
      "learning_rate": 5.26278582559519e-06,
      "loss": 0.1911,
      "step": 21792
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5485433048425001,
      "learning_rate": 5.2619123221322235e-06,
      "loss": 0.8261,
      "step": 21793
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2903280604016302,
      "learning_rate": 5.26103886528412e-06,
      "loss": 0.2441,
      "step": 21794
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9218685023550891,
      "learning_rate": 5.26016545505948e-06,
      "loss": 0.513,
      "step": 21795
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2744085892197861,
      "learning_rate": 5.259292091466885e-06,
      "loss": 0.1828,
      "step": 21796
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8551811070026839,
      "learning_rate": 5.258418774514943e-06,
      "loss": 0.4866,
      "step": 21797
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6041483375889203,
      "learning_rate": 5.257545504212232e-06,
      "loss": 0.2859,
      "step": 21798
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31122826995800523,
      "learning_rate": 5.25667228056735e-06,
      "loss": 0.2151,
      "step": 21799
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29571455972949384,
      "learning_rate": 5.255799103588889e-06,
      "loss": 0.1766,
      "step": 21800
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21097084154300028,
      "learning_rate": 5.254925973285434e-06,
      "loss": 0.1464,
      "step": 21801
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2778951382934414,
      "learning_rate": 5.25405288966558e-06,
      "loss": 0.7184,
      "step": 21802
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9004112514096085,
      "learning_rate": 5.253179852737913e-06,
      "loss": 0.5344,
      "step": 21803
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8571561020302839,
      "learning_rate": 5.25230686251103e-06,
      "loss": 0.4589,
      "step": 21804
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28037532880929855,
      "learning_rate": 5.25143391899351e-06,
      "loss": 0.1779,
      "step": 21805
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44867768220534215,
      "learning_rate": 5.2505610221939464e-06,
      "loss": 0.3286,
      "step": 21806
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37836740030678145,
      "learning_rate": 5.249688172120929e-06,
      "loss": 0.2453,
      "step": 21807
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8791680839958727,
      "learning_rate": 5.248815368783036e-06,
      "loss": 0.8374,
      "step": 21808
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.24235636810081165,
      "learning_rate": 5.2479426121888665e-06,
      "loss": 0.0847,
      "step": 21809
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.28139665630424293,
      "learning_rate": 5.2470699023469985e-06,
      "loss": 0.2269,
      "step": 21810
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0730228722871125,
      "learning_rate": 5.24619723926602e-06,
      "loss": 0.4071,
      "step": 21811
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5468589423311317,
      "learning_rate": 5.245324622954522e-06,
      "loss": 0.353,
      "step": 21812
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41250738769700546,
      "learning_rate": 5.244452053421078e-06,
      "loss": 0.265,
      "step": 21813
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2841242014908128,
      "learning_rate": 5.243579530674285e-06,
      "loss": 0.1818,
      "step": 21814
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.773652176146961,
      "learning_rate": 5.24270705472272e-06,
      "loss": 0.397,
      "step": 21815
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.31167982376526,
      "learning_rate": 5.241834625574966e-06,
      "loss": 0.1981,
      "step": 21816
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4677082076634051,
      "learning_rate": 5.24096224323961e-06,
      "loss": 0.2845,
      "step": 21817
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.25627595826925453,
      "learning_rate": 5.240089907725238e-06,
      "loss": 0.1846,
      "step": 21818
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.23218335094519324,
      "learning_rate": 5.239217619040422e-06,
      "loss": 0.1979,
      "step": 21819
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.954484698165557,
      "learning_rate": 5.238345377193751e-06,
      "loss": 0.3338,
      "step": 21820
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2483180434121777,
      "learning_rate": 5.2374731821938085e-06,
      "loss": 0.8511,
      "step": 21821
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7226831326048314,
      "learning_rate": 5.236601034049164e-06,
      "loss": 0.2709,
      "step": 21822
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35672152619846426,
      "learning_rate": 5.235728932768414e-06,
      "loss": 0.2549,
      "step": 21823
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.423180210279325,
      "learning_rate": 5.234856878360126e-06,
      "loss": 0.2521,
      "step": 21824
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5386616355293768,
      "learning_rate": 5.233984870832885e-06,
      "loss": 0.3575,
      "step": 21825
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.106701485220901,
      "learning_rate": 5.23311291019527e-06,
      "loss": 0.2174,
      "step": 21826
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20244145216067058,
      "learning_rate": 5.232240996455858e-06,
      "loss": 0.0725,
      "step": 21827
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36254183410469853,
      "learning_rate": 5.231369129623227e-06,
      "loss": 0.2606,
      "step": 21828
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.23121136666269332,
      "learning_rate": 5.230497309705955e-06,
      "loss": 0.1464,
      "step": 21829
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.320765206037349,
      "learning_rate": 5.229625536712623e-06,
      "loss": 0.2565,
      "step": 21830
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8870541893258193,
      "learning_rate": 5.2287538106518035e-06,
      "loss": 0.2957,
      "step": 21831
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5853175660923534,
      "learning_rate": 5.227882131532071e-06,
      "loss": 0.3264,
      "step": 21832
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3099328156679569,
      "learning_rate": 5.227010499362011e-06,
      "loss": 0.2126,
      "step": 21833
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0902642815120454,
      "learning_rate": 5.2261389141501825e-06,
      "loss": 0.4735,
      "step": 21834
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4190087278102843,
      "learning_rate": 5.225267375905179e-06,
      "loss": 0.2075,
      "step": 21835
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3016331737159123,
      "learning_rate": 5.22439588463556e-06,
      "loss": 0.2623,
      "step": 21836
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2544313551216352,
      "learning_rate": 5.223524440349907e-06,
      "loss": 0.1569,
      "step": 21837
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.46668481545626955,
      "learning_rate": 5.222653043056798e-06,
      "loss": 0.3063,
      "step": 21838
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2314262401791067,
      "learning_rate": 5.22178169276479e-06,
      "loss": 0.3296,
      "step": 21839
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0775976294260197,
      "learning_rate": 5.220910389482476e-06,
      "loss": 0.3023,
      "step": 21840
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37493975829013576,
      "learning_rate": 5.220039133218412e-06,
      "loss": 0.2424,
      "step": 21841
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2586115412185602,
      "learning_rate": 5.2191679239811764e-06,
      "loss": 0.2184,
      "step": 21842
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3244407772058273,
      "learning_rate": 5.218296761779345e-06,
      "loss": 0.565,
      "step": 21843
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5458052733388996,
      "learning_rate": 5.217425646621479e-06,
      "loss": 0.1502,
      "step": 21844
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.348378250873855,
      "learning_rate": 5.216554578516152e-06,
      "loss": 0.1839,
      "step": 21845
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33343097998185695,
      "learning_rate": 5.215683557471936e-06,
      "loss": 0.2125,
      "step": 21846
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.432592835175339,
      "learning_rate": 5.214812583497404e-06,
      "loss": 0.712,
      "step": 21847
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5972929502628505,
      "learning_rate": 5.213941656601111e-06,
      "loss": 0.2838,
      "step": 21848
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5892615557490875,
      "learning_rate": 5.213070776791643e-06,
      "loss": 0.476,
      "step": 21849
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.26997225290460963,
      "learning_rate": 5.2121999440775565e-06,
      "loss": 0.1732,
      "step": 21850
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5742151509726192,
      "learning_rate": 5.211329158467422e-06,
      "loss": 0.8465,
      "step": 21851
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29086133693981187,
      "learning_rate": 5.210458419969811e-06,
      "loss": 0.0808,
      "step": 21852
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3085994310128656,
      "learning_rate": 5.209587728593282e-06,
      "loss": 0.2459,
      "step": 21853
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9239139999733433,
      "learning_rate": 5.2087170843464055e-06,
      "loss": 0.4205,
      "step": 21854
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29949079529472195,
      "learning_rate": 5.207846487237746e-06,
      "loss": 0.1765,
      "step": 21855
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8277693150149326,
      "learning_rate": 5.206975937275874e-06,
      "loss": 0.4704,
      "step": 21856
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7857010514362648,
      "learning_rate": 5.206105434469346e-06,
      "loss": 0.3104,
      "step": 21857
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8989407280483783,
      "learning_rate": 5.20523497882673e-06,
      "loss": 0.8111,
      "step": 21858
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19344320727735323,
      "learning_rate": 5.204364570356595e-06,
      "loss": 0.0716,
      "step": 21859
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2956181484435893,
      "learning_rate": 5.20349420906749e-06,
      "loss": 0.267,
      "step": 21860
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3174017620095562,
      "learning_rate": 5.202623894967996e-06,
      "loss": 0.1691,
      "step": 21861
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2694644616912265,
      "learning_rate": 5.201753628066661e-06,
      "loss": 0.5498,
      "step": 21862
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.364561860248169,
      "learning_rate": 5.2008834083720546e-06,
      "loss": 0.2812,
      "step": 21863
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.388933121689564,
      "learning_rate": 5.2000132358927406e-06,
      "loss": 0.2554,
      "step": 21864
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34795581575302414,
      "learning_rate": 5.199143110637266e-06,
      "loss": 0.2405,
      "step": 21865
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.606419196883205,
      "learning_rate": 5.198273032614209e-06,
      "loss": 0.2528,
      "step": 21866
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9287128614170864,
      "learning_rate": 5.19740300183212e-06,
      "loss": 0.5479,
      "step": 21867
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2916107292452294,
      "learning_rate": 5.1965330182995586e-06,
      "loss": 0.1683,
      "step": 21868
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2449489407243723,
      "learning_rate": 5.195663082025091e-06,
      "loss": 0.2004,
      "step": 21869
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5431809426159739,
      "learning_rate": 5.194793193017267e-06,
      "loss": 0.0145,
      "step": 21870
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48484697188466674,
      "learning_rate": 5.193923351284649e-06,
      "loss": 0.3503,
      "step": 21871
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3309475345804925,
      "learning_rate": 5.193053556835793e-06,
      "loss": 0.1871,
      "step": 21872
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34994454186742036,
      "learning_rate": 5.192183809679263e-06,
      "loss": 0.2425,
      "step": 21873
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5606814553856097,
      "learning_rate": 5.1913141098236066e-06,
      "loss": 0.291,
      "step": 21874
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7324597920393332,
      "learning_rate": 5.190444457277384e-06,
      "loss": 0.4322,
      "step": 21875
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.031908367918381,
      "learning_rate": 5.189574852049155e-06,
      "loss": 0.4074,
      "step": 21876
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3786196584115973,
      "learning_rate": 5.188705294147463e-06,
      "loss": 0.2555,
      "step": 21877
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3200965771224598,
      "learning_rate": 5.187835783580881e-06,
      "loss": 0.2002,
      "step": 21878
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2205004805875561,
      "learning_rate": 5.186966320357949e-06,
      "loss": 0.1246,
      "step": 21879
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.59992288521551,
      "learning_rate": 5.186096904487225e-06,
      "loss": 0.4975,
      "step": 21880
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5781936460457354,
      "learning_rate": 5.18522753597727e-06,
      "loss": 0.2329,
      "step": 21881
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48232797096108,
      "learning_rate": 5.184358214836625e-06,
      "loss": 0.298,
      "step": 21882
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.32340680671344024,
      "learning_rate": 5.183488941073849e-06,
      "loss": 0.2477,
      "step": 21883
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4470059330575814,
      "learning_rate": 5.182619714697495e-06,
      "loss": 0.334,
      "step": 21884
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8583950150679728,
      "learning_rate": 5.181750535716115e-06,
      "loss": 0.5153,
      "step": 21885
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9284728559218464,
      "learning_rate": 5.1808814041382516e-06,
      "loss": 0.3938,
      "step": 21886
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.22024705350992868,
      "learning_rate": 5.180012319972471e-06,
      "loss": 0.1542,
      "step": 21887
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.461047771176241,
      "learning_rate": 5.17914328322731e-06,
      "loss": 0.2446,
      "step": 21888
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3133470048221257,
      "learning_rate": 5.178274293911325e-06,
      "loss": 0.2273,
      "step": 21889
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.068941201575254,
      "learning_rate": 5.177405352033067e-06,
      "loss": 0.49,
      "step": 21890
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3195911585057568,
      "learning_rate": 5.176536457601076e-06,
      "loss": 0.1905,
      "step": 21891
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39786742655637214,
      "learning_rate": 5.175667610623912e-06,
      "loss": 0.2475,
      "step": 21892
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7257697738916546,
      "learning_rate": 5.174798811110115e-06,
      "loss": 0.8117,
      "step": 21893
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1892899594108135,
      "learning_rate": 5.173930059068235e-06,
      "loss": 0.3755,
      "step": 21894
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3269761943048859,
      "learning_rate": 5.173061354506822e-06,
      "loss": 0.2321,
      "step": 21895
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2579759796957344,
      "learning_rate": 5.172192697434416e-06,
      "loss": 0.2105,
      "step": 21896
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.44359996794850876,
      "learning_rate": 5.171324087859567e-06,
      "loss": 0.226,
      "step": 21897
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3812354726908065,
      "learning_rate": 5.170455525790821e-06,
      "loss": 0.1524,
      "step": 21898
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6309677047235499,
      "learning_rate": 5.169587011236725e-06,
      "loss": 0.3958,
      "step": 21899
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2650062621730836,
      "learning_rate": 5.168718544205819e-06,
      "loss": 0.1776,
      "step": 21900
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.4853947773813325,
      "learning_rate": 5.16785012470665e-06,
      "loss": 0.5055,
      "step": 21901
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3145825944435453,
      "learning_rate": 5.166981752747764e-06,
      "loss": 0.2301,
      "step": 21902
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3225149324795125,
      "learning_rate": 5.1661134283376955e-06,
      "loss": 0.8764,
      "step": 21903
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19938280666875216,
      "learning_rate": 5.165245151485e-06,
      "loss": 0.0694,
      "step": 21904
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3417285749485018,
      "learning_rate": 5.164376922198211e-06,
      "loss": 0.2025,
      "step": 21905
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2577343354671949,
      "learning_rate": 5.163508740485872e-06,
      "loss": 0.1393,
      "step": 21906
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3361726104593039,
      "learning_rate": 5.1626406063565285e-06,
      "loss": 0.2463,
      "step": 21907
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.711815304966188,
      "learning_rate": 5.161772519818715e-06,
      "loss": 0.4136,
      "step": 21908
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2753232797155895,
      "learning_rate": 5.160904480880977e-06,
      "loss": 0.069,
      "step": 21909
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40133953603446637,
      "learning_rate": 5.1600364895518516e-06,
      "loss": 0.2748,
      "step": 21910
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.802986218922769,
      "learning_rate": 5.159168545839884e-06,
      "loss": 0.4088,
      "step": 21911
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.49834029149079273,
      "learning_rate": 5.158300649753605e-06,
      "loss": 0.3395,
      "step": 21912
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3237498815605645,
      "learning_rate": 5.157432801301559e-06,
      "loss": 0.1901,
      "step": 21913
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.355415353096178,
      "learning_rate": 5.1565650004922794e-06,
      "loss": 0.2604,
      "step": 21914
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.19374167839686965,
      "learning_rate": 5.155697247334309e-06,
      "loss": 0.0695,
      "step": 21915
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5517123640089762,
      "learning_rate": 5.154829541836187e-06,
      "loss": 0.4013,
      "step": 21916
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8301887134139813,
      "learning_rate": 5.153961884006438e-06,
      "loss": 0.2751,
      "step": 21917
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3271767790101217,
      "learning_rate": 5.153094273853614e-06,
      "loss": 0.1617,
      "step": 21918
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3778322793566526,
      "learning_rate": 5.1522267113862415e-06,
      "loss": 0.2677,
      "step": 21919
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4679937601416872,
      "learning_rate": 5.151359196612855e-06,
      "loss": 0.2547,
      "step": 21920
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2987375559583136,
      "learning_rate": 5.150491729541998e-06,
      "loss": 0.8576,
      "step": 21921
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7018572224249037,
      "learning_rate": 5.149624310182193e-06,
      "loss": 0.1209,
      "step": 21922
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33793399567710874,
      "learning_rate": 5.1487569385419825e-06,
      "loss": 0.2587,
      "step": 21923
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9000186152062903,
      "learning_rate": 5.147889614629896e-06,
      "loss": 0.2931,
      "step": 21924
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3069257290789368,
      "learning_rate": 5.1470223384544725e-06,
      "loss": 0.258,
      "step": 21925
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31382463032497876,
      "learning_rate": 5.146155110024236e-06,
      "loss": 0.1603,
      "step": 21926
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39151308907326415,
      "learning_rate": 5.145287929347722e-06,
      "loss": 0.2435,
      "step": 21927
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5973785621750259,
      "learning_rate": 5.144420796433468e-06,
      "loss": 0.2014,
      "step": 21928
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1528088125192666,
      "learning_rate": 5.1435537112899916e-06,
      "loss": 0.612,
      "step": 21929
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4741793520991722,
      "learning_rate": 5.1426866739258385e-06,
      "loss": 0.2426,
      "step": 21930
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3232832184720211,
      "learning_rate": 5.141819684349529e-06,
      "loss": 0.1822,
      "step": 21931
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6188083179120925,
      "learning_rate": 5.140952742569596e-06,
      "loss": 0.338,
      "step": 21932
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4447400605779666,
      "learning_rate": 5.1400858485945736e-06,
      "loss": 0.2596,
      "step": 21933
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.783556877797699,
      "learning_rate": 5.13921900243298e-06,
      "loss": 0.4036,
      "step": 21934
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1885921595904323,
      "learning_rate": 5.138352204093351e-06,
      "loss": 0.0684,
      "step": 21935
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9175710285574732,
      "learning_rate": 5.137485453584212e-06,
      "loss": 0.4734,
      "step": 21936
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.246682613069573,
      "learning_rate": 5.136618750914096e-06,
      "loss": 0.2091,
      "step": 21937
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4905919258296887,
      "learning_rate": 5.13575209609152e-06,
      "loss": 0.3704,
      "step": 21938
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1203216922592871,
      "learning_rate": 5.134885489125014e-06,
      "loss": 0.3549,
      "step": 21939
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2340519684769522,
      "learning_rate": 5.134018930023108e-06,
      "loss": 0.7259,
      "step": 21940
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3256250349338442,
      "learning_rate": 5.133152418794324e-06,
      "loss": 0.1958,
      "step": 21941
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.48718721100676743,
      "learning_rate": 5.132285955447192e-06,
      "loss": 0.3688,
      "step": 21942
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29966180628098416,
      "learning_rate": 5.131419539990228e-06,
      "loss": 0.2518,
      "step": 21943
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2695312781097627,
      "learning_rate": 5.13055317243196e-06,
      "loss": 0.0905,
      "step": 21944
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3047163667357216,
      "learning_rate": 5.129686852780917e-06,
      "loss": 0.1839,
      "step": 21945
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33263981304068735,
      "learning_rate": 5.128820581045609e-06,
      "loss": 0.2095,
      "step": 21946
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.795144729166314,
      "learning_rate": 5.127954357234575e-06,
      "loss": 0.8482,
      "step": 21947
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41352396772857025,
      "learning_rate": 5.127088181356325e-06,
      "loss": 0.2687,
      "step": 21948
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.42602028803198005,
      "learning_rate": 5.126222053419385e-06,
      "loss": 0.3357,
      "step": 21949
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2553938211628642,
      "learning_rate": 5.125355973432277e-06,
      "loss": 0.1756,
      "step": 21950
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7140821474590137,
      "learning_rate": 5.124489941403524e-06,
      "loss": 0.3711,
      "step": 21951
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7421216839793605,
      "learning_rate": 5.123623957341639e-06,
      "loss": 0.2556,
      "step": 21952
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4714604115128191,
      "learning_rate": 5.122758021255148e-06,
      "loss": 0.2357,
      "step": 21953
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.1994500662137747,
      "learning_rate": 5.1218921331525695e-06,
      "loss": 0.1489,
      "step": 21954
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37108077914766463,
      "learning_rate": 5.1210262930424145e-06,
      "loss": 0.2863,
      "step": 21955
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3937473485145581,
      "learning_rate": 5.120160500933216e-06,
      "loss": 0.2349,
      "step": 21956
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.349434857383362,
      "learning_rate": 5.1192947568334816e-06,
      "loss": 0.7661,
      "step": 21957
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6505344125057319,
      "learning_rate": 5.11842906075173e-06,
      "loss": 0.2639,
      "step": 21958
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39660192982596865,
      "learning_rate": 5.117563412696484e-06,
      "loss": 0.1717,
      "step": 21959
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.469488062805493,
      "learning_rate": 5.11669781267625e-06,
      "loss": 0.3126,
      "step": 21960
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2757262795293174,
      "learning_rate": 5.115832260699552e-06,
      "loss": 0.2116,
      "step": 21961
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47352298610952764,
      "learning_rate": 5.1149667567749024e-06,
      "loss": 0.2339,
      "step": 21962
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.18511878935148007,
      "learning_rate": 5.114101300910821e-06,
      "loss": 0.0706,
      "step": 21963
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.37402630143394,
      "learning_rate": 5.113235893115816e-06,
      "loss": 0.2662,
      "step": 21964
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.1167151003017182,
      "learning_rate": 5.112370533398403e-06,
      "loss": 0.4133,
      "step": 21965
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.34338216398591337,
      "learning_rate": 5.111505221767097e-06,
      "loss": 0.2973,
      "step": 21966
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7570015331825092,
      "learning_rate": 5.110639958230413e-06,
      "loss": 0.2781,
      "step": 21967
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5178904683600702,
      "learning_rate": 5.109774742796862e-06,
      "loss": 0.3072,
      "step": 21968
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3476870549191175,
      "learning_rate": 5.108909575474955e-06,
      "loss": 0.2342,
      "step": 21969
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.5589647794123018,
      "learning_rate": 5.108044456273205e-06,
      "loss": 0.4704,
      "step": 21970
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5454542820146787,
      "learning_rate": 5.107179385200126e-06,
      "loss": 0.0229,
      "step": 21971
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.23437817385344817,
      "learning_rate": 5.10631436226422e-06,
      "loss": 0.1783,
      "step": 21972
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.35259086302597387,
      "learning_rate": 5.10544938747401e-06,
      "loss": 0.255,
      "step": 21973
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2797483671103861,
      "learning_rate": 5.104584460837996e-06,
      "loss": 0.1236,
      "step": 21974
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9123432168507764,
      "learning_rate": 5.10371958236469e-06,
      "loss": 0.5091,
      "step": 21975
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8799288445934429,
      "learning_rate": 5.102854752062607e-06,
      "loss": 0.321,
      "step": 21976
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.36089261495912395,
      "learning_rate": 5.101989969940246e-06,
      "loss": 0.2195,
      "step": 21977
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4762224307988373,
      "learning_rate": 5.101125236006119e-06,
      "loss": 0.2065,
      "step": 21978
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4216684339149619,
      "learning_rate": 5.100260550268734e-06,
      "loss": 0.2801,
      "step": 21979
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8862781146800812,
      "learning_rate": 5.099395912736602e-06,
      "loss": 0.1417,
      "step": 21980
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.263059830436143,
      "learning_rate": 5.098531323418222e-06,
      "loss": 0.1912,
      "step": 21981
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.38792282273502476,
      "learning_rate": 5.097666782322104e-06,
      "loss": 0.182,
      "step": 21982
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5263299089497258,
      "learning_rate": 5.096802289456755e-06,
      "loss": 0.3482,
      "step": 21983
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.6748885533816503,
      "learning_rate": 5.095937844830673e-06,
      "loss": 0.2661,
      "step": 21984
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4791562523537312,
      "learning_rate": 5.095073448452375e-06,
      "loss": 0.2594,
      "step": 21985
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9186554034514188,
      "learning_rate": 5.094209100330354e-06,
      "loss": 0.4069,
      "step": 21986
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2821043725811628,
      "learning_rate": 5.093344800473119e-06,
      "loss": 0.1881,
      "step": 21987
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.715303545387504,
      "learning_rate": 5.092480548889171e-06,
      "loss": 0.7313,
      "step": 21988
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.5217160689804544,
      "learning_rate": 5.091616345587019e-06,
      "loss": 0.2504,
      "step": 21989
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4500869508660216,
      "learning_rate": 5.090752190575157e-06,
      "loss": 0.3265,
      "step": 21990
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2873045747402873,
      "learning_rate": 5.089888083862089e-06,
      "loss": 0.1892,
      "step": 21991
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.43326705624721323,
      "learning_rate": 5.08902402545632e-06,
      "loss": 0.2746,
      "step": 21992
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7826972309744501,
      "learning_rate": 5.088160015366346e-06,
      "loss": 0.3018,
      "step": 21993
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.334830473303694,
      "learning_rate": 5.087296053600675e-06,
      "loss": 0.7809,
      "step": 21994
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.21923811187075162,
      "learning_rate": 5.086432140167798e-06,
      "loss": 0.092,
      "step": 21995
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3875665841037314,
      "learning_rate": 5.085568275076218e-06,
      "loss": 0.2977,
      "step": 21996
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3117850097989245,
      "learning_rate": 5.08470445833444e-06,
      "loss": 0.2286,
      "step": 21997
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.8122168091668578,
      "learning_rate": 5.083840689950947e-06,
      "loss": 0.12,
      "step": 21998
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.534444319549221,
      "learning_rate": 5.0829769699342545e-06,
      "loss": 0.7312,
      "step": 21999
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3015810644797846,
      "learning_rate": 5.08211329829285e-06,
      "loss": 0.1753,
      "step": 22000
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7094132907580971,
      "learning_rate": 5.081249675035233e-06,
      "loss": 0.3596,
      "step": 22001
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.41205950677042985,
      "learning_rate": 5.080386100169903e-06,
      "loss": 0.2426,
      "step": 22002
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47829050224222897,
      "learning_rate": 5.079522573705351e-06,
      "loss": 0.3594,
      "step": 22003
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.20142010737366398,
      "learning_rate": 5.078659095650074e-06,
      "loss": 0.0945,
      "step": 22004
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.2744403438936678,
      "learning_rate": 5.077795666012568e-06,
      "loss": 0.2313,
      "step": 22005
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.7632097868228251,
      "learning_rate": 5.0769322848013315e-06,
      "loss": 0.1449,
      "step": 22006
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.3810238838244973,
      "learning_rate": 5.076068952024851e-06,
      "loss": 0.6191,
      "step": 22007
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.30358153166153734,
      "learning_rate": 5.075205667691625e-06,
      "loss": 0.2117,
      "step": 22008
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.492904630091312,
      "learning_rate": 5.074342431810151e-06,
      "loss": 0.2967,
      "step": 22009
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.4614410229054855,
      "learning_rate": 5.073479244388908e-06,
      "loss": 0.2409,
      "step": 22010
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8778269668184732,
      "learning_rate": 5.072616105436406e-06,
      "loss": 0.3135,
      "step": 22011
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.40247720487233246,
      "learning_rate": 5.071753014961122e-06,
      "loss": 0.2581,
      "step": 22012
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3114660000416734,
      "learning_rate": 5.070889972971554e-06,
      "loss": 0.0712,
      "step": 22013
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3797550948922339,
      "learning_rate": 5.070026979476197e-06,
      "loss": 0.2886,
      "step": 22014
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3461286405858374,
      "learning_rate": 5.069164034483529e-06,
      "loss": 0.2269,
      "step": 22015
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.2852399673797954,
      "learning_rate": 5.0683011380020565e-06,
      "loss": 0.5734,
      "step": 22016
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.8370494299184272,
      "learning_rate": 5.067438290040254e-06,
      "loss": 0.258,
      "step": 22017
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.542386357998477,
      "learning_rate": 5.066575490606618e-06,
      "loss": 0.2877,
      "step": 22018
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3140689627252972,
      "learning_rate": 5.065712739709636e-06,
      "loss": 0.2359,
      "step": 22019
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.33180852926580096,
      "learning_rate": 5.064850037357798e-06,
      "loss": 0.2846,
      "step": 22020
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.9050287374930117,
      "learning_rate": 5.063987383559588e-06,
      "loss": 0.45,
      "step": 22021
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7509530623108286,
      "learning_rate": 5.063124778323492e-06,
      "loss": 0.4049,
      "step": 22022
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.23406852377043633,
      "learning_rate": 5.062262221658003e-06,
      "loss": 0.1569,
      "step": 22023
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47597839390113733,
      "learning_rate": 5.061399713571595e-06,
      "loss": 0.1048,
      "step": 22024
    },
    {
      "epoch": 0.67,
      "grad_norm": 2.249979523095704,
      "learning_rate": 5.06053725407277e-06,
      "loss": 0.6384,
      "step": 22025
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.39453838659855217,
      "learning_rate": 5.0596748431700016e-06,
      "loss": 0.2602,
      "step": 22026
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3966119427125113,
      "learning_rate": 5.058812480871777e-06,
      "loss": 0.2607,
      "step": 22027
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.582287366207051,
      "learning_rate": 5.057950167186585e-06,
      "loss": 0.2552,
      "step": 22028
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.831286987389966,
      "learning_rate": 5.057087902122902e-06,
      "loss": 0.8238,
      "step": 22029
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.312008616374489,
      "learning_rate": 5.056225685689214e-06,
      "loss": 0.3906,
      "step": 22030
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.29637384662181215,
      "learning_rate": 5.055363517894003e-06,
      "loss": 0.2183,
      "step": 22031
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3101589451944086,
      "learning_rate": 5.054501398745759e-06,
      "loss": 0.188,
      "step": 22032
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.31486841382868763,
      "learning_rate": 5.0536393282529526e-06,
      "loss": 0.2261,
      "step": 22033
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.7733916379435336,
      "learning_rate": 5.05277730642407e-06,
      "loss": 0.2652,
      "step": 22034
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.609301234169071,
      "learning_rate": 5.051915333267597e-06,
      "loss": 0.3644,
      "step": 22035
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3390550723912323,
      "learning_rate": 5.051053408792e-06,
      "loss": 0.1628,
      "step": 22036
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.47941823345017337,
      "learning_rate": 5.050191533005776e-06,
      "loss": 0.2102,
      "step": 22037
    },
    {
      "epoch": 0.67,
      "grad_norm": 0.3012196375140651,
      "learning_rate": 5.049329705917393e-06,
      "loss": 0.2726,
      "step": 22038
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0878751550948231,
      "learning_rate": 5.0484679275353336e-06,
      "loss": 0.3875,
      "step": 22039
    },
    {
      "epoch": 0.67,
      "grad_norm": 1.0255484613655124,
      "learning_rate": 5.047606197868078e-06,
      "loss": 0.4435,
      "step": 22040
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.30379078644598645,
      "learning_rate": 5.046744516924096e-06,
      "loss": 0.1934,
      "step": 22041
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3109175204458503,
      "learning_rate": 5.045882884711879e-06,
      "loss": 0.1869,
      "step": 22042
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.48710215616885677,
      "learning_rate": 5.045021301239892e-06,
      "loss": 0.2606,
      "step": 22043
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44996501997585836,
      "learning_rate": 5.044159766516616e-06,
      "loss": 0.3062,
      "step": 22044
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6284423613363788,
      "learning_rate": 5.043298280550532e-06,
      "loss": 0.2315,
      "step": 22045
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3613856500052489,
      "learning_rate": 5.042436843350105e-06,
      "loss": 0.2725,
      "step": 22046
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0391303539738794,
      "learning_rate": 5.041575454923816e-06,
      "loss": 0.3936,
      "step": 22047
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.618412762603035,
      "learning_rate": 5.0407141152801396e-06,
      "loss": 0.6094,
      "step": 22048
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27522179538881475,
      "learning_rate": 5.039852824427553e-06,
      "loss": 0.2047,
      "step": 22049
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2785268330487597,
      "learning_rate": 5.0389915823745195e-06,
      "loss": 0.1841,
      "step": 22050
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.28502532114411155,
      "learning_rate": 5.038130389129525e-06,
      "loss": 0.1671,
      "step": 22051
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.676840456726596,
      "learning_rate": 5.037269244701035e-06,
      "loss": 0.2932,
      "step": 22052
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6998551728541195,
      "learning_rate": 5.036408149097521e-06,
      "loss": 0.3973,
      "step": 22053
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2928644126332998,
      "learning_rate": 5.035547102327462e-06,
      "loss": 0.1516,
      "step": 22054
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5423311098509188,
      "learning_rate": 5.03468610439932e-06,
      "loss": 0.3727,
      "step": 22055
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27641120383466455,
      "learning_rate": 5.03382515532157e-06,
      "loss": 0.2091,
      "step": 22056
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1721950679513042,
      "learning_rate": 5.032964255102681e-06,
      "loss": 0.7647,
      "step": 22057
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7415436233870834,
      "learning_rate": 5.0321034037511296e-06,
      "loss": 0.137,
      "step": 22058
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34257668161501786,
      "learning_rate": 5.031242601275375e-06,
      "loss": 0.2415,
      "step": 22059
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.33480001830542694,
      "learning_rate": 5.03038184768389e-06,
      "loss": 0.1526,
      "step": 22060
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.742702154531495,
      "learning_rate": 5.029521142985149e-06,
      "loss": 0.4123,
      "step": 22061
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21527906407435055,
      "learning_rate": 5.028660487187607e-06,
      "loss": 0.1737,
      "step": 22062
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.583740279499883,
      "learning_rate": 5.027799880299744e-06,
      "loss": 0.113,
      "step": 22063
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4044227671170905,
      "learning_rate": 5.02693932233002e-06,
      "loss": 0.2987,
      "step": 22064
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9282960448164547,
      "learning_rate": 5.026078813286903e-06,
      "loss": 0.355,
      "step": 22065
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5137493311738577,
      "learning_rate": 5.025218353178863e-06,
      "loss": 0.6847,
      "step": 22066
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2838414746174046,
      "learning_rate": 5.024357942014353e-06,
      "loss": 0.2149,
      "step": 22067
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4670898646393466,
      "learning_rate": 5.023497579801856e-06,
      "loss": 0.301,
      "step": 22068
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5104819037046953,
      "learning_rate": 5.022637266549824e-06,
      "loss": 0.2419,
      "step": 22069
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7929507353727058,
      "learning_rate": 5.021777002266722e-06,
      "loss": 0.4136,
      "step": 22070
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1496601638011627,
      "learning_rate": 5.020916786961021e-06,
      "loss": 0.0674,
      "step": 22071
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8796854915173676,
      "learning_rate": 5.020056620641176e-06,
      "loss": 0.3762,
      "step": 22072
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.29712464888612955,
      "learning_rate": 5.019196503315651e-06,
      "loss": 0.2064,
      "step": 22073
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3233512297625316,
      "learning_rate": 5.0183364349929095e-06,
      "loss": 0.2927,
      "step": 22074
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4014117476930885,
      "learning_rate": 5.01747641568142e-06,
      "loss": 0.4607,
      "step": 22075
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9478130049627913,
      "learning_rate": 5.016616445389631e-06,
      "loss": 0.2643,
      "step": 22076
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40527931542708934,
      "learning_rate": 5.015756524126009e-06,
      "loss": 0.272,
      "step": 22077
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5241563459624242,
      "learning_rate": 5.014896651899019e-06,
      "loss": 0.265,
      "step": 22078
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.8145692023357847,
      "learning_rate": 5.014036828717107e-06,
      "loss": 0.7585,
      "step": 22079
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.1849017940026705,
      "learning_rate": 5.01317705458875e-06,
      "loss": 0.159,
      "step": 22080
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9237035291980774,
      "learning_rate": 5.012317329522394e-06,
      "loss": 0.3927,
      "step": 22081
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3073861123310662,
      "learning_rate": 5.011457653526501e-06,
      "loss": 0.1973,
      "step": 22082
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.776212580977404,
      "learning_rate": 5.010598026609534e-06,
      "loss": 0.9225,
      "step": 22083
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.996584605947062,
      "learning_rate": 5.0097384487799405e-06,
      "loss": 0.4337,
      "step": 22084
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3304448646427779,
      "learning_rate": 5.008878920046183e-06,
      "loss": 0.2773,
      "step": 22085
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2814091288535947,
      "learning_rate": 5.008019440416716e-06,
      "loss": 0.1748,
      "step": 22086
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0161949051192953,
      "learning_rate": 5.007160009900001e-06,
      "loss": 0.4679,
      "step": 22087
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7108785916985934,
      "learning_rate": 5.00630062850448e-06,
      "loss": 0.2546,
      "step": 22088
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.20414037820872874,
      "learning_rate": 5.005441296238625e-06,
      "loss": 0.082,
      "step": 22089
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6093813672499211,
      "learning_rate": 5.004582013110878e-06,
      "loss": 0.3291,
      "step": 22090
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31946169233164856,
      "learning_rate": 5.0037227791296975e-06,
      "loss": 0.1973,
      "step": 22091
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.33816725312264145,
      "learning_rate": 5.002863594303538e-06,
      "loss": 0.2781,
      "step": 22092
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2882345065170213,
      "learning_rate": 5.002004458640845e-06,
      "loss": 0.3512,
      "step": 22093
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9699589886714591,
      "learning_rate": 5.001145372150085e-06,
      "loss": 0.4193,
      "step": 22094
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2306765285428388,
      "learning_rate": 5.000286334839696e-06,
      "loss": 0.0706,
      "step": 22095
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35945785398583213,
      "learning_rate": 4.9994273467181355e-06,
      "loss": 0.2722,
      "step": 22096
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3042153597089538,
      "learning_rate": 4.998568407793859e-06,
      "loss": 0.2188,
      "step": 22097
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.9914834980723581,
      "learning_rate": 4.997709518075308e-06,
      "loss": 0.663,
      "step": 22098
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.14662576387248824,
      "learning_rate": 4.996850677570936e-06,
      "loss": 0.0691,
      "step": 22099
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35112512123468215,
      "learning_rate": 4.995991886289193e-06,
      "loss": 0.2676,
      "step": 22100
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9628498319230193,
      "learning_rate": 4.995133144238533e-06,
      "loss": 0.6017,
      "step": 22101
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1279843734730999,
      "learning_rate": 4.994274451427396e-06,
      "loss": 0.2743,
      "step": 22102
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3476541790155963,
      "learning_rate": 4.993415807864235e-06,
      "loss": 0.3084,
      "step": 22103
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36086661996294106,
      "learning_rate": 4.992557213557499e-06,
      "loss": 0.1983,
      "step": 22104
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7302226915664604,
      "learning_rate": 4.991698668515625e-06,
      "loss": 0.2991,
      "step": 22105
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7239424958443084,
      "learning_rate": 4.9908401727470765e-06,
      "loss": 0.1668,
      "step": 22106
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40886234924676695,
      "learning_rate": 4.989981726260287e-06,
      "loss": 0.2069,
      "step": 22107
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.24976133672066173,
      "learning_rate": 4.989123329063704e-06,
      "loss": 0.1533,
      "step": 22108
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3558615405758393,
      "learning_rate": 4.988264981165779e-06,
      "loss": 0.26,
      "step": 22109
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3939112200367487,
      "learning_rate": 4.987406682574949e-06,
      "loss": 0.2221,
      "step": 22110
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1778130239965867,
      "learning_rate": 4.986548433299662e-06,
      "loss": 0.5018,
      "step": 22111
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9329216686336032,
      "learning_rate": 4.985690233348359e-06,
      "loss": 0.2965,
      "step": 22112
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.602574216389057,
      "learning_rate": 4.98483208272949e-06,
      "loss": 0.2912,
      "step": 22113
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3484166221093689,
      "learning_rate": 4.983973981451491e-06,
      "loss": 0.211,
      "step": 22114
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3200495946165473,
      "learning_rate": 4.983115929522805e-06,
      "loss": 0.2292,
      "step": 22115
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7209649186856377,
      "learning_rate": 4.982257926951879e-06,
      "loss": 0.8651,
      "step": 22116
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.16435568438358397,
      "learning_rate": 4.981399973747143e-06,
      "loss": 0.0868,
      "step": 22117
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.520114698659065,
      "learning_rate": 4.980542069917054e-06,
      "loss": 0.3177,
      "step": 22118
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44641119417556896,
      "learning_rate": 4.9796842154700355e-06,
      "loss": 0.2633,
      "step": 22119
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2374554611134185,
      "learning_rate": 4.9788264104145425e-06,
      "loss": 0.3948,
      "step": 22120
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.37052018573060086,
      "learning_rate": 4.977968654759005e-06,
      "loss": 0.2386,
      "step": 22121
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.41025888576279795,
      "learning_rate": 4.977110948511863e-06,
      "loss": 0.2299,
      "step": 22122
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3243564323158852,
      "learning_rate": 4.976253291681561e-06,
      "loss": 0.1947,
      "step": 22123
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.11526800772356,
      "learning_rate": 4.975395684276527e-06,
      "loss": 0.4414,
      "step": 22124
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32563947159097945,
      "learning_rate": 4.9745381263052055e-06,
      "loss": 0.0874,
      "step": 22125
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2886392850573982,
      "learning_rate": 4.97368061777603e-06,
      "loss": 0.7932,
      "step": 22126
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2892615456311463,
      "learning_rate": 4.972823158697443e-06,
      "loss": 0.1957,
      "step": 22127
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.33028682876527177,
      "learning_rate": 4.971965749077871e-06,
      "loss": 0.2397,
      "step": 22128
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7801159249630946,
      "learning_rate": 4.971108388925756e-06,
      "loss": 0.398,
      "step": 22129
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3546124336794177,
      "learning_rate": 4.970251078249535e-06,
      "loss": 0.1622,
      "step": 22130
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6747878312326662,
      "learning_rate": 4.9693938170576306e-06,
      "loss": 0.3777,
      "step": 22131
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3265319967826668,
      "learning_rate": 4.968536605358493e-06,
      "loss": 0.1958,
      "step": 22132
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5106077569291904,
      "learning_rate": 4.967679443160543e-06,
      "loss": 0.3342,
      "step": 22133
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4216604245198054,
      "learning_rate": 4.96682233047222e-06,
      "loss": 0.2158,
      "step": 22134
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2925641597984285,
      "learning_rate": 4.965965267301959e-06,
      "loss": 0.6199,
      "step": 22135
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25463062996666036,
      "learning_rate": 4.965108253658183e-06,
      "loss": 0.177,
      "step": 22136
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8800460467604345,
      "learning_rate": 4.964251289549329e-06,
      "loss": 0.4983,
      "step": 22137
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5704943152794234,
      "learning_rate": 4.963394374983829e-06,
      "loss": 0.2979,
      "step": 22138
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32176765805736174,
      "learning_rate": 4.962537509970114e-06,
      "loss": 0.2902,
      "step": 22139
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.14169137550308955,
      "learning_rate": 4.96168069451661e-06,
      "loss": 0.0715,
      "step": 22140
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3289728091771623,
      "learning_rate": 4.960823928631751e-06,
      "loss": 0.195,
      "step": 22141
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4327798042124693,
      "learning_rate": 4.959967212323961e-06,
      "loss": 0.6421,
      "step": 22142
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0540078696680841,
      "learning_rate": 4.959110545601674e-06,
      "loss": 0.2481,
      "step": 22143
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3108039089416349,
      "learning_rate": 4.958253928473319e-06,
      "loss": 0.258,
      "step": 22144
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2340758110113896,
      "learning_rate": 4.957397360947318e-06,
      "loss": 0.0655,
      "step": 22145
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3556016531538141,
      "learning_rate": 4.956540843032101e-06,
      "loss": 0.297,
      "step": 22146
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.68635996398085,
      "learning_rate": 4.955684374736097e-06,
      "loss": 0.3106,
      "step": 22147
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3999384069860834,
      "learning_rate": 4.954827956067723e-06,
      "loss": 0.1943,
      "step": 22148
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2162879266327421,
      "learning_rate": 4.95397158703542e-06,
      "loss": 0.0751,
      "step": 22149
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3487047557589296,
      "learning_rate": 4.9531152676476e-06,
      "loss": 0.2776,
      "step": 22150
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3054655242734205,
      "learning_rate": 4.9522589979126935e-06,
      "loss": 0.2254,
      "step": 22151
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3597593265064538,
      "learning_rate": 4.951402777839124e-06,
      "loss": 0.6021,
      "step": 22152
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2533042490569524,
      "learning_rate": 4.950546607435318e-06,
      "loss": 0.31,
      "step": 22153
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6520954063046217,
      "learning_rate": 4.949690486709693e-06,
      "loss": 0.2962,
      "step": 22154
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35220241293475574,
      "learning_rate": 4.948834415670675e-06,
      "loss": 0.237,
      "step": 22155
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43187115461106995,
      "learning_rate": 4.947978394326689e-06,
      "loss": 0.2776,
      "step": 22156
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.528163285505055,
      "learning_rate": 4.947122422686146e-06,
      "loss": 0.3227,
      "step": 22157
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21083584433100472,
      "learning_rate": 4.946266500757484e-06,
      "loss": 0.0772,
      "step": 22158
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3256170187120666,
      "learning_rate": 4.945410628549111e-06,
      "loss": 0.2449,
      "step": 22159
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7647193151960694,
      "learning_rate": 4.944554806069451e-06,
      "loss": 0.1813,
      "step": 22160
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4820826677976255,
      "learning_rate": 4.9436990333269284e-06,
      "loss": 0.6287,
      "step": 22161
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3423015120664466,
      "learning_rate": 4.942843310329954e-06,
      "loss": 0.2479,
      "step": 22162
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39782815873467975,
      "learning_rate": 4.9419876370869515e-06,
      "loss": 0.2323,
      "step": 22163
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.44141707483831616,
      "learning_rate": 4.941132013606339e-06,
      "loss": 0.2323,
      "step": 22164
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8582170902764986,
      "learning_rate": 4.940276439896538e-06,
      "loss": 0.5412,
      "step": 22165
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9884497158894623,
      "learning_rate": 4.939420915965959e-06,
      "loss": 0.4321,
      "step": 22166
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.28306478449387623,
      "learning_rate": 4.938565441823022e-06,
      "loss": 0.1463,
      "step": 22167
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4001167896653588,
      "learning_rate": 4.937710017476144e-06,
      "loss": 0.1803,
      "step": 22168
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2578249940119372,
      "learning_rate": 4.936854642933739e-06,
      "loss": 0.2276,
      "step": 22169
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2068043620286477,
      "learning_rate": 4.93599931820423e-06,
      "loss": 0.3863,
      "step": 22170
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6801534154711302,
      "learning_rate": 4.93514404329602e-06,
      "loss": 0.2974,
      "step": 22171
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9424383355171576,
      "learning_rate": 4.934288818217531e-06,
      "loss": 0.442,
      "step": 22172
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3102984919437721,
      "learning_rate": 4.933433642977179e-06,
      "loss": 0.2055,
      "step": 22173
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5008517223961969,
      "learning_rate": 4.932578517583365e-06,
      "loss": 0.375,
      "step": 22174
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4471269750011812,
      "learning_rate": 4.931723442044521e-06,
      "loss": 0.2569,
      "step": 22175
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5027610320123822,
      "learning_rate": 4.930868416369043e-06,
      "loss": 0.2411,
      "step": 22176
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2656637919034583,
      "learning_rate": 4.930013440565351e-06,
      "loss": 0.1722,
      "step": 22177
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5420549126429538,
      "learning_rate": 4.929158514641859e-06,
      "loss": 0.2113,
      "step": 22178
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6348324899411848,
      "learning_rate": 4.928303638606969e-06,
      "loss": 0.2839,
      "step": 22179
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4822824239676892,
      "learning_rate": 4.927448812469098e-06,
      "loss": 0.3261,
      "step": 22180
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3429427106861638,
      "learning_rate": 4.926594036236654e-06,
      "loss": 0.2198,
      "step": 22181
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3135118740154803,
      "learning_rate": 4.925739309918052e-06,
      "loss": 0.2144,
      "step": 22182
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.488308638507155,
      "learning_rate": 4.9248846335216915e-06,
      "loss": 0.833,
      "step": 22183
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.144711275560465,
      "learning_rate": 4.924030007055985e-06,
      "loss": 0.3208,
      "step": 22184
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.26670907054325005,
      "learning_rate": 4.923175430529347e-06,
      "loss": 0.1367,
      "step": 22185
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27207844478361015,
      "learning_rate": 4.9223209039501715e-06,
      "loss": 0.1753,
      "step": 22186
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32146714268364024,
      "learning_rate": 4.921466427326881e-06,
      "loss": 0.2752,
      "step": 22187
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6933059056968836,
      "learning_rate": 4.920612000667871e-06,
      "loss": 0.2788,
      "step": 22188
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7109810722800654,
      "learning_rate": 4.919757623981553e-06,
      "loss": 0.4446,
      "step": 22189
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4683268217696703,
      "learning_rate": 4.918903297276329e-06,
      "loss": 0.1375,
      "step": 22190
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39125228669551354,
      "learning_rate": 4.918049020560611e-06,
      "loss": 0.297,
      "step": 22191
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4689825467622395,
      "learning_rate": 4.917194793842795e-06,
      "loss": 0.3029,
      "step": 22192
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.45995179692279586,
      "learning_rate": 4.9163406171312885e-06,
      "loss": 0.3525,
      "step": 22193
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.70689852247022,
      "learning_rate": 4.915486490434497e-06,
      "loss": 0.1632,
      "step": 22194
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3724598630322518,
      "learning_rate": 4.914632413760823e-06,
      "loss": 0.174,
      "step": 22195
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23470290477385936,
      "learning_rate": 4.913778387118671e-06,
      "loss": 0.1604,
      "step": 22196
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.010508527300828,
      "learning_rate": 4.912924410516436e-06,
      "loss": 0.2983,
      "step": 22197
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3354580665309763,
      "learning_rate": 4.9120704839625264e-06,
      "loss": 0.2884,
      "step": 22198
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3314973057198477,
      "learning_rate": 4.911216607465344e-06,
      "loss": 0.0681,
      "step": 22199
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.35315584575283054,
      "learning_rate": 4.910362781033281e-06,
      "loss": 0.2902,
      "step": 22200
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8132791426487932,
      "learning_rate": 4.909509004674751e-06,
      "loss": 0.5436,
      "step": 22201
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4215824247996252,
      "learning_rate": 4.908655278398141e-06,
      "loss": 0.5484,
      "step": 22202
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2818387350501594,
      "learning_rate": 4.907801602211857e-06,
      "loss": 0.0771,
      "step": 22203
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.36972317047997133,
      "learning_rate": 4.9069479761243e-06,
      "loss": 0.2631,
      "step": 22204
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32382317312738956,
      "learning_rate": 4.9060944001438604e-06,
      "loss": 0.2298,
      "step": 22205
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8787460248048122,
      "learning_rate": 4.9052408742789405e-06,
      "loss": 0.3949,
      "step": 22206
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5743809156908153,
      "learning_rate": 4.904387398537937e-06,
      "loss": 0.0901,
      "step": 22207
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.25585874951085785,
      "learning_rate": 4.903533972929252e-06,
      "loss": 0.0709,
      "step": 22208
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3569611838206654,
      "learning_rate": 4.902680597461272e-06,
      "loss": 0.258,
      "step": 22209
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3778642735332076,
      "learning_rate": 4.9018272721423965e-06,
      "loss": 0.2538,
      "step": 22210
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4888846554478206,
      "learning_rate": 4.900973996981026e-06,
      "loss": 0.3287,
      "step": 22211
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7782248031519546,
      "learning_rate": 4.900120771985545e-06,
      "loss": 0.2251,
      "step": 22212
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4202189265498609,
      "learning_rate": 4.899267597164361e-06,
      "loss": 0.2951,
      "step": 22213
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5809629045898685,
      "learning_rate": 4.898414472525856e-06,
      "loss": 0.2543,
      "step": 22214
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9725744186360012,
      "learning_rate": 4.897561398078429e-06,
      "loss": 0.4853,
      "step": 22215
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.16610284298623815,
      "learning_rate": 4.896708373830476e-06,
      "loss": 0.1446,
      "step": 22216
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8931084573180208,
      "learning_rate": 4.8958553997903805e-06,
      "loss": 0.423,
      "step": 22217
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2941587034069177,
      "learning_rate": 4.8950024759665395e-06,
      "loss": 0.1899,
      "step": 22218
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5788801315941146,
      "learning_rate": 4.8941496023673425e-06,
      "loss": 0.833,
      "step": 22219
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.327483019419177,
      "learning_rate": 4.893296779001185e-06,
      "loss": 0.368,
      "step": 22220
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3218202315705776,
      "learning_rate": 4.8924440058764515e-06,
      "loss": 0.2281,
      "step": 22221
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9599439122681075,
      "learning_rate": 4.891591283001539e-06,
      "loss": 0.3793,
      "step": 22222
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32077827179179125,
      "learning_rate": 4.890738610384828e-06,
      "loss": 0.2068,
      "step": 22223
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8894232942668826,
      "learning_rate": 4.889885988034711e-06,
      "loss": 0.3863,
      "step": 22224
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3010101995496895,
      "learning_rate": 4.889033415959583e-06,
      "loss": 0.0918,
      "step": 22225
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.922939577742292,
      "learning_rate": 4.888180894167818e-06,
      "loss": 0.732,
      "step": 22226
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2671381706564666,
      "learning_rate": 4.887328422667818e-06,
      "loss": 0.1776,
      "step": 22227
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4593177851996713,
      "learning_rate": 4.8864760014679605e-06,
      "loss": 0.347,
      "step": 22228
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4382766890184893,
      "learning_rate": 4.885623630576634e-06,
      "loss": 0.2344,
      "step": 22229
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0229236372830355,
      "learning_rate": 4.884771310002229e-06,
      "loss": 0.4944,
      "step": 22230
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4811890931566887,
      "learning_rate": 4.8839190397531225e-06,
      "loss": 0.1984,
      "step": 22231
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3691435252437793,
      "learning_rate": 4.883066819837705e-06,
      "loss": 0.2628,
      "step": 22232
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3554817711107611,
      "learning_rate": 4.882214650264359e-06,
      "loss": 0.1896,
      "step": 22233
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27145358874469205,
      "learning_rate": 4.881362531041473e-06,
      "loss": 0.1554,
      "step": 22234
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.48156600673407113,
      "learning_rate": 4.880510462177424e-06,
      "loss": 0.2941,
      "step": 22235
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2786939699257859,
      "learning_rate": 4.879658443680596e-06,
      "loss": 0.1767,
      "step": 22236
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4855473157132595,
      "learning_rate": 4.878806475559378e-06,
      "loss": 0.8588,
      "step": 22237
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0428261500526808,
      "learning_rate": 4.87795455782214e-06,
      "loss": 0.2971,
      "step": 22238
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4261377000184869,
      "learning_rate": 4.8771026904772765e-06,
      "loss": 0.3121,
      "step": 22239
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2714322506294729,
      "learning_rate": 4.876250873533159e-06,
      "loss": 0.1941,
      "step": 22240
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4526176644918193,
      "learning_rate": 4.875399106998172e-06,
      "loss": 0.3055,
      "step": 22241
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3528882664254553,
      "learning_rate": 4.874547390880698e-06,
      "loss": 0.1825,
      "step": 22242
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4832294024303884,
      "learning_rate": 4.8736957251891095e-06,
      "loss": 0.2616,
      "step": 22243
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.21491319300772402,
      "learning_rate": 4.872844109931789e-06,
      "loss": 0.1164,
      "step": 22244
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3540220466075035,
      "learning_rate": 4.8719925451171155e-06,
      "loss": 0.2597,
      "step": 22245
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39311491122429487,
      "learning_rate": 4.871141030753468e-06,
      "loss": 0.2391,
      "step": 22246
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4197272670700935,
      "learning_rate": 4.8702895668492245e-06,
      "loss": 0.2603,
      "step": 22247
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7749903784721472,
      "learning_rate": 4.869438153412758e-06,
      "loss": 0.3729,
      "step": 22248
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6248426543211791,
      "learning_rate": 4.868586790452445e-06,
      "loss": 0.1007,
      "step": 22249
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.33775358482306994,
      "learning_rate": 4.867735477976665e-06,
      "loss": 0.2508,
      "step": 22250
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2673127151249715,
      "learning_rate": 4.866884215993796e-06,
      "loss": 0.2152,
      "step": 22251
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.24145702684336853,
      "learning_rate": 4.8660330045122e-06,
      "loss": 0.1998,
      "step": 22252
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2409451707446217,
      "learning_rate": 4.8651818435402706e-06,
      "loss": 0.0995,
      "step": 22253
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5599070256907623,
      "learning_rate": 4.864330733086371e-06,
      "loss": 0.3124,
      "step": 22254
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.344315548207509,
      "learning_rate": 4.863479673158867e-06,
      "loss": 0.2337,
      "step": 22255
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0825354733405008,
      "learning_rate": 4.862628663766148e-06,
      "loss": 0.4146,
      "step": 22256
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42236220808775593,
      "learning_rate": 4.8617777049165746e-06,
      "loss": 0.2488,
      "step": 22257
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.42387082978559354,
      "learning_rate": 4.8609267966185235e-06,
      "loss": 0.2804,
      "step": 22258
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2909884270707949,
      "learning_rate": 4.860075938880365e-06,
      "loss": 0.1951,
      "step": 22259
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2391415274836344,
      "learning_rate": 4.859225131710476e-06,
      "loss": 0.1847,
      "step": 22260
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.944488780063132,
      "learning_rate": 4.8583743751172165e-06,
      "loss": 0.8057,
      "step": 22261
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.23054307684453723,
      "learning_rate": 4.857523669108962e-06,
      "loss": 0.1495,
      "step": 22262
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40070016221410737,
      "learning_rate": 4.856673013694086e-06,
      "loss": 0.2439,
      "step": 22263
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.30348972170971983,
      "learning_rate": 4.855822408880945e-06,
      "loss": 0.2425,
      "step": 22264
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9857833270345866,
      "learning_rate": 4.8549718546779235e-06,
      "loss": 0.4625,
      "step": 22265
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.46859601799484246,
      "learning_rate": 4.854121351093377e-06,
      "loss": 0.1984,
      "step": 22266
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8063176109408825,
      "learning_rate": 4.853270898135679e-06,
      "loss": 0.3257,
      "step": 22267
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3209330644948115,
      "learning_rate": 4.852420495813199e-06,
      "loss": 0.1942,
      "step": 22268
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.7012655153388607,
      "learning_rate": 4.851570144134297e-06,
      "loss": 0.6881,
      "step": 22269
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3171153964064211,
      "learning_rate": 4.850719843107341e-06,
      "loss": 0.2416,
      "step": 22270
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4481834134808441,
      "learning_rate": 4.849869592740698e-06,
      "loss": 0.2826,
      "step": 22271
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.30009659432194624,
      "learning_rate": 4.849019393042732e-06,
      "loss": 0.1462,
      "step": 22272
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4742414998118958,
      "learning_rate": 4.848169244021812e-06,
      "loss": 0.2325,
      "step": 22273
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9590560280474157,
      "learning_rate": 4.8473191456862945e-06,
      "loss": 0.3533,
      "step": 22274
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2164536324491888,
      "learning_rate": 4.846469098044548e-06,
      "loss": 0.1858,
      "step": 22275
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9130644413127248,
      "learning_rate": 4.8456191011049325e-06,
      "loss": 0.4225,
      "step": 22276
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.30361074382115477,
      "learning_rate": 4.844769154875817e-06,
      "loss": 0.1946,
      "step": 22277
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9913959272460294,
      "learning_rate": 4.843919259365554e-06,
      "loss": 0.446,
      "step": 22278
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8963452525809475,
      "learning_rate": 4.843069414582511e-06,
      "loss": 0.3028,
      "step": 22279
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8502143582325313,
      "learning_rate": 4.842219620535052e-06,
      "loss": 0.4782,
      "step": 22280
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.29779010514952764,
      "learning_rate": 4.841369877231524e-06,
      "loss": 0.1752,
      "step": 22281
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2665876951014407,
      "learning_rate": 4.840520184680306e-06,
      "loss": 0.2547,
      "step": 22282
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6524649678751957,
      "learning_rate": 4.839670542889744e-06,
      "loss": 0.3022,
      "step": 22283
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4430379736055309,
      "learning_rate": 4.838820951868201e-06,
      "loss": 0.2578,
      "step": 22284
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.30415129971427224,
      "learning_rate": 4.837971411624039e-06,
      "loss": 0.0861,
      "step": 22285
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27189395215237466,
      "learning_rate": 4.83712192216561e-06,
      "loss": 0.1856,
      "step": 22286
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5287265839265015,
      "learning_rate": 4.836272483501273e-06,
      "loss": 0.314,
      "step": 22287
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3743156461963936,
      "learning_rate": 4.835423095639388e-06,
      "loss": 0.2058,
      "step": 22288
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7920293735313313,
      "learning_rate": 4.834573758588312e-06,
      "loss": 0.4492,
      "step": 22289
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3567921521454048,
      "learning_rate": 4.833724472356392e-06,
      "loss": 0.1859,
      "step": 22290
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4786983229045858,
      "learning_rate": 4.832875236951999e-06,
      "loss": 0.3148,
      "step": 22291
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9947879114731032,
      "learning_rate": 4.832026052383475e-06,
      "loss": 0.4679,
      "step": 22292
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.507249620872654,
      "learning_rate": 4.83117691865918e-06,
      "loss": 0.3594,
      "step": 22293
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.17187965607346886,
      "learning_rate": 4.830327835787471e-06,
      "loss": 0.1252,
      "step": 22294
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34233444256725926,
      "learning_rate": 4.829478803776689e-06,
      "loss": 0.2586,
      "step": 22295
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3474591345665505,
      "learning_rate": 4.828629822635205e-06,
      "loss": 0.1255,
      "step": 22296
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4033754004201617,
      "learning_rate": 4.827780892371357e-06,
      "loss": 0.5418,
      "step": 22297
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.39702203439228234,
      "learning_rate": 4.8269320129935036e-06,
      "loss": 0.2775,
      "step": 22298
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31452945713552277,
      "learning_rate": 4.826083184509999e-06,
      "loss": 0.191,
      "step": 22299
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.38305731340011917,
      "learning_rate": 4.8252344069291855e-06,
      "loss": 0.2791,
      "step": 22300
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6693661234047429,
      "learning_rate": 4.8243856802594204e-06,
      "loss": 0.3308,
      "step": 22301
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43640309803363775,
      "learning_rate": 4.82353700450905e-06,
      "loss": 0.208,
      "step": 22302
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3307589158687878,
      "learning_rate": 4.822688379686431e-06,
      "loss": 0.0993,
      "step": 22303
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.4225409140386362,
      "learning_rate": 4.821839805799904e-06,
      "loss": 0.244,
      "step": 22304
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.31709550541510506,
      "learning_rate": 4.8209912828578195e-06,
      "loss": 0.2113,
      "step": 22305
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5068224130688812,
      "learning_rate": 4.820142810868531e-06,
      "loss": 0.3455,
      "step": 22306
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8057964762972841,
      "learning_rate": 4.819294389840374e-06,
      "loss": 0.2972,
      "step": 22307
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.166588324811617,
      "learning_rate": 4.818446019781712e-06,
      "loss": 0.3335,
      "step": 22308
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.27838353729618326,
      "learning_rate": 4.817597700700879e-06,
      "loss": 0.196,
      "step": 22309
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3009870790156204,
      "learning_rate": 4.816749432606225e-06,
      "loss": 0.7523,
      "step": 22310
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.41978783626743654,
      "learning_rate": 4.815901215506099e-06,
      "loss": 0.2031,
      "step": 22311
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19472187655859138,
      "learning_rate": 4.81505304940884e-06,
      "loss": 0.128,
      "step": 22312
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.40751971615653915,
      "learning_rate": 4.814204934322794e-06,
      "loss": 0.2679,
      "step": 22313
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.46642687224090307,
      "learning_rate": 4.813356870256306e-06,
      "loss": 0.2071,
      "step": 22314
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9963057212849381,
      "learning_rate": 4.812508857217723e-06,
      "loss": 0.4753,
      "step": 22315
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.47288015423719026,
      "learning_rate": 4.811660895215383e-06,
      "loss": 0.2525,
      "step": 22316
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3946561326658815,
      "learning_rate": 4.810812984257629e-06,
      "loss": 0.2891,
      "step": 22317
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3104263097865832,
      "learning_rate": 4.809965124352809e-06,
      "loss": 0.2073,
      "step": 22318
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.6287986149663904,
      "learning_rate": 4.809117315509252e-06,
      "loss": 0.7456,
      "step": 22319
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5372644643883613,
      "learning_rate": 4.808269557735313e-06,
      "loss": 0.2948,
      "step": 22320
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.29058893077702547,
      "learning_rate": 4.8074218510393215e-06,
      "loss": 0.1759,
      "step": 22321
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2794791085090085,
      "learning_rate": 4.806574195429627e-06,
      "loss": 0.168,
      "step": 22322
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4574789502453882,
      "learning_rate": 4.805726590914562e-06,
      "loss": 0.557,
      "step": 22323
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.33343179661629146,
      "learning_rate": 4.804879037502469e-06,
      "loss": 0.2297,
      "step": 22324
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8226802405849669,
      "learning_rate": 4.804031535201688e-06,
      "loss": 0.2763,
      "step": 22325
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6601031398649571,
      "learning_rate": 4.803184084020551e-06,
      "loss": 0.3168,
      "step": 22326
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3526847794348904,
      "learning_rate": 4.802336683967399e-06,
      "loss": 0.2044,
      "step": 22327
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.3272368997894881,
      "learning_rate": 4.801489335050568e-06,
      "loss": 0.6964,
      "step": 22328
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6395131541549348,
      "learning_rate": 4.800642037278401e-06,
      "loss": 0.2456,
      "step": 22329
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.398483979982309,
      "learning_rate": 4.799794790659224e-06,
      "loss": 0.1687,
      "step": 22330
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3381536927656333,
      "learning_rate": 4.798947595201376e-06,
      "loss": 0.0712,
      "step": 22331
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3916393776692805,
      "learning_rate": 4.798100450913198e-06,
      "loss": 0.2861,
      "step": 22332
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2571949530443367,
      "learning_rate": 4.797253357803011e-06,
      "loss": 0.3117,
      "step": 22333
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7773756060703814,
      "learning_rate": 4.796406315879165e-06,
      "loss": 0.3185,
      "step": 22334
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.32588458815639076,
      "learning_rate": 4.7955593251499825e-06,
      "loss": 0.1868,
      "step": 22335
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.34508731926853176,
      "learning_rate": 4.7947123856237995e-06,
      "loss": 0.2584,
      "step": 22336
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8823323898170823,
      "learning_rate": 4.793865497308952e-06,
      "loss": 0.5915,
      "step": 22337
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2392802838879549,
      "learning_rate": 4.7930186602137674e-06,
      "loss": 0.5371,
      "step": 22338
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.19945930798065228,
      "learning_rate": 4.792171874346576e-06,
      "loss": 0.0675,
      "step": 22339
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43263265842884224,
      "learning_rate": 4.791325139715712e-06,
      "loss": 0.1894,
      "step": 22340
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2925894059487825,
      "learning_rate": 4.79047845632951e-06,
      "loss": 0.2716,
      "step": 22341
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.2495513529272415,
      "learning_rate": 4.789631824196291e-06,
      "loss": 0.3033,
      "step": 22342
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.0030916547252178,
      "learning_rate": 4.788785243324388e-06,
      "loss": 0.547,
      "step": 22343
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7482918772126356,
      "learning_rate": 4.787938713722133e-06,
      "loss": 0.2099,
      "step": 22344
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3692996404333443,
      "learning_rate": 4.7870922353978465e-06,
      "loss": 0.2841,
      "step": 22345
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.1473451001432653,
      "learning_rate": 4.786245808359868e-06,
      "loss": 0.472,
      "step": 22346
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3336180270621781,
      "learning_rate": 4.785399432616517e-06,
      "loss": 0.2931,
      "step": 22347
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.371287548622036,
      "learning_rate": 4.78455310817612e-06,
      "loss": 0.1063,
      "step": 22348
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3966191923105117,
      "learning_rate": 4.78370683504701e-06,
      "loss": 0.2757,
      "step": 22349
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.7048838919675864,
      "learning_rate": 4.782860613237501e-06,
      "loss": 0.236,
      "step": 22350
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.520811664518212,
      "learning_rate": 4.782014442755934e-06,
      "loss": 0.4529,
      "step": 22351
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.2536794008079638,
      "learning_rate": 4.781168323610622e-06,
      "loss": 0.1508,
      "step": 22352
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3023884440086867,
      "learning_rate": 4.780322255809893e-06,
      "loss": 0.1972,
      "step": 22353
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5729734299797236,
      "learning_rate": 4.779476239362075e-06,
      "loss": 0.3324,
      "step": 22354
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.45504552051076713,
      "learning_rate": 4.7786302742754845e-06,
      "loss": 0.2684,
      "step": 22355
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.185125554626982,
      "learning_rate": 4.7777843605584465e-06,
      "loss": 0.458,
      "step": 22356
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.43242454583977424,
      "learning_rate": 4.776938498219284e-06,
      "loss": 0.2578,
      "step": 22357
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.9482198544956807,
      "learning_rate": 4.7760926872663226e-06,
      "loss": 0.3484,
      "step": 22358
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.26618010486726673,
      "learning_rate": 4.775246927707874e-06,
      "loss": 0.2024,
      "step": 22359
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.8211013973880834,
      "learning_rate": 4.774401219552273e-06,
      "loss": 0.3986,
      "step": 22360
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.15544706435150543,
      "learning_rate": 4.773555562807828e-06,
      "loss": 0.07,
      "step": 22361
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.4619700132281102,
      "learning_rate": 4.772709957482862e-06,
      "loss": 0.7643,
      "step": 22362
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.3181305369480769,
      "learning_rate": 4.7718644035857e-06,
      "loss": 0.1778,
      "step": 22363
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.5499963806557183,
      "learning_rate": 4.771018901124653e-06,
      "loss": 0.3278,
      "step": 22364
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.29633015225305775,
      "learning_rate": 4.770173450108042e-06,
      "loss": 0.2226,
      "step": 22365
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.6167594194314805,
      "learning_rate": 4.769328050544187e-06,
      "loss": 0.3032,
      "step": 22366
    },
    {
      "epoch": 0.68,
      "grad_norm": 0.519187626107139,
      "learning_rate": 4.768482702441405e-06,
      "loss": 0.3572,
      "step": 22367
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.32799613399441796,
      "learning_rate": 4.767637405808009e-06,
      "loss": 0.2319,
      "step": 22368
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4348867885191756,
      "learning_rate": 4.766792160652318e-06,
      "loss": 0.2138,
      "step": 22369
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25015914178182047,
      "learning_rate": 4.765946966982652e-06,
      "loss": 0.1497,
      "step": 22370
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43517988676961206,
      "learning_rate": 4.765101824807313e-06,
      "loss": 0.3247,
      "step": 22371
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25730849778292225,
      "learning_rate": 4.764256734134634e-06,
      "loss": 0.1731,
      "step": 22372
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5959064355554144,
      "learning_rate": 4.763411694972915e-06,
      "loss": 0.731,
      "step": 22373
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4787268935046047,
      "learning_rate": 4.7625667073304745e-06,
      "loss": 0.1911,
      "step": 22374
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7577866846526314,
      "learning_rate": 4.761721771215631e-06,
      "loss": 0.399,
      "step": 22375
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3242505910638993,
      "learning_rate": 4.760876886636684e-06,
      "loss": 0.1994,
      "step": 22376
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5990412259963561,
      "learning_rate": 4.760032053601962e-06,
      "loss": 0.3602,
      "step": 22377
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3155432025151011,
      "learning_rate": 4.759187272119765e-06,
      "loss": 0.2189,
      "step": 22378
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.15277689023187602,
      "learning_rate": 4.7583425421984075e-06,
      "loss": 0.0845,
      "step": 22379
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7785553167389535,
      "learning_rate": 4.7574978638462056e-06,
      "loss": 0.4378,
      "step": 22380
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3702006433034263,
      "learning_rate": 4.75665323707146e-06,
      "loss": 0.1797,
      "step": 22381
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7818557980527563,
      "learning_rate": 4.755808661882486e-06,
      "loss": 0.3129,
      "step": 22382
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3126126377680618,
      "learning_rate": 4.754964138287592e-06,
      "loss": 0.2455,
      "step": 22383
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6433363015772627,
      "learning_rate": 4.754119666295089e-06,
      "loss": 0.3507,
      "step": 22384
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5592257835894221,
      "learning_rate": 4.753275245913282e-06,
      "loss": 0.0859,
      "step": 22385
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35265443409801617,
      "learning_rate": 4.752430877150477e-06,
      "loss": 0.2496,
      "step": 22386
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3154955017981792,
      "learning_rate": 4.751586560014988e-06,
      "loss": 0.0845,
      "step": 22387
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25073546571241756,
      "learning_rate": 4.750742294515109e-06,
      "loss": 0.2178,
      "step": 22388
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9756221953163904,
      "learning_rate": 4.7498980806591634e-06,
      "loss": 0.4633,
      "step": 22389
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6088060939408143,
      "learning_rate": 4.7490539184554444e-06,
      "loss": 0.321,
      "step": 22390
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3389909444316299,
      "learning_rate": 4.74820980791226e-06,
      "loss": 0.211,
      "step": 22391
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5829763183802096,
      "learning_rate": 4.7473657490379155e-06,
      "loss": 0.2747,
      "step": 22392
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8593703993587839,
      "learning_rate": 4.746521741840719e-06,
      "loss": 0.4316,
      "step": 22393
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26751146014535937,
      "learning_rate": 4.745677786328967e-06,
      "loss": 0.1985,
      "step": 22394
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34577131421245993,
      "learning_rate": 4.744833882510966e-06,
      "loss": 0.2674,
      "step": 22395
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.960362465898429,
      "learning_rate": 4.74399003039502e-06,
      "loss": 0.163,
      "step": 22396
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43585788513250073,
      "learning_rate": 4.7431462299894245e-06,
      "loss": 0.1917,
      "step": 22397
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9457105391499898,
      "learning_rate": 4.7423024813024936e-06,
      "loss": 0.5761,
      "step": 22398
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38739246727776205,
      "learning_rate": 4.741458784342515e-06,
      "loss": 0.2767,
      "step": 22399
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2638049259171708,
      "learning_rate": 4.740615139117797e-06,
      "loss": 0.1286,
      "step": 22400
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3701293873340362,
      "learning_rate": 4.739771545636641e-06,
      "loss": 0.3099,
      "step": 22401
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8029755612746906,
      "learning_rate": 4.738928003907336e-06,
      "loss": 0.2958,
      "step": 22402
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8319839561506049,
      "learning_rate": 4.738084513938196e-06,
      "loss": 0.3232,
      "step": 22403
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.391606443450561,
      "learning_rate": 4.7372410757375075e-06,
      "loss": 0.1766,
      "step": 22404
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5364705419927379,
      "learning_rate": 4.736397689313575e-06,
      "loss": 0.2109,
      "step": 22405
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34768058073428015,
      "learning_rate": 4.735554354674695e-06,
      "loss": 0.2917,
      "step": 22406
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0305151755395265,
      "learning_rate": 4.734711071829161e-06,
      "loss": 0.6077,
      "step": 22407
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9507462631020815,
      "learning_rate": 4.733867840785272e-06,
      "loss": 0.395,
      "step": 22408
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3213148421625751,
      "learning_rate": 4.733024661551324e-06,
      "loss": 0.2118,
      "step": 22409
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7327506200660227,
      "learning_rate": 4.732181534135616e-06,
      "loss": 0.4077,
      "step": 22410
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19681758108462918,
      "learning_rate": 4.7313384585464365e-06,
      "loss": 0.1283,
      "step": 22411
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5047389805008656,
      "learning_rate": 4.730495434792082e-06,
      "loss": 0.3677,
      "step": 22412
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.30100785006509817,
      "learning_rate": 4.729652462880852e-06,
      "loss": 0.1875,
      "step": 22413
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7308601098293535,
      "learning_rate": 4.728809542821025e-06,
      "loss": 0.8179,
      "step": 22414
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0947646961549196,
      "learning_rate": 4.727966674620915e-06,
      "loss": 0.3008,
      "step": 22415
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.841350947219009,
      "learning_rate": 4.7271238582887975e-06,
      "loss": 0.7574,
      "step": 22416
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.43883010218397234,
      "learning_rate": 4.726281093832972e-06,
      "loss": 0.2138,
      "step": 22417
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25988337739535144,
      "learning_rate": 4.725438381261732e-06,
      "loss": 0.2208,
      "step": 22418
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.446318952671195,
      "learning_rate": 4.72459572058336e-06,
      "loss": 0.3194,
      "step": 22419
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.17029549900260138,
      "learning_rate": 4.723753111806152e-06,
      "loss": 0.0678,
      "step": 22420
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1708758185087331,
      "learning_rate": 4.7229105549383965e-06,
      "loss": 0.4163,
      "step": 22421
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2552777669867372,
      "learning_rate": 4.722068049988384e-06,
      "loss": 0.1781,
      "step": 22422
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6051368846002467,
      "learning_rate": 4.721225596964403e-06,
      "loss": 0.7337,
      "step": 22423
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2954308619342662,
      "learning_rate": 4.7203831958747435e-06,
      "loss": 0.2087,
      "step": 22424
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2712556762562415,
      "learning_rate": 4.719540846727689e-06,
      "loss": 0.7707,
      "step": 22425
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4084319323186882,
      "learning_rate": 4.718698549531529e-06,
      "loss": 0.2064,
      "step": 22426
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3750208237259298,
      "learning_rate": 4.717856304294555e-06,
      "loss": 0.272,
      "step": 22427
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5827897211573524,
      "learning_rate": 4.717014111025039e-06,
      "loss": 0.306,
      "step": 22428
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.29020273024366156,
      "learning_rate": 4.716171969731286e-06,
      "loss": 0.1702,
      "step": 22429
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.30274271026826766,
      "learning_rate": 4.715329880421568e-06,
      "loss": 0.2187,
      "step": 22430
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3683152982420057,
      "learning_rate": 4.714487843104172e-06,
      "loss": 0.1809,
      "step": 22431
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.495798039661526,
      "learning_rate": 4.713645857787388e-06,
      "loss": 0.303,
      "step": 22432
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.280861784400498,
      "learning_rate": 4.712803924479492e-06,
      "loss": 0.3262,
      "step": 22433
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5714498966574421,
      "learning_rate": 4.711962043188771e-06,
      "loss": 0.3776,
      "step": 22434
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5441161852675584,
      "learning_rate": 4.7111202139235065e-06,
      "loss": 0.2157,
      "step": 22435
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36166569774752094,
      "learning_rate": 4.710278436691986e-06,
      "loss": 0.2615,
      "step": 22436
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.33100004271996275,
      "learning_rate": 4.7094367115024846e-06,
      "loss": 0.2332,
      "step": 22437
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28298819076062165,
      "learning_rate": 4.708595038363282e-06,
      "loss": 0.1677,
      "step": 22438
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.42166874106867,
      "learning_rate": 4.70775341728267e-06,
      "loss": 0.2196,
      "step": 22439
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6393544427894611,
      "learning_rate": 4.706911848268911e-06,
      "loss": 0.3071,
      "step": 22440
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3441476430627406,
      "learning_rate": 4.7060703313303045e-06,
      "loss": 0.2036,
      "step": 22441
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3424981897154624,
      "learning_rate": 4.705228866475116e-06,
      "loss": 0.2979,
      "step": 22442
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8607897518778432,
      "learning_rate": 4.704387453711627e-06,
      "loss": 0.2959,
      "step": 22443
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5385272564011794,
      "learning_rate": 4.70354609304812e-06,
      "loss": 0.2203,
      "step": 22444
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39113508336120995,
      "learning_rate": 4.702704784492868e-06,
      "loss": 0.2828,
      "step": 22445
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0643543957600103,
      "learning_rate": 4.701863528054147e-06,
      "loss": 0.4496,
      "step": 22446
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4670835752563422,
      "learning_rate": 4.701022323740237e-06,
      "loss": 0.2412,
      "step": 22447
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.22572141722479583,
      "learning_rate": 4.700181171559413e-06,
      "loss": 0.181,
      "step": 22448
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.39427371630617647,
      "learning_rate": 4.699340071519953e-06,
      "loss": 0.2679,
      "step": 22449
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4880146506616783,
      "learning_rate": 4.698499023630128e-06,
      "loss": 0.1905,
      "step": 22450
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.365496533948799,
      "learning_rate": 4.697658027898212e-06,
      "loss": 0.5366,
      "step": 22451
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.015487330330111,
      "learning_rate": 4.696817084332483e-06,
      "loss": 0.3091,
      "step": 22452
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4067704285683972,
      "learning_rate": 4.695976192941214e-06,
      "loss": 0.302,
      "step": 22453
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3241697846674741,
      "learning_rate": 4.695135353732674e-06,
      "loss": 0.1905,
      "step": 22454
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46115021561859154,
      "learning_rate": 4.694294566715137e-06,
      "loss": 0.3488,
      "step": 22455
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1274788862852945,
      "learning_rate": 4.69345383189688e-06,
      "loss": 0.3317,
      "step": 22456
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.13683834732101638,
      "learning_rate": 4.692613149286164e-06,
      "loss": 0.0718,
      "step": 22457
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41863015020073036,
      "learning_rate": 4.691772518891272e-06,
      "loss": 0.2476,
      "step": 22458
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5189133897519022,
      "learning_rate": 4.690931940720466e-06,
      "loss": 0.2111,
      "step": 22459
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3401831241243218,
      "learning_rate": 4.690091414782019e-06,
      "loss": 0.303,
      "step": 22460
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0363948290563236,
      "learning_rate": 4.6892509410841976e-06,
      "loss": 0.2919,
      "step": 22461
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2785468097560968,
      "learning_rate": 4.688410519635279e-06,
      "loss": 0.4463,
      "step": 22462
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.31679049593705766,
      "learning_rate": 4.687570150443521e-06,
      "loss": 0.1853,
      "step": 22463
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3343918365161116,
      "learning_rate": 4.686729833517195e-06,
      "loss": 0.765,
      "step": 22464
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3133430685223632,
      "learning_rate": 4.685889568864573e-06,
      "loss": 0.2512,
      "step": 22465
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.49399975899312376,
      "learning_rate": 4.685049356493912e-06,
      "loss": 0.2567,
      "step": 22466
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2611566035273352,
      "learning_rate": 4.684209196413491e-06,
      "loss": 0.0823,
      "step": 22467
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4140982490540512,
      "learning_rate": 4.683369088631565e-06,
      "loss": 0.3002,
      "step": 22468
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0575133359278526,
      "learning_rate": 4.682529033156403e-06,
      "loss": 0.3153,
      "step": 22469
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34704683194165054,
      "learning_rate": 4.6816890299962746e-06,
      "loss": 0.012,
      "step": 22470
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35228398621039636,
      "learning_rate": 4.680849079159435e-06,
      "loss": 0.291,
      "step": 22471
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25990273129331715,
      "learning_rate": 4.680009180654153e-06,
      "loss": 0.1784,
      "step": 22472
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4481178332385425,
      "learning_rate": 4.679169334488691e-06,
      "loss": 0.7682,
      "step": 22473
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.336243066635696,
      "learning_rate": 4.678329540671312e-06,
      "loss": 0.2049,
      "step": 22474
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2680667650435841,
      "learning_rate": 4.6774897992102815e-06,
      "loss": 0.1668,
      "step": 22475
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3375766102074465,
      "learning_rate": 4.676650110113854e-06,
      "loss": 0.1581,
      "step": 22476
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5051319394032849,
      "learning_rate": 4.6758104733902945e-06,
      "loss": 0.3145,
      "step": 22477
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.30452912520025754,
      "learning_rate": 4.674970889047863e-06,
      "loss": 0.2173,
      "step": 22478
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8256800340312601,
      "learning_rate": 4.6741313570948245e-06,
      "loss": 0.3917,
      "step": 22479
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6626578760513401,
      "learning_rate": 4.673291877539429e-06,
      "loss": 0.2356,
      "step": 22480
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7126302952711363,
      "learning_rate": 4.6724524503899425e-06,
      "loss": 0.2472,
      "step": 22481
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.918747035719671,
      "learning_rate": 4.671613075654626e-06,
      "loss": 0.44,
      "step": 22482
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2841574886713235,
      "learning_rate": 4.670773753341724e-06,
      "loss": 0.2193,
      "step": 22483
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26214707957865285,
      "learning_rate": 4.669934483459512e-06,
      "loss": 0.1567,
      "step": 22484
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4789051588739805,
      "learning_rate": 4.669095266016234e-06,
      "loss": 0.1761,
      "step": 22485
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3725605839574842,
      "learning_rate": 4.668256101020151e-06,
      "loss": 0.2812,
      "step": 22486
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7390400103415832,
      "learning_rate": 4.667416988479522e-06,
      "loss": 0.2816,
      "step": 22487
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0677535858184233,
      "learning_rate": 4.666577928402597e-06,
      "loss": 0.5184,
      "step": 22488
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2927561015893257,
      "learning_rate": 4.665738920797632e-06,
      "loss": 0.2123,
      "step": 22489
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.387221434549358,
      "learning_rate": 4.664899965672884e-06,
      "loss": 0.2477,
      "step": 22490
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.44858209123518217,
      "learning_rate": 4.664061063036608e-06,
      "loss": 0.2649,
      "step": 22491
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5589762665774851,
      "learning_rate": 4.663222212897047e-06,
      "loss": 0.4465,
      "step": 22492
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8343201601950933,
      "learning_rate": 4.662383415262471e-06,
      "loss": 0.2635,
      "step": 22493
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5989389599176851,
      "learning_rate": 4.66154467014112e-06,
      "loss": 0.3424,
      "step": 22494
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2977614953723192,
      "learning_rate": 4.660705977541247e-06,
      "loss": 0.2146,
      "step": 22495
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19525732595497308,
      "learning_rate": 4.65986733747111e-06,
      "loss": 0.1603,
      "step": 22496
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.0488591268518404,
      "learning_rate": 4.659028749938952e-06,
      "loss": 0.784,
      "step": 22497
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1296319468974425,
      "learning_rate": 4.658190214953026e-06,
      "loss": 0.4855,
      "step": 22498
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41263918553802864,
      "learning_rate": 4.657351732521584e-06,
      "loss": 0.273,
      "step": 22499
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4206474791766778,
      "learning_rate": 4.656513302652872e-06,
      "loss": 0.1912,
      "step": 22500
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4482692432160655,
      "learning_rate": 4.6556749253551446e-06,
      "loss": 0.5687,
      "step": 22501
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.342220376242174,
      "learning_rate": 4.6548366006366415e-06,
      "loss": 0.241,
      "step": 22502
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7107681632210027,
      "learning_rate": 4.653998328505616e-06,
      "loss": 0.3655,
      "step": 22503
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.33003138218244255,
      "learning_rate": 4.653160108970313e-06,
      "loss": 0.2144,
      "step": 22504
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.787224931577651,
      "learning_rate": 4.6523219420389835e-06,
      "loss": 0.8095,
      "step": 22505
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1411780922243066,
      "learning_rate": 4.651483827719867e-06,
      "loss": 0.084,
      "step": 22506
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3519636534563187,
      "learning_rate": 4.6506457660212124e-06,
      "loss": 0.2853,
      "step": 22507
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28353770366727427,
      "learning_rate": 4.649807756951269e-06,
      "loss": 0.1719,
      "step": 22508
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0196039502258254,
      "learning_rate": 4.64896980051827e-06,
      "loss": 0.4503,
      "step": 22509
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1578351524509591,
      "learning_rate": 4.6481318967304745e-06,
      "loss": 0.5408,
      "step": 22510
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7051477531803381,
      "learning_rate": 4.6472940455961145e-06,
      "loss": 0.2855,
      "step": 22511
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6225888605238002,
      "learning_rate": 4.646456247123436e-06,
      "loss": 0.2248,
      "step": 22512
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3438821268113135,
      "learning_rate": 4.645618501320688e-06,
      "loss": 0.1988,
      "step": 22513
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3387741394842754,
      "learning_rate": 4.644780808196101e-06,
      "loss": 0.2829,
      "step": 22514
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.1947662110101553,
      "learning_rate": 4.6439431677579235e-06,
      "loss": 0.0853,
      "step": 22515
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5561863527657376,
      "learning_rate": 4.643105580014396e-06,
      "loss": 0.8358,
      "step": 22516
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2933233000262826,
      "learning_rate": 4.642268044973762e-06,
      "loss": 0.1115,
      "step": 22517
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3881189325263687,
      "learning_rate": 4.641430562644255e-06,
      "loss": 0.293,
      "step": 22518
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3048539677565219,
      "learning_rate": 4.640593133034116e-06,
      "loss": 0.2495,
      "step": 22519
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9004484291572452,
      "learning_rate": 4.639755756151589e-06,
      "loss": 0.4016,
      "step": 22520
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.739815160380748,
      "learning_rate": 4.6389184320049015e-06,
      "loss": 0.2135,
      "step": 22521
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.349251396200244,
      "learning_rate": 4.638081160602306e-06,
      "loss": 0.2663,
      "step": 22522
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.40179791499819356,
      "learning_rate": 4.637243941952029e-06,
      "loss": 0.1047,
      "step": 22523
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19559094107909278,
      "learning_rate": 4.63640677606231e-06,
      "loss": 0.09,
      "step": 22524
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3490426091961013,
      "learning_rate": 4.63556966294139e-06,
      "loss": 0.3046,
      "step": 22525
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3032152067594418,
      "learning_rate": 4.634732602597494e-06,
      "loss": 0.1676,
      "step": 22526
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5786132772697441,
      "learning_rate": 4.633895595038871e-06,
      "loss": 0.3422,
      "step": 22527
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.0817200645889398,
      "learning_rate": 4.633058640273745e-06,
      "loss": 0.3003,
      "step": 22528
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7616370192394908,
      "learning_rate": 4.632221738310353e-06,
      "loss": 0.3633,
      "step": 22529
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3590913945809708,
      "learning_rate": 4.6313848891569314e-06,
      "loss": 0.2005,
      "step": 22530
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35491688884936345,
      "learning_rate": 4.630548092821715e-06,
      "loss": 0.2678,
      "step": 22531
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4329029277361639,
      "learning_rate": 4.629711349312929e-06,
      "loss": 0.2356,
      "step": 22532
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2892382585031328,
      "learning_rate": 4.62887465863881e-06,
      "loss": 0.1489,
      "step": 22533
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8743438630394131,
      "learning_rate": 4.628038020807594e-06,
      "loss": 0.5927,
      "step": 22534
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5911677470760307,
      "learning_rate": 4.627201435827498e-06,
      "loss": 0.2599,
      "step": 22535
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.32831052215067635,
      "learning_rate": 4.626364903706772e-06,
      "loss": 0.2013,
      "step": 22536
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.42687365747064565,
      "learning_rate": 4.625528424453632e-06,
      "loss": 0.2389,
      "step": 22537
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.016339367456734,
      "learning_rate": 4.624691998076311e-06,
      "loss": 0.4082,
      "step": 22538
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7688023186414825,
      "learning_rate": 4.623855624583043e-06,
      "loss": 0.108,
      "step": 22539
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4657331395085201,
      "learning_rate": 4.623019303982048e-06,
      "loss": 0.2932,
      "step": 22540
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4778683922115133,
      "learning_rate": 4.622183036281559e-06,
      "loss": 0.2002,
      "step": 22541
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5221869796998999,
      "learning_rate": 4.621346821489802e-06,
      "loss": 0.2944,
      "step": 22542
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.292438918561777,
      "learning_rate": 4.620510659615011e-06,
      "loss": 0.228,
      "step": 22543
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.884502035752749,
      "learning_rate": 4.6196745506654005e-06,
      "loss": 0.4888,
      "step": 22544
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.32299553185369617,
      "learning_rate": 4.6188384946492014e-06,
      "loss": 0.2037,
      "step": 22545
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8090829562819358,
      "learning_rate": 4.6180024915746444e-06,
      "loss": 0.4114,
      "step": 22546
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.23305938879989088,
      "learning_rate": 4.617166541449943e-06,
      "loss": 0.0661,
      "step": 22547
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.49477322628507736,
      "learning_rate": 4.616330644283336e-06,
      "loss": 0.3371,
      "step": 22548
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24498735885943757,
      "learning_rate": 4.6154948000830345e-06,
      "loss": 0.1765,
      "step": 22549
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.44876486859632264,
      "learning_rate": 4.614659008857267e-06,
      "loss": 0.1943,
      "step": 22550
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.617772722448667,
      "learning_rate": 4.613823270614261e-06,
      "loss": 0.7359,
      "step": 22551
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8982230809889048,
      "learning_rate": 4.612987585362227e-06,
      "loss": 0.5923,
      "step": 22552
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5849799764253254,
      "learning_rate": 4.6121519531093996e-06,
      "loss": 0.3062,
      "step": 22553
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.371500811831045,
      "learning_rate": 4.61131637386399e-06,
      "loss": 0.2361,
      "step": 22554
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5011098789980831,
      "learning_rate": 4.610480847634224e-06,
      "loss": 0.3269,
      "step": 22555
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2647175718303505,
      "learning_rate": 4.609645374428325e-06,
      "loss": 0.1434,
      "step": 22556
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1519359474815392,
      "learning_rate": 4.608809954254504e-06,
      "loss": 0.5336,
      "step": 22557
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2964676376647323,
      "learning_rate": 4.607974587120984e-06,
      "loss": 0.1829,
      "step": 22558
    },
    {
      "epoch": 0.69,
      "grad_norm": 2.4077671795514184,
      "learning_rate": 4.607139273035985e-06,
      "loss": 0.6989,
      "step": 22559
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.074986277936297,
      "learning_rate": 4.606304012007728e-06,
      "loss": 0.31,
      "step": 22560
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3248682822410856,
      "learning_rate": 4.605468804044421e-06,
      "loss": 0.2737,
      "step": 22561
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.449307275656063,
      "learning_rate": 4.604633649154292e-06,
      "loss": 0.2006,
      "step": 22562
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3384735637657371,
      "learning_rate": 4.603798547345549e-06,
      "loss": 0.1992,
      "step": 22563
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6808236436746198,
      "learning_rate": 4.602963498626412e-06,
      "loss": 0.431,
      "step": 22564
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24875722345310483,
      "learning_rate": 4.602128503005101e-06,
      "loss": 0.0863,
      "step": 22565
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.33884522878813506,
      "learning_rate": 4.60129356048982e-06,
      "loss": 0.282,
      "step": 22566
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2934756111500061,
      "learning_rate": 4.600458671088791e-06,
      "loss": 0.0703,
      "step": 22567
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4216855740505256,
      "learning_rate": 4.599623834810225e-06,
      "loss": 0.2903,
      "step": 22568
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2019886139752047,
      "learning_rate": 4.598789051662341e-06,
      "loss": 0.3218,
      "step": 22569
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.823929346904807,
      "learning_rate": 4.597954321653344e-06,
      "loss": 0.3744,
      "step": 22570
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6043441584505039,
      "learning_rate": 4.59711964479145e-06,
      "loss": 0.2093,
      "step": 22571
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3437008341835308,
      "learning_rate": 4.5962850210848745e-06,
      "loss": 0.2736,
      "step": 22572
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3024865161755041,
      "learning_rate": 4.595450450541817e-06,
      "loss": 0.2501,
      "step": 22573
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2966890635424135,
      "learning_rate": 4.5946159331705045e-06,
      "loss": 0.1602,
      "step": 22574
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9697074645087732,
      "learning_rate": 4.593781468979134e-06,
      "loss": 0.082,
      "step": 22575
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.30281533738779726,
      "learning_rate": 4.592947057975921e-06,
      "loss": 0.152,
      "step": 22576
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5161391797774045,
      "learning_rate": 4.592112700169078e-06,
      "loss": 0.2847,
      "step": 22577
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.052189324694874,
      "learning_rate": 4.5912783955668015e-06,
      "loss": 0.4561,
      "step": 22578
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36696297267383937,
      "learning_rate": 4.590444144177316e-06,
      "loss": 0.2808,
      "step": 22579
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8494560049023618,
      "learning_rate": 4.589609946008817e-06,
      "loss": 0.268,
      "step": 22580
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35204304678597215,
      "learning_rate": 4.588775801069516e-06,
      "loss": 0.265,
      "step": 22581
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1235644709717567,
      "learning_rate": 4.587941709367623e-06,
      "loss": 0.4651,
      "step": 22582
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3352430600134919,
      "learning_rate": 4.587107670911336e-06,
      "loss": 0.1808,
      "step": 22583
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3226976894303475,
      "learning_rate": 4.5862736857088655e-06,
      "loss": 0.2211,
      "step": 22584
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41080260957503856,
      "learning_rate": 4.585439753768415e-06,
      "loss": 0.2445,
      "step": 22585
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4826169364985932,
      "learning_rate": 4.5846058750981935e-06,
      "loss": 0.1941,
      "step": 22586
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5103535478823065,
      "learning_rate": 4.583772049706399e-06,
      "loss": 0.6683,
      "step": 22587
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9532153662700404,
      "learning_rate": 4.5829382776012355e-06,
      "loss": 0.2937,
      "step": 22588
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.32352636172000243,
      "learning_rate": 4.5821045587909115e-06,
      "loss": 0.1493,
      "step": 22589
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5633154239667826,
      "learning_rate": 4.5812708932836195e-06,
      "loss": 0.3704,
      "step": 22590
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.25553831788136294,
      "learning_rate": 4.580437281087574e-06,
      "loss": 0.227,
      "step": 22591
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4982936130950827,
      "learning_rate": 4.579603722210967e-06,
      "loss": 0.2392,
      "step": 22592
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3321828905980994,
      "learning_rate": 4.578770216662001e-06,
      "loss": 0.083,
      "step": 22593
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.038225075546539,
      "learning_rate": 4.577936764448877e-06,
      "loss": 0.3596,
      "step": 22594
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.34351215095911314,
      "learning_rate": 4.5771033655798e-06,
      "loss": 0.2003,
      "step": 22595
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5060858158358308,
      "learning_rate": 4.576270020062961e-06,
      "loss": 0.3012,
      "step": 22596
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.44004679137621416,
      "learning_rate": 4.575436727906562e-06,
      "loss": 0.2628,
      "step": 22597
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6361004650541544,
      "learning_rate": 4.574603489118804e-06,
      "loss": 0.7798,
      "step": 22598
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2665821318352104,
      "learning_rate": 4.573770303707875e-06,
      "loss": 0.1667,
      "step": 22599
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3554801880455711,
      "learning_rate": 4.572937171681987e-06,
      "loss": 0.7259,
      "step": 22600
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.17046029689435171,
      "learning_rate": 4.5721040930493254e-06,
      "loss": 0.0689,
      "step": 22601
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.29241122896104554,
      "learning_rate": 4.57127106781809e-06,
      "loss": 0.2235,
      "step": 22602
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6608611084092211,
      "learning_rate": 4.570438095996479e-06,
      "loss": 0.3061,
      "step": 22603
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.353834141902716,
      "learning_rate": 4.569605177592677e-06,
      "loss": 0.2004,
      "step": 22604
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7113980190177634,
      "learning_rate": 4.568772312614894e-06,
      "loss": 0.4249,
      "step": 22605
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6361102764770693,
      "learning_rate": 4.5679395010713125e-06,
      "loss": 0.3107,
      "step": 22606
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46795320365488086,
      "learning_rate": 4.56710674297013e-06,
      "loss": 0.3109,
      "step": 22607
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.27413611185777315,
      "learning_rate": 4.566274038319541e-06,
      "loss": 0.1721,
      "step": 22608
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46066476643121773,
      "learning_rate": 4.565441387127735e-06,
      "loss": 0.3233,
      "step": 22609
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2292891254804903,
      "learning_rate": 4.564608789402902e-06,
      "loss": 0.1077,
      "step": 22610
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28924654469782385,
      "learning_rate": 4.563776245153237e-06,
      "loss": 0.1768,
      "step": 22611
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4694770167596445,
      "learning_rate": 4.562943754386935e-06,
      "loss": 0.165,
      "step": 22612
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4271940526576306,
      "learning_rate": 4.562111317112177e-06,
      "loss": 0.2895,
      "step": 22613
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4182556372810825,
      "learning_rate": 4.561278933337158e-06,
      "loss": 0.2605,
      "step": 22614
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4448918558785542,
      "learning_rate": 4.560446603070069e-06,
      "loss": 0.2565,
      "step": 22615
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8607443895805854,
      "learning_rate": 4.559614326319088e-06,
      "loss": 0.4919,
      "step": 22616
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3775836055063552,
      "learning_rate": 4.558782103092419e-06,
      "loss": 0.16,
      "step": 22617
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5510119980033511,
      "learning_rate": 4.55794993339824e-06,
      "loss": 0.3559,
      "step": 22618
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.382677355600926,
      "learning_rate": 4.55711781724474e-06,
      "loss": 0.084,
      "step": 22619
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24723466137268882,
      "learning_rate": 4.556285754640108e-06,
      "loss": 0.2131,
      "step": 22620
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6629152167083855,
      "learning_rate": 4.555453745592525e-06,
      "loss": 0.1825,
      "step": 22621
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.38959361234958356,
      "learning_rate": 4.55462179011018e-06,
      "loss": 0.2835,
      "step": 22622
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6728472741586539,
      "learning_rate": 4.553789888201258e-06,
      "loss": 0.2907,
      "step": 22623
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3956303396334668,
      "learning_rate": 4.552958039873946e-06,
      "loss": 0.6881,
      "step": 22624
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9587069470022965,
      "learning_rate": 4.552126245136421e-06,
      "loss": 0.4636,
      "step": 22625
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3745829451577595,
      "learning_rate": 4.551294503996871e-06,
      "loss": 0.2755,
      "step": 22626
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.27669121501817573,
      "learning_rate": 4.550462816463479e-06,
      "loss": 0.2155,
      "step": 22627
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.5655906037892244,
      "learning_rate": 4.5496311825444265e-06,
      "loss": 0.1676,
      "step": 22628
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.26499322856336927,
      "learning_rate": 4.5487996022479e-06,
      "loss": 0.1433,
      "step": 22629
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6446929415718518,
      "learning_rate": 4.547968075582069e-06,
      "loss": 0.2202,
      "step": 22630
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4015143297549844,
      "learning_rate": 4.547136602555129e-06,
      "loss": 0.285,
      "step": 22631
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2890036390790254,
      "learning_rate": 4.54630518317525e-06,
      "loss": 0.2255,
      "step": 22632
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.3419556296286692,
      "learning_rate": 4.545473817450615e-06,
      "loss": 0.633,
      "step": 22633
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8589757635514462,
      "learning_rate": 4.544642505389407e-06,
      "loss": 0.5801,
      "step": 22634
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3745300529763357,
      "learning_rate": 4.543811246999798e-06,
      "loss": 0.2242,
      "step": 22635
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5242969051767206,
      "learning_rate": 4.542980042289968e-06,
      "loss": 0.2203,
      "step": 22636
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4265369499215388,
      "learning_rate": 4.5421488912680975e-06,
      "loss": 0.6776,
      "step": 22637
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.31370598531659677,
      "learning_rate": 4.5413177939423645e-06,
      "loss": 0.2273,
      "step": 22638
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6082623868059774,
      "learning_rate": 4.54048675032094e-06,
      "loss": 0.3501,
      "step": 22639
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3540965573539544,
      "learning_rate": 4.539655760412004e-06,
      "loss": 0.1777,
      "step": 22640
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19377756688595585,
      "learning_rate": 4.538824824223734e-06,
      "loss": 0.1392,
      "step": 22641
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.2749485887721446,
      "learning_rate": 4.537993941764295e-06,
      "loss": 0.7872,
      "step": 22642
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.29054175467548754,
      "learning_rate": 4.537163113041877e-06,
      "loss": 0.2354,
      "step": 22643
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8804236585374876,
      "learning_rate": 4.536332338064642e-06,
      "loss": 0.3705,
      "step": 22644
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2934143658417523,
      "learning_rate": 4.5355016168407666e-06,
      "loss": 0.1912,
      "step": 22645
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.1649147921992606,
      "learning_rate": 4.534670949378428e-06,
      "loss": 0.4302,
      "step": 22646
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5104658987166852,
      "learning_rate": 4.533840335685792e-06,
      "loss": 0.2967,
      "step": 22647
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6009127793865974,
      "learning_rate": 4.533009775771032e-06,
      "loss": 0.7664,
      "step": 22648
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2785589850678294,
      "learning_rate": 4.532179269642322e-06,
      "loss": 0.1788,
      "step": 22649
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3094276137953034,
      "learning_rate": 4.531348817307834e-06,
      "loss": 0.2652,
      "step": 22650
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.19868620338938575,
      "learning_rate": 4.530518418775734e-06,
      "loss": 0.1022,
      "step": 22651
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.6413473856635217,
      "learning_rate": 4.529688074054191e-06,
      "loss": 0.8377,
      "step": 22652
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.37052820417498333,
      "learning_rate": 4.528857783151379e-06,
      "loss": 0.1137,
      "step": 22653
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.328345834903526,
      "learning_rate": 4.528027546075463e-06,
      "loss": 0.1914,
      "step": 22654
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.46889709495198517,
      "learning_rate": 4.527197362834616e-06,
      "loss": 0.2742,
      "step": 22655
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41061344355054635,
      "learning_rate": 4.526367233437e-06,
      "loss": 0.2431,
      "step": 22656
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6748811157123109,
      "learning_rate": 4.525537157890784e-06,
      "loss": 0.3997,
      "step": 22657
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28003875759081287,
      "learning_rate": 4.524707136204137e-06,
      "loss": 0.1741,
      "step": 22658
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2556367310070353,
      "learning_rate": 4.523877168385216e-06,
      "loss": 0.1726,
      "step": 22659
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.939978916008888,
      "learning_rate": 4.523047254442201e-06,
      "loss": 0.4331,
      "step": 22660
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4737136614449283,
      "learning_rate": 4.522217394383246e-06,
      "loss": 0.3739,
      "step": 22661
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2581295947790259,
      "learning_rate": 4.5213875882165185e-06,
      "loss": 0.1826,
      "step": 22662
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5109688588796042,
      "learning_rate": 4.520557835950181e-06,
      "loss": 0.3234,
      "step": 22663
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.8380463050760073,
      "learning_rate": 4.519728137592404e-06,
      "loss": 0.2478,
      "step": 22664
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5987284028036762,
      "learning_rate": 4.518898493151341e-06,
      "loss": 0.3434,
      "step": 22665
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7091395604750056,
      "learning_rate": 4.518068902635158e-06,
      "loss": 0.152,
      "step": 22666
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.35990309620457667,
      "learning_rate": 4.51723936605202e-06,
      "loss": 0.1779,
      "step": 22667
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.36672129822668253,
      "learning_rate": 4.5164098834100775e-06,
      "loss": 0.2827,
      "step": 22668
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.24106040567238712,
      "learning_rate": 4.515580454717507e-06,
      "loss": 0.1727,
      "step": 22669
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.42500038013328933,
      "learning_rate": 4.514751079982457e-06,
      "loss": 0.2412,
      "step": 22670
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3852985054740883,
      "learning_rate": 4.5139217592130906e-06,
      "loss": 0.0612,
      "step": 22671
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.3485591434105517,
      "learning_rate": 4.5130924924175715e-06,
      "loss": 0.2631,
      "step": 22672
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7136423439441187,
      "learning_rate": 4.512263279604049e-06,
      "loss": 0.2665,
      "step": 22673
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5031558839791723,
      "learning_rate": 4.511434120780686e-06,
      "loss": 0.3085,
      "step": 22674
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.5702737137675501,
      "learning_rate": 4.510605015955642e-06,
      "loss": 0.0362,
      "step": 22675
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7145150506676331,
      "learning_rate": 4.50977596513707e-06,
      "loss": 0.3687,
      "step": 22676
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2525162703479756,
      "learning_rate": 4.508946968333132e-06,
      "loss": 0.165,
      "step": 22677
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.406355193322711,
      "learning_rate": 4.508118025551979e-06,
      "loss": 0.7836,
      "step": 22678
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.4320031580183242,
      "learning_rate": 4.507289136801767e-06,
      "loss": 0.2475,
      "step": 22679
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.20775484315086087,
      "learning_rate": 4.5064603020906505e-06,
      "loss": 0.1352,
      "step": 22680
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.41099182561337916,
      "learning_rate": 4.50563152142679e-06,
      "loss": 0.3004,
      "step": 22681
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9753806082197419,
      "learning_rate": 4.5048027948183315e-06,
      "loss": 0.2877,
      "step": 22682
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.4090952296785182,
      "learning_rate": 4.50397412227343e-06,
      "loss": 0.5466,
      "step": 22683
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.288194049765999,
      "learning_rate": 4.503145503800243e-06,
      "loss": 0.1306,
      "step": 22684
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.45174887498511807,
      "learning_rate": 4.5023169394069136e-06,
      "loss": 0.2989,
      "step": 22685
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.28307292645798204,
      "learning_rate": 4.5014884291016044e-06,
      "loss": 0.2348,
      "step": 22686
    },
    {
      "epoch": 0.69,
      "grad_norm": 1.7874361809738788,
      "learning_rate": 4.500659972892459e-06,
      "loss": 0.7967,
      "step": 22687
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2631367823557474,
      "learning_rate": 4.49983157078763e-06,
      "loss": 0.1576,
      "step": 22688
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.7272528594787238,
      "learning_rate": 4.499003222795272e-06,
      "loss": 0.3452,
      "step": 22689
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2964287683338239,
      "learning_rate": 4.498174928923525e-06,
      "loss": 0.1776,
      "step": 22690
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.9539526898699281,
      "learning_rate": 4.4973466891805455e-06,
      "loss": 0.3975,
      "step": 22691
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.2377381405361453,
      "learning_rate": 4.496518503574479e-06,
      "loss": 0.1811,
      "step": 22692
    },
    {
      "epoch": 0.69,
      "grad_norm": 0.6660013961345729,
      "learning_rate": 4.495690372113478e-06,
      "loss": 0.1073,
      "step": 22693
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.48776923092726593,
      "learning_rate": 4.494862294805679e-06,
      "loss": 0.2771,
      "step": 22694
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4530094233871378,
      "learning_rate": 4.494034271659243e-06,
      "loss": 0.2049,
      "step": 22695
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3202628617258845,
      "learning_rate": 4.493206302682306e-06,
      "loss": 0.7651,
      "step": 22696
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3387306929929359,
      "learning_rate": 4.4923783878830175e-06,
      "loss": 0.236,
      "step": 22697
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9344035607857436,
      "learning_rate": 4.491550527269527e-06,
      "loss": 0.3942,
      "step": 22698
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2955620221116774,
      "learning_rate": 4.490722720849971e-06,
      "loss": 0.1855,
      "step": 22699
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8907611711000051,
      "learning_rate": 4.489894968632497e-06,
      "loss": 0.41,
      "step": 22700
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22025918290721688,
      "learning_rate": 4.489067270625249e-06,
      "loss": 0.0856,
      "step": 22701
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8195054357272145,
      "learning_rate": 4.488239626836371e-06,
      "loss": 0.7599,
      "step": 22702
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31473751100125447,
      "learning_rate": 4.487412037274009e-06,
      "loss": 0.176,
      "step": 22703
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30137648879207196,
      "learning_rate": 4.486584501946298e-06,
      "loss": 0.2696,
      "step": 22704
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0479420756808135,
      "learning_rate": 4.485757020861383e-06,
      "loss": 0.542,
      "step": 22705
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1899580013798108,
      "learning_rate": 4.484929594027404e-06,
      "loss": 0.3311,
      "step": 22706
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0917439184651438,
      "learning_rate": 4.4841022214525065e-06,
      "loss": 0.3762,
      "step": 22707
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2610942902697341,
      "learning_rate": 4.483274903144823e-06,
      "loss": 0.175,
      "step": 22708
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2899893097144387,
      "learning_rate": 4.482447639112496e-06,
      "loss": 0.2304,
      "step": 22709
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.249882540937713,
      "learning_rate": 4.4816204293636676e-06,
      "loss": 0.1603,
      "step": 22710
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3162909517594084,
      "learning_rate": 4.480793273906467e-06,
      "loss": 0.4697,
      "step": 22711
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.362120829754605,
      "learning_rate": 4.479966172749044e-06,
      "loss": 0.1692,
      "step": 22712
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5665840465640759,
      "learning_rate": 4.479139125899527e-06,
      "loss": 0.2997,
      "step": 22713
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0375896408668537,
      "learning_rate": 4.478312133366057e-06,
      "loss": 0.4701,
      "step": 22714
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42665312815508355,
      "learning_rate": 4.477485195156772e-06,
      "loss": 0.2858,
      "step": 22715
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3890440477428676,
      "learning_rate": 4.476658311279801e-06,
      "loss": 0.2873,
      "step": 22716
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34708777423377973,
      "learning_rate": 4.475831481743284e-06,
      "loss": 0.2864,
      "step": 22717
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0228433676462974,
      "learning_rate": 4.475004706555352e-06,
      "loss": 0.4828,
      "step": 22718
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2523122514937202,
      "learning_rate": 4.4741779857241475e-06,
      "loss": 0.1428,
      "step": 22719
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42031530864008276,
      "learning_rate": 4.473351319257795e-06,
      "loss": 0.2163,
      "step": 22720
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3220990474857037,
      "learning_rate": 4.472524707164429e-06,
      "loss": 0.1792,
      "step": 22721
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42671667465612223,
      "learning_rate": 4.471698149452188e-06,
      "loss": 0.3189,
      "step": 22722
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9259026194326753,
      "learning_rate": 4.4708716461291925e-06,
      "loss": 0.3269,
      "step": 22723
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7301837122042278,
      "learning_rate": 4.4700451972035885e-06,
      "loss": 0.3918,
      "step": 22724
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42909110139119583,
      "learning_rate": 4.469218802683496e-06,
      "loss": 0.2107,
      "step": 22725
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4421284808277825,
      "learning_rate": 4.468392462577048e-06,
      "loss": 0.2681,
      "step": 22726
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31294643647633585,
      "learning_rate": 4.467566176892379e-06,
      "loss": 0.217,
      "step": 22727
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24731700887703165,
      "learning_rate": 4.466739945637608e-06,
      "loss": 0.2082,
      "step": 22728
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.393678174934705,
      "learning_rate": 4.465913768820876e-06,
      "loss": 0.1926,
      "step": 22729
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8551844993585982,
      "learning_rate": 4.465087646450302e-06,
      "loss": 0.3791,
      "step": 22730
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2991648213399724,
      "learning_rate": 4.464261578534018e-06,
      "loss": 0.1824,
      "step": 22731
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8260983636565794,
      "learning_rate": 4.46343556508015e-06,
      "loss": 0.3623,
      "step": 22732
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.29700747281768297,
      "learning_rate": 4.4626096060968275e-06,
      "loss": 0.2185,
      "step": 22733
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7810445517119805,
      "learning_rate": 4.4617837015921695e-06,
      "loss": 0.3178,
      "step": 22734
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3880810537471549,
      "learning_rate": 4.460957851574307e-06,
      "loss": 0.2595,
      "step": 22735
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4014737881036466,
      "learning_rate": 4.4601320560513676e-06,
      "loss": 0.1988,
      "step": 22736
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49937059926815563,
      "learning_rate": 4.459306315031463e-06,
      "loss": 0.247,
      "step": 22737
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2973937353365242,
      "learning_rate": 4.458480628522735e-06,
      "loss": 0.1668,
      "step": 22738
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42684137911044184,
      "learning_rate": 4.457654996533292e-06,
      "loss": 0.2981,
      "step": 22739
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28567904133188077,
      "learning_rate": 4.4568294190712655e-06,
      "loss": 0.186,
      "step": 22740
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9377457507151644,
      "learning_rate": 4.456003896144777e-06,
      "loss": 0.4883,
      "step": 22741
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8740545800512193,
      "learning_rate": 4.455178427761943e-06,
      "loss": 0.3056,
      "step": 22742
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.355476702295045,
      "learning_rate": 4.454353013930889e-06,
      "loss": 0.779,
      "step": 22743
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2720000244988331,
      "learning_rate": 4.453527654659735e-06,
      "loss": 0.1783,
      "step": 22744
    },
    {
      "epoch": 0.7,
      "grad_norm": 2.3822317858134676,
      "learning_rate": 4.452702349956604e-06,
      "loss": 0.7527,
      "step": 22745
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3085704337696233,
      "learning_rate": 4.4518770998296075e-06,
      "loss": 0.2223,
      "step": 22746
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22615869493715998,
      "learning_rate": 4.451051904286872e-06,
      "loss": 0.0895,
      "step": 22747
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5857796337513067,
      "learning_rate": 4.450226763336516e-06,
      "loss": 0.3361,
      "step": 22748
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32921220139988777,
      "learning_rate": 4.449401676986649e-06,
      "loss": 0.2025,
      "step": 22749
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8653036754878212,
      "learning_rate": 4.448576645245401e-06,
      "loss": 0.3978,
      "step": 22750
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.29416189521057556,
      "learning_rate": 4.447751668120879e-06,
      "loss": 0.2269,
      "step": 22751
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3639703828502945,
      "learning_rate": 4.446926745621203e-06,
      "loss": 0.8205,
      "step": 22752
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3470435127647991,
      "learning_rate": 4.4461018777544925e-06,
      "loss": 0.0741,
      "step": 22753
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.38777942406131827,
      "learning_rate": 4.445277064528852e-06,
      "loss": 0.2787,
      "step": 22754
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.15157698974377237,
      "learning_rate": 4.444452305952411e-06,
      "loss": 0.0726,
      "step": 22755
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49682692336860124,
      "learning_rate": 4.443627602033273e-06,
      "loss": 0.3613,
      "step": 22756
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35016231196689046,
      "learning_rate": 4.442802952779555e-06,
      "loss": 0.1866,
      "step": 22757
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33551496845369433,
      "learning_rate": 4.4419783581993725e-06,
      "loss": 0.2497,
      "step": 22758
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.718250346333942,
      "learning_rate": 4.441153818300833e-06,
      "loss": 0.2932,
      "step": 22759
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8100047677425678,
      "learning_rate": 4.440329333092051e-06,
      "loss": 0.3077,
      "step": 22760
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.41429776376298,
      "learning_rate": 4.439504902581139e-06,
      "loss": 0.8442,
      "step": 22761
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3499502571248081,
      "learning_rate": 4.438680526776211e-06,
      "loss": 0.163,
      "step": 22762
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26682838504163137,
      "learning_rate": 4.4378562056853654e-06,
      "loss": 0.2598,
      "step": 22763
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.295649486578737,
      "learning_rate": 4.437031939316729e-06,
      "loss": 0.1745,
      "step": 22764
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24127370376449556,
      "learning_rate": 4.436207727678399e-06,
      "loss": 0.1329,
      "step": 22765
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4795640661258653,
      "learning_rate": 4.435383570778489e-06,
      "loss": 0.2552,
      "step": 22766
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32041057579436405,
      "learning_rate": 4.434559468625108e-06,
      "loss": 0.2472,
      "step": 22767
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7783855377236736,
      "learning_rate": 4.43373542122636e-06,
      "loss": 0.2685,
      "step": 22768
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3444267434587055,
      "learning_rate": 4.432911428590354e-06,
      "loss": 0.2511,
      "step": 22769
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8689457679964273,
      "learning_rate": 4.432087490725198e-06,
      "loss": 0.5853,
      "step": 22770
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42106572682995524,
      "learning_rate": 4.431263607639e-06,
      "loss": 0.2898,
      "step": 22771
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49167966983840905,
      "learning_rate": 4.4304397793398605e-06,
      "loss": 0.1998,
      "step": 22772
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35088806231420105,
      "learning_rate": 4.429616005835886e-06,
      "loss": 0.0945,
      "step": 22773
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3254839607442477,
      "learning_rate": 4.428792287135186e-06,
      "loss": 0.2749,
      "step": 22774
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4858536055648442,
      "learning_rate": 4.427968623245853e-06,
      "loss": 0.2137,
      "step": 22775
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5548335717767436,
      "learning_rate": 4.427145014176006e-06,
      "loss": 0.3321,
      "step": 22776
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2734390243931586,
      "learning_rate": 4.426321459933736e-06,
      "loss": 0.1618,
      "step": 22777
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.603630488526838,
      "learning_rate": 4.42549796052715e-06,
      "loss": 0.8129,
      "step": 22778
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1693729457590558,
      "learning_rate": 4.424674515964352e-06,
      "loss": 0.4513,
      "step": 22779
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41226697379425975,
      "learning_rate": 4.423851126253435e-06,
      "loss": 0.2818,
      "step": 22780
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25493361982372464,
      "learning_rate": 4.423027791402511e-06,
      "loss": 0.2031,
      "step": 22781
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9343815286091913,
      "learning_rate": 4.422204511419672e-06,
      "loss": 0.3873,
      "step": 22782
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.657193205804415,
      "learning_rate": 4.42138128631302e-06,
      "loss": 0.2592,
      "step": 22783
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6707562639968166,
      "learning_rate": 4.4205581160906585e-06,
      "loss": 0.3938,
      "step": 22784
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22616562316812455,
      "learning_rate": 4.419735000760679e-06,
      "loss": 0.1287,
      "step": 22785
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24122451169054923,
      "learning_rate": 4.418911940331181e-06,
      "loss": 0.1786,
      "step": 22786
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.48578509367082984,
      "learning_rate": 4.418088934810265e-06,
      "loss": 0.3582,
      "step": 22787
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0308987911451426,
      "learning_rate": 4.417265984206031e-06,
      "loss": 0.5432,
      "step": 22788
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6452428042360426,
      "learning_rate": 4.416443088526565e-06,
      "loss": 0.3028,
      "step": 22789
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32998846511106017,
      "learning_rate": 4.415620247779972e-06,
      "loss": 0.1939,
      "step": 22790
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9505880512896011,
      "learning_rate": 4.414797461974346e-06,
      "loss": 0.4937,
      "step": 22791
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3773715582378982,
      "learning_rate": 4.413974731117775e-06,
      "loss": 0.2466,
      "step": 22792
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4044228077042223,
      "learning_rate": 4.413152055218365e-06,
      "loss": 0.2862,
      "step": 22793
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28046883009331347,
      "learning_rate": 4.412329434284199e-06,
      "loss": 0.1782,
      "step": 22794
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4482812651438854,
      "learning_rate": 4.411506868323375e-06,
      "loss": 0.5645,
      "step": 22795
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.18112488676246824,
      "learning_rate": 4.410684357343986e-06,
      "loss": 0.1022,
      "step": 22796
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1897899962152008,
      "learning_rate": 4.409861901354126e-06,
      "loss": 0.6081,
      "step": 22797
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2576881038199592,
      "learning_rate": 4.4090395003618815e-06,
      "loss": 0.2068,
      "step": 22798
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3173830692584147,
      "learning_rate": 4.408217154375346e-06,
      "loss": 0.1961,
      "step": 22799
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8225083152890892,
      "learning_rate": 4.4073948634026134e-06,
      "loss": 0.3662,
      "step": 22800
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6347690709426747,
      "learning_rate": 4.406572627451763e-06,
      "loss": 0.2764,
      "step": 22801
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9032352209307394,
      "learning_rate": 4.4057504465308985e-06,
      "loss": 0.4363,
      "step": 22802
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34174024944219,
      "learning_rate": 4.404928320648099e-06,
      "loss": 0.1661,
      "step": 22803
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.22694078971900705,
      "learning_rate": 4.404106249811455e-06,
      "loss": 0.1671,
      "step": 22804
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3197759398639179,
      "learning_rate": 4.403284234029058e-06,
      "loss": 0.2372,
      "step": 22805
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2491902908354566,
      "learning_rate": 4.402462273308986e-06,
      "loss": 0.6253,
      "step": 22806
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31033585111483997,
      "learning_rate": 4.40164036765934e-06,
      "loss": 0.0658,
      "step": 22807
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3656695369299,
      "learning_rate": 4.4008185170881935e-06,
      "loss": 0.284,
      "step": 22808
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9593481507975749,
      "learning_rate": 4.399996721603637e-06,
      "loss": 0.2872,
      "step": 22809
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3211877548384919,
      "learning_rate": 4.399174981213759e-06,
      "loss": 0.2869,
      "step": 22810
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6894958036416339,
      "learning_rate": 4.398353295926636e-06,
      "loss": 0.0331,
      "step": 22811
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.339960192258334,
      "learning_rate": 4.3975316657503575e-06,
      "loss": 0.1561,
      "step": 22812
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.29351902970595284,
      "learning_rate": 4.396710090693006e-06,
      "loss": 0.2014,
      "step": 22813
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0628523716277545,
      "learning_rate": 4.395888570762667e-06,
      "loss": 0.5512,
      "step": 22814
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4620691787245209,
      "learning_rate": 4.3950671059674176e-06,
      "loss": 0.2623,
      "step": 22815
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.27428431934486014,
      "learning_rate": 4.394245696315343e-06,
      "loss": 0.2018,
      "step": 22816
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35770862340979026,
      "learning_rate": 4.393424341814526e-06,
      "loss": 0.2629,
      "step": 22817
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0823421295425288,
      "learning_rate": 4.392603042473038e-06,
      "loss": 0.2993,
      "step": 22818
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.818563529461808,
      "learning_rate": 4.391781798298974e-06,
      "loss": 0.3816,
      "step": 22819
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2858272247717508,
      "learning_rate": 4.390960609300402e-06,
      "loss": 0.1656,
      "step": 22820
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6884862010276167,
      "learning_rate": 4.390139475485406e-06,
      "loss": 0.2584,
      "step": 22821
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2397626864244352,
      "learning_rate": 4.389318396862066e-06,
      "loss": 0.1316,
      "step": 22822
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3174050916528758,
      "learning_rate": 4.388497373438455e-06,
      "loss": 0.2849,
      "step": 22823
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24044743798121268,
      "learning_rate": 4.387676405222653e-06,
      "loss": 0.1546,
      "step": 22824
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7825704226993901,
      "learning_rate": 4.386855492222738e-06,
      "loss": 0.226,
      "step": 22825
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5789237474402138,
      "learning_rate": 4.386034634446789e-06,
      "loss": 0.3423,
      "step": 22826
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7086646997124079,
      "learning_rate": 4.3852138319028734e-06,
      "loss": 0.2443,
      "step": 22827
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32510679110822466,
      "learning_rate": 4.384393084599073e-06,
      "loss": 0.2914,
      "step": 22828
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0957135530392383,
      "learning_rate": 4.383572392543461e-06,
      "loss": 0.1474,
      "step": 22829
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9242849453627018,
      "learning_rate": 4.382751755744113e-06,
      "loss": 0.4544,
      "step": 22830
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.462561527733392,
      "learning_rate": 4.381931174209103e-06,
      "loss": 0.1925,
      "step": 22831
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3465504367571077,
      "learning_rate": 4.381110647946496e-06,
      "loss": 0.8217,
      "step": 22832
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.26213334259457965,
      "learning_rate": 4.380290176964379e-06,
      "loss": 0.1849,
      "step": 22833
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.46687709474438077,
      "learning_rate": 4.379469761270813e-06,
      "loss": 0.3364,
      "step": 22834
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.267341837307869,
      "learning_rate": 4.378649400873872e-06,
      "loss": 0.1736,
      "step": 22835
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6092454132559225,
      "learning_rate": 4.377829095781633e-06,
      "loss": 0.4195,
      "step": 22836
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35630340122142157,
      "learning_rate": 4.3770088460021564e-06,
      "loss": 0.0839,
      "step": 22837
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8613362444377008,
      "learning_rate": 4.376188651543517e-06,
      "loss": 0.0418,
      "step": 22838
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6321230426870795,
      "learning_rate": 4.375368512413785e-06,
      "loss": 0.2721,
      "step": 22839
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.21756961601668118,
      "learning_rate": 4.374548428621032e-06,
      "loss": 0.2095,
      "step": 22840
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3829204203629828,
      "learning_rate": 4.373728400173318e-06,
      "loss": 0.8018,
      "step": 22841
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8342633448432036,
      "learning_rate": 4.372908427078716e-06,
      "loss": 0.3105,
      "step": 22842
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6668980593322362,
      "learning_rate": 4.3720885093452944e-06,
      "loss": 0.3669,
      "step": 22843
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2820536806554277,
      "learning_rate": 4.371268646981113e-06,
      "loss": 0.177,
      "step": 22844
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4494967132171946,
      "learning_rate": 4.370448839994249e-06,
      "loss": 0.2088,
      "step": 22845
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25603308705611966,
      "learning_rate": 4.3696290883927574e-06,
      "loss": 0.1902,
      "step": 22846
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.152948394668242,
      "learning_rate": 4.368809392184709e-06,
      "loss": 0.4414,
      "step": 22847
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.266319919032165,
      "learning_rate": 4.367989751378169e-06,
      "loss": 0.0663,
      "step": 22848
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3944517788873962,
      "learning_rate": 4.367170165981196e-06,
      "loss": 0.298,
      "step": 22849
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.881523278281078,
      "learning_rate": 4.366350636001856e-06,
      "loss": 0.5378,
      "step": 22850
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4751682858959096,
      "learning_rate": 4.365531161448212e-06,
      "loss": 0.2647,
      "step": 22851
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40293414700826946,
      "learning_rate": 4.364711742328331e-06,
      "loss": 0.2829,
      "step": 22852
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3940551922434903,
      "learning_rate": 4.3638923786502654e-06,
      "loss": 0.1622,
      "step": 22853
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5631607735669495,
      "learning_rate": 4.363073070422081e-06,
      "loss": 0.3425,
      "step": 22854
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2394349604418763,
      "learning_rate": 4.3622538176518395e-06,
      "loss": 0.077,
      "step": 22855
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3827797087177178,
      "learning_rate": 4.3614346203475995e-06,
      "loss": 0.6287,
      "step": 22856
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35978956728400985,
      "learning_rate": 4.360615478517425e-06,
      "loss": 0.178,
      "step": 22857
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3468577580143816,
      "learning_rate": 4.359796392169367e-06,
      "loss": 0.2555,
      "step": 22858
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.418486056581916,
      "learning_rate": 4.358977361311488e-06,
      "loss": 0.2753,
      "step": 22859
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5534989383527141,
      "learning_rate": 4.358158385951849e-06,
      "loss": 0.3909,
      "step": 22860
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7270643572414015,
      "learning_rate": 4.357339466098498e-06,
      "loss": 0.2805,
      "step": 22861
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6554370071645834,
      "learning_rate": 4.356520601759505e-06,
      "loss": 0.3037,
      "step": 22862
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2622013713005157,
      "learning_rate": 4.355701792942916e-06,
      "loss": 0.1682,
      "step": 22863
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2592286388594667,
      "learning_rate": 4.354883039656789e-06,
      "loss": 0.2028,
      "step": 22864
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3723875942925332,
      "learning_rate": 4.354064341909181e-06,
      "loss": 0.5554,
      "step": 22865
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.713867946619785,
      "learning_rate": 4.353245699708148e-06,
      "loss": 0.1288,
      "step": 22866
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34936350175188996,
      "learning_rate": 4.35242711306174e-06,
      "loss": 0.2549,
      "step": 22867
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8852618180363938,
      "learning_rate": 4.351608581978012e-06,
      "loss": 0.3178,
      "step": 22868
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7329153680453289,
      "learning_rate": 4.3507901064650205e-06,
      "loss": 0.4822,
      "step": 22869
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3025374921840034,
      "learning_rate": 4.3499716865308074e-06,
      "loss": 0.2394,
      "step": 22870
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4038562718512427,
      "learning_rate": 4.349153322183439e-06,
      "loss": 0.2733,
      "step": 22871
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4606982045200341,
      "learning_rate": 4.348335013430956e-06,
      "loss": 0.1885,
      "step": 22872
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3000337225931605,
      "learning_rate": 4.347516760281414e-06,
      "loss": 0.1952,
      "step": 22873
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4996681166591357,
      "learning_rate": 4.346698562742864e-06,
      "loss": 0.1518,
      "step": 22874
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3129922568553471,
      "learning_rate": 4.345880420823351e-06,
      "loss": 0.2736,
      "step": 22875
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31209909774138705,
      "learning_rate": 4.345062334530926e-06,
      "loss": 0.1864,
      "step": 22876
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1842294042339612,
      "learning_rate": 4.34424430387364e-06,
      "loss": 0.3004,
      "step": 22877
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8583123035943072,
      "learning_rate": 4.343426328859542e-06,
      "loss": 0.44,
      "step": 22878
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1622115810475053,
      "learning_rate": 4.342608409496674e-06,
      "loss": 0.348,
      "step": 22879
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6596236496653076,
      "learning_rate": 4.341790545793085e-06,
      "loss": 0.3089,
      "step": 22880
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36111383453822393,
      "learning_rate": 4.340972737756823e-06,
      "loss": 0.2007,
      "step": 22881
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2423876669791724,
      "learning_rate": 4.340154985395935e-06,
      "loss": 0.2182,
      "step": 22882
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3079766550676169,
      "learning_rate": 4.339337288718466e-06,
      "loss": 0.0781,
      "step": 22883
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8138544244789143,
      "learning_rate": 4.338519647732456e-06,
      "loss": 0.8575,
      "step": 22884
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2906724573123517,
      "learning_rate": 4.337702062445953e-06,
      "loss": 0.177,
      "step": 22885
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2138548143098298,
      "learning_rate": 4.336884532867004e-06,
      "loss": 0.4176,
      "step": 22886
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.36842804466994317,
      "learning_rate": 4.336067059003641e-06,
      "loss": 0.2594,
      "step": 22887
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.6374770662136344,
      "learning_rate": 4.335249640863921e-06,
      "loss": 0.777,
      "step": 22888
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41214235009683015,
      "learning_rate": 4.334432278455876e-06,
      "loss": 0.1231,
      "step": 22889
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3594573725384671,
      "learning_rate": 4.33361497178755e-06,
      "loss": 0.2154,
      "step": 22890
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2926868836398946,
      "learning_rate": 4.332797720866988e-06,
      "loss": 0.1679,
      "step": 22891
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4156016546031356,
      "learning_rate": 4.331980525702223e-06,
      "loss": 0.1813,
      "step": 22892
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3279622920786458,
      "learning_rate": 4.331163386301298e-06,
      "loss": 0.3008,
      "step": 22893
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28679872655699834,
      "learning_rate": 4.330346302672255e-06,
      "loss": 0.1749,
      "step": 22894
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7658273521836674,
      "learning_rate": 4.329529274823132e-06,
      "loss": 0.3741,
      "step": 22895
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.990230722014133,
      "learning_rate": 4.328712302761959e-06,
      "loss": 0.4718,
      "step": 22896
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4327989456761239,
      "learning_rate": 4.327895386496789e-06,
      "loss": 0.763,
      "step": 22897
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.32350905489760773,
      "learning_rate": 4.32707852603565e-06,
      "loss": 0.1586,
      "step": 22898
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3678183203201682,
      "learning_rate": 4.326261721386571e-06,
      "loss": 0.3048,
      "step": 22899
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42187997780990383,
      "learning_rate": 4.3254449725576045e-06,
      "loss": 0.2015,
      "step": 22900
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33114489282793114,
      "learning_rate": 4.324628279556773e-06,
      "loss": 0.1909,
      "step": 22901
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5465916533334557,
      "learning_rate": 4.323811642392116e-06,
      "loss": 0.2544,
      "step": 22902
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3639988945185147,
      "learning_rate": 4.322995061071669e-06,
      "loss": 0.2881,
      "step": 22903
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7282030110312881,
      "learning_rate": 4.3221785356034666e-06,
      "loss": 0.2714,
      "step": 22904
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2765397706051135,
      "learning_rate": 4.321362065995537e-06,
      "loss": 0.2253,
      "step": 22905
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.583576538555836,
      "learning_rate": 4.320545652255916e-06,
      "loss": 0.6786,
      "step": 22906
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5226238443418799,
      "learning_rate": 4.319729294392636e-06,
      "loss": 0.0746,
      "step": 22907
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.34231004875348897,
      "learning_rate": 4.318912992413727e-06,
      "loss": 0.2637,
      "step": 22908
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35966193933290935,
      "learning_rate": 4.318096746327226e-06,
      "loss": 0.0888,
      "step": 22909
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40502994709382834,
      "learning_rate": 4.317280556141156e-06,
      "loss": 0.2646,
      "step": 22910
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3080658695500884,
      "learning_rate": 4.316464421863549e-06,
      "loss": 0.2324,
      "step": 22911
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41211058282260987,
      "learning_rate": 4.3156483435024374e-06,
      "loss": 0.2954,
      "step": 22912
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.43714593599219415,
      "learning_rate": 4.314832321065842e-06,
      "loss": 0.2475,
      "step": 22913
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2602406945624456,
      "learning_rate": 4.314016354561804e-06,
      "loss": 0.737,
      "step": 22914
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0766695388598508,
      "learning_rate": 4.31320044399834e-06,
      "loss": 0.3103,
      "step": 22915
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8522939327298099,
      "learning_rate": 4.312384589383482e-06,
      "loss": 0.3357,
      "step": 22916
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2539792381138997,
      "learning_rate": 4.311568790725259e-06,
      "loss": 0.2058,
      "step": 22917
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2740678266807683,
      "learning_rate": 4.3107530480316894e-06,
      "loss": 0.1499,
      "step": 22918
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9342871815866506,
      "learning_rate": 4.309937361310804e-06,
      "loss": 0.515,
      "step": 22919
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3740856501611141,
      "learning_rate": 4.309121730570628e-06,
      "loss": 0.2032,
      "step": 22920
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3881782375088359,
      "learning_rate": 4.308306155819189e-06,
      "loss": 0.2498,
      "step": 22921
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4371426115485429,
      "learning_rate": 4.307490637064501e-06,
      "loss": 0.2535,
      "step": 22922
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.49264844682916503,
      "learning_rate": 4.306675174314595e-06,
      "loss": 0.3403,
      "step": 22923
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4349899815830328,
      "learning_rate": 4.305859767577495e-06,
      "loss": 0.2724,
      "step": 22924
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8446545605707906,
      "learning_rate": 4.305044416861213e-06,
      "loss": 0.3582,
      "step": 22925
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3041816098515996,
      "learning_rate": 4.304229122173784e-06,
      "loss": 0.1995,
      "step": 22926
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.350465397478481,
      "learning_rate": 4.3034138835232205e-06,
      "loss": 0.5888,
      "step": 22927
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5923665910232145,
      "learning_rate": 4.302598700917546e-06,
      "loss": 0.289,
      "step": 22928
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33078092626818834,
      "learning_rate": 4.301783574364784e-06,
      "loss": 0.2708,
      "step": 22929
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.25319362660750855,
      "learning_rate": 4.300968503872942e-06,
      "loss": 0.1662,
      "step": 22930
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.17672019819766976,
      "learning_rate": 4.300153489450055e-06,
      "loss": 0.0735,
      "step": 22931
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2295699719827011,
      "learning_rate": 4.299338531104131e-06,
      "loss": 0.7733,
      "step": 22932
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9298517034014043,
      "learning_rate": 4.298523628843188e-06,
      "loss": 0.3014,
      "step": 22933
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8793564987167626,
      "learning_rate": 4.2977087826752465e-06,
      "loss": 0.3445,
      "step": 22934
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28897168752875246,
      "learning_rate": 4.296893992608325e-06,
      "loss": 0.1868,
      "step": 22935
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33595995763083236,
      "learning_rate": 4.296079258650434e-06,
      "loss": 0.2808,
      "step": 22936
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6836435089342053,
      "learning_rate": 4.2952645808095924e-06,
      "loss": 0.2889,
      "step": 22937
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.42529688339303484,
      "learning_rate": 4.2944499590938195e-06,
      "loss": 0.205,
      "step": 22938
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.185331579516365,
      "learning_rate": 4.2936353935111164e-06,
      "loss": 0.0705,
      "step": 22939
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37451058365853956,
      "learning_rate": 4.292820884069514e-06,
      "loss": 0.305,
      "step": 22940
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2996726497238999,
      "learning_rate": 4.292006430777013e-06,
      "loss": 0.2405,
      "step": 22941
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.219757551714907,
      "learning_rate": 4.2911920336416324e-06,
      "loss": 0.4667,
      "step": 22942
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6780744539719713,
      "learning_rate": 4.290377692671386e-06,
      "loss": 0.1114,
      "step": 22943
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2963032521031666,
      "learning_rate": 4.289563407874279e-06,
      "loss": 0.1793,
      "step": 22944
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.579121263201325,
      "learning_rate": 4.2887491792583255e-06,
      "loss": 0.3971,
      "step": 22945
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9781325840780885,
      "learning_rate": 4.287935006831538e-06,
      "loss": 0.2691,
      "step": 22946
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3366142123708311,
      "learning_rate": 4.287120890601929e-06,
      "loss": 0.298,
      "step": 22947
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.317832656983147,
      "learning_rate": 4.2863068305775e-06,
      "loss": 0.1522,
      "step": 22948
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33150299606999145,
      "learning_rate": 4.285492826766264e-06,
      "loss": 0.2338,
      "step": 22949
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0350343860046716,
      "learning_rate": 4.284678879176235e-06,
      "loss": 0.4584,
      "step": 22950
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.44633231913655996,
      "learning_rate": 4.283864987815408e-06,
      "loss": 0.2548,
      "step": 22951
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4113594885041813,
      "learning_rate": 4.283051152691805e-06,
      "loss": 0.2382,
      "step": 22952
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.30571696849163793,
      "learning_rate": 4.282237373813423e-06,
      "loss": 0.2407,
      "step": 22953
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4656251494969968,
      "learning_rate": 4.28142365118827e-06,
      "loss": 0.2374,
      "step": 22954
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9404698414951379,
      "learning_rate": 4.280609984824358e-06,
      "loss": 0.4746,
      "step": 22955
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4071950687846797,
      "learning_rate": 4.279796374729678e-06,
      "loss": 0.119,
      "step": 22956
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3612930005022071,
      "learning_rate": 4.278982820912251e-06,
      "loss": 0.1136,
      "step": 22957
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41028386773130593,
      "learning_rate": 4.27816932338007e-06,
      "loss": 0.3055,
      "step": 22958
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3036618969901887,
      "learning_rate": 4.277355882141142e-06,
      "loss": 0.2518,
      "step": 22959
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2833026771250564,
      "learning_rate": 4.276542497203474e-06,
      "loss": 0.1987,
      "step": 22960
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.707358105212325,
      "learning_rate": 4.27572916857506e-06,
      "loss": 0.2173,
      "step": 22961
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4781326329974936,
      "learning_rate": 4.274915896263906e-06,
      "loss": 0.2963,
      "step": 22962
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.660714079465027,
      "learning_rate": 4.274102680278013e-06,
      "loss": 0.273,
      "step": 22963
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.46694011499152055,
      "learning_rate": 4.273289520625387e-06,
      "loss": 0.3365,
      "step": 22964
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.40582696462504403,
      "learning_rate": 4.272476417314013e-06,
      "loss": 0.2085,
      "step": 22965
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.0524424604063431,
      "learning_rate": 4.271663370351911e-06,
      "loss": 0.4607,
      "step": 22966
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24035049046751542,
      "learning_rate": 4.270850379747063e-06,
      "loss": 0.1648,
      "step": 22967
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2590122176448433,
      "learning_rate": 4.270037445507477e-06,
      "loss": 0.7423,
      "step": 22968
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3259708149537797,
      "learning_rate": 4.269224567641149e-06,
      "loss": 0.1666,
      "step": 22969
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2860291724643319,
      "learning_rate": 4.268411746156074e-06,
      "loss": 0.2233,
      "step": 22970
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3816620277905466,
      "learning_rate": 4.267598981060248e-06,
      "loss": 0.2414,
      "step": 22971
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.43851574040307384,
      "learning_rate": 4.266786272361672e-06,
      "loss": 0.2614,
      "step": 22972
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.845684323994499,
      "learning_rate": 4.26597362006834e-06,
      "loss": 0.8568,
      "step": 22973
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.3509138768011832,
      "learning_rate": 4.265161024188245e-06,
      "loss": 0.2129,
      "step": 22974
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6600526891665133,
      "learning_rate": 4.264348484729382e-06,
      "loss": 0.3191,
      "step": 22975
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.31967671925507823,
      "learning_rate": 4.263536001699749e-06,
      "loss": 0.2203,
      "step": 22976
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4625593292211742,
      "learning_rate": 4.26272357510733e-06,
      "loss": 0.3468,
      "step": 22977
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.35100596807868545,
      "learning_rate": 4.26191120496013e-06,
      "loss": 0.2641,
      "step": 22978
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5434461139679141,
      "learning_rate": 4.261098891266134e-06,
      "loss": 0.3464,
      "step": 22979
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.29754145215439864,
      "learning_rate": 4.260286634033333e-06,
      "loss": 0.18,
      "step": 22980
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.45148779967993674,
      "learning_rate": 4.259474433269727e-06,
      "loss": 0.2651,
      "step": 22981
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2683699376329027,
      "learning_rate": 4.258662288983292e-06,
      "loss": 0.1663,
      "step": 22982
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4850920409831923,
      "learning_rate": 4.257850201182033e-06,
      "loss": 0.2384,
      "step": 22983
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6516924032654084,
      "learning_rate": 4.25703816987393e-06,
      "loss": 0.3329,
      "step": 22984
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3188697211304766,
      "learning_rate": 4.256226195066976e-06,
      "loss": 0.1908,
      "step": 22985
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.2600409101831935,
      "learning_rate": 4.255414276769162e-06,
      "loss": 0.8286,
      "step": 22986
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5893255863113961,
      "learning_rate": 4.254602414988468e-06,
      "loss": 0.2931,
      "step": 22987
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3094313351912337,
      "learning_rate": 4.253790609732886e-06,
      "loss": 0.2869,
      "step": 22988
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2725024119012814,
      "learning_rate": 4.252978861010404e-06,
      "loss": 0.0677,
      "step": 22989
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.28040194881216485,
      "learning_rate": 4.25216716882901e-06,
      "loss": 0.2334,
      "step": 22990
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2834231079391988,
      "learning_rate": 4.251355533196683e-06,
      "loss": 0.076,
      "step": 22991
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1588188089432112,
      "learning_rate": 4.250543954121412e-06,
      "loss": 0.4819,
      "step": 22992
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37890473974283645,
      "learning_rate": 4.249732431611184e-06,
      "loss": 0.1407,
      "step": 22993
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.33145967825726114,
      "learning_rate": 4.248920965673975e-06,
      "loss": 0.2421,
      "step": 22994
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.29781702657520626,
      "learning_rate": 4.24810955631778e-06,
      "loss": 0.2321,
      "step": 22995
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.7731448967484663,
      "learning_rate": 4.247298203550572e-06,
      "loss": 0.2824,
      "step": 22996
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.6249870646400738,
      "learning_rate": 4.246486907380337e-06,
      "loss": 0.4138,
      "step": 22997
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2943867866965115,
      "learning_rate": 4.2456756678150615e-06,
      "loss": 0.1479,
      "step": 22998
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24643403100597014,
      "learning_rate": 4.2448644848627184e-06,
      "loss": 0.1774,
      "step": 22999
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.41459591333587076,
      "learning_rate": 4.244053358531291e-06,
      "loss": 0.2218,
      "step": 23000
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5287912745955936,
      "learning_rate": 4.243242288828763e-06,
      "loss": 0.3528,
      "step": 23001
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4078219123451108,
      "learning_rate": 4.2424312757631135e-06,
      "loss": 0.1413,
      "step": 23002
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3404843569469961,
      "learning_rate": 4.241620319342312e-06,
      "loss": 0.2759,
      "step": 23003
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.093578539176219,
      "learning_rate": 4.240809419574352e-06,
      "loss": 0.2919,
      "step": 23004
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9423472025229972,
      "learning_rate": 4.239998576467201e-06,
      "loss": 0.3664,
      "step": 23005
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.286938024178596,
      "learning_rate": 4.239187790028839e-06,
      "loss": 0.228,
      "step": 23006
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3907870408168681,
      "learning_rate": 4.238377060267247e-06,
      "loss": 0.245,
      "step": 23007
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.4960142106258088,
      "learning_rate": 4.23756638719039e-06,
      "loss": 0.1996,
      "step": 23008
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2784802421631547,
      "learning_rate": 4.236755770806258e-06,
      "loss": 0.0864,
      "step": 23009
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.1938120134187704,
      "learning_rate": 4.235945211122815e-06,
      "loss": 0.5186,
      "step": 23010
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.3590042404755711,
      "learning_rate": 4.235134708148041e-06,
      "loss": 0.1543,
      "step": 23011
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.5726955755247365,
      "learning_rate": 4.234324261889911e-06,
      "loss": 0.3121,
      "step": 23012
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.37136621461317204,
      "learning_rate": 4.233513872356393e-06,
      "loss": 0.2449,
      "step": 23013
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.4596861332555486,
      "learning_rate": 4.2327035395554626e-06,
      "loss": 0.4342,
      "step": 23014
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.8425094312758913,
      "learning_rate": 4.231893263495092e-06,
      "loss": 0.4218,
      "step": 23015
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.9293862912753765,
      "learning_rate": 4.231083044183258e-06,
      "loss": 0.4851,
      "step": 23016
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.24139111214866552,
      "learning_rate": 4.2302728816279235e-06,
      "loss": 0.1568,
      "step": 23017
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.2967947311269018,
      "learning_rate": 4.2294627758370635e-06,
      "loss": 0.2217,
      "step": 23018
    },
    {
      "epoch": 0.7,
      "grad_norm": 0.48707931481427813,
      "learning_rate": 4.228652726818649e-06,
      "loss": 0.2581,
      "step": 23019
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.8082794510380602,
      "learning_rate": 4.2278427345806426e-06,
      "loss": 0.7002,
      "step": 23020
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37125502876909944,
      "learning_rate": 4.227032799131025e-06,
      "loss": 0.1889,
      "step": 23021
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5098388946653352,
      "learning_rate": 4.226222920477756e-06,
      "loss": 0.2634,
      "step": 23022
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4074297379138911,
      "learning_rate": 4.225413098628804e-06,
      "loss": 0.8062,
      "step": 23023
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.29987903598909493,
      "learning_rate": 4.224603333592142e-06,
      "loss": 0.2199,
      "step": 23024
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6402943340092824,
      "learning_rate": 4.2237936253757304e-06,
      "loss": 0.2476,
      "step": 23025
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36185361801093674,
      "learning_rate": 4.2229839739875365e-06,
      "loss": 0.2027,
      "step": 23026
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30118972560040846,
      "learning_rate": 4.222174379435528e-06,
      "loss": 0.1756,
      "step": 23027
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4347705664872419,
      "learning_rate": 4.221364841727672e-06,
      "loss": 0.1747,
      "step": 23028
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.47944152035113735,
      "learning_rate": 4.220555360871926e-06,
      "loss": 0.3396,
      "step": 23029
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2655386277202751,
      "learning_rate": 4.219745936876259e-06,
      "loss": 0.1759,
      "step": 23030
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.42663090780797425,
      "learning_rate": 4.218936569748637e-06,
      "loss": 0.2989,
      "step": 23031
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0679428446587484,
      "learning_rate": 4.218127259497012e-06,
      "loss": 0.3575,
      "step": 23032
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2266948410504588,
      "learning_rate": 4.21731800612936e-06,
      "loss": 0.6428,
      "step": 23033
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5515656142960719,
      "learning_rate": 4.21650880965363e-06,
      "loss": 0.1115,
      "step": 23034
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36117478623848687,
      "learning_rate": 4.215699670077797e-06,
      "loss": 0.1973,
      "step": 23035
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22888494316234612,
      "learning_rate": 4.214890587409809e-06,
      "loss": 0.1703,
      "step": 23036
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4048652084903013,
      "learning_rate": 4.214081561657632e-06,
      "loss": 0.2138,
      "step": 23037
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5640079312682793,
      "learning_rate": 4.213272592829227e-06,
      "loss": 0.3148,
      "step": 23038
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3856220440661358,
      "learning_rate": 4.212463680932547e-06,
      "loss": 0.1755,
      "step": 23039
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.49464228757781864,
      "learning_rate": 4.211654825975554e-06,
      "loss": 0.3526,
      "step": 23040
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3748600836819384,
      "learning_rate": 4.210846027966205e-06,
      "loss": 0.2803,
      "step": 23041
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.46237220426150105,
      "learning_rate": 4.2100372869124615e-06,
      "loss": 0.3376,
      "step": 23042
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.778644678131574,
      "learning_rate": 4.209228602822273e-06,
      "loss": 0.1141,
      "step": 23043
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32009689968480715,
      "learning_rate": 4.2084199757035985e-06,
      "loss": 0.2449,
      "step": 23044
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22656275820230037,
      "learning_rate": 4.207611405564398e-06,
      "loss": 0.0772,
      "step": 23045
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.725250414962111,
      "learning_rate": 4.206802892412614e-06,
      "loss": 0.4206,
      "step": 23046
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39296499264191515,
      "learning_rate": 4.205994436256218e-06,
      "loss": 0.2363,
      "step": 23047
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.27216640445282586,
      "learning_rate": 4.205186037103152e-06,
      "loss": 0.1865,
      "step": 23048
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.49805200123374443,
      "learning_rate": 4.204377694961371e-06,
      "loss": 0.3336,
      "step": 23049
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0027402184859764,
      "learning_rate": 4.203569409838834e-06,
      "loss": 0.5289,
      "step": 23050
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2365413408889332,
      "learning_rate": 4.2027611817434835e-06,
      "loss": 0.605,
      "step": 23051
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28243910797402266,
      "learning_rate": 4.201953010683277e-06,
      "loss": 0.062,
      "step": 23052
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.371105667624191,
      "learning_rate": 4.201144896666165e-06,
      "loss": 0.279,
      "step": 23053
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2946087879205612,
      "learning_rate": 4.2003368397000996e-06,
      "loss": 0.2124,
      "step": 23054
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3667647358859084,
      "learning_rate": 4.1995288397930264e-06,
      "loss": 0.232,
      "step": 23055
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35082335606312787,
      "learning_rate": 4.198720896952897e-06,
      "loss": 0.2029,
      "step": 23056
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3833345283419914,
      "learning_rate": 4.197913011187664e-06,
      "loss": 0.2281,
      "step": 23057
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.525845351763351,
      "learning_rate": 4.197105182505265e-06,
      "loss": 0.2053,
      "step": 23058
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2026934007120578,
      "learning_rate": 4.19629741091366e-06,
      "loss": 0.7835,
      "step": 23059
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2816832116586865,
      "learning_rate": 4.195489696420789e-06,
      "loss": 0.2196,
      "step": 23060
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5021267425456132,
      "learning_rate": 4.1946820390346e-06,
      "loss": 0.1101,
      "step": 23061
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5740395736237563,
      "learning_rate": 4.1938744387630424e-06,
      "loss": 0.3294,
      "step": 23062
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25459168286042744,
      "learning_rate": 4.193066895614052e-06,
      "loss": 0.1396,
      "step": 23063
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6754412425551014,
      "learning_rate": 4.192259409595586e-06,
      "loss": 0.3533,
      "step": 23064
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23211289230732593,
      "learning_rate": 4.1914519807155805e-06,
      "loss": 0.1906,
      "step": 23065
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9305979113192203,
      "learning_rate": 4.190644608981982e-06,
      "loss": 0.4186,
      "step": 23066
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30541979051187196,
      "learning_rate": 4.189837294402732e-06,
      "loss": 0.1899,
      "step": 23067
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3433204615184007,
      "learning_rate": 4.18903003698578e-06,
      "loss": 0.8276,
      "step": 23068
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9645824739820285,
      "learning_rate": 4.188222836739058e-06,
      "loss": 0.2988,
      "step": 23069
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6663938887064529,
      "learning_rate": 4.187415693670511e-06,
      "loss": 0.7465,
      "step": 23070
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2546560640418267,
      "learning_rate": 4.1866086077880865e-06,
      "loss": 0.1668,
      "step": 23071
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.340534068766306,
      "learning_rate": 4.185801579099711e-06,
      "loss": 0.2628,
      "step": 23072
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7155845536494436,
      "learning_rate": 4.184994607613339e-06,
      "loss": 0.2899,
      "step": 23073
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2024617497606104,
      "learning_rate": 4.184187693336902e-06,
      "loss": 0.0707,
      "step": 23074
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3145537651315111,
      "learning_rate": 4.183380836278339e-06,
      "loss": 0.1718,
      "step": 23075
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3465955717604178,
      "learning_rate": 4.1825740364455925e-06,
      "loss": 0.2225,
      "step": 23076
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4723949261214366,
      "learning_rate": 4.1817672938465935e-06,
      "loss": 0.3577,
      "step": 23077
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.42774948562413245,
      "learning_rate": 4.180960608489283e-06,
      "loss": 0.2179,
      "step": 23078
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0046638259015828,
      "learning_rate": 4.180153980381596e-06,
      "loss": 0.4179,
      "step": 23079
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2686884581618699,
      "learning_rate": 4.179347409531474e-06,
      "loss": 0.1697,
      "step": 23080
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7211328113336462,
      "learning_rate": 4.178540895946842e-06,
      "loss": 0.3944,
      "step": 23081
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9829522446253587,
      "learning_rate": 4.17773443963564e-06,
      "loss": 0.2809,
      "step": 23082
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.386530661697314,
      "learning_rate": 4.176928040605805e-06,
      "loss": 0.2811,
      "step": 23083
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2323172933510436,
      "learning_rate": 4.176121698865266e-06,
      "loss": 0.1266,
      "step": 23084
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33571106115196614,
      "learning_rate": 4.175315414421961e-06,
      "loss": 0.2316,
      "step": 23085
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9565895614715236,
      "learning_rate": 4.174509187283817e-06,
      "loss": 0.5329,
      "step": 23086
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5590125735480096,
      "learning_rate": 4.173703017458768e-06,
      "loss": 0.7967,
      "step": 23087
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9973121939922581,
      "learning_rate": 4.172896904954749e-06,
      "loss": 0.2138,
      "step": 23088
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35517488370287914,
      "learning_rate": 4.1720908497796815e-06,
      "loss": 0.1636,
      "step": 23089
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28480515295147185,
      "learning_rate": 4.17128485194151e-06,
      "loss": 0.2681,
      "step": 23090
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0057786758498277,
      "learning_rate": 4.17047891144815e-06,
      "loss": 0.2906,
      "step": 23091
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4478959236071244,
      "learning_rate": 4.169673028307538e-06,
      "loss": 0.2272,
      "step": 23092
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39921402115602594,
      "learning_rate": 4.168867202527604e-06,
      "loss": 0.1403,
      "step": 23093
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33574049429487074,
      "learning_rate": 4.168061434116271e-06,
      "loss": 0.2617,
      "step": 23094
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.421531486634359,
      "learning_rate": 4.167255723081468e-06,
      "loss": 0.276,
      "step": 23095
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.29674536391360645,
      "learning_rate": 4.166450069431122e-06,
      "loss": 0.2173,
      "step": 23096
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0536159737581245,
      "learning_rate": 4.165644473173163e-06,
      "loss": 0.2818,
      "step": 23097
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6824162861842019,
      "learning_rate": 4.164838934315511e-06,
      "loss": 0.3838,
      "step": 23098
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3777620949896715,
      "learning_rate": 4.164033452866092e-06,
      "loss": 0.2358,
      "step": 23099
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.543823635067446,
      "learning_rate": 4.1632280288328365e-06,
      "loss": 0.7086,
      "step": 23100
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43566751651856944,
      "learning_rate": 4.162422662223658e-06,
      "loss": 0.2462,
      "step": 23101
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3152085225585723,
      "learning_rate": 4.161617353046492e-06,
      "loss": 0.191,
      "step": 23102
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3501311222009138,
      "learning_rate": 4.160812101309253e-06,
      "loss": 0.2447,
      "step": 23103
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22185975006912934,
      "learning_rate": 4.160006907019864e-06,
      "loss": 0.1477,
      "step": 23104
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4609575011936074,
      "learning_rate": 4.159201770186251e-06,
      "loss": 0.2542,
      "step": 23105
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43341585775247776,
      "learning_rate": 4.158396690816335e-06,
      "loss": 0.2566,
      "step": 23106
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38774752442294025,
      "learning_rate": 4.157591668918032e-06,
      "loss": 0.2622,
      "step": 23107
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.337119992217046,
      "learning_rate": 4.156786704499264e-06,
      "loss": 0.2229,
      "step": 23108
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6126798449701252,
      "learning_rate": 4.1559817975679504e-06,
      "loss": 0.633,
      "step": 23109
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1004896475156665,
      "learning_rate": 4.155176948132011e-06,
      "loss": 0.1339,
      "step": 23110
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8901460792511949,
      "learning_rate": 4.1543721561993685e-06,
      "loss": 0.4543,
      "step": 23111
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.27186500548739606,
      "learning_rate": 4.153567421777932e-06,
      "loss": 0.187,
      "step": 23112
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5228268268972789,
      "learning_rate": 4.152762744875623e-06,
      "loss": 0.8086,
      "step": 23113
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.203951928229166,
      "learning_rate": 4.151958125500362e-06,
      "loss": 0.1624,
      "step": 23114
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4909660689256897,
      "learning_rate": 4.151153563660053e-06,
      "loss": 0.242,
      "step": 23115
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.46665362281527917,
      "learning_rate": 4.1503490593626284e-06,
      "loss": 0.2833,
      "step": 23116
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4857591355133546,
      "learning_rate": 4.14954461261599e-06,
      "loss": 0.2233,
      "step": 23117
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3100085657606302,
      "learning_rate": 4.148740223428057e-06,
      "loss": 0.5159,
      "step": 23118
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30231979900782985,
      "learning_rate": 4.147935891806747e-06,
      "loss": 0.2175,
      "step": 23119
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7419266324102444,
      "learning_rate": 4.147131617759965e-06,
      "loss": 0.3597,
      "step": 23120
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3108598267268271,
      "learning_rate": 4.146327401295628e-06,
      "loss": 0.1976,
      "step": 23121
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4855081218963722,
      "learning_rate": 4.1455232424216485e-06,
      "loss": 0.8722,
      "step": 23122
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6657734927110828,
      "learning_rate": 4.1447191411459406e-06,
      "loss": 0.291,
      "step": 23123
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8905162960093521,
      "learning_rate": 4.14391509747641e-06,
      "loss": 0.4614,
      "step": 23124
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23386727485785333,
      "learning_rate": 4.143111111420969e-06,
      "loss": 0.1933,
      "step": 23125
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28374443198585186,
      "learning_rate": 4.142307182987531e-06,
      "loss": 0.2397,
      "step": 23126
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.48010770111566997,
      "learning_rate": 4.141503312183996e-06,
      "loss": 0.1101,
      "step": 23127
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1883899982827386,
      "learning_rate": 4.140699499018287e-06,
      "loss": 0.1239,
      "step": 23128
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3192751946936094,
      "learning_rate": 4.1398957434983005e-06,
      "loss": 0.7262,
      "step": 23129
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28637736986519285,
      "learning_rate": 4.139092045631947e-06,
      "loss": 0.1757,
      "step": 23130
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.34710089250401566,
      "learning_rate": 4.1382884054271376e-06,
      "loss": 0.2901,
      "step": 23131
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8868398554300694,
      "learning_rate": 4.137484822891773e-06,
      "loss": 0.2809,
      "step": 23132
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7291210736710562,
      "learning_rate": 4.136681298033761e-06,
      "loss": 0.3591,
      "step": 23133
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20640782725214238,
      "learning_rate": 4.135877830861009e-06,
      "loss": 0.1156,
      "step": 23134
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32531451293759805,
      "learning_rate": 4.13507442138142e-06,
      "loss": 0.2207,
      "step": 23135
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7168710404215793,
      "learning_rate": 4.134271069602897e-06,
      "loss": 0.0461,
      "step": 23136
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31386602172634126,
      "learning_rate": 4.13346777553335e-06,
      "loss": 0.2589,
      "step": 23137
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7157246224986158,
      "learning_rate": 4.132664539180675e-06,
      "loss": 0.2329,
      "step": 23138
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3673045484621633,
      "learning_rate": 4.131861360552776e-06,
      "loss": 0.2797,
      "step": 23139
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9846096506942793,
      "learning_rate": 4.131058239657559e-06,
      "loss": 0.4473,
      "step": 23140
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.216583995924118,
      "learning_rate": 4.130255176502915e-06,
      "loss": 0.3067,
      "step": 23141
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45847110649954764,
      "learning_rate": 4.129452171096759e-06,
      "loss": 0.3152,
      "step": 23142
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31197072252080893,
      "learning_rate": 4.128649223446981e-06,
      "loss": 0.1809,
      "step": 23143
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3921999173460958,
      "learning_rate": 4.1278463335614845e-06,
      "loss": 0.2999,
      "step": 23144
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19144432268517386,
      "learning_rate": 4.127043501448172e-06,
      "loss": 0.0717,
      "step": 23145
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2404649814454234,
      "learning_rate": 4.126240727114932e-06,
      "loss": 0.5528,
      "step": 23146
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9738220413386364,
      "learning_rate": 4.12543801056967e-06,
      "loss": 0.4423,
      "step": 23147
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39749357952809145,
      "learning_rate": 4.124635351820282e-06,
      "loss": 0.2635,
      "step": 23148
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2848894684104733,
      "learning_rate": 4.123832750874668e-06,
      "loss": 0.2328,
      "step": 23149
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9845770300307845,
      "learning_rate": 4.123030207740717e-06,
      "loss": 0.3924,
      "step": 23150
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2372058793960186,
      "learning_rate": 4.122227722426328e-06,
      "loss": 0.2083,
      "step": 23151
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9221770471242576,
      "learning_rate": 4.1214252949394016e-06,
      "loss": 0.4423,
      "step": 23152
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2098497649798293,
      "learning_rate": 4.120622925287819e-06,
      "loss": 0.1468,
      "step": 23153
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.340158306688462,
      "learning_rate": 4.119820613479492e-06,
      "loss": 0.1583,
      "step": 23154
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4389166155697345,
      "learning_rate": 4.119018359522298e-06,
      "loss": 0.302,
      "step": 23155
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1763324345501573,
      "learning_rate": 4.118216163424139e-06,
      "loss": 0.3324,
      "step": 23156
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40511815524504713,
      "learning_rate": 4.117414025192908e-06,
      "loss": 0.286,
      "step": 23157
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43755001246200814,
      "learning_rate": 4.116611944836488e-06,
      "loss": 0.2478,
      "step": 23158
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7079431161124572,
      "learning_rate": 4.115809922362776e-06,
      "loss": 0.6954,
      "step": 23159
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4977429008657796,
      "learning_rate": 4.1150079577796646e-06,
      "loss": 0.2589,
      "step": 23160
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40269845955442346,
      "learning_rate": 4.11420605109504e-06,
      "loss": 0.2852,
      "step": 23161
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22084286547119936,
      "learning_rate": 4.1134042023167955e-06,
      "loss": 0.1572,
      "step": 23162
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.46117608714129765,
      "learning_rate": 4.112602411452815e-06,
      "loss": 0.2627,
      "step": 23163
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1935562339191812,
      "learning_rate": 4.111800678510989e-06,
      "loss": 0.2327,
      "step": 23164
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.837639133977783,
      "learning_rate": 4.110999003499204e-06,
      "loss": 0.424,
      "step": 23165
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2800912581584032,
      "learning_rate": 4.110197386425354e-06,
      "loss": 0.1729,
      "step": 23166
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30023672963774056,
      "learning_rate": 4.109395827297312e-06,
      "loss": 0.2241,
      "step": 23167
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6748740118765828,
      "learning_rate": 4.10859432612298e-06,
      "loss": 0.4099,
      "step": 23168
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8610213689067873,
      "learning_rate": 4.107792882910234e-06,
      "loss": 0.284,
      "step": 23169
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.597810515375884,
      "learning_rate": 4.106991497666954e-06,
      "loss": 0.2804,
      "step": 23170
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37476391177553525,
      "learning_rate": 4.106190170401039e-06,
      "loss": 0.2124,
      "step": 23171
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2822333050935059,
      "learning_rate": 4.105388901120361e-06,
      "loss": 0.1526,
      "step": 23172
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30972562251446945,
      "learning_rate": 4.1045876898328055e-06,
      "loss": 0.2347,
      "step": 23173
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8585648921592548,
      "learning_rate": 4.103786536546256e-06,
      "loss": 0.4077,
      "step": 23174
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4730249989894128,
      "learning_rate": 4.102985441268598e-06,
      "loss": 0.1602,
      "step": 23175
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3957943149550934,
      "learning_rate": 4.102184404007706e-06,
      "loss": 0.2836,
      "step": 23176
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9440328349287601,
      "learning_rate": 4.101383424771466e-06,
      "loss": 0.5279,
      "step": 23177
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.47772036759259934,
      "learning_rate": 4.100582503567759e-06,
      "loss": 0.3065,
      "step": 23178
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3630908673170209,
      "learning_rate": 4.099781640404455e-06,
      "loss": 0.194,
      "step": 23179
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3309060538090397,
      "learning_rate": 4.098980835289449e-06,
      "loss": 0.1958,
      "step": 23180
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25766115421068225,
      "learning_rate": 4.098180088230607e-06,
      "loss": 0.1371,
      "step": 23181
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0510632759025764,
      "learning_rate": 4.097379399235812e-06,
      "loss": 0.2806,
      "step": 23182
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8162897842110884,
      "learning_rate": 4.096578768312943e-06,
      "loss": 0.3514,
      "step": 23183
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28894176314737424,
      "learning_rate": 4.095778195469872e-06,
      "loss": 0.1764,
      "step": 23184
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26639235817922713,
      "learning_rate": 4.094977680714477e-06,
      "loss": 0.261,
      "step": 23185
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9649776898820276,
      "learning_rate": 4.094177224054635e-06,
      "loss": 0.4374,
      "step": 23186
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5215812503522395,
      "learning_rate": 4.093376825498221e-06,
      "loss": 0.7889,
      "step": 23187
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7365904919823699,
      "learning_rate": 4.092576485053112e-06,
      "loss": 0.1394,
      "step": 23188
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33481819272636515,
      "learning_rate": 4.091776202727178e-06,
      "loss": 0.2399,
      "step": 23189
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1528054764882837,
      "learning_rate": 4.090975978528292e-06,
      "loss": 0.0668,
      "step": 23190
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3424310961324444,
      "learning_rate": 4.090175812464328e-06,
      "loss": 0.2832,
      "step": 23191
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.777544886470906,
      "learning_rate": 4.089375704543163e-06,
      "loss": 0.3409,
      "step": 23192
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.29558388105086963,
      "learning_rate": 4.088575654772663e-06,
      "loss": 0.1779,
      "step": 23193
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5957587723703566,
      "learning_rate": 4.0877756631606995e-06,
      "loss": 0.7452,
      "step": 23194
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9541215403277669,
      "learning_rate": 4.086975729715147e-06,
      "loss": 0.5208,
      "step": 23195
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3320780848326032,
      "learning_rate": 4.086175854443867e-06,
      "loss": 0.2683,
      "step": 23196
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7526274875515936,
      "learning_rate": 4.085376037354742e-06,
      "loss": 0.1166,
      "step": 23197
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4607752951246412,
      "learning_rate": 4.084576278455629e-06,
      "loss": 0.2762,
      "step": 23198
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2522899527823769,
      "learning_rate": 4.083776577754401e-06,
      "loss": 0.1228,
      "step": 23199
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9092693172316091,
      "learning_rate": 4.082976935258931e-06,
      "loss": 0.38,
      "step": 23200
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3792798635603949,
      "learning_rate": 4.082177350977076e-06,
      "loss": 0.1589,
      "step": 23201
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3829339561931485,
      "learning_rate": 4.081377824916708e-06,
      "loss": 0.2689,
      "step": 23202
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2503154980878492,
      "learning_rate": 4.080578357085692e-06,
      "loss": 0.2103,
      "step": 23203
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3435497827522236,
      "learning_rate": 4.079778947491898e-06,
      "loss": 0.8036,
      "step": 23204
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5480634312156771,
      "learning_rate": 4.0789795961431786e-06,
      "loss": 0.1493,
      "step": 23205
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6145743175907268,
      "learning_rate": 4.078180303047414e-06,
      "loss": 0.1673,
      "step": 23206
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39235184772782766,
      "learning_rate": 4.077381068212457e-06,
      "loss": 0.2427,
      "step": 23207
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.27018728514537216,
      "learning_rate": 4.076581891646173e-06,
      "loss": 0.2291,
      "step": 23208
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7234476498289992,
      "learning_rate": 4.075782773356429e-06,
      "loss": 0.3711,
      "step": 23209
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2316316324869341,
      "learning_rate": 4.074983713351077e-06,
      "loss": 0.1017,
      "step": 23210
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6296082210865506,
      "learning_rate": 4.074184711637992e-06,
      "loss": 0.3087,
      "step": 23211
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38427198455614103,
      "learning_rate": 4.0733857682250235e-06,
      "loss": 0.2329,
      "step": 23212
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4289154211142814,
      "learning_rate": 4.072586883120037e-06,
      "loss": 0.7412,
      "step": 23213
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3100934633332762,
      "learning_rate": 4.071788056330893e-06,
      "loss": 0.2132,
      "step": 23214
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2334601398534504,
      "learning_rate": 4.070989287865445e-06,
      "loss": 0.5439,
      "step": 23215
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2631183039522809,
      "learning_rate": 4.070190577731557e-06,
      "loss": 0.1737,
      "step": 23216
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7237077079270136,
      "learning_rate": 4.069391925937085e-06,
      "loss": 0.8226,
      "step": 23217
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5738454604709804,
      "learning_rate": 4.068593332489889e-06,
      "loss": 0.286,
      "step": 23218
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.22136420472185153,
      "learning_rate": 4.067794797397822e-06,
      "loss": 0.135,
      "step": 23219
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36823403522374015,
      "learning_rate": 4.066996320668739e-06,
      "loss": 0.2758,
      "step": 23220
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.34636472244464056,
      "learning_rate": 4.066197902310504e-06,
      "loss": 0.2061,
      "step": 23221
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3794472566752598,
      "learning_rate": 4.065399542330959e-06,
      "loss": 0.7495,
      "step": 23222
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1784190654008255,
      "learning_rate": 4.064601240737972e-06,
      "loss": 0.203,
      "step": 23223
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6469741389208253,
      "learning_rate": 4.063802997539389e-06,
      "loss": 0.3342,
      "step": 23224
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.35576631622788385,
      "learning_rate": 4.063004812743065e-06,
      "loss": 0.1716,
      "step": 23225
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.38923545035292034,
      "learning_rate": 4.062206686356856e-06,
      "loss": 0.3093,
      "step": 23226
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2880952622183271,
      "learning_rate": 4.061408618388608e-06,
      "loss": 0.2496,
      "step": 23227
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3244009022330497,
      "learning_rate": 4.0606106088461775e-06,
      "loss": 0.2126,
      "step": 23228
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3886758084628996,
      "learning_rate": 4.059812657737412e-06,
      "loss": 0.1435,
      "step": 23229
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3690366109997502,
      "learning_rate": 4.05901476507017e-06,
      "loss": 0.2736,
      "step": 23230
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9180038769483031,
      "learning_rate": 4.05821693085229e-06,
      "loss": 0.5101,
      "step": 23231
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33318642210654037,
      "learning_rate": 4.057419155091627e-06,
      "loss": 0.2167,
      "step": 23232
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.857556772967971,
      "learning_rate": 4.056621437796035e-06,
      "loss": 0.5582,
      "step": 23233
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31554350241066076,
      "learning_rate": 4.055823778973348e-06,
      "loss": 0.18,
      "step": 23234
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45675324168725856,
      "learning_rate": 4.05502617863143e-06,
      "loss": 0.3203,
      "step": 23235
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7026878347093333,
      "learning_rate": 4.054228636778115e-06,
      "loss": 0.1711,
      "step": 23236
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5035970592128549,
      "learning_rate": 4.0534311534212615e-06,
      "loss": 0.213,
      "step": 23237
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2694395961321205,
      "learning_rate": 4.052633728568704e-06,
      "loss": 0.2179,
      "step": 23238
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33513657438384037,
      "learning_rate": 4.051836362228294e-06,
      "loss": 0.2518,
      "step": 23239
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0688349700693662,
      "learning_rate": 4.05103905440788e-06,
      "loss": 0.4593,
      "step": 23240
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.41047499036412954,
      "learning_rate": 4.050241805115296e-06,
      "loss": 0.2207,
      "step": 23241
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6843432559948986,
      "learning_rate": 4.049444614358391e-06,
      "loss": 0.2824,
      "step": 23242
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37907234942668944,
      "learning_rate": 4.0486474821450075e-06,
      "loss": 0.2455,
      "step": 23243
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4041344172899677,
      "learning_rate": 4.047850408482993e-06,
      "loss": 0.2357,
      "step": 23244
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.33721864610694846,
      "learning_rate": 4.04705339338018e-06,
      "loss": 0.2437,
      "step": 23245
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3827758125508567,
      "learning_rate": 4.046256436844414e-06,
      "loss": 0.5487,
      "step": 23246
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32461339785038396,
      "learning_rate": 4.045459538883542e-06,
      "loss": 0.16,
      "step": 23247
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5533801684594344,
      "learning_rate": 4.04466269950539e-06,
      "loss": 0.3207,
      "step": 23248
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2680229210015603,
      "learning_rate": 4.043865918717813e-06,
      "loss": 0.1653,
      "step": 23249
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2591984859392996,
      "learning_rate": 4.043069196528639e-06,
      "loss": 0.2266,
      "step": 23250
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7004728661066739,
      "learning_rate": 4.0422725329457116e-06,
      "loss": 0.2895,
      "step": 23251
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8737925974506731,
      "learning_rate": 4.041475927976869e-06,
      "loss": 0.3465,
      "step": 23252
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2941627030208918,
      "learning_rate": 4.040679381629945e-06,
      "loss": 0.191,
      "step": 23253
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.418954839326017,
      "learning_rate": 4.039882893912778e-06,
      "loss": 0.507,
      "step": 23254
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3256474641405557,
      "learning_rate": 4.039086464833203e-06,
      "loss": 0.2144,
      "step": 23255
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.6873678092468496,
      "learning_rate": 4.038290094399061e-06,
      "loss": 0.8289,
      "step": 23256
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25307176528715036,
      "learning_rate": 4.0374937826181795e-06,
      "loss": 0.1697,
      "step": 23257
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.23387350064615448,
      "learning_rate": 4.0366975294983955e-06,
      "loss": 0.1452,
      "step": 23258
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6677914793443686,
      "learning_rate": 4.035901335047547e-06,
      "loss": 0.3489,
      "step": 23259
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.323534098519349,
      "learning_rate": 4.0351051992734566e-06,
      "loss": 0.1284,
      "step": 23260
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7038528276530277,
      "learning_rate": 4.034309122183972e-06,
      "loss": 0.2964,
      "step": 23261
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2340224035543275,
      "learning_rate": 4.0335131037869125e-06,
      "loss": 0.2132,
      "step": 23262
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1114888217879415,
      "learning_rate": 4.032717144090115e-06,
      "loss": 0.4543,
      "step": 23263
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7336840096555247,
      "learning_rate": 4.031921243101412e-06,
      "loss": 0.0317,
      "step": 23264
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3412019396502968,
      "learning_rate": 4.031125400828626e-06,
      "loss": 0.7941,
      "step": 23265
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.28031848803557374,
      "learning_rate": 4.030329617279599e-06,
      "loss": 0.1866,
      "step": 23266
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5431563282102447,
      "learning_rate": 4.0295338924621495e-06,
      "loss": 0.8231,
      "step": 23267
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30653669204637674,
      "learning_rate": 4.02873822638411e-06,
      "loss": 0.2362,
      "step": 23268
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8316690912097962,
      "learning_rate": 4.027942619053314e-06,
      "loss": 0.3988,
      "step": 23269
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.13141666471692973,
      "learning_rate": 4.027147070477577e-06,
      "loss": 0.0666,
      "step": 23270
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3905154898878273,
      "learning_rate": 4.026351580664735e-06,
      "loss": 0.3153,
      "step": 23271
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0460782799139945,
      "learning_rate": 4.02555614962261e-06,
      "loss": 0.1369,
      "step": 23272
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30626106723598606,
      "learning_rate": 4.024760777359035e-06,
      "loss": 0.2406,
      "step": 23273
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7402577995736533,
      "learning_rate": 4.02396546388182e-06,
      "loss": 0.4053,
      "step": 23274
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3882982841110107,
      "learning_rate": 4.023170209198808e-06,
      "loss": 0.1853,
      "step": 23275
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.533708356328197,
      "learning_rate": 4.022375013317811e-06,
      "loss": 0.3531,
      "step": 23276
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.68557554084757,
      "learning_rate": 4.021579876246654e-06,
      "loss": 0.2962,
      "step": 23277
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.25181183503742444,
      "learning_rate": 4.020784797993166e-06,
      "loss": 0.1423,
      "step": 23278
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3417680641243935,
      "learning_rate": 4.0199897785651595e-06,
      "loss": 0.1854,
      "step": 23279
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3366535459516042,
      "learning_rate": 4.019194817970463e-06,
      "loss": 0.2547,
      "step": 23280
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.49873700471368887,
      "learning_rate": 4.018399916216895e-06,
      "loss": 0.2366,
      "step": 23281
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.7716530479325867,
      "learning_rate": 4.01760507331228e-06,
      "loss": 0.7902,
      "step": 23282
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.0528851082605875,
      "learning_rate": 4.0168102892644324e-06,
      "loss": 0.4538,
      "step": 23283
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5378010543593035,
      "learning_rate": 4.016015564081172e-06,
      "loss": 0.3274,
      "step": 23284
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31519135812067905,
      "learning_rate": 4.015220897770326e-06,
      "loss": 0.232,
      "step": 23285
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.31010868794434204,
      "learning_rate": 4.014426290339697e-06,
      "loss": 0.2394,
      "step": 23286
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.226993092867928,
      "learning_rate": 4.013631741797118e-06,
      "loss": 0.5684,
      "step": 23287
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2780487784330612,
      "learning_rate": 4.0128372521503974e-06,
      "loss": 0.0721,
      "step": 23288
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2699155775457047,
      "learning_rate": 4.0120428214073535e-06,
      "loss": 0.2105,
      "step": 23289
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5956252577185128,
      "learning_rate": 4.011248449575807e-06,
      "loss": 0.1668,
      "step": 23290
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.500893398876139,
      "learning_rate": 4.010454136663561e-06,
      "loss": 0.2755,
      "step": 23291
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.44315514947045065,
      "learning_rate": 4.009659882678445e-06,
      "loss": 0.2803,
      "step": 23292
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.40313738000889904,
      "learning_rate": 4.0088656876282626e-06,
      "loss": 0.2328,
      "step": 23293
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.43448471948083917,
      "learning_rate": 4.008071551520831e-06,
      "loss": 0.2543,
      "step": 23294
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8425855639980997,
      "learning_rate": 4.0072774743639675e-06,
      "loss": 0.3925,
      "step": 23295
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.5847792494529969,
      "learning_rate": 4.006483456165475e-06,
      "loss": 0.306,
      "step": 23296
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3170070522443049,
      "learning_rate": 4.005689496933172e-06,
      "loss": 0.2762,
      "step": 23297
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3200572276945496,
      "learning_rate": 4.004895596674867e-06,
      "loss": 0.1925,
      "step": 23298
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.1583650674409883,
      "learning_rate": 4.004101755398375e-06,
      "loss": 0.0718,
      "step": 23299
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.466129244754481,
      "learning_rate": 4.0033079731114995e-06,
      "loss": 0.6973,
      "step": 23300
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7231247278271737,
      "learning_rate": 4.002514249822054e-06,
      "loss": 0.2907,
      "step": 23301
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9363627505425717,
      "learning_rate": 4.0017205855378504e-06,
      "loss": 0.4185,
      "step": 23302
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32547898111960843,
      "learning_rate": 4.000926980266686e-06,
      "loss": 0.2161,
      "step": 23303
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32568546477738486,
      "learning_rate": 4.0001334340163824e-06,
      "loss": 0.2716,
      "step": 23304
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.1705844040512903,
      "learning_rate": 3.999339946794737e-06,
      "loss": 0.2326,
      "step": 23305
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4217545218533683,
      "learning_rate": 3.998546518609561e-06,
      "loss": 0.2089,
      "step": 23306
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26749932764858314,
      "learning_rate": 3.997753149468657e-06,
      "loss": 0.1707,
      "step": 23307
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.47587937417970266,
      "learning_rate": 3.996959839379837e-06,
      "loss": 0.241,
      "step": 23308
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30901982387386806,
      "learning_rate": 3.996166588350898e-06,
      "loss": 0.2467,
      "step": 23309
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.8966941449430341,
      "learning_rate": 3.995373396389648e-06,
      "loss": 0.4682,
      "step": 23310
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.46555522261642546,
      "learning_rate": 3.994580263503893e-06,
      "loss": 0.2055,
      "step": 23311
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3028969946726719,
      "learning_rate": 3.993787189701427e-06,
      "loss": 0.234,
      "step": 23312
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4205295089149417,
      "learning_rate": 3.9929941749900654e-06,
      "loss": 0.7891,
      "step": 23313
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.3763326291135978,
      "learning_rate": 3.9922012193776005e-06,
      "loss": 0.1463,
      "step": 23314
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3587035818311083,
      "learning_rate": 3.991408322871837e-06,
      "loss": 0.3098,
      "step": 23315
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.26010394897486677,
      "learning_rate": 3.990615485480579e-06,
      "loss": 0.1691,
      "step": 23316
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.2500766908603725,
      "learning_rate": 3.989822707211616e-06,
      "loss": 0.1609,
      "step": 23317
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.2586808202176654,
      "learning_rate": 3.989029988072761e-06,
      "loss": 0.2296,
      "step": 23318
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7146769331045956,
      "learning_rate": 3.988237328071805e-06,
      "loss": 0.4003,
      "step": 23319
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3243757654704541,
      "learning_rate": 3.987444727216548e-06,
      "loss": 0.1801,
      "step": 23320
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3467288287182066,
      "learning_rate": 3.98665218551479e-06,
      "loss": 0.2793,
      "step": 23321
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3532241339292436,
      "learning_rate": 3.985859702974324e-06,
      "loss": 0.2298,
      "step": 23322
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.071290595545842,
      "learning_rate": 3.98506727960295e-06,
      "loss": 0.4608,
      "step": 23323
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.19569787571691918,
      "learning_rate": 3.984274915408461e-06,
      "loss": 0.0712,
      "step": 23324
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.37588743547863607,
      "learning_rate": 3.9834826103986614e-06,
      "loss": 0.1751,
      "step": 23325
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3210613931235584,
      "learning_rate": 3.982690364581334e-06,
      "loss": 0.2133,
      "step": 23326
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.32887649650743117,
      "learning_rate": 3.98189817796428e-06,
      "loss": 0.2575,
      "step": 23327
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.7455044838710483,
      "learning_rate": 3.981106050555293e-06,
      "loss": 0.3886,
      "step": 23328
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.45122449875251647,
      "learning_rate": 3.980313982362161e-06,
      "loss": 0.1411,
      "step": 23329
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3231009660859838,
      "learning_rate": 3.979521973392686e-06,
      "loss": 0.256,
      "step": 23330
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.037321169174245,
      "learning_rate": 3.978730023654652e-06,
      "loss": 0.5128,
      "step": 23331
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4963143058383695,
      "learning_rate": 3.977938133155853e-06,
      "loss": 0.3224,
      "step": 23332
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4345568050078745,
      "learning_rate": 3.9771463019040846e-06,
      "loss": 0.2184,
      "step": 23333
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.48911094260390464,
      "learning_rate": 3.9763545299071285e-06,
      "loss": 0.2625,
      "step": 23334
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.20720758962511862,
      "learning_rate": 3.9755628171727776e-06,
      "loss": 0.1135,
      "step": 23335
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.6806113678931589,
      "learning_rate": 3.974771163708824e-06,
      "loss": 0.3637,
      "step": 23336
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.9502758696369301,
      "learning_rate": 3.9739795695230525e-06,
      "loss": 0.2425,
      "step": 23337
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.3473598899887084,
      "learning_rate": 3.9731880346232535e-06,
      "loss": 0.1638,
      "step": 23338
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.30928083139568996,
      "learning_rate": 3.972396559017218e-06,
      "loss": 0.2715,
      "step": 23339
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4284321324232248,
      "learning_rate": 3.971605142712724e-06,
      "loss": 0.2512,
      "step": 23340
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.4303449906928687,
      "learning_rate": 3.970813785717562e-06,
      "loss": 0.518,
      "step": 23341
    },
    {
      "epoch": 0.71,
      "grad_norm": 1.284832033990258,
      "learning_rate": 3.970022488039522e-06,
      "loss": 0.1752,
      "step": 23342
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.39034663161141947,
      "learning_rate": 3.969231249686377e-06,
      "loss": 0.2403,
      "step": 23343
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.4334333557102443,
      "learning_rate": 3.968440070665928e-06,
      "loss": 0.1984,
      "step": 23344
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.5256230540159903,
      "learning_rate": 3.967648950985945e-06,
      "loss": 0.3149,
      "step": 23345
    },
    {
      "epoch": 0.71,
      "grad_norm": 0.36622866695125067,
      "learning_rate": 3.966857890654217e-06,
      "loss": 0.2007,
      "step": 23346
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8279970699630268,
      "learning_rate": 3.966066889678531e-06,
      "loss": 0.5377,
      "step": 23347
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.287215639711715,
      "learning_rate": 3.965275948066659e-06,
      "loss": 0.1918,
      "step": 23348
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.80881763288635,
      "learning_rate": 3.964485065826389e-06,
      "loss": 0.5491,
      "step": 23349
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6094953679061438,
      "learning_rate": 3.9636942429655e-06,
      "loss": 0.0401,
      "step": 23350
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2957769935331069,
      "learning_rate": 3.962903479491777e-06,
      "loss": 0.1985,
      "step": 23351
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.48636171324142846,
      "learning_rate": 3.962112775412994e-06,
      "loss": 0.293,
      "step": 23352
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4698956107700504,
      "learning_rate": 3.961322130736931e-06,
      "loss": 0.2223,
      "step": 23353
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7781606936772609,
      "learning_rate": 3.960531545471371e-06,
      "loss": 0.3736,
      "step": 23354
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.1795759594194569,
      "learning_rate": 3.959741019624082e-06,
      "loss": 0.1027,
      "step": 23355
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4167592831614549,
      "learning_rate": 3.958950553202857e-06,
      "loss": 0.2935,
      "step": 23356
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.297077130563372,
      "learning_rate": 3.9581601462154595e-06,
      "loss": 0.1864,
      "step": 23357
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.454749854839555,
      "learning_rate": 3.9573697986696705e-06,
      "loss": 0.3413,
      "step": 23358
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2559956809776873,
      "learning_rate": 3.9565795105732685e-06,
      "loss": 0.1439,
      "step": 23359
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0131902639340022,
      "learning_rate": 3.955789281934024e-06,
      "loss": 0.4984,
      "step": 23360
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.44850353436955354,
      "learning_rate": 3.954999112759712e-06,
      "loss": 0.1777,
      "step": 23361
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36565605794910916,
      "learning_rate": 3.954209003058108e-06,
      "loss": 0.2674,
      "step": 23362
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30830054443828725,
      "learning_rate": 3.953418952836986e-06,
      "loss": 0.2408,
      "step": 23363
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24626238045259183,
      "learning_rate": 3.9526289621041206e-06,
      "loss": 0.0665,
      "step": 23364
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7906031153226218,
      "learning_rate": 3.951839030867277e-06,
      "loss": 0.4785,
      "step": 23365
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3193079341258246,
      "learning_rate": 3.951049159134233e-06,
      "loss": 0.2044,
      "step": 23366
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.42717746647133187,
      "learning_rate": 3.9502593469127574e-06,
      "loss": 0.2394,
      "step": 23367
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3936959351299414,
      "learning_rate": 3.949469594210623e-06,
      "loss": 0.1779,
      "step": 23368
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35276671999235265,
      "learning_rate": 3.948679901035595e-06,
      "loss": 0.2878,
      "step": 23369
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3637674785256853,
      "learning_rate": 3.947890267395445e-06,
      "loss": 0.1953,
      "step": 23370
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.47856891238352134,
      "learning_rate": 3.947100693297946e-06,
      "loss": 0.3383,
      "step": 23371
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9181092220757724,
      "learning_rate": 3.946311178750856e-06,
      "loss": 0.2705,
      "step": 23372
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4755081921605098,
      "learning_rate": 3.9455217237619534e-06,
      "loss": 0.2492,
      "step": 23373
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27071643645617144,
      "learning_rate": 3.944732328338998e-06,
      "loss": 0.2195,
      "step": 23374
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3453595567258478,
      "learning_rate": 3.9439429924897576e-06,
      "loss": 0.2437,
      "step": 23375
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9098814886164662,
      "learning_rate": 3.943153716222e-06,
      "loss": 0.4403,
      "step": 23376
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40792556602427643,
      "learning_rate": 3.942364499543492e-06,
      "loss": 0.0975,
      "step": 23377
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8731545980395459,
      "learning_rate": 3.9415753424619915e-06,
      "loss": 0.375,
      "step": 23378
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5313199424676904,
      "learning_rate": 3.940786244985266e-06,
      "loss": 0.1795,
      "step": 23379
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37995819086744564,
      "learning_rate": 3.9399972071210835e-06,
      "loss": 0.2716,
      "step": 23380
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30443289831001036,
      "learning_rate": 3.939208228877195e-06,
      "loss": 0.2241,
      "step": 23381
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4636071714292032,
      "learning_rate": 3.938419310261378e-06,
      "loss": 0.6943,
      "step": 23382
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4617788535344533,
      "learning_rate": 3.937630451281382e-06,
      "loss": 0.2105,
      "step": 23383
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4561013054666452,
      "learning_rate": 3.9368416519449726e-06,
      "loss": 0.3049,
      "step": 23384
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23228578630379187,
      "learning_rate": 3.9360529122599135e-06,
      "loss": 0.1417,
      "step": 23385
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4945564831609447,
      "learning_rate": 3.935264232233957e-06,
      "loss": 0.3256,
      "step": 23386
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5600068733195824,
      "learning_rate": 3.934475611874867e-06,
      "loss": 0.2571,
      "step": 23387
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8303784275336766,
      "learning_rate": 3.933687051190403e-06,
      "loss": 0.371,
      "step": 23388
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.31812043161151216,
      "learning_rate": 3.93289855018832e-06,
      "loss": 0.191,
      "step": 23389
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.969313654048659,
      "learning_rate": 3.9321101088763825e-06,
      "loss": 0.1529,
      "step": 23390
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5891459809324655,
      "learning_rate": 3.9313217272623385e-06,
      "loss": 0.8292,
      "step": 23391
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30883626604320363,
      "learning_rate": 3.930533405353947e-06,
      "loss": 0.2431,
      "step": 23392
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37076404309919914,
      "learning_rate": 3.929745143158966e-06,
      "loss": 0.2278,
      "step": 23393
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25998489293012517,
      "learning_rate": 3.928956940685152e-06,
      "loss": 0.1463,
      "step": 23394
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38179825609653584,
      "learning_rate": 3.928168797940256e-06,
      "loss": 0.1817,
      "step": 23395
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9452746587921661,
      "learning_rate": 3.927380714932032e-06,
      "loss": 0.312,
      "step": 23396
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5396256436688465,
      "learning_rate": 3.9265926916682375e-06,
      "loss": 0.2078,
      "step": 23397
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28882560089958237,
      "learning_rate": 3.9258047281566155e-06,
      "loss": 0.2036,
      "step": 23398
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.49432358588160447,
      "learning_rate": 3.925016824404934e-06,
      "loss": 0.3392,
      "step": 23399
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.025881505817231,
      "learning_rate": 3.924228980420931e-06,
      "loss": 0.3294,
      "step": 23400
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3626458219371735,
      "learning_rate": 3.923441196212363e-06,
      "loss": 0.8417,
      "step": 23401
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28023481324729416,
      "learning_rate": 3.9226534717869834e-06,
      "loss": 0.1809,
      "step": 23402
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2344052432547766,
      "learning_rate": 3.921865807152535e-06,
      "loss": 0.0672,
      "step": 23403
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4451356283501317,
      "learning_rate": 3.921078202316771e-06,
      "loss": 0.3045,
      "step": 23404
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.40600984116776107,
      "learning_rate": 3.920290657287439e-06,
      "loss": 0.2082,
      "step": 23405
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6800329548866172,
      "learning_rate": 3.919503172072292e-06,
      "loss": 0.3039,
      "step": 23406
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.32481426530419866,
      "learning_rate": 3.918715746679067e-06,
      "loss": 0.2011,
      "step": 23407
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4977378241790689,
      "learning_rate": 3.917928381115524e-06,
      "loss": 0.5203,
      "step": 23408
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1445861369563237,
      "learning_rate": 3.9171410753894e-06,
      "loss": 0.3888,
      "step": 23409
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3242042160353666,
      "learning_rate": 3.916353829508444e-06,
      "loss": 0.2929,
      "step": 23410
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25354584090882953,
      "learning_rate": 3.915566643480403e-06,
      "loss": 0.0698,
      "step": 23411
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35529770616811673,
      "learning_rate": 3.914779517313016e-06,
      "loss": 0.2638,
      "step": 23412
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5084591481237003,
      "learning_rate": 3.913992451014031e-06,
      "loss": 0.2879,
      "step": 23413
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4225176359980384,
      "learning_rate": 3.913205444591191e-06,
      "loss": 0.2185,
      "step": 23414
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21172241744303233,
      "learning_rate": 3.912418498052238e-06,
      "loss": 0.0713,
      "step": 23415
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25736228340037975,
      "learning_rate": 3.911631611404919e-06,
      "loss": 0.1892,
      "step": 23416
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5074373919198613,
      "learning_rate": 3.910844784656968e-06,
      "loss": 0.3622,
      "step": 23417
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9405430240244355,
      "learning_rate": 3.910058017816129e-06,
      "loss": 0.3025,
      "step": 23418
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4312839910487485,
      "learning_rate": 3.909271310890142e-06,
      "loss": 0.8292,
      "step": 23419
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35251008032131964,
      "learning_rate": 3.908484663886754e-06,
      "loss": 0.1667,
      "step": 23420
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4761369503936596,
      "learning_rate": 3.907698076813694e-06,
      "loss": 0.3125,
      "step": 23421
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3871504799113047,
      "learning_rate": 3.906911549678705e-06,
      "loss": 0.25,
      "step": 23422
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43580602994307066,
      "learning_rate": 3.906125082489527e-06,
      "loss": 0.3306,
      "step": 23423
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18216801971976707,
      "learning_rate": 3.90533867525389e-06,
      "loss": 0.0745,
      "step": 23424
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3402599723798319,
      "learning_rate": 3.904552327979543e-06,
      "loss": 0.2688,
      "step": 23425
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0981891312185332,
      "learning_rate": 3.903766040674213e-06,
      "loss": 0.1231,
      "step": 23426
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.111812756103485,
      "learning_rate": 3.902979813345639e-06,
      "loss": 0.6319,
      "step": 23427
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2802290640414107,
      "learning_rate": 3.902193646001557e-06,
      "loss": 0.2268,
      "step": 23428
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27361896009278114,
      "learning_rate": 3.9014075386496985e-06,
      "loss": 0.1788,
      "step": 23429
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9708676000195279,
      "learning_rate": 3.9006214912978e-06,
      "loss": 0.5162,
      "step": 23430
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6411651557896275,
      "learning_rate": 3.899835503953593e-06,
      "loss": 0.2774,
      "step": 23431
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4145148502882435,
      "learning_rate": 3.899049576624816e-06,
      "loss": 0.212,
      "step": 23432
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19734233856343225,
      "learning_rate": 3.8982637093191925e-06,
      "loss": 0.0744,
      "step": 23433
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5532012444455019,
      "learning_rate": 3.897477902044458e-06,
      "loss": 0.3918,
      "step": 23434
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26791698715372,
      "learning_rate": 3.896692154808348e-06,
      "loss": 0.2147,
      "step": 23435
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4260891613992253,
      "learning_rate": 3.895906467618582e-06,
      "loss": 0.7489,
      "step": 23436
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8031709127376637,
      "learning_rate": 3.895120840482903e-06,
      "loss": 0.3069,
      "step": 23437
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8535823448667154,
      "learning_rate": 3.894335273409031e-06,
      "loss": 0.4122,
      "step": 23438
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2959460200432666,
      "learning_rate": 3.893549766404698e-06,
      "loss": 0.1985,
      "step": 23439
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3375790135702705,
      "learning_rate": 3.892764319477631e-06,
      "loss": 0.2761,
      "step": 23440
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5781719566237142,
      "learning_rate": 3.891978932635558e-06,
      "loss": 0.2062,
      "step": 23441
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.15711982339225236,
      "learning_rate": 3.891193605886211e-06,
      "loss": 0.0714,
      "step": 23442
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4086359805619151,
      "learning_rate": 3.890408339237307e-06,
      "loss": 0.2595,
      "step": 23443
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4629303519238014,
      "learning_rate": 3.889623132696576e-06,
      "loss": 0.2151,
      "step": 23444
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4706868271874047,
      "learning_rate": 3.888837986271743e-06,
      "loss": 0.8735,
      "step": 23445
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2867324504225926,
      "learning_rate": 3.888052899970536e-06,
      "loss": 0.2199,
      "step": 23446
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5681472975925164,
      "learning_rate": 3.887267873800673e-06,
      "loss": 0.3284,
      "step": 23447
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3391462796839173,
      "learning_rate": 3.886482907769879e-06,
      "loss": 0.237,
      "step": 23448
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.5413193042002544,
      "learning_rate": 3.885698001885881e-06,
      "loss": 0.7784,
      "step": 23449
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4370791720034877,
      "learning_rate": 3.8849131561563925e-06,
      "loss": 0.1322,
      "step": 23450
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.239060348292042,
      "learning_rate": 3.884128370589146e-06,
      "loss": 0.2036,
      "step": 23451
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26457986813526,
      "learning_rate": 3.883343645191855e-06,
      "loss": 0.1777,
      "step": 23452
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.48752362034797175,
      "learning_rate": 3.882558979972241e-06,
      "loss": 0.3388,
      "step": 23453
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9020648235105145,
      "learning_rate": 3.881774374938029e-06,
      "loss": 0.4138,
      "step": 23454
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.677394589395832,
      "learning_rate": 3.880989830096928e-06,
      "loss": 0.2929,
      "step": 23455
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6066107319243904,
      "learning_rate": 3.880205345456664e-06,
      "loss": 0.3397,
      "step": 23456
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36377917967286594,
      "learning_rate": 3.879420921024955e-06,
      "loss": 0.2102,
      "step": 23457
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3826586276917292,
      "learning_rate": 3.878636556809518e-06,
      "loss": 0.2929,
      "step": 23458
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2163043433814131,
      "learning_rate": 3.877852252818066e-06,
      "loss": 0.1748,
      "step": 23459
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.425624088053713,
      "learning_rate": 3.877068009058319e-06,
      "loss": 0.231,
      "step": 23460
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3479083892632899,
      "learning_rate": 3.876283825537994e-06,
      "loss": 0.1606,
      "step": 23461
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3119779411489312,
      "learning_rate": 3.875499702264798e-06,
      "loss": 0.2014,
      "step": 23462
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43063293931001145,
      "learning_rate": 3.874715639246457e-06,
      "loss": 0.2648,
      "step": 23463
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4874746849417513,
      "learning_rate": 3.8739316364906755e-06,
      "loss": 0.3234,
      "step": 23464
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4823944098924528,
      "learning_rate": 3.87314769400517e-06,
      "loss": 0.1526,
      "step": 23465
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.359772761906292,
      "learning_rate": 3.872363811797658e-06,
      "loss": 0.2837,
      "step": 23466
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8044959742262303,
      "learning_rate": 3.8715799898758385e-06,
      "loss": 0.4223,
      "step": 23467
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6786484283964573,
      "learning_rate": 3.87079622824744e-06,
      "loss": 0.5947,
      "step": 23468
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24758724725102618,
      "learning_rate": 3.870012526920162e-06,
      "loss": 0.1408,
      "step": 23469
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24780878856596905,
      "learning_rate": 3.869228885901717e-06,
      "loss": 0.1806,
      "step": 23470
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3399318037708641,
      "learning_rate": 3.868445305199818e-06,
      "loss": 0.2263,
      "step": 23471
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7442202098739076,
      "learning_rate": 3.86766178482217e-06,
      "loss": 0.2649,
      "step": 23472
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7144997824869315,
      "learning_rate": 3.866878324776482e-06,
      "loss": 0.4484,
      "step": 23473
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27330413697216976,
      "learning_rate": 3.866094925070462e-06,
      "loss": 0.1529,
      "step": 23474
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3655327101633222,
      "learning_rate": 3.865311585711824e-06,
      "loss": 0.2732,
      "step": 23475
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.29481400694345306,
      "learning_rate": 3.864528306708262e-06,
      "loss": 0.2306,
      "step": 23476
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2443261525111697,
      "learning_rate": 3.863745088067495e-06,
      "loss": 0.4466,
      "step": 23477
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2093850433017133,
      "learning_rate": 3.862961929797221e-06,
      "loss": 0.066,
      "step": 23478
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3902059327819561,
      "learning_rate": 3.862178831905145e-06,
      "loss": 0.2462,
      "step": 23479
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24996417771313953,
      "learning_rate": 3.8613957943989785e-06,
      "loss": 0.1363,
      "step": 23480
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9857982186482233,
      "learning_rate": 3.860612817286417e-06,
      "loss": 0.3982,
      "step": 23481
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4083222265768388,
      "learning_rate": 3.859829900575165e-06,
      "loss": 0.2399,
      "step": 23482
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3330784670419039,
      "learning_rate": 3.859047044272927e-06,
      "loss": 0.158,
      "step": 23483
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.648111527325604,
      "learning_rate": 3.85826424838741e-06,
      "loss": 0.3062,
      "step": 23484
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46623262005918376,
      "learning_rate": 3.857481512926306e-06,
      "loss": 0.2252,
      "step": 23485
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4657307433343754,
      "learning_rate": 3.856698837897319e-06,
      "loss": 0.5183,
      "step": 23486
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.219175306338189,
      "learning_rate": 3.855916223308154e-06,
      "loss": 0.1718,
      "step": 23487
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9042938228196942,
      "learning_rate": 3.855133669166499e-06,
      "loss": 0.3935,
      "step": 23488
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3716777062906162,
      "learning_rate": 3.85435117548007e-06,
      "loss": 0.213,
      "step": 23489
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0122536320505997,
      "learning_rate": 3.853568742256553e-06,
      "loss": 0.3935,
      "step": 23490
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2771800520998947,
      "learning_rate": 3.852786369503647e-06,
      "loss": 0.1231,
      "step": 23491
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8885388941644163,
      "learning_rate": 3.852004057229056e-06,
      "loss": 0.7542,
      "step": 23492
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26763076861234636,
      "learning_rate": 3.851221805440465e-06,
      "loss": 0.1745,
      "step": 23493
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3449405717874563,
      "learning_rate": 3.850439614145585e-06,
      "loss": 0.2779,
      "step": 23494
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.473497589658232,
      "learning_rate": 3.849657483352099e-06,
      "loss": 0.1752,
      "step": 23495
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1923031493344651,
      "learning_rate": 3.848875413067706e-06,
      "loss": 0.142,
      "step": 23496
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5774010594097071,
      "learning_rate": 3.848093403300105e-06,
      "loss": 0.32,
      "step": 23497
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37763358954147236,
      "learning_rate": 3.8473114540569814e-06,
      "loss": 0.2267,
      "step": 23498
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34150603421526426,
      "learning_rate": 3.846529565346032e-06,
      "loss": 0.2736,
      "step": 23499
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21514303271801663,
      "learning_rate": 3.8457477371749506e-06,
      "loss": 0.0715,
      "step": 23500
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.471746568593255,
      "learning_rate": 3.8449659695514305e-06,
      "loss": 0.817,
      "step": 23501
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28690514386814925,
      "learning_rate": 3.844184262483157e-06,
      "loss": 0.1855,
      "step": 23502
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6449324060038384,
      "learning_rate": 3.843402615977825e-06,
      "loss": 0.8272,
      "step": 23503
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3251078888596663,
      "learning_rate": 3.842621030043126e-06,
      "loss": 0.145,
      "step": 23504
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3240123722357196,
      "learning_rate": 3.841839504686742e-06,
      "loss": 0.2661,
      "step": 23505
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3265422153604051,
      "learning_rate": 3.841058039916373e-06,
      "loss": 0.1778,
      "step": 23506
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4619088787782548,
      "learning_rate": 3.840276635739698e-06,
      "loss": 0.3257,
      "step": 23507
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6101889590452476,
      "learning_rate": 3.839495292164409e-06,
      "loss": 0.2643,
      "step": 23508
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3139879695736419,
      "learning_rate": 3.838714009198193e-06,
      "loss": 0.1292,
      "step": 23509
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8535564144678778,
      "learning_rate": 3.837932786848738e-06,
      "loss": 0.4331,
      "step": 23510
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35270537490096476,
      "learning_rate": 3.837151625123725e-06,
      "loss": 0.1567,
      "step": 23511
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.29253529936266404,
      "learning_rate": 3.836370524030843e-06,
      "loss": 0.2542,
      "step": 23512
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.21430437466809046,
      "learning_rate": 3.8355894835777775e-06,
      "loss": 0.0716,
      "step": 23513
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6658883336086073,
      "learning_rate": 3.8348085037722045e-06,
      "loss": 0.3886,
      "step": 23514
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7478312049954733,
      "learning_rate": 3.834027584621822e-06,
      "loss": 0.2147,
      "step": 23515
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35575164752492355,
      "learning_rate": 3.833246726134301e-06,
      "loss": 0.2505,
      "step": 23516
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.45799877011650597,
      "learning_rate": 3.832465928317328e-06,
      "loss": 0.205,
      "step": 23517
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2800507605758967,
      "learning_rate": 3.831685191178588e-06,
      "loss": 0.2076,
      "step": 23518
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.013176370111778,
      "learning_rate": 3.8309045147257515e-06,
      "loss": 0.4548,
      "step": 23519
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5744864768728604,
      "learning_rate": 3.830123898966514e-06,
      "loss": 0.2763,
      "step": 23520
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25435112400665844,
      "learning_rate": 3.829343343908544e-06,
      "loss": 0.162,
      "step": 23521
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.48257309167480195,
      "learning_rate": 3.828562849559524e-06,
      "loss": 0.2419,
      "step": 23522
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5216613897724685,
      "learning_rate": 3.827782415927137e-06,
      "loss": 0.3555,
      "step": 23523
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5630001095367823,
      "learning_rate": 3.827002043019054e-06,
      "loss": 0.2163,
      "step": 23524
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38934503464007575,
      "learning_rate": 3.826221730842956e-06,
      "loss": 0.2614,
      "step": 23525
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.193570038757558,
      "learning_rate": 3.825441479406521e-06,
      "loss": 0.0767,
      "step": 23526
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7564807394960937,
      "learning_rate": 3.824661288717427e-06,
      "loss": 0.6051,
      "step": 23527
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4090980505064038,
      "learning_rate": 3.823881158783343e-06,
      "loss": 0.2606,
      "step": 23528
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38327889891278216,
      "learning_rate": 3.82310108961195e-06,
      "loss": 0.2555,
      "step": 23529
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24310729945740556,
      "learning_rate": 3.8223210812109234e-06,
      "loss": 0.1727,
      "step": 23530
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4787257216612674,
      "learning_rate": 3.821541133587928e-06,
      "loss": 0.2415,
      "step": 23531
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7185044349865611,
      "learning_rate": 3.820761246750652e-06,
      "loss": 0.2976,
      "step": 23532
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8183693965576289,
      "learning_rate": 3.819981420706756e-06,
      "loss": 0.3713,
      "step": 23533
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3088908845924756,
      "learning_rate": 3.819201655463914e-06,
      "loss": 0.1933,
      "step": 23534
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4566945277009361,
      "learning_rate": 3.818421951029806e-06,
      "loss": 0.2301,
      "step": 23535
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.488307447892675,
      "learning_rate": 3.8176423074120925e-06,
      "loss": 0.3288,
      "step": 23536
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5157978775725521,
      "learning_rate": 3.816862724618449e-06,
      "loss": 0.2642,
      "step": 23537
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.43747178170671,
      "learning_rate": 3.816083202656543e-06,
      "loss": 0.2382,
      "step": 23538
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20180903331101346,
      "learning_rate": 3.81530374153405e-06,
      "loss": 0.1178,
      "step": 23539
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8524438965068608,
      "learning_rate": 3.8145243412586264e-06,
      "loss": 0.3774,
      "step": 23540
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37491461043067337,
      "learning_rate": 3.813745001837955e-06,
      "loss": 0.2483,
      "step": 23541
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6090522437754884,
      "learning_rate": 3.812965723279691e-06,
      "loss": 0.6823,
      "step": 23542
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27625863519355925,
      "learning_rate": 3.8121865055915074e-06,
      "loss": 0.1825,
      "step": 23543
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0990216094766194,
      "learning_rate": 3.811407348781072e-06,
      "loss": 0.4123,
      "step": 23544
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9294289325853475,
      "learning_rate": 3.810628252856041e-06,
      "loss": 0.5145,
      "step": 23545
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7368624956348686,
      "learning_rate": 3.809849217824092e-06,
      "loss": 0.7813,
      "step": 23546
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.252051058474677,
      "learning_rate": 3.8090702436928804e-06,
      "loss": 0.2062,
      "step": 23547
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2453784479500954,
      "learning_rate": 3.808291330470072e-06,
      "loss": 0.1606,
      "step": 23548
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0637712311543275,
      "learning_rate": 3.8075124781633356e-06,
      "loss": 0.3778,
      "step": 23549
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4577185486331678,
      "learning_rate": 3.806733686780324e-06,
      "loss": 0.1419,
      "step": 23550
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.399336195928118,
      "learning_rate": 3.8059549563287057e-06,
      "loss": 0.445,
      "step": 23551
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2657872354146772,
      "learning_rate": 3.80517628681614e-06,
      "loss": 0.17,
      "step": 23552
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33392187943745566,
      "learning_rate": 3.8043976782502912e-06,
      "loss": 0.2563,
      "step": 23553
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7145312664222827,
      "learning_rate": 3.8036191306388136e-06,
      "loss": 0.3218,
      "step": 23554
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2390829142015105,
      "learning_rate": 3.8028406439893705e-06,
      "loss": 0.7703,
      "step": 23555
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3340248670433016,
      "learning_rate": 3.8020622183096235e-06,
      "loss": 0.1593,
      "step": 23556
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7451554910334385,
      "learning_rate": 3.8012838536072207e-06,
      "loss": 0.3222,
      "step": 23557
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2828453538965572,
      "learning_rate": 3.8005055498898335e-06,
      "loss": 0.2851,
      "step": 23558
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33441690419559483,
      "learning_rate": 3.79972730716511e-06,
      "loss": 0.2811,
      "step": 23559
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.16290849712771552,
      "learning_rate": 3.7989491254407087e-06,
      "loss": 0.0695,
      "step": 23560
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.31222578096792447,
      "learning_rate": 3.7981710047242904e-06,
      "loss": 0.1847,
      "step": 23561
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.6178166740951558,
      "learning_rate": 3.7973929450235026e-06,
      "loss": 0.7969,
      "step": 23562
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.003230621243744,
      "learning_rate": 3.7966149463460032e-06,
      "loss": 0.5291,
      "step": 23563
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34155000948446645,
      "learning_rate": 3.7958370086994478e-06,
      "loss": 0.2877,
      "step": 23564
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9810466474384496,
      "learning_rate": 3.795059132091493e-06,
      "loss": 0.2085,
      "step": 23565
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35289138116382496,
      "learning_rate": 3.794281316529784e-06,
      "loss": 0.2496,
      "step": 23566
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3677547671331565,
      "learning_rate": 3.7935035620219786e-06,
      "loss": 0.3042,
      "step": 23567
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.25839090878738397,
      "learning_rate": 3.792725868575726e-06,
      "loss": 0.1329,
      "step": 23568
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4346398727553473,
      "learning_rate": 3.79194823619868e-06,
      "loss": 0.1448,
      "step": 23569
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.39405000195842094,
      "learning_rate": 3.7911706648984924e-06,
      "loss": 0.2497,
      "step": 23570
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26044797587425805,
      "learning_rate": 3.7903931546828086e-06,
      "loss": 0.2183,
      "step": 23571
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.471992748814273,
      "learning_rate": 3.7896157055592796e-06,
      "loss": 0.7277,
      "step": 23572
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.005490056110411,
      "learning_rate": 3.788838317535558e-06,
      "loss": 0.4638,
      "step": 23573
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0658627780240333,
      "learning_rate": 3.788060990619281e-06,
      "loss": 0.2122,
      "step": 23574
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.49222720502041845,
      "learning_rate": 3.7872837248181127e-06,
      "loss": 0.2627,
      "step": 23575
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.526546667880047,
      "learning_rate": 3.7865065201396868e-06,
      "loss": 0.2272,
      "step": 23576
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4525084357892464,
      "learning_rate": 3.7857293765916536e-06,
      "loss": 0.2915,
      "step": 23577
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.18449744151356282,
      "learning_rate": 3.7849522941816608e-06,
      "loss": 0.0745,
      "step": 23578
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.353794929390207,
      "learning_rate": 3.7841752729173555e-06,
      "loss": 0.2385,
      "step": 23579
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46944695023616945,
      "learning_rate": 3.7833983128063757e-06,
      "loss": 0.2172,
      "step": 23580
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.7244463033165054,
      "learning_rate": 3.782621413856369e-06,
      "loss": 0.7483,
      "step": 23581
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3009855901657344,
      "learning_rate": 3.7818445760749813e-06,
      "loss": 0.2502,
      "step": 23582
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5751074684902961,
      "learning_rate": 3.7810677994698465e-06,
      "loss": 0.3227,
      "step": 23583
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3187609928479136,
      "learning_rate": 3.7802910840486196e-06,
      "loss": 0.2127,
      "step": 23584
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.8505971292939631,
      "learning_rate": 3.7795144298189323e-06,
      "loss": 0.8314,
      "step": 23585
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.26602940070317177,
      "learning_rate": 3.778737836788429e-06,
      "loss": 0.0703,
      "step": 23586
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4363013615255023,
      "learning_rate": 3.777961304964752e-06,
      "loss": 0.0741,
      "step": 23587
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3324336243608915,
      "learning_rate": 3.7771848343555373e-06,
      "loss": 0.2417,
      "step": 23588
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28856690490512427,
      "learning_rate": 3.776408424968425e-06,
      "loss": 0.2223,
      "step": 23589
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4506734453273025,
      "learning_rate": 3.775632076811054e-06,
      "loss": 0.8313,
      "step": 23590
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5964569582101434,
      "learning_rate": 3.7748557898910644e-06,
      "loss": 0.2818,
      "step": 23591
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8218072835541296,
      "learning_rate": 3.774079564216094e-06,
      "loss": 0.351,
      "step": 23592
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2827543376314151,
      "learning_rate": 3.773303399793774e-06,
      "loss": 0.1855,
      "step": 23593
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.3132514965239654,
      "learning_rate": 3.772527296631744e-06,
      "loss": 0.5848,
      "step": 23594
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2895715018162031,
      "learning_rate": 3.77175125473764e-06,
      "loss": 0.2069,
      "step": 23595
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24011415343379336,
      "learning_rate": 3.7709752741191e-06,
      "loss": 0.1443,
      "step": 23596
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.19445038688954294,
      "learning_rate": 3.770199354783751e-06,
      "loss": 0.0686,
      "step": 23597
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38384308016556296,
      "learning_rate": 3.7694234967392306e-06,
      "loss": 0.2667,
      "step": 23598
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.667785692129204,
      "learning_rate": 3.7686476999931763e-06,
      "loss": 0.3266,
      "step": 23599
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4078892118658624,
      "learning_rate": 3.767871964553208e-06,
      "loss": 0.2266,
      "step": 23600
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3962708355640972,
      "learning_rate": 3.767096290426974e-06,
      "loss": 0.299,
      "step": 23601
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.30678821205004436,
      "learning_rate": 3.766320677622094e-06,
      "loss": 0.1961,
      "step": 23602
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.211548345119367,
      "learning_rate": 3.765545126146202e-06,
      "loss": 0.497,
      "step": 23603
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.654815018102703,
      "learning_rate": 3.764769636006932e-06,
      "loss": 0.2077,
      "step": 23604
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2650076891685019,
      "learning_rate": 3.763994207211906e-06,
      "loss": 0.1397,
      "step": 23605
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.27333998656987807,
      "learning_rate": 3.7632188397687574e-06,
      "loss": 0.185,
      "step": 23606
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.35545351460270347,
      "learning_rate": 3.762443533685114e-06,
      "loss": 0.2908,
      "step": 23607
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8865432589643178,
      "learning_rate": 3.7616682889686072e-06,
      "loss": 0.293,
      "step": 23608
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7907460041592554,
      "learning_rate": 3.7608931056268537e-06,
      "loss": 0.3936,
      "step": 23609
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.395435180904283,
      "learning_rate": 3.760117983667494e-06,
      "loss": 0.2523,
      "step": 23610
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34806937211975725,
      "learning_rate": 3.7593429230981416e-06,
      "loss": 0.2571,
      "step": 23611
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6225061149039224,
      "learning_rate": 3.7585679239264284e-06,
      "loss": 0.0223,
      "step": 23612
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3159746015463188,
      "learning_rate": 3.7577929861599804e-06,
      "loss": 0.2148,
      "step": 23613
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33398730482416084,
      "learning_rate": 3.757018109806416e-06,
      "loss": 0.1768,
      "step": 23614
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.34912069732559003,
      "learning_rate": 3.756243294873362e-06,
      "loss": 0.154,
      "step": 23615
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.564951035928054,
      "learning_rate": 3.75546854136844e-06,
      "loss": 0.3613,
      "step": 23616
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0974786233967417,
      "learning_rate": 3.7546938492992735e-06,
      "loss": 0.3185,
      "step": 23617
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.33497792336479393,
      "learning_rate": 3.7539192186734884e-06,
      "loss": 0.2907,
      "step": 23618
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.394035568672946,
      "learning_rate": 3.7531446494986968e-06,
      "loss": 0.2023,
      "step": 23619
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4032404272353337,
      "learning_rate": 3.752370141782523e-06,
      "loss": 0.2588,
      "step": 23620
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4295085199830431,
      "learning_rate": 3.7515956955325884e-06,
      "loss": 0.2042,
      "step": 23621
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1915736886053203,
      "learning_rate": 3.750821310756515e-06,
      "loss": 0.4848,
      "step": 23622
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2681464214859015,
      "learning_rate": 3.7500469874619137e-06,
      "loss": 0.1599,
      "step": 23623
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4158071155039463,
      "learning_rate": 3.7492727256564076e-06,
      "loss": 0.2781,
      "step": 23624
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.220745818175614,
      "learning_rate": 3.748498525347616e-06,
      "loss": 0.1534,
      "step": 23625
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.1415237218456704,
      "learning_rate": 3.7477243865431458e-06,
      "loss": 0.2879,
      "step": 23626
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4384941582873543,
      "learning_rate": 3.7469503092506277e-06,
      "loss": 0.7477,
      "step": 23627
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.9457419341237812,
      "learning_rate": 3.7461762934776657e-06,
      "loss": 0.4239,
      "step": 23628
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3943490265671187,
      "learning_rate": 3.7454023392318794e-06,
      "loss": 0.2345,
      "step": 23629
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.24900162805010134,
      "learning_rate": 3.744628446520887e-06,
      "loss": 0.2089,
      "step": 23630
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.850812451323234,
      "learning_rate": 3.7438546153522946e-06,
      "loss": 0.9072,
      "step": 23631
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3565526251247163,
      "learning_rate": 3.7430808457337187e-06,
      "loss": 0.0827,
      "step": 23632
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.48833459265297563,
      "learning_rate": 3.7423071376727725e-06,
      "loss": 0.2551,
      "step": 23633
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.4038501280037945,
      "learning_rate": 3.7415334911770707e-06,
      "loss": 0.2476,
      "step": 23634
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.0510807219337472,
      "learning_rate": 3.740759906254219e-06,
      "loss": 0.4892,
      "step": 23635
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2598014779350755,
      "learning_rate": 3.73998638291183e-06,
      "loss": 0.2048,
      "step": 23636
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.2642027624356647,
      "learning_rate": 3.739212921157519e-06,
      "loss": 0.7096,
      "step": 23637
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.28358130580146645,
      "learning_rate": 3.7384395209988842e-06,
      "loss": 0.1692,
      "step": 23638
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4017330771259966,
      "learning_rate": 3.737666182443549e-06,
      "loss": 0.18,
      "step": 23639
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4826305601416505,
      "learning_rate": 3.736892905499111e-06,
      "loss": 0.6264,
      "step": 23640
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.36718249932360536,
      "learning_rate": 3.736119690173181e-06,
      "loss": 0.2701,
      "step": 23641
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37012682063171976,
      "learning_rate": 3.7353465364733697e-06,
      "loss": 0.2603,
      "step": 23642
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3410055066844776,
      "learning_rate": 3.7345734444072746e-06,
      "loss": 0.1973,
      "step": 23643
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.725361260977173,
      "learning_rate": 3.7338004139825136e-06,
      "loss": 0.4071,
      "step": 23644
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2631300838861128,
      "learning_rate": 3.733027445206683e-06,
      "loss": 0.143,
      "step": 23645
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.350936241309666,
      "learning_rate": 3.73225453808739e-06,
      "loss": 0.2052,
      "step": 23646
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37967398510936723,
      "learning_rate": 3.7314816926322396e-06,
      "loss": 0.1751,
      "step": 23647
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3557882718855849,
      "learning_rate": 3.730708908848838e-06,
      "loss": 0.2845,
      "step": 23648
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.45652301396957223,
      "learning_rate": 3.7299361867447823e-06,
      "loss": 0.2099,
      "step": 23649
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7318343664312109,
      "learning_rate": 3.729163526327677e-06,
      "loss": 0.3919,
      "step": 23650
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5595564638742883,
      "learning_rate": 3.728390927605128e-06,
      "loss": 0.206,
      "step": 23651
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3469283560616812,
      "learning_rate": 3.727618390584725e-06,
      "loss": 0.2601,
      "step": 23652
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.7072541634534227,
      "learning_rate": 3.7268459152740845e-06,
      "loss": 0.2697,
      "step": 23653
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3059999384232405,
      "learning_rate": 3.7260735016807935e-06,
      "loss": 0.2254,
      "step": 23654
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.378422681324349,
      "learning_rate": 3.7253011498124558e-06,
      "loss": 0.8414,
      "step": 23655
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.20564856219966907,
      "learning_rate": 3.7245288596766737e-06,
      "loss": 0.1132,
      "step": 23656
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.37313450219716987,
      "learning_rate": 3.7237566312810393e-06,
      "loss": 0.2301,
      "step": 23657
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.4012847881514647,
      "learning_rate": 3.7229844646331505e-06,
      "loss": 0.2368,
      "step": 23658
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.6479643126347815,
      "learning_rate": 3.722212359740607e-06,
      "loss": 0.3772,
      "step": 23659
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.2911628220024661,
      "learning_rate": 3.721440316611007e-06,
      "loss": 0.2114,
      "step": 23660
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3605749355306128,
      "learning_rate": 3.72066833525194e-06,
      "loss": 0.2897,
      "step": 23661
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.8924713784986555,
      "learning_rate": 3.719896415671004e-06,
      "loss": 0.0584,
      "step": 23662
    },
    {
      "epoch": 0.72,
      "grad_norm": 1.511348095967301,
      "learning_rate": 3.719124557875796e-06,
      "loss": 0.7714,
      "step": 23663
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.16211954530089265,
      "learning_rate": 3.718352761873901e-06,
      "loss": 0.0954,
      "step": 23664
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3807787561138092,
      "learning_rate": 3.7175810276729263e-06,
      "loss": 0.2522,
      "step": 23665
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.32701138506789784,
      "learning_rate": 3.7168093552804516e-06,
      "loss": 0.2261,
      "step": 23666
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.38280358281462395,
      "learning_rate": 3.716037744704075e-06,
      "loss": 0.2179,
      "step": 23667
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5651037523179254,
      "learning_rate": 3.71526619595139e-06,
      "loss": 0.3751,
      "step": 23668
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.41501281507784654,
      "learning_rate": 3.7144947090299777e-06,
      "loss": 0.0766,
      "step": 23669
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.5731936654821032,
      "learning_rate": 3.7137232839474402e-06,
      "loss": 0.3683,
      "step": 23670
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.46152278638761285,
      "learning_rate": 3.7129519207113594e-06,
      "loss": 0.2026,
      "step": 23671
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.3038884453757449,
      "learning_rate": 3.7121806193293255e-06,
      "loss": 0.2735,
      "step": 23672
    },
    {
      "epoch": 0.72,
      "grad_norm": 0.23465166728646517,
      "learning_rate": 3.7114093798089313e-06,
      "loss": 0.1226,
      "step": 23673
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9398766869086982,
      "learning_rate": 3.710638202157757e-06,
      "loss": 0.4124,
      "step": 23674
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2327137382962464,
      "learning_rate": 3.7098670863833943e-06,
      "loss": 0.1545,
      "step": 23675
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6737571403850648,
      "learning_rate": 3.7090960324934287e-06,
      "loss": 0.4487,
      "step": 23676
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31110895886031353,
      "learning_rate": 3.70832504049545e-06,
      "loss": 0.2424,
      "step": 23677
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.8026934550731961,
      "learning_rate": 3.7075541103970324e-06,
      "loss": 0.7475,
      "step": 23678
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27227853234402716,
      "learning_rate": 3.7067832422057746e-06,
      "loss": 0.1725,
      "step": 23679
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2436132997141423,
      "learning_rate": 3.7060124359292515e-06,
      "loss": 0.2654,
      "step": 23680
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.47438140393696,
      "learning_rate": 3.705241691575048e-06,
      "loss": 0.8174,
      "step": 23681
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31808207696449836,
      "learning_rate": 3.704471009150752e-06,
      "loss": 0.1086,
      "step": 23682
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.384359300659401,
      "learning_rate": 3.703700388663938e-06,
      "loss": 0.229,
      "step": 23683
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21043963740013003,
      "learning_rate": 3.702929830122192e-06,
      "loss": 0.1781,
      "step": 23684
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7247593228588729,
      "learning_rate": 3.702159333533093e-06,
      "loss": 0.3938,
      "step": 23685
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5900289234750902,
      "learning_rate": 3.701388898904228e-06,
      "loss": 0.2474,
      "step": 23686
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1303928161802523,
      "learning_rate": 3.700618526243167e-06,
      "loss": 0.3318,
      "step": 23687
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2708297495421638,
      "learning_rate": 3.699848215557493e-06,
      "loss": 0.1743,
      "step": 23688
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.655108314249325,
      "learning_rate": 3.6990779668547903e-06,
      "loss": 0.7503,
      "step": 23689
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30615683156122203,
      "learning_rate": 3.698307780142624e-06,
      "loss": 0.2313,
      "step": 23690
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3765211102455515,
      "learning_rate": 3.697537655428587e-06,
      "loss": 0.7227,
      "step": 23691
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.445797653978132,
      "learning_rate": 3.696767592720244e-06,
      "loss": 0.1608,
      "step": 23692
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3472138826013803,
      "learning_rate": 3.6959975920251755e-06,
      "loss": 0.2027,
      "step": 23693
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7270920187732997,
      "learning_rate": 3.6952276533509614e-06,
      "loss": 0.3501,
      "step": 23694
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18292887371504415,
      "learning_rate": 3.6944577767051635e-06,
      "loss": 0.148,
      "step": 23695
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8870007435465831,
      "learning_rate": 3.6936879620953734e-06,
      "loss": 0.4001,
      "step": 23696
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4330211872975687,
      "learning_rate": 3.6929182095291518e-06,
      "loss": 0.1787,
      "step": 23697
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5324458720493577,
      "learning_rate": 3.692148519014076e-06,
      "loss": 0.3218,
      "step": 23698
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9817978575289323,
      "learning_rate": 3.691378890557723e-06,
      "loss": 0.5739,
      "step": 23699
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9489381532440851,
      "learning_rate": 3.6906093241676564e-06,
      "loss": 0.4228,
      "step": 23700
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3671302536607117,
      "learning_rate": 3.689839819851452e-06,
      "loss": 0.1989,
      "step": 23701
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3350183953339407,
      "learning_rate": 3.689070377616678e-06,
      "loss": 0.2364,
      "step": 23702
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4545873097706119,
      "learning_rate": 3.6883009974709105e-06,
      "loss": 0.2572,
      "step": 23703
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27789963120775096,
      "learning_rate": 3.687531679421712e-06,
      "loss": 0.1646,
      "step": 23704
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4621735229049886,
      "learning_rate": 3.686762423476653e-06,
      "loss": 0.1737,
      "step": 23705
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3521386607425769,
      "learning_rate": 3.6859932296433065e-06,
      "loss": 0.144,
      "step": 23706
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.34302813214932903,
      "learning_rate": 3.6852240979292285e-06,
      "loss": 0.2819,
      "step": 23707
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4409232578914099,
      "learning_rate": 3.6844550283420022e-06,
      "loss": 0.2502,
      "step": 23708
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5350660754500796,
      "learning_rate": 3.683686020889181e-06,
      "loss": 0.6567,
      "step": 23709
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.48616527806602805,
      "learning_rate": 3.682917075578335e-06,
      "loss": 0.2007,
      "step": 23710
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37856930424496626,
      "learning_rate": 3.682148192417029e-06,
      "loss": 0.2647,
      "step": 23711
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8495340145624218,
      "learning_rate": 3.6813793714128323e-06,
      "loss": 0.2752,
      "step": 23712
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30813507274518576,
      "learning_rate": 3.6806106125732998e-06,
      "loss": 0.2219,
      "step": 23713
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2575069695001503,
      "learning_rate": 3.6798419159060004e-06,
      "loss": 0.1581,
      "step": 23714
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3821434282755238,
      "learning_rate": 3.679073281418499e-06,
      "loss": 0.2455,
      "step": 23715
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43541793019809477,
      "learning_rate": 3.678304709118348e-06,
      "loss": 0.2028,
      "step": 23716
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4991300794404085,
      "learning_rate": 3.677536199013122e-06,
      "loss": 0.7304,
      "step": 23717
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7113880357005755,
      "learning_rate": 3.6767677511103715e-06,
      "loss": 0.2922,
      "step": 23718
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2772497788177291,
      "learning_rate": 3.6759993654176607e-06,
      "loss": 0.1955,
      "step": 23719
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5003557098558423,
      "learning_rate": 3.675231041942553e-06,
      "loss": 0.2895,
      "step": 23720
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4235394581922607,
      "learning_rate": 3.6744627806925958e-06,
      "loss": 0.2274,
      "step": 23721
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4375288974073751,
      "learning_rate": 3.673694581675362e-06,
      "loss": 0.2122,
      "step": 23722
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38781557319527504,
      "learning_rate": 3.6729264448983994e-06,
      "loss": 0.1083,
      "step": 23723
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9225879649915448,
      "learning_rate": 3.6721583703692676e-06,
      "loss": 0.5049,
      "step": 23724
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2593920458591406,
      "learning_rate": 3.671390358095529e-06,
      "loss": 0.1969,
      "step": 23725
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4502956598414895,
      "learning_rate": 3.6706224080847296e-06,
      "loss": 0.3225,
      "step": 23726
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.633491012387481,
      "learning_rate": 3.6698545203444303e-06,
      "loss": 0.2766,
      "step": 23727
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.947628748744147,
      "learning_rate": 3.6690866948821848e-06,
      "loss": 0.4853,
      "step": 23728
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2720377979677964,
      "learning_rate": 3.66831893170555e-06,
      "loss": 0.1789,
      "step": 23729
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.732882442010401,
      "learning_rate": 3.6675512308220753e-06,
      "loss": 0.6651,
      "step": 23730
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2457244674190315,
      "learning_rate": 3.6667835922393147e-06,
      "loss": 0.1743,
      "step": 23731
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3846850852478673,
      "learning_rate": 3.6660160159648238e-06,
      "loss": 0.1007,
      "step": 23732
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6309449994858316,
      "learning_rate": 3.6652485020061444e-06,
      "loss": 0.2805,
      "step": 23733
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37204866976595896,
      "learning_rate": 3.6644810503708428e-06,
      "loss": 0.2067,
      "step": 23734
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8589299864312716,
      "learning_rate": 3.663713661066457e-06,
      "loss": 0.434,
      "step": 23735
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.940387178185205,
      "learning_rate": 3.662946334100541e-06,
      "loss": 0.2958,
      "step": 23736
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3360242937606305,
      "learning_rate": 3.662179069480648e-06,
      "loss": 0.2821,
      "step": 23737
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.293861472454093,
      "learning_rate": 3.6614118672143194e-06,
      "loss": 0.1774,
      "step": 23738
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3436687663147777,
      "learning_rate": 3.6606447273091063e-06,
      "loss": 0.6461,
      "step": 23739
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6141152345934908,
      "learning_rate": 3.6598776497725563e-06,
      "loss": 0.1392,
      "step": 23740
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2691153871794027,
      "learning_rate": 3.6591106346122194e-06,
      "loss": 0.1553,
      "step": 23741
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33937056345415795,
      "learning_rate": 3.6583436818356354e-06,
      "loss": 0.1548,
      "step": 23742
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5579417722100202,
      "learning_rate": 3.6575767914503523e-06,
      "loss": 0.4098,
      "step": 23743
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3960544605072089,
      "learning_rate": 3.6568099634639167e-06,
      "loss": 0.2448,
      "step": 23744
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6610083883820845,
      "learning_rate": 3.656043197883872e-06,
      "loss": 0.329,
      "step": 23745
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7917938745377984,
      "learning_rate": 3.6552764947177643e-06,
      "loss": 0.4571,
      "step": 23746
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36495281851639655,
      "learning_rate": 3.6545098539731273e-06,
      "loss": 0.1655,
      "step": 23747
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5723654442755108,
      "learning_rate": 3.6537432756575176e-06,
      "loss": 0.3308,
      "step": 23748
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33527633135000284,
      "learning_rate": 3.652976759778466e-06,
      "loss": 0.2373,
      "step": 23749
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3495941740224241,
      "learning_rate": 3.6522103063435166e-06,
      "loss": 0.1718,
      "step": 23750
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7890649352140967,
      "learning_rate": 3.6514439153602156e-06,
      "loss": 0.1038,
      "step": 23751
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.35354539446044875,
      "learning_rate": 3.650677586836094e-06,
      "loss": 0.2584,
      "step": 23752
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2087034000225871,
      "learning_rate": 3.6499113207786964e-06,
      "loss": 0.3314,
      "step": 23753
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5503141825317648,
      "learning_rate": 3.6491451171955604e-06,
      "loss": 0.3462,
      "step": 23754
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41589218003172684,
      "learning_rate": 3.648378976094228e-06,
      "loss": 0.2374,
      "step": 23755
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6241499518134205,
      "learning_rate": 3.64761289748223e-06,
      "loss": 0.3438,
      "step": 23756
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3458556186346162,
      "learning_rate": 3.6468468813671054e-06,
      "loss": 0.1908,
      "step": 23757
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41353675225328873,
      "learning_rate": 3.646080927756396e-06,
      "loss": 0.0868,
      "step": 23758
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4757355645102899,
      "learning_rate": 3.645315036657626e-06,
      "loss": 0.2306,
      "step": 23759
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3955377477217164,
      "learning_rate": 3.6445492080783452e-06,
      "loss": 0.1669,
      "step": 23760
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3056308719062829,
      "learning_rate": 3.6437834420260764e-06,
      "loss": 0.2642,
      "step": 23761
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44327148012216194,
      "learning_rate": 3.6430177385083577e-06,
      "loss": 0.2598,
      "step": 23762
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5589114542531386,
      "learning_rate": 3.642252097532725e-06,
      "loss": 0.7987,
      "step": 23763
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.887216087042526,
      "learning_rate": 3.641486519106705e-06,
      "loss": 0.5147,
      "step": 23764
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43009079253306126,
      "learning_rate": 3.640721003237834e-06,
      "loss": 0.2581,
      "step": 23765
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3133633765255886,
      "learning_rate": 3.6399555499336403e-06,
      "loss": 0.2004,
      "step": 23766
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.45714594415768195,
      "learning_rate": 3.6391901592016623e-06,
      "loss": 0.3031,
      "step": 23767
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3298221524021902,
      "learning_rate": 3.63842483104942e-06,
      "loss": 0.0785,
      "step": 23768
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5207751304061905,
      "learning_rate": 3.637659565484447e-06,
      "loss": 0.3639,
      "step": 23769
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3056150708685246,
      "learning_rate": 3.6368943625142726e-06,
      "loss": 0.1862,
      "step": 23770
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3438744847236149,
      "learning_rate": 3.6361292221464263e-06,
      "loss": 0.1817,
      "step": 23771
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.451324934676012,
      "learning_rate": 3.635364144388437e-06,
      "loss": 0.3361,
      "step": 23772
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.34434877348687726,
      "learning_rate": 3.6345991292478266e-06,
      "loss": 0.2188,
      "step": 23773
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44933368952116,
      "learning_rate": 3.6338341767321238e-06,
      "loss": 0.283,
      "step": 23774
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.47725393667019583,
      "learning_rate": 3.6330692868488593e-06,
      "loss": 0.185,
      "step": 23775
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.857854209793224,
      "learning_rate": 3.632304459605547e-06,
      "loss": 0.7147,
      "step": 23776
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3731339393081215,
      "learning_rate": 3.631539695009726e-06,
      "loss": 0.1858,
      "step": 23777
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43059215804977896,
      "learning_rate": 3.6307749930689094e-06,
      "loss": 0.2826,
      "step": 23778
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3150847013487798,
      "learning_rate": 3.630010353790624e-06,
      "loss": 0.1923,
      "step": 23779
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4083501052027478,
      "learning_rate": 3.629245777182392e-06,
      "loss": 0.3129,
      "step": 23780
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25614346075141653,
      "learning_rate": 3.6284812632517407e-06,
      "loss": 0.1472,
      "step": 23781
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5982204680276861,
      "learning_rate": 3.6277168120061834e-06,
      "loss": 0.6904,
      "step": 23782
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.44737720964758,
      "learning_rate": 3.626952423453245e-06,
      "loss": 0.0919,
      "step": 23783
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3543133171048738,
      "learning_rate": 3.6261880976004494e-06,
      "loss": 0.2006,
      "step": 23784
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36025474326939794,
      "learning_rate": 3.625423834455305e-06,
      "loss": 0.2809,
      "step": 23785
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7629776416631929,
      "learning_rate": 3.624659634025346e-06,
      "loss": 0.2455,
      "step": 23786
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6267738674111025,
      "learning_rate": 3.6238954963180794e-06,
      "loss": 0.3697,
      "step": 23787
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30853515650019075,
      "learning_rate": 3.6231314213410274e-06,
      "loss": 0.2062,
      "step": 23788
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.431574066777507,
      "learning_rate": 3.62236740910171e-06,
      "loss": 0.2009,
      "step": 23789
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41452604798657056,
      "learning_rate": 3.6216034596076366e-06,
      "loss": 0.2361,
      "step": 23790
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5385191243437756,
      "learning_rate": 3.620839572866328e-06,
      "loss": 0.3649,
      "step": 23791
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.22774878154686146,
      "learning_rate": 3.6200757488852976e-06,
      "loss": 0.1273,
      "step": 23792
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5789853098174689,
      "learning_rate": 3.619311987672065e-06,
      "loss": 0.3145,
      "step": 23793
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3544611432877165,
      "learning_rate": 3.6185482892341384e-06,
      "loss": 0.1584,
      "step": 23794
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6433752012871559,
      "learning_rate": 3.6177846535790318e-06,
      "loss": 0.3694,
      "step": 23795
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30994885519418,
      "learning_rate": 3.6170210807142615e-06,
      "loss": 0.2296,
      "step": 23796
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3162419894015419,
      "learning_rate": 3.616257570647337e-06,
      "loss": 0.1943,
      "step": 23797
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6259491021887094,
      "learning_rate": 3.6154941233857743e-06,
      "loss": 0.7712,
      "step": 23798
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.948230001153982,
      "learning_rate": 3.614730738937078e-06,
      "loss": 0.5126,
      "step": 23799
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.26212195668585286,
      "learning_rate": 3.613967417308761e-06,
      "loss": 0.1617,
      "step": 23800
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30841433974581434,
      "learning_rate": 3.613204158508339e-06,
      "loss": 0.0699,
      "step": 23801
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38605467751030786,
      "learning_rate": 3.6124409625433078e-06,
      "loss": 0.2786,
      "step": 23802
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33875073462580324,
      "learning_rate": 3.611677829421192e-06,
      "loss": 0.2354,
      "step": 23803
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8703868951748873,
      "learning_rate": 3.6109147591494876e-06,
      "loss": 0.3718,
      "step": 23804
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7632704026870012,
      "learning_rate": 3.6101517517357067e-06,
      "loss": 0.1896,
      "step": 23805
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43694375089536047,
      "learning_rate": 3.6093888071873585e-06,
      "loss": 0.2748,
      "step": 23806
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4232863711240987,
      "learning_rate": 3.6086259255119416e-06,
      "loss": 0.2146,
      "step": 23807
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33424843084601147,
      "learning_rate": 3.607863106716967e-06,
      "loss": 0.2867,
      "step": 23808
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0416798063021486,
      "learning_rate": 3.6071003508099388e-06,
      "loss": 0.4527,
      "step": 23809
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1418969750733601,
      "learning_rate": 3.6063376577983635e-06,
      "loss": 0.0711,
      "step": 23810
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4110246380760809,
      "learning_rate": 3.605575027689735e-06,
      "loss": 0.307,
      "step": 23811
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3252901633176835,
      "learning_rate": 3.6048124604915714e-06,
      "loss": 0.3176,
      "step": 23812
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.163517154229942,
      "learning_rate": 3.604049956211366e-06,
      "loss": 0.4783,
      "step": 23813
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31688420217200064,
      "learning_rate": 3.603287514856616e-06,
      "loss": 0.2076,
      "step": 23814
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3782402287538942,
      "learning_rate": 3.6025251364348336e-06,
      "loss": 0.238,
      "step": 23815
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.48737201536617114,
      "learning_rate": 3.6017628209535115e-06,
      "loss": 0.2153,
      "step": 23816
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6472011729549305,
      "learning_rate": 3.601000568420152e-06,
      "loss": 0.7392,
      "step": 23817
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.261326406065225,
      "learning_rate": 3.6002383788422548e-06,
      "loss": 0.1204,
      "step": 23818
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.29275160884097995,
      "learning_rate": 3.5994762522273206e-06,
      "loss": 0.177,
      "step": 23819
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32781873015314344,
      "learning_rate": 3.5987141885828427e-06,
      "loss": 0.2197,
      "step": 23820
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5057872218211611,
      "learning_rate": 3.5979521879163202e-06,
      "loss": 0.3346,
      "step": 23821
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.631051875263481,
      "learning_rate": 3.5971902502352507e-06,
      "loss": 0.3181,
      "step": 23822
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7512139862478967,
      "learning_rate": 3.596428375547131e-06,
      "loss": 0.1134,
      "step": 23823
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3502840093436134,
      "learning_rate": 3.595666563859459e-06,
      "loss": 0.2712,
      "step": 23824
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9961680732825735,
      "learning_rate": 3.594904815179723e-06,
      "loss": 0.3341,
      "step": 23825
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4844839619880205,
      "learning_rate": 3.5941431295154205e-06,
      "loss": 0.3341,
      "step": 23826
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2619401383657469,
      "learning_rate": 3.5933815068740506e-06,
      "loss": 0.1708,
      "step": 23827
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3659384158561714,
      "learning_rate": 3.592619947263094e-06,
      "loss": 0.1991,
      "step": 23828
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3531774432012235,
      "learning_rate": 3.5918584506900567e-06,
      "loss": 0.2157,
      "step": 23829
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8607436585032985,
      "learning_rate": 3.5910970171624214e-06,
      "loss": 0.3964,
      "step": 23830
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4502696592824958,
      "learning_rate": 3.5903356466876837e-06,
      "loss": 0.2253,
      "step": 23831
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4502922104410417,
      "learning_rate": 3.589574339273335e-06,
      "loss": 0.3281,
      "step": 23832
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2720506623434302,
      "learning_rate": 3.588813094926861e-06,
      "loss": 0.0676,
      "step": 23833
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3642144744330665,
      "learning_rate": 3.5880519136557525e-06,
      "loss": 0.2746,
      "step": 23834
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9101130525595305,
      "learning_rate": 3.5872907954674984e-06,
      "loss": 0.6085,
      "step": 23835
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6714574301704284,
      "learning_rate": 3.586529740369592e-06,
      "loss": 0.459,
      "step": 23836
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4894516674174505,
      "learning_rate": 3.5857687483695124e-06,
      "loss": 0.2431,
      "step": 23837
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23775656134749928,
      "learning_rate": 3.5850078194747506e-06,
      "loss": 0.2055,
      "step": 23838
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37587285903220746,
      "learning_rate": 3.584246953692797e-06,
      "loss": 0.316,
      "step": 23839
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.14748797618535867,
      "learning_rate": 3.5834861510311246e-06,
      "loss": 0.0703,
      "step": 23840
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3290097975569,
      "learning_rate": 3.582725411497234e-06,
      "loss": 0.5701,
      "step": 23841
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32998683516774957,
      "learning_rate": 3.5819647350985996e-06,
      "loss": 0.1574,
      "step": 23842
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5179755888308443,
      "learning_rate": 3.5812041218427073e-06,
      "loss": 0.3609,
      "step": 23843
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41428127295549944,
      "learning_rate": 3.5804435717370447e-06,
      "loss": 0.2397,
      "step": 23844
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.46119572961527694,
      "learning_rate": 3.5796830847890873e-06,
      "loss": 0.356,
      "step": 23845
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.558482351220606,
      "learning_rate": 3.57892266100632e-06,
      "loss": 0.246,
      "step": 23846
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32314146677133787,
      "learning_rate": 3.578162300396224e-06,
      "loss": 0.2454,
      "step": 23847
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6890913129872926,
      "learning_rate": 3.5774020029662813e-06,
      "loss": 0.2629,
      "step": 23848
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31846664451777346,
      "learning_rate": 3.5766417687239706e-06,
      "loss": 0.1789,
      "step": 23849
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.29794029656240534,
      "learning_rate": 3.5758815976767747e-06,
      "loss": 0.2178,
      "step": 23850
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2707019415671591,
      "learning_rate": 3.575121489832166e-06,
      "loss": 0.1811,
      "step": 23851
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3446975343048002,
      "learning_rate": 3.574361445197627e-06,
      "loss": 0.6592,
      "step": 23852
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9600182674198318,
      "learning_rate": 3.5736014637806373e-06,
      "loss": 0.585,
      "step": 23853
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7147593075045577,
      "learning_rate": 3.5728415455886643e-06,
      "loss": 0.3652,
      "step": 23854
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33066099308093544,
      "learning_rate": 3.5720816906291976e-06,
      "loss": 0.1983,
      "step": 23855
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5432137755186152,
      "learning_rate": 3.5713218989097033e-06,
      "loss": 0.3274,
      "step": 23856
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32019727017429245,
      "learning_rate": 3.570562170437658e-06,
      "loss": 0.2355,
      "step": 23857
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3410053247535427,
      "learning_rate": 3.569802505220543e-06,
      "loss": 0.1817,
      "step": 23858
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.36797272859301,
      "learning_rate": 3.5690429032658226e-06,
      "loss": 0.1173,
      "step": 23859
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.889314946100797,
      "learning_rate": 3.5682833645809735e-06,
      "loss": 0.4578,
      "step": 23860
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.29147584951408,
      "learning_rate": 3.5675238891734697e-06,
      "loss": 0.1869,
      "step": 23861
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36530367842139444,
      "learning_rate": 3.566764477050786e-06,
      "loss": 0.302,
      "step": 23862
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6720834925026056,
      "learning_rate": 3.5660051282203868e-06,
      "loss": 0.2906,
      "step": 23863
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6372170709077182,
      "learning_rate": 3.565245842689745e-06,
      "loss": 0.2591,
      "step": 23864
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36812807012301135,
      "learning_rate": 3.5644866204663366e-06,
      "loss": 0.2281,
      "step": 23865
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.501256565401922,
      "learning_rate": 3.563727461557619e-06,
      "loss": 0.1993,
      "step": 23866
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2784628867227453,
      "learning_rate": 3.5629683659710758e-06,
      "loss": 0.1629,
      "step": 23867
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.29587916589395613,
      "learning_rate": 3.562209333714164e-06,
      "loss": 0.2287,
      "step": 23868
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.674207153319339,
      "learning_rate": 3.5614503647943556e-06,
      "loss": 0.3051,
      "step": 23869
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3331355479506116,
      "learning_rate": 3.56069145921912e-06,
      "loss": 0.2146,
      "step": 23870
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7018048727798067,
      "learning_rate": 3.5599326169959135e-06,
      "loss": 0.8221,
      "step": 23871
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7024412574516391,
      "learning_rate": 3.5591738381322163e-06,
      "loss": 0.2733,
      "step": 23872
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3532126474050306,
      "learning_rate": 3.5584151226354836e-06,
      "loss": 0.3054,
      "step": 23873
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2713261531050271,
      "learning_rate": 3.557656470513181e-06,
      "loss": 0.1672,
      "step": 23874
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6775770079068508,
      "learning_rate": 3.5568978817727783e-06,
      "loss": 0.8172,
      "step": 23875
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32688841548447706,
      "learning_rate": 3.5561393564217307e-06,
      "loss": 0.0775,
      "step": 23876
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4064356815889936,
      "learning_rate": 3.555380894467504e-06,
      "loss": 0.109,
      "step": 23877
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7063792506466762,
      "learning_rate": 3.55462249591756e-06,
      "loss": 0.4107,
      "step": 23878
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3354742430256102,
      "learning_rate": 3.5538641607793656e-06,
      "loss": 0.2024,
      "step": 23879
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3318377516557831,
      "learning_rate": 3.553105889060369e-06,
      "loss": 0.3029,
      "step": 23880
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5145018925987759,
      "learning_rate": 3.5523476807680446e-06,
      "loss": 0.2732,
      "step": 23881
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.913533094984673,
      "learning_rate": 3.551589535909842e-06,
      "loss": 0.4139,
      "step": 23882
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.39645811210462756,
      "learning_rate": 3.5508314544932242e-06,
      "loss": 0.1791,
      "step": 23883
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6021332293246213,
      "learning_rate": 3.5500734365256507e-06,
      "loss": 0.3524,
      "step": 23884
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27730976805726126,
      "learning_rate": 3.5493154820145746e-06,
      "loss": 0.1679,
      "step": 23885
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3201292168301627,
      "learning_rate": 3.548557590967455e-06,
      "loss": 0.2404,
      "step": 23886
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3842893373015159,
      "learning_rate": 3.5477997633917483e-06,
      "loss": 0.1426,
      "step": 23887
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3622185655491347,
      "learning_rate": 3.5470419992949145e-06,
      "loss": 0.2716,
      "step": 23888
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7260399163501968,
      "learning_rate": 3.5462842986844016e-06,
      "loss": 0.2898,
      "step": 23889
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9513310226787065,
      "learning_rate": 3.5455266615676665e-06,
      "loss": 0.3317,
      "step": 23890
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32611696288439873,
      "learning_rate": 3.544769087952168e-06,
      "loss": 0.2621,
      "step": 23891
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30943954724834977,
      "learning_rate": 3.544011577845349e-06,
      "loss": 0.1592,
      "step": 23892
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6024365454697012,
      "learning_rate": 3.543254131254674e-06,
      "loss": 0.4049,
      "step": 23893
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1997259919434844,
      "learning_rate": 3.542496748187586e-06,
      "loss": 0.0888,
      "step": 23894
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.23337972061189474,
      "learning_rate": 3.5417394286515405e-06,
      "loss": 0.135,
      "step": 23895
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0310991808139396,
      "learning_rate": 3.5409821726539905e-06,
      "loss": 0.2606,
      "step": 23896
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32699436986531805,
      "learning_rate": 3.540224980202377e-06,
      "loss": 0.2607,
      "step": 23897
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2889653015879698,
      "learning_rate": 3.539467851304161e-06,
      "loss": 0.2229,
      "step": 23898
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8088778397214789,
      "learning_rate": 3.5387107859667845e-06,
      "loss": 0.4445,
      "step": 23899
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2257754762165438,
      "learning_rate": 3.537953784197696e-06,
      "loss": 0.2503,
      "step": 23900
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.45291740764214766,
      "learning_rate": 3.5371968460043495e-06,
      "loss": 0.2931,
      "step": 23901
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4213591746993919,
      "learning_rate": 3.536439971394182e-06,
      "loss": 0.191,
      "step": 23902
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4305315413010613,
      "learning_rate": 3.5356831603746454e-06,
      "loss": 0.2189,
      "step": 23903
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2480533158745236,
      "learning_rate": 3.5349264129531847e-06,
      "loss": 0.1841,
      "step": 23904
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5903760553071531,
      "learning_rate": 3.534169729137249e-06,
      "loss": 0.2017,
      "step": 23905
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5736234489884543,
      "learning_rate": 3.533413108934276e-06,
      "loss": 0.3476,
      "step": 23906
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5831191491885297,
      "learning_rate": 3.532656552351712e-06,
      "loss": 0.26,
      "step": 23907
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2957293413266942,
      "learning_rate": 3.5319000593970043e-06,
      "loss": 0.7892,
      "step": 23908
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31308280649807785,
      "learning_rate": 3.531143630077585e-06,
      "loss": 0.2161,
      "step": 23909
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0470864743964006,
      "learning_rate": 3.530387264400911e-06,
      "loss": 0.3749,
      "step": 23910
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.28992246620769746,
      "learning_rate": 3.5296309623744105e-06,
      "loss": 0.184,
      "step": 23911
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.6427175171934807,
      "learning_rate": 3.528874724005532e-06,
      "loss": 0.41,
      "step": 23912
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3335240905074433,
      "learning_rate": 3.5281185493017157e-06,
      "loss": 0.1719,
      "step": 23913
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.0364423791982516,
      "learning_rate": 3.527362438270395e-06,
      "loss": 0.4251,
      "step": 23914
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27408513405443263,
      "learning_rate": 3.526606390919013e-06,
      "loss": 0.1832,
      "step": 23915
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.28616964242225834,
      "learning_rate": 3.525850407255006e-06,
      "loss": 0.2451,
      "step": 23916
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.2328977665422278,
      "learning_rate": 3.5250944872858172e-06,
      "loss": 0.7675,
      "step": 23917
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1937317128451173,
      "learning_rate": 3.524338631018873e-06,
      "loss": 0.066,
      "step": 23918
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6811043939368975,
      "learning_rate": 3.523582838461622e-06,
      "loss": 0.3208,
      "step": 23919
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36335769752244396,
      "learning_rate": 3.5228271096214917e-06,
      "loss": 0.2128,
      "step": 23920
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5074167753027266,
      "learning_rate": 3.522071444505919e-06,
      "loss": 0.3037,
      "step": 23921
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.37274468034026764,
      "learning_rate": 3.5213158431223425e-06,
      "loss": 0.2399,
      "step": 23922
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6940335433721534,
      "learning_rate": 3.520560305478186e-06,
      "loss": 0.3821,
      "step": 23923
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2537298268779111,
      "learning_rate": 3.519804831580895e-06,
      "loss": 0.1692,
      "step": 23924
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.403105636490447,
      "learning_rate": 3.5190494214378946e-06,
      "loss": 0.2243,
      "step": 23925
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8781299215389675,
      "learning_rate": 3.5182940750566174e-06,
      "loss": 0.5114,
      "step": 23926
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2888547029863776,
      "learning_rate": 3.517538792444499e-06,
      "loss": 0.2236,
      "step": 23927
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3316403728173597,
      "learning_rate": 3.516783573608964e-06,
      "loss": 0.1657,
      "step": 23928
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.40152303472093576,
      "learning_rate": 3.5160284185574458e-06,
      "loss": 0.1836,
      "step": 23929
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.296349305976521,
      "learning_rate": 3.515273327297373e-06,
      "loss": 0.4785,
      "step": 23930
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6400638566685536,
      "learning_rate": 3.514518299836178e-06,
      "loss": 0.2685,
      "step": 23931
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.43427878851540225,
      "learning_rate": 3.513763336181283e-06,
      "loss": 0.2718,
      "step": 23932
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3769748323012758,
      "learning_rate": 3.5130084363401186e-06,
      "loss": 0.1764,
      "step": 23933
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.36125920389950256,
      "learning_rate": 3.512253600320116e-06,
      "loss": 0.2759,
      "step": 23934
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.060290784177128,
      "learning_rate": 3.5114988281286886e-06,
      "loss": 0.5982,
      "step": 23935
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3018021356530739,
      "learning_rate": 3.510744119773278e-06,
      "loss": 0.1591,
      "step": 23936
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.31772814915095354,
      "learning_rate": 3.5099894752612994e-06,
      "loss": 0.0606,
      "step": 23937
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3565329843976899,
      "learning_rate": 3.5092348946001795e-06,
      "loss": 0.2513,
      "step": 23938
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.30047787796401426,
      "learning_rate": 3.5084803777973462e-06,
      "loss": 0.2378,
      "step": 23939
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5930432296780394,
      "learning_rate": 3.5077259248602157e-06,
      "loss": 0.3662,
      "step": 23940
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8220266437379906,
      "learning_rate": 3.506971535796213e-06,
      "loss": 0.2835,
      "step": 23941
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3695302833010303,
      "learning_rate": 3.5062172106127614e-06,
      "loss": 0.1601,
      "step": 23942
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5471216056008187,
      "learning_rate": 3.5054629493172845e-06,
      "loss": 0.3875,
      "step": 23943
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4248749697660498,
      "learning_rate": 3.5047087519171975e-06,
      "loss": 0.2462,
      "step": 23944
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.47405564446411147,
      "learning_rate": 3.5039546184199223e-06,
      "loss": 0.3469,
      "step": 23945
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.172619411369806,
      "learning_rate": 3.5032005488328825e-06,
      "loss": 0.0718,
      "step": 23946
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.400316409305373,
      "learning_rate": 3.5024465431634868e-06,
      "loss": 0.2723,
      "step": 23947
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6940890214291079,
      "learning_rate": 3.5016926014191666e-06,
      "loss": 0.2747,
      "step": 23948
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7229251037968297,
      "learning_rate": 3.5009387236073254e-06,
      "loss": 0.3754,
      "step": 23949
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.41750374519632166,
      "learning_rate": 3.500184909735396e-06,
      "loss": 0.2151,
      "step": 23950
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38526726178034076,
      "learning_rate": 3.499431159810781e-06,
      "loss": 0.259,
      "step": 23951
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3104915167530969,
      "learning_rate": 3.4986774738409025e-06,
      "loss": 0.2182,
      "step": 23952
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.5646947619998237,
      "learning_rate": 3.497923851833177e-06,
      "loss": 0.7983,
      "step": 23953
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.16099298492894557,
      "learning_rate": 3.4971702937950123e-06,
      "loss": 0.088,
      "step": 23954
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3787208445261835,
      "learning_rate": 3.496416799733827e-06,
      "loss": 0.0694,
      "step": 23955
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3916955726703076,
      "learning_rate": 3.495663369657033e-06,
      "loss": 0.2694,
      "step": 23956
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33133946158014593,
      "learning_rate": 3.4949100035720464e-06,
      "loss": 0.2451,
      "step": 23957
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8297448489137771,
      "learning_rate": 3.494156701486272e-06,
      "loss": 0.4001,
      "step": 23958
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7970847374264504,
      "learning_rate": 3.4934034634071245e-06,
      "loss": 0.1449,
      "step": 23959
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9324540404476575,
      "learning_rate": 3.492650289342018e-06,
      "loss": 0.3928,
      "step": 23960
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32473950164080134,
      "learning_rate": 3.4918971792983525e-06,
      "loss": 0.2234,
      "step": 23961
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4440668696261239,
      "learning_rate": 3.4911441332835504e-06,
      "loss": 0.8323,
      "step": 23962
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21580723310167327,
      "learning_rate": 3.4903911513050113e-06,
      "loss": 0.1741,
      "step": 23963
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.45127446504382207,
      "learning_rate": 3.4896382333701463e-06,
      "loss": 0.2157,
      "step": 23964
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.279346795762276,
      "learning_rate": 3.4888853794863652e-06,
      "loss": 0.1756,
      "step": 23965
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9541851633418601,
      "learning_rate": 3.4881325896610695e-06,
      "loss": 0.3816,
      "step": 23966
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.8969407116282031,
      "learning_rate": 3.4873798639016676e-06,
      "loss": 0.2278,
      "step": 23967
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4399034382697847,
      "learning_rate": 3.4866272022155656e-06,
      "loss": 0.2272,
      "step": 23968
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.38277729427172924,
      "learning_rate": 3.485874604610171e-06,
      "loss": 0.2573,
      "step": 23969
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3484375340846264,
      "learning_rate": 3.4851220710928835e-06,
      "loss": 0.23,
      "step": 23970
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.3944218039543583,
      "learning_rate": 3.484369601671107e-06,
      "loss": 0.7075,
      "step": 23971
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.1920141646269339,
      "learning_rate": 3.4836171963522515e-06,
      "loss": 0.0658,
      "step": 23972
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9080844201957969,
      "learning_rate": 3.482864855143706e-06,
      "loss": 0.3627,
      "step": 23973
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27587823367247244,
      "learning_rate": 3.4821125780528877e-06,
      "loss": 0.1759,
      "step": 23974
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.32628803811470203,
      "learning_rate": 3.4813603650871876e-06,
      "loss": 0.2901,
      "step": 23975
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.18070048415304787,
      "learning_rate": 3.480608216254009e-06,
      "loss": 0.065,
      "step": 23976
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7355400516496284,
      "learning_rate": 3.479856131560755e-06,
      "loss": 0.784,
      "step": 23977
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3037939156558511,
      "learning_rate": 3.479104111014815e-06,
      "loss": 0.1691,
      "step": 23978
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.5025642088873452,
      "learning_rate": 3.478352154623601e-06,
      "loss": 0.3372,
      "step": 23979
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.1066386689533008,
      "learning_rate": 3.4776002623945016e-06,
      "loss": 0.2206,
      "step": 23980
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3323636780697244,
      "learning_rate": 3.4768484343349173e-06,
      "loss": 0.2347,
      "step": 23981
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7195681206470985,
      "learning_rate": 3.4760966704522436e-06,
      "loss": 0.3106,
      "step": 23982
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3249695636540924,
      "learning_rate": 3.475344970753881e-06,
      "loss": 0.1935,
      "step": 23983
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.7729934851071147,
      "learning_rate": 3.4745933352472182e-06,
      "loss": 0.3701,
      "step": 23984
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.27217077158206954,
      "learning_rate": 3.473841763939654e-06,
      "loss": 0.0928,
      "step": 23985
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.6555123912484266,
      "learning_rate": 3.473090256838585e-06,
      "loss": 0.2968,
      "step": 23986
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3979524311916083,
      "learning_rate": 3.472338813951395e-06,
      "loss": 0.0639,
      "step": 23987
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.33771818062189957,
      "learning_rate": 3.4715874352854893e-06,
      "loss": 0.2465,
      "step": 23988
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9779776396493876,
      "learning_rate": 3.4708361208482523e-06,
      "loss": 0.5689,
      "step": 23989
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9026522962804704,
      "learning_rate": 3.470084870647077e-06,
      "loss": 0.3921,
      "step": 23990
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.9380813469122491,
      "learning_rate": 3.469333684689359e-06,
      "loss": 0.2803,
      "step": 23991
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.3900636671019824,
      "learning_rate": 3.468582562982482e-06,
      "loss": 0.2505,
      "step": 23992
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.25742125364037755,
      "learning_rate": 3.4678315055338394e-06,
      "loss": 0.2133,
      "step": 23993
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.21147007435220525,
      "learning_rate": 3.467080512350819e-06,
      "loss": 0.0815,
      "step": 23994
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.4211493002703899,
      "learning_rate": 3.4663295834408137e-06,
      "loss": 0.5017,
      "step": 23995
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.4955518017129953,
      "learning_rate": 3.4655787188112043e-06,
      "loss": 0.1225,
      "step": 23996
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.40123481667701444,
      "learning_rate": 3.464827918469382e-06,
      "loss": 0.3066,
      "step": 23997
    },
    {
      "epoch": 0.73,
      "grad_norm": 0.2862906358494236,
      "learning_rate": 3.4640771824227326e-06,
      "loss": 0.2275,
      "step": 23998
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.42723400602861,
      "learning_rate": 3.463326510678643e-06,
      "loss": 0.5513,
      "step": 23999
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4970857929907233,
      "learning_rate": 3.4625759032445016e-06,
      "loss": 0.305,
      "step": 24000
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3914522465979143,
      "learning_rate": 3.4618253601276853e-06,
      "loss": 0.2401,
      "step": 24001
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.30434855870938865,
      "learning_rate": 3.461074881335583e-06,
      "loss": 0.1399,
      "step": 24002
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4936524457643495,
      "learning_rate": 3.460324466875581e-06,
      "loss": 0.2652,
      "step": 24003
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3226257451193271,
      "learning_rate": 3.459574116755051e-06,
      "loss": 0.2091,
      "step": 24004
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8954776180942897,
      "learning_rate": 3.4588238309813902e-06,
      "loss": 0.4112,
      "step": 24005
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35493381435746074,
      "learning_rate": 3.4580736095619706e-06,
      "loss": 0.2008,
      "step": 24006
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0320167081336973,
      "learning_rate": 3.457323452504174e-06,
      "loss": 0.4573,
      "step": 24007
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.9831938874014126,
      "learning_rate": 3.456573359815386e-06,
      "loss": 0.406,
      "step": 24008
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.242820841543765,
      "learning_rate": 3.45582333150298e-06,
      "loss": 0.2213,
      "step": 24009
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5111835618392383,
      "learning_rate": 3.455073367574335e-06,
      "loss": 0.3497,
      "step": 24010
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2708945873813005,
      "learning_rate": 3.454323468036834e-06,
      "loss": 0.2207,
      "step": 24011
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4473337912830879,
      "learning_rate": 3.4535736328978553e-06,
      "loss": 0.2471,
      "step": 24012
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40908429899254506,
      "learning_rate": 3.4528238621647692e-06,
      "loss": 0.1207,
      "step": 24013
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.933313452012259,
      "learning_rate": 3.4520741558449568e-06,
      "loss": 0.4368,
      "step": 24014
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33876924159629607,
      "learning_rate": 3.451324513945797e-06,
      "loss": 0.1904,
      "step": 24015
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6245526039998985,
      "learning_rate": 3.450574936474654e-06,
      "loss": 0.3494,
      "step": 24016
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.052879923629337,
      "learning_rate": 3.4498254234389172e-06,
      "loss": 0.2901,
      "step": 24017
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3319573223887367,
      "learning_rate": 3.44907597484595e-06,
      "loss": 0.5968,
      "step": 24018
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5405603525577334,
      "learning_rate": 3.448326590703128e-06,
      "loss": 0.0938,
      "step": 24019
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37186374974074254,
      "learning_rate": 3.447577271017826e-06,
      "loss": 0.2058,
      "step": 24020
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.301547056831642,
      "learning_rate": 3.4468280157974166e-06,
      "loss": 0.1592,
      "step": 24021
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27195745508725183,
      "learning_rate": 3.4460788250492673e-06,
      "loss": 0.2095,
      "step": 24022
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8127656458749138,
      "learning_rate": 3.4453296987807507e-06,
      "loss": 0.414,
      "step": 24023
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29560366555494133,
      "learning_rate": 3.4445806369992372e-06,
      "loss": 0.2021,
      "step": 24024
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.835936202694665,
      "learning_rate": 3.4438316397120964e-06,
      "loss": 0.3623,
      "step": 24025
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8534002064628126,
      "learning_rate": 3.4430827069267005e-06,
      "loss": 0.31,
      "step": 24026
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4012691794044967,
      "learning_rate": 3.4423338386504114e-06,
      "loss": 0.6166,
      "step": 24027
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2785860635885546,
      "learning_rate": 3.4415850348906e-06,
      "loss": 0.1893,
      "step": 24028
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34691988662248885,
      "learning_rate": 3.440836295654636e-06,
      "loss": 0.2732,
      "step": 24029
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6257529778774433,
      "learning_rate": 3.4400876209498758e-06,
      "loss": 0.1935,
      "step": 24030
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2825757575375442,
      "learning_rate": 3.4393390107836987e-06,
      "loss": 0.1275,
      "step": 24031
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6582145766249187,
      "learning_rate": 3.4385904651634604e-06,
      "loss": 0.2752,
      "step": 24032
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36977090921617584,
      "learning_rate": 3.437841984096528e-06,
      "loss": 0.2841,
      "step": 24033
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1214346054549422,
      "learning_rate": 3.437093567590268e-06,
      "loss": 0.2987,
      "step": 24034
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3114374660935911,
      "learning_rate": 3.436345215652038e-06,
      "loss": 0.2473,
      "step": 24035
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.632238966930007,
      "learning_rate": 3.4355969282892042e-06,
      "loss": 0.6898,
      "step": 24036
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4964475072621165,
      "learning_rate": 3.4348487055091275e-06,
      "loss": 0.0942,
      "step": 24037
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3787933533185623,
      "learning_rate": 3.4341005473191724e-06,
      "loss": 0.2666,
      "step": 24038
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1607257886651473,
      "learning_rate": 3.433352453726694e-06,
      "loss": 0.0706,
      "step": 24039
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33681946790200556,
      "learning_rate": 3.432604424739053e-06,
      "loss": 0.2762,
      "step": 24040
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6243746631127735,
      "learning_rate": 3.431856460363616e-06,
      "loss": 0.2272,
      "step": 24041
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44926692251149153,
      "learning_rate": 3.431108560607729e-06,
      "loss": 0.3133,
      "step": 24042
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5492816897553241,
      "learning_rate": 3.430360725478764e-06,
      "loss": 0.254,
      "step": 24043
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3589729192434448,
      "learning_rate": 3.4296129549840684e-06,
      "loss": 0.7913,
      "step": 24044
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.45412264272597963,
      "learning_rate": 3.4288652491310025e-06,
      "loss": 0.2264,
      "step": 24045
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43223541087244194,
      "learning_rate": 3.4281176079269262e-06,
      "loss": 0.2784,
      "step": 24046
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3105484393824293,
      "learning_rate": 3.427370031379188e-06,
      "loss": 0.1948,
      "step": 24047
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.38669501639501147,
      "learning_rate": 3.4266225194951462e-06,
      "loss": 0.0874,
      "step": 24048
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44891452264336623,
      "learning_rate": 3.425875072282154e-06,
      "loss": 0.2536,
      "step": 24049
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4884537772845849,
      "learning_rate": 3.425127689747567e-06,
      "loss": 0.2933,
      "step": 24050
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4267520838364163,
      "learning_rate": 3.424380371898737e-06,
      "loss": 0.275,
      "step": 24051
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2419804258455325,
      "learning_rate": 3.423633118743019e-06,
      "loss": 0.2222,
      "step": 24052
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7580899858535015,
      "learning_rate": 3.4228859302877603e-06,
      "loss": 0.7666,
      "step": 24053
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1035853787902934,
      "learning_rate": 3.422138806540314e-06,
      "loss": 0.2563,
      "step": 24054
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6764074156809116,
      "learning_rate": 3.4213917475080328e-06,
      "loss": 0.312,
      "step": 24055
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34967986647328847,
      "learning_rate": 3.4206447531982577e-06,
      "loss": 0.183,
      "step": 24056
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.798141075192955,
      "learning_rate": 3.419897823618351e-06,
      "loss": 0.7336,
      "step": 24057
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2229179717095614,
      "learning_rate": 3.419150958775652e-06,
      "loss": 0.1846,
      "step": 24058
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6472203257141071,
      "learning_rate": 3.4184041586775105e-06,
      "loss": 0.3867,
      "step": 24059
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2945782059287872,
      "learning_rate": 3.4176574233312785e-06,
      "loss": 0.1707,
      "step": 24060
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7573561732952021,
      "learning_rate": 3.4169107527442958e-06,
      "loss": 0.2872,
      "step": 24061
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5502522665095753,
      "learning_rate": 3.4161641469239103e-06,
      "loss": 0.8266,
      "step": 24062
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2446050233842922,
      "learning_rate": 3.415417605877469e-06,
      "loss": 0.1843,
      "step": 24063
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7616947763378036,
      "learning_rate": 3.414671129612318e-06,
      "loss": 0.3527,
      "step": 24064
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.337467044711944,
      "learning_rate": 3.413924718135797e-06,
      "loss": 0.1948,
      "step": 24065
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1892526290293717,
      "learning_rate": 3.413178371455251e-06,
      "loss": 0.4009,
      "step": 24066
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6945623500559897,
      "learning_rate": 3.4124320895780284e-06,
      "loss": 0.2937,
      "step": 24067
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40644389092483557,
      "learning_rate": 3.4116858725114587e-06,
      "loss": 0.2466,
      "step": 24068
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3828076175138048,
      "learning_rate": 3.410939720262898e-06,
      "loss": 0.1813,
      "step": 24069
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3715854408427831,
      "learning_rate": 3.4101936328396766e-06,
      "loss": 0.2954,
      "step": 24070
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4214885554266783,
      "learning_rate": 3.409447610249139e-06,
      "loss": 0.2964,
      "step": 24071
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.409454648828336,
      "learning_rate": 3.408701652498628e-06,
      "loss": 0.1949,
      "step": 24072
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43824872887269334,
      "learning_rate": 3.4079557595954747e-06,
      "loss": 0.0696,
      "step": 24073
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3172940303535407,
      "learning_rate": 3.407209931547023e-06,
      "loss": 0.1832,
      "step": 24074
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6785964785381097,
      "learning_rate": 3.4064641683606094e-06,
      "loss": 0.3705,
      "step": 24075
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2864181737947072,
      "learning_rate": 3.40571847004357e-06,
      "loss": 0.2211,
      "step": 24076
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.849181191118372,
      "learning_rate": 3.404972836603245e-06,
      "loss": 0.4104,
      "step": 24077
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.30238077168849875,
      "learning_rate": 3.4042272680469648e-06,
      "loss": 0.1438,
      "step": 24078
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5052184912531495,
      "learning_rate": 3.403481764382066e-06,
      "loss": 0.3403,
      "step": 24079
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2801516531621195,
      "learning_rate": 3.4027363256158853e-06,
      "loss": 0.4622,
      "step": 24080
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23421584625868125,
      "learning_rate": 3.401990951755758e-06,
      "loss": 0.1853,
      "step": 24081
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33512962431452764,
      "learning_rate": 3.401245642809009e-06,
      "loss": 0.1969,
      "step": 24082
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34899383949737883,
      "learning_rate": 3.400500398782983e-06,
      "loss": 0.2651,
      "step": 24083
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2095509513782658,
      "learning_rate": 3.3997552196850048e-06,
      "loss": 0.1217,
      "step": 24084
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7344978036778511,
      "learning_rate": 3.3990101055224013e-06,
      "loss": 0.3663,
      "step": 24085
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7743376578325641,
      "learning_rate": 3.398265056302513e-06,
      "loss": 0.295,
      "step": 24086
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.279207988050606,
      "learning_rate": 3.397520072032664e-06,
      "loss": 0.1803,
      "step": 24087
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4626970490426806,
      "learning_rate": 3.396775152720184e-06,
      "loss": 0.3353,
      "step": 24088
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4095901058184695,
      "learning_rate": 3.3960302983724037e-06,
      "loss": 0.2722,
      "step": 24089
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.32106877121938165,
      "learning_rate": 3.395285508996652e-06,
      "loss": 0.1784,
      "step": 24090
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8269568529027204,
      "learning_rate": 3.3945407846002533e-06,
      "loss": 0.0882,
      "step": 24091
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5936365642922824,
      "learning_rate": 3.393796125190534e-06,
      "loss": 0.3751,
      "step": 24092
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5599252820037248,
      "learning_rate": 3.3930515307748267e-06,
      "loss": 0.2528,
      "step": 24093
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3368409138996005,
      "learning_rate": 3.392307001360445e-06,
      "loss": 0.2866,
      "step": 24094
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.323617853876737,
      "learning_rate": 3.391562536954729e-06,
      "loss": 0.2789,
      "step": 24095
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9213395306465673,
      "learning_rate": 3.3908181375649907e-06,
      "loss": 0.3874,
      "step": 24096
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3203714004342671,
      "learning_rate": 3.3900738031985592e-06,
      "loss": 0.202,
      "step": 24097
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5595189598014871,
      "learning_rate": 3.3893295338627596e-06,
      "loss": 0.6542,
      "step": 24098
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.22303113704813765,
      "learning_rate": 3.3885853295649075e-06,
      "loss": 0.1801,
      "step": 24099
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33343826892997236,
      "learning_rate": 3.3878411903123287e-06,
      "loss": 0.0814,
      "step": 24100
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.32563949699190803,
      "learning_rate": 3.3870971161123435e-06,
      "loss": 0.2423,
      "step": 24101
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9168466710124611,
      "learning_rate": 3.3863531069722734e-06,
      "loss": 0.2945,
      "step": 24102
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3909681400940481,
      "learning_rate": 3.3856091628994412e-06,
      "loss": 0.6159,
      "step": 24103
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9643299730603493,
      "learning_rate": 3.3848652839011587e-06,
      "loss": 0.2979,
      "step": 24104
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36563977229527145,
      "learning_rate": 3.384121469984748e-06,
      "loss": 0.2609,
      "step": 24105
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2881946418469078,
      "learning_rate": 3.383377721157527e-06,
      "loss": 0.2158,
      "step": 24106
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4173207909854937,
      "learning_rate": 3.3826340374268175e-06,
      "loss": 0.7935,
      "step": 24107
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3944589865527503,
      "learning_rate": 3.381890418799928e-06,
      "loss": 0.0711,
      "step": 24108
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.46396840842610965,
      "learning_rate": 3.381146865284177e-06,
      "loss": 0.2855,
      "step": 24109
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.28402260300161514,
      "learning_rate": 3.380403376886886e-06,
      "loss": 0.1798,
      "step": 24110
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.050120874718848,
      "learning_rate": 3.3796599536153575e-06,
      "loss": 0.4472,
      "step": 24111
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3853041377404089,
      "learning_rate": 3.3789165954769186e-06,
      "loss": 0.2505,
      "step": 24112
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1449993071332125,
      "learning_rate": 3.3781733024788744e-06,
      "loss": 0.218,
      "step": 24113
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6183598010508135,
      "learning_rate": 3.3774300746285405e-06,
      "loss": 0.2888,
      "step": 24114
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34339286003090025,
      "learning_rate": 3.376686911933231e-06,
      "loss": 0.2077,
      "step": 24115
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3442453533773142,
      "learning_rate": 3.375943814400252e-06,
      "loss": 0.5872,
      "step": 24116
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2341741365561822,
      "learning_rate": 3.3752007820369183e-06,
      "loss": 0.1857,
      "step": 24117
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5733224521390156,
      "learning_rate": 3.374457814850538e-06,
      "loss": 0.3005,
      "step": 24118
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.37372160069678306,
      "learning_rate": 3.3737149128484258e-06,
      "loss": 0.1777,
      "step": 24119
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4066061076637984,
      "learning_rate": 3.3729720760378794e-06,
      "loss": 0.3184,
      "step": 24120
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.22937199897949792,
      "learning_rate": 3.372229304426221e-06,
      "loss": 0.0671,
      "step": 24121
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5267739974040098,
      "learning_rate": 3.371486598020749e-06,
      "loss": 0.3336,
      "step": 24122
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3284520409540387,
      "learning_rate": 3.3707439568287713e-06,
      "loss": 0.1844,
      "step": 24123
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36193701631327607,
      "learning_rate": 3.3700013808576004e-06,
      "loss": 0.2537,
      "step": 24124
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8747062687813159,
      "learning_rate": 3.3692588701145313e-06,
      "loss": 0.3362,
      "step": 24125
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6219412770310161,
      "learning_rate": 3.368516424606881e-06,
      "loss": 0.2935,
      "step": 24126
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7403466059352476,
      "learning_rate": 3.3677740443419447e-06,
      "loss": 0.4123,
      "step": 24127
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3154062404869039,
      "learning_rate": 3.3670317293270304e-06,
      "loss": 0.1533,
      "step": 24128
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2738006101253976,
      "learning_rate": 3.3662894795694432e-06,
      "loss": 0.2326,
      "step": 24129
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26244115136986396,
      "learning_rate": 3.3655472950764802e-06,
      "loss": 0.1518,
      "step": 24130
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7943158228481388,
      "learning_rate": 3.3648051758554446e-06,
      "loss": 0.8205,
      "step": 24131
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3108302080684512,
      "learning_rate": 3.36406312191364e-06,
      "loss": 0.1519,
      "step": 24132
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3985530599952516,
      "learning_rate": 3.363321133258369e-06,
      "loss": 0.2936,
      "step": 24133
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.4612932382628951,
      "learning_rate": 3.362579209896926e-06,
      "loss": 0.2145,
      "step": 24134
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4422627770403685,
      "learning_rate": 3.361837351836611e-06,
      "loss": 0.3222,
      "step": 24135
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4822768845707195,
      "learning_rate": 3.3610955590847295e-06,
      "loss": 0.2636,
      "step": 24136
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.388932832320885,
      "learning_rate": 3.3603538316485663e-06,
      "loss": 0.2568,
      "step": 24137
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4887386832807737,
      "learning_rate": 3.3596121695354345e-06,
      "loss": 0.2061,
      "step": 24138
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.22068744454197659,
      "learning_rate": 3.3588705727526194e-06,
      "loss": 0.0784,
      "step": 24139
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5168713715602808,
      "learning_rate": 3.3581290413074207e-06,
      "loss": 0.3445,
      "step": 24140
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2962874604976787,
      "learning_rate": 3.357387575207137e-06,
      "loss": 0.1844,
      "step": 24141
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5154444939658048,
      "learning_rate": 3.3566461744590573e-06,
      "loss": 0.2782,
      "step": 24142
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4479976459911975,
      "learning_rate": 3.3559048390704784e-06,
      "loss": 0.2284,
      "step": 24143
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6406318549293432,
      "learning_rate": 3.3551635690486938e-06,
      "loss": 0.3864,
      "step": 24144
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8737622567392986,
      "learning_rate": 3.354422364401e-06,
      "loss": 0.2459,
      "step": 24145
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9514051704804561,
      "learning_rate": 3.3536812251346818e-06,
      "loss": 0.432,
      "step": 24146
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2521445606297078,
      "learning_rate": 3.352940151257035e-06,
      "loss": 0.2001,
      "step": 24147
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.22978236249621173,
      "learning_rate": 3.352199142775353e-06,
      "loss": 0.193,
      "step": 24148
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.227339012878508,
      "learning_rate": 3.3514581996969176e-06,
      "loss": 0.2929,
      "step": 24149
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6304194436094726,
      "learning_rate": 3.3507173220290303e-06,
      "loss": 0.6906,
      "step": 24150
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2927406998144607,
      "learning_rate": 3.349976509778967e-06,
      "loss": 0.1763,
      "step": 24151
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8832386524191247,
      "learning_rate": 3.3492357629540297e-06,
      "loss": 0.3045,
      "step": 24152
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.343414646703846,
      "learning_rate": 3.3484950815614956e-06,
      "loss": 0.2918,
      "step": 24153
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.691875744330752,
      "learning_rate": 3.3477544656086557e-06,
      "loss": 0.0423,
      "step": 24154
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6747865653976252,
      "learning_rate": 3.3470139151028002e-06,
      "loss": 0.2811,
      "step": 24155
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3328809781711675,
      "learning_rate": 3.3462734300512065e-06,
      "loss": 0.1923,
      "step": 24156
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.301522771314385,
      "learning_rate": 3.345533010461165e-06,
      "loss": 0.1475,
      "step": 24157
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9743259211064932,
      "learning_rate": 3.3447926563399602e-06,
      "loss": 0.4461,
      "step": 24158
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33919491732540674,
      "learning_rate": 3.3440523676948766e-06,
      "loss": 0.2933,
      "step": 24159
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.28379546390581456,
      "learning_rate": 3.343312144533194e-06,
      "loss": 0.1746,
      "step": 24160
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9207917817936284,
      "learning_rate": 3.3425719868621963e-06,
      "loss": 0.4769,
      "step": 24161
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8816588048857824,
      "learning_rate": 3.341831894689169e-06,
      "loss": 0.3016,
      "step": 24162
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3045092368097668,
      "learning_rate": 3.3410918680213844e-06,
      "loss": 0.6494,
      "step": 24163
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3122920885109005,
      "learning_rate": 3.340351906866135e-06,
      "loss": 0.1548,
      "step": 24164
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.32068308447965443,
      "learning_rate": 3.3396120112306918e-06,
      "loss": 0.2188,
      "step": 24165
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29694323964518,
      "learning_rate": 3.3388721811223367e-06,
      "loss": 0.2207,
      "step": 24166
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4193999778768539,
      "learning_rate": 3.338132416548353e-06,
      "loss": 0.1527,
      "step": 24167
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7810582727555548,
      "learning_rate": 3.3373927175160114e-06,
      "loss": 0.3498,
      "step": 24168
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3717802322609691,
      "learning_rate": 3.3366530840325915e-06,
      "loss": 0.1704,
      "step": 24169
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5669146147492788,
      "learning_rate": 3.3359135161053726e-06,
      "loss": 0.3042,
      "step": 24170
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.32825213495671,
      "learning_rate": 3.335174013741631e-06,
      "loss": 0.2479,
      "step": 24171
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.276324091025836,
      "learning_rate": 3.3344345769486386e-06,
      "loss": 0.5359,
      "step": 24172
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9855499918199949,
      "learning_rate": 3.3336952057336714e-06,
      "loss": 0.1156,
      "step": 24173
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35909364984994996,
      "learning_rate": 3.332955900104008e-06,
      "loss": 0.2608,
      "step": 24174
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24513797332703982,
      "learning_rate": 3.3322166600669104e-06,
      "loss": 0.0766,
      "step": 24175
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4835952957285788,
      "learning_rate": 3.331477485629667e-06,
      "loss": 0.363,
      "step": 24176
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4111917069891832,
      "learning_rate": 3.330738376799538e-06,
      "loss": 0.2166,
      "step": 24177
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39858593761778227,
      "learning_rate": 3.329999333583801e-06,
      "loss": 0.1819,
      "step": 24178
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.48926548826680716,
      "learning_rate": 3.3292603559897274e-06,
      "loss": 0.3199,
      "step": 24179
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3199170631809671,
      "learning_rate": 3.3285214440245795e-06,
      "loss": 0.3032,
      "step": 24180
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.6206645298513571,
      "learning_rate": 3.327782597695638e-06,
      "loss": 0.5677,
      "step": 24181
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3038353232748719,
      "learning_rate": 3.327043817010165e-06,
      "loss": 0.1684,
      "step": 24182
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3002940239131903,
      "learning_rate": 3.32630510197543e-06,
      "loss": 0.2662,
      "step": 24183
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.1287858577200624,
      "learning_rate": 3.3255664525987043e-06,
      "loss": 0.071,
      "step": 24184
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7032669940502918,
      "learning_rate": 3.3248278688872494e-06,
      "loss": 0.8117,
      "step": 24185
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6377702748131294,
      "learning_rate": 3.324089350848334e-06,
      "loss": 0.2858,
      "step": 24186
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4281443834629608,
      "learning_rate": 3.3233508984892236e-06,
      "loss": 0.2338,
      "step": 24187
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.45490917493958477,
      "learning_rate": 3.322612511817188e-06,
      "loss": 0.2428,
      "step": 24188
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3233589046579089,
      "learning_rate": 3.3218741908394813e-06,
      "loss": 0.2802,
      "step": 24189
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2022359443332067,
      "learning_rate": 3.3211359355633797e-06,
      "loss": 0.3175,
      "step": 24190
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4401849522144071,
      "learning_rate": 3.320397745996138e-06,
      "loss": 0.076,
      "step": 24191
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.374263314994175,
      "learning_rate": 3.3196596221450205e-06,
      "loss": 0.2778,
      "step": 24192
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4010270651860307,
      "learning_rate": 3.3189215640172934e-06,
      "loss": 0.0835,
      "step": 24193
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4588416093002333,
      "learning_rate": 3.3181835716202103e-06,
      "loss": 0.3131,
      "step": 24194
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40032886584389016,
      "learning_rate": 3.317445644961036e-06,
      "loss": 0.2277,
      "step": 24195
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.44252923009459444,
      "learning_rate": 3.316707784047031e-06,
      "loss": 0.2682,
      "step": 24196
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41103805076721867,
      "learning_rate": 3.315969988885457e-06,
      "loss": 0.2486,
      "step": 24197
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4079977167524943,
      "learning_rate": 3.3152322594835663e-06,
      "loss": 0.2366,
      "step": 24198
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5282686248185795,
      "learning_rate": 3.3144945958486196e-06,
      "loss": 0.2146,
      "step": 24199
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5023638361238003,
      "learning_rate": 3.313756997987878e-06,
      "loss": 0.3362,
      "step": 24200
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27752514850116244,
      "learning_rate": 3.3130194659085903e-06,
      "loss": 0.1779,
      "step": 24201
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5267053060466627,
      "learning_rate": 3.3122819996180224e-06,
      "loss": 0.3328,
      "step": 24202
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6796553682216542,
      "learning_rate": 3.3115445991234217e-06,
      "loss": 0.2652,
      "step": 24203
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7546551577515002,
      "learning_rate": 3.3108072644320455e-06,
      "loss": 0.2844,
      "step": 24204
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.588320817380345,
      "learning_rate": 3.310069995551153e-06,
      "loss": 0.2867,
      "step": 24205
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.31698689680298253,
      "learning_rate": 3.309332792487986e-06,
      "loss": 0.2303,
      "step": 24206
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2965044927789555,
      "learning_rate": 3.308595655249811e-06,
      "loss": 0.2487,
      "step": 24207
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39304083562723535,
      "learning_rate": 3.3078585838438705e-06,
      "loss": 0.0826,
      "step": 24208
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8541089277238517,
      "learning_rate": 3.307121578277419e-06,
      "loss": 0.3179,
      "step": 24209
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3323522715851663,
      "learning_rate": 3.3063846385577112e-06,
      "loss": 0.1993,
      "step": 24210
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9604860656902425,
      "learning_rate": 3.3056477646919906e-06,
      "loss": 0.4283,
      "step": 24211
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.45647024225169724,
      "learning_rate": 3.3049109566875102e-06,
      "loss": 0.2259,
      "step": 24212
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.43203223376774974,
      "learning_rate": 3.3041742145515187e-06,
      "loss": 0.3116,
      "step": 24213
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.31027608971810117,
      "learning_rate": 3.3034375382912675e-06,
      "loss": 0.1649,
      "step": 24214
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.504245067120271,
      "learning_rate": 3.3027009279139975e-06,
      "loss": 0.3238,
      "step": 24215
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.23592573240572645,
      "learning_rate": 3.30196438342696e-06,
      "loss": 0.0917,
      "step": 24216
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4701349778539075,
      "learning_rate": 3.301227904837405e-06,
      "loss": 0.247,
      "step": 24217
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2820743576106196,
      "learning_rate": 3.3004914921525668e-06,
      "loss": 0.2034,
      "step": 24218
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29952723292750133,
      "learning_rate": 3.2997551453797037e-06,
      "loss": 0.1849,
      "step": 24219
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5659165932926447,
      "learning_rate": 3.2990188645260514e-06,
      "loss": 0.5431,
      "step": 24220
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7231033635073647,
      "learning_rate": 3.2982826495988564e-06,
      "loss": 0.2805,
      "step": 24221
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3928622523017948,
      "learning_rate": 3.297546500605362e-06,
      "loss": 0.7706,
      "step": 24222
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2782804375061481,
      "learning_rate": 3.296810417552814e-06,
      "loss": 0.1138,
      "step": 24223
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4344744683040885,
      "learning_rate": 3.2960744004484467e-06,
      "loss": 0.2906,
      "step": 24224
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.31195419922686657,
      "learning_rate": 3.2953384492995054e-06,
      "loss": 0.2349,
      "step": 24225
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.30941742885451445,
      "learning_rate": 3.294602564113234e-06,
      "loss": 0.1605,
      "step": 24226
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.348381983409376,
      "learning_rate": 3.293866744896863e-06,
      "loss": 0.1271,
      "step": 24227
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4042165296874942,
      "learning_rate": 3.2931309916576437e-06,
      "loss": 0.2553,
      "step": 24228
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4829850109423627,
      "learning_rate": 3.2923953044028043e-06,
      "loss": 0.2411,
      "step": 24229
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3298788640504088,
      "learning_rate": 3.2916596831395886e-06,
      "loss": 0.2835,
      "step": 24230
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7058323102238848,
      "learning_rate": 3.2909241278752346e-06,
      "loss": 0.1943,
      "step": 24231
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.31562245907172787,
      "learning_rate": 3.290188638616971e-06,
      "loss": 0.1574,
      "step": 24232
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3674955500713293,
      "learning_rate": 3.289453215372045e-06,
      "loss": 0.2869,
      "step": 24233
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.14115480724138854,
      "learning_rate": 3.288717858147683e-06,
      "loss": 0.0836,
      "step": 24234
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5962751949418865,
      "learning_rate": 3.2879825669511233e-06,
      "loss": 0.7948,
      "step": 24235
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.301221377105057,
      "learning_rate": 3.2872473417896034e-06,
      "loss": 0.2217,
      "step": 24236
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3808436900074356,
      "learning_rate": 3.2865121826703496e-06,
      "loss": 0.2288,
      "step": 24237
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40108639643180527,
      "learning_rate": 3.285777089600597e-06,
      "loss": 0.2567,
      "step": 24238
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2349193989514462,
      "learning_rate": 3.28504206258758e-06,
      "loss": 0.3837,
      "step": 24239
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1201812889918654,
      "learning_rate": 3.284307101638531e-06,
      "loss": 0.1868,
      "step": 24240
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6520264873450555,
      "learning_rate": 3.283572206760676e-06,
      "loss": 0.3252,
      "step": 24241
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27127576501744444,
      "learning_rate": 3.2828373779612465e-06,
      "loss": 0.2212,
      "step": 24242
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4901959655613381,
      "learning_rate": 3.2821026152474765e-06,
      "loss": 0.3019,
      "step": 24243
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21939226659444336,
      "learning_rate": 3.2813679186265858e-06,
      "loss": 0.0857,
      "step": 24244
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5361173304727358,
      "learning_rate": 3.2806332881058144e-06,
      "loss": 0.2137,
      "step": 24245
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3532550120609427,
      "learning_rate": 3.2798987236923807e-06,
      "loss": 0.2286,
      "step": 24246
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7780894778574937,
      "learning_rate": 3.279164225393514e-06,
      "loss": 0.2834,
      "step": 24247
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.32448060315660476,
      "learning_rate": 3.2784297932164442e-06,
      "loss": 0.2706,
      "step": 24248
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.3732359236778495,
      "learning_rate": 3.2776954271683914e-06,
      "loss": 0.2798,
      "step": 24249
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8219692054380371,
      "learning_rate": 3.2769611272565826e-06,
      "loss": 0.4261,
      "step": 24250
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.31242501055744154,
      "learning_rate": 3.2762268934882426e-06,
      "loss": 0.1907,
      "step": 24251
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.571713516086043,
      "learning_rate": 3.2754927258705936e-06,
      "loss": 0.7346,
      "step": 24252
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2894914535999652,
      "learning_rate": 3.2747586244108608e-06,
      "loss": 0.175,
      "step": 24253
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5780624561187993,
      "learning_rate": 3.2740245891162693e-06,
      "loss": 0.3238,
      "step": 24254
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4960743545938286,
      "learning_rate": 3.273290619994033e-06,
      "loss": 0.1708,
      "step": 24255
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.371479787299291,
      "learning_rate": 3.272556717051377e-06,
      "loss": 0.2701,
      "step": 24256
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.41081094354237857,
      "learning_rate": 3.2718228802955253e-06,
      "loss": 0.0918,
      "step": 24257
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0075915074567705,
      "learning_rate": 3.2710891097336873e-06,
      "loss": 0.3306,
      "step": 24258
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1892489569208764,
      "learning_rate": 3.2703554053730955e-06,
      "loss": 0.5092,
      "step": 24259
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26947990691649826,
      "learning_rate": 3.2696217672209573e-06,
      "loss": 0.1976,
      "step": 24260
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4949028487044833,
      "learning_rate": 3.2688881952844953e-06,
      "loss": 0.3984,
      "step": 24261
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4041837154479515,
      "learning_rate": 3.2681546895709304e-06,
      "loss": 0.0863,
      "step": 24262
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6751495033222578,
      "learning_rate": 3.26742125008747e-06,
      "loss": 0.3526,
      "step": 24263
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35352357772243814,
      "learning_rate": 3.266687876841336e-06,
      "loss": 0.1876,
      "step": 24264
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34156678120630724,
      "learning_rate": 3.265954569839741e-06,
      "loss": 0.1902,
      "step": 24265
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.28254161260114224,
      "learning_rate": 3.265221329089904e-06,
      "loss": 0.2096,
      "step": 24266
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.7702534221307331,
      "learning_rate": 3.2644881545990326e-06,
      "loss": 0.7999,
      "step": 24267
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4434389842774901,
      "learning_rate": 3.263755046374343e-06,
      "loss": 0.2629,
      "step": 24268
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3647091606129617,
      "learning_rate": 3.263022004423051e-06,
      "loss": 0.2519,
      "step": 24269
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0038737017589006,
      "learning_rate": 3.262289028752359e-06,
      "loss": 0.3076,
      "step": 24270
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39357800151061334,
      "learning_rate": 3.261556119369491e-06,
      "loss": 0.2604,
      "step": 24271
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4541677395131696,
      "learning_rate": 3.260823276281648e-06,
      "loss": 0.3312,
      "step": 24272
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.26781745335713375,
      "learning_rate": 3.2600904994960435e-06,
      "loss": 0.1772,
      "step": 24273
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.73347755759711,
      "learning_rate": 3.259357789019889e-06,
      "loss": 0.771,
      "step": 24274
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.21616123363180204,
      "learning_rate": 3.2586251448603866e-06,
      "loss": 0.0807,
      "step": 24275
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.2036981796949018,
      "learning_rate": 3.2578925670247495e-06,
      "loss": 0.6679,
      "step": 24276
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3461182554111915,
      "learning_rate": 3.2571600555201832e-06,
      "loss": 0.2096,
      "step": 24277
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4639375764071935,
      "learning_rate": 3.256427610353895e-06,
      "loss": 0.2685,
      "step": 24278
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3370395537614272,
      "learning_rate": 3.2556952315330948e-06,
      "loss": 0.2171,
      "step": 24279
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.651947599321028,
      "learning_rate": 3.25496291906498e-06,
      "loss": 0.4047,
      "step": 24280
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7563368674493882,
      "learning_rate": 3.2542306729567597e-06,
      "loss": 0.2287,
      "step": 24281
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.39158629815342605,
      "learning_rate": 3.253498493215638e-06,
      "loss": 0.2204,
      "step": 24282
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.25790116144604425,
      "learning_rate": 3.2527663798488207e-06,
      "loss": 0.16,
      "step": 24283
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2738268269946829,
      "learning_rate": 3.2520343328635005e-06,
      "loss": 0.2088,
      "step": 24284
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.70423250776933,
      "learning_rate": 3.251302352266895e-06,
      "loss": 0.6675,
      "step": 24285
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9797778832680887,
      "learning_rate": 3.2505704380661973e-06,
      "loss": 0.5497,
      "step": 24286
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5458483248122262,
      "learning_rate": 3.2498385902686016e-06,
      "loss": 0.2765,
      "step": 24287
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4996832165437363,
      "learning_rate": 3.2491068088813218e-06,
      "loss": 0.2442,
      "step": 24288
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.1183057485504928,
      "learning_rate": 3.2483750939115466e-06,
      "loss": 0.444,
      "step": 24289
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.29476081718758007,
      "learning_rate": 3.2476434453664794e-06,
      "loss": 0.2329,
      "step": 24290
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7270065307321669,
      "learning_rate": 3.246911863253318e-06,
      "loss": 0.3553,
      "step": 24291
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.24111080219402328,
      "learning_rate": 3.246180347579263e-06,
      "loss": 0.1636,
      "step": 24292
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4553402196502619,
      "learning_rate": 3.2454488983515043e-06,
      "loss": 0.2436,
      "step": 24293
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5010410163546812,
      "learning_rate": 3.2447175155772417e-06,
      "loss": 0.3383,
      "step": 24294
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.40385746530307126,
      "learning_rate": 3.2439861992636744e-06,
      "loss": 0.2965,
      "step": 24295
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27959613542987016,
      "learning_rate": 3.2432549494179878e-06,
      "loss": 0.1692,
      "step": 24296
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0734651630144807,
      "learning_rate": 3.2425237660473884e-06,
      "loss": 0.3175,
      "step": 24297
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7964921180832767,
      "learning_rate": 3.2417926491590613e-06,
      "loss": 0.3877,
      "step": 24298
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.146199252965124,
      "learning_rate": 3.2410615987602002e-06,
      "loss": 0.1225,
      "step": 24299
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.6029399262152284,
      "learning_rate": 3.2403306148580037e-06,
      "loss": 0.249,
      "step": 24300
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3251721664667507,
      "learning_rate": 3.2395996974596555e-06,
      "loss": 0.1926,
      "step": 24301
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.261792298905441,
      "learning_rate": 3.2388688465723494e-06,
      "loss": 0.244,
      "step": 24302
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33515078033556184,
      "learning_rate": 3.2381380622032767e-06,
      "loss": 0.1346,
      "step": 24303
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.7519201408574873,
      "learning_rate": 3.2374073443596266e-06,
      "loss": 0.4109,
      "step": 24304
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3962339550402218,
      "learning_rate": 3.2366766930485904e-06,
      "loss": 0.1722,
      "step": 24305
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5916839365568881,
      "learning_rate": 3.2359461082773527e-06,
      "loss": 0.3205,
      "step": 24306
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.34208465332342314,
      "learning_rate": 3.235215590053102e-06,
      "loss": 0.2434,
      "step": 24307
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.50253181593649,
      "learning_rate": 3.234485138383027e-06,
      "loss": 0.4989,
      "step": 24308
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.8695336978393944,
      "learning_rate": 3.2337547532743154e-06,
      "loss": 0.0952,
      "step": 24309
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.30572369565200525,
      "learning_rate": 3.233024434734149e-06,
      "loss": 0.178,
      "step": 24310
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.27066500856051373,
      "learning_rate": 3.232294182769714e-06,
      "loss": 0.1413,
      "step": 24311
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9994818869022408,
      "learning_rate": 3.2315639973881984e-06,
      "loss": 0.559,
      "step": 24312
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3492685655742518,
      "learning_rate": 3.2308338785967774e-06,
      "loss": 0.2957,
      "step": 24313
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.36565222999578084,
      "learning_rate": 3.2301038264026476e-06,
      "loss": 0.1683,
      "step": 24314
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.5873833342768483,
      "learning_rate": 3.229373840812979e-06,
      "loss": 0.3401,
      "step": 24315
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9108025271146343,
      "learning_rate": 3.2286439218349596e-06,
      "loss": 0.3104,
      "step": 24316
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.5373109349800862,
      "learning_rate": 3.2279140694757718e-06,
      "loss": 0.4083,
      "step": 24317
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.35503569338122687,
      "learning_rate": 3.2271842837425917e-06,
      "loss": 0.064,
      "step": 24318
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.3492358371078106,
      "learning_rate": 3.2264545646425993e-06,
      "loss": 0.2434,
      "step": 24319
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.33002008654431986,
      "learning_rate": 3.225724912182977e-06,
      "loss": 0.2347,
      "step": 24320
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.25926905271573625,
      "learning_rate": 3.2249953263709045e-06,
      "loss": 0.1689,
      "step": 24321
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.9057593469804343,
      "learning_rate": 3.2242658072135513e-06,
      "loss": 0.2854,
      "step": 24322
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.2954573890137714,
      "learning_rate": 3.2235363547181074e-06,
      "loss": 0.1829,
      "step": 24323
    },
    {
      "epoch": 0.74,
      "grad_norm": 1.0323390022938232,
      "learning_rate": 3.2228069688917387e-06,
      "loss": 0.4065,
      "step": 24324
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4410735435662351,
      "learning_rate": 3.222077649741625e-06,
      "loss": 0.2691,
      "step": 24325
    },
    {
      "epoch": 0.74,
      "grad_norm": 0.4933270744530096,
      "learning_rate": 3.221348397274946e-06,
      "loss": 0.2796,
      "step": 24326
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3823259483182369,
      "learning_rate": 3.2206192114988677e-06,
      "loss": 0.0779,
      "step": 24327
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5480882131329051,
      "learning_rate": 3.219890092420568e-06,
      "loss": 0.3197,
      "step": 24328
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.17974534504698766,
      "learning_rate": 3.2191610400472205e-06,
      "loss": 0.114,
      "step": 24329
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7992898695715155,
      "learning_rate": 3.2184320543859993e-06,
      "loss": 0.4108,
      "step": 24330
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35152392282075134,
      "learning_rate": 3.2177031354440767e-06,
      "loss": 0.2444,
      "step": 24331
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9912496785373488,
      "learning_rate": 3.21697428322862e-06,
      "loss": 0.4566,
      "step": 24332
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.28506412551456545,
      "learning_rate": 3.2162454977468014e-06,
      "loss": 0.1894,
      "step": 24333
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4839837578167736,
      "learning_rate": 3.215516779005793e-06,
      "loss": 0.7281,
      "step": 24334
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1981087060175644,
      "learning_rate": 3.2147881270127647e-06,
      "loss": 0.1562,
      "step": 24335
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3551353323989553,
      "learning_rate": 3.214059541774881e-06,
      "loss": 0.0608,
      "step": 24336
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36408884310078987,
      "learning_rate": 3.213331023299312e-06,
      "loss": 0.2637,
      "step": 24337
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2263918932329085,
      "learning_rate": 3.2126025715932298e-06,
      "loss": 0.1711,
      "step": 24338
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.181691173625843,
      "learning_rate": 3.211874186663789e-06,
      "loss": 0.4824,
      "step": 24339
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7669418870170996,
      "learning_rate": 3.2111458685181707e-06,
      "loss": 0.2733,
      "step": 24340
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5836819929211579,
      "learning_rate": 3.210417617163529e-06,
      "loss": 0.3112,
      "step": 24341
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31030752061503986,
      "learning_rate": 3.2096894326070348e-06,
      "loss": 0.2104,
      "step": 24342
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31532100265837004,
      "learning_rate": 3.208961314855852e-06,
      "loss": 0.2284,
      "step": 24343
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42166681987053406,
      "learning_rate": 3.20823326391714e-06,
      "loss": 0.2218,
      "step": 24344
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1144142604875993,
      "learning_rate": 3.2075052797980634e-06,
      "loss": 0.414,
      "step": 24345
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30777339942495,
      "learning_rate": 3.206777362505784e-06,
      "loss": 0.1827,
      "step": 24346
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4154881883632306,
      "learning_rate": 3.206049512047469e-06,
      "loss": 0.5704,
      "step": 24347
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9333019359017289,
      "learning_rate": 3.205321728430271e-06,
      "loss": 0.2915,
      "step": 24348
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30535221467758644,
      "learning_rate": 3.204594011661353e-06,
      "loss": 0.247,
      "step": 24349
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42947481861304865,
      "learning_rate": 3.203866361747877e-06,
      "loss": 0.2573,
      "step": 24350
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43813943559594126,
      "learning_rate": 3.203138778696996e-06,
      "loss": 0.2079,
      "step": 24351
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.45207742942549467,
      "learning_rate": 3.2024112625158765e-06,
      "loss": 0.2567,
      "step": 24352
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21963639220528153,
      "learning_rate": 3.20168381321167e-06,
      "loss": 0.067,
      "step": 24353
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43957832444254785,
      "learning_rate": 3.200956430791534e-06,
      "loss": 0.2905,
      "step": 24354
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3830047832619551,
      "learning_rate": 3.2002291152626265e-06,
      "loss": 0.1695,
      "step": 24355
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.37262543408968224,
      "learning_rate": 3.199501866632102e-06,
      "loss": 0.2725,
      "step": 24356
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.885538047026436,
      "learning_rate": 3.19877468490712e-06,
      "loss": 0.2924,
      "step": 24357
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.392415072078266,
      "learning_rate": 3.1980475700948267e-06,
      "loss": 0.7501,
      "step": 24358
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3649755331144598,
      "learning_rate": 3.1973205222023794e-06,
      "loss": 0.1348,
      "step": 24359
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34795985761711684,
      "learning_rate": 3.196593541236931e-06,
      "loss": 0.2493,
      "step": 24360
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2322033854489974,
      "learning_rate": 3.195866627205637e-06,
      "loss": 0.1898,
      "step": 24361
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38315474933860255,
      "learning_rate": 3.1951397801156437e-06,
      "loss": 0.0817,
      "step": 24362
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7535534971804516,
      "learning_rate": 3.1944129999741037e-06,
      "loss": 0.7999,
      "step": 24363
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3354879703268053,
      "learning_rate": 3.1936862867881714e-06,
      "loss": 0.151,
      "step": 24364
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4977992887338921,
      "learning_rate": 3.1929596405649865e-06,
      "loss": 0.2995,
      "step": 24365
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7996144196675348,
      "learning_rate": 3.1922330613117103e-06,
      "loss": 0.2816,
      "step": 24366
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34578491611979906,
      "learning_rate": 3.1915065490354824e-06,
      "loss": 0.2811,
      "step": 24367
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3388976747270111,
      "learning_rate": 3.190780103743454e-06,
      "loss": 0.16,
      "step": 24368
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3971874457790628,
      "learning_rate": 3.1900537254427745e-06,
      "loss": 0.314,
      "step": 24369
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21567348734549519,
      "learning_rate": 3.189327414140584e-06,
      "loss": 0.1019,
      "step": 24370
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5306689424405197,
      "learning_rate": 3.1886011698440312e-06,
      "loss": 0.2045,
      "step": 24371
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31761606001540116,
      "learning_rate": 3.1878749925602605e-06,
      "loss": 0.2189,
      "step": 24372
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.44243755634092585,
      "learning_rate": 3.1871488822964213e-06,
      "loss": 0.2902,
      "step": 24373
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6090424634098406,
      "learning_rate": 3.18642283905965e-06,
      "loss": 0.2451,
      "step": 24374
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9811882493539212,
      "learning_rate": 3.1856968628570916e-06,
      "loss": 0.2884,
      "step": 24375
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2190145309667368,
      "learning_rate": 3.184970953695894e-06,
      "loss": 0.6032,
      "step": 24376
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3163632301287405,
      "learning_rate": 3.1842451115831883e-06,
      "loss": 0.1421,
      "step": 24377
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5860602675633194,
      "learning_rate": 3.1835193365261273e-06,
      "loss": 0.3341,
      "step": 24378
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.28995308932103514,
      "learning_rate": 3.182793628531844e-06,
      "loss": 0.2174,
      "step": 24379
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3426865288147033,
      "learning_rate": 3.18206798760748e-06,
      "loss": 0.2088,
      "step": 24380
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8148156983235367,
      "learning_rate": 3.181342413760177e-06,
      "loss": 0.2231,
      "step": 24381
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9687815477205868,
      "learning_rate": 3.1806169069970647e-06,
      "loss": 0.4669,
      "step": 24382
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4852330350596814,
      "learning_rate": 3.1798914673252936e-06,
      "loss": 0.2406,
      "step": 24383
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5528361877884695,
      "learning_rate": 3.179166094751991e-06,
      "loss": 0.5595,
      "step": 24384
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31241074899248855,
      "learning_rate": 3.1784407892842973e-06,
      "loss": 0.2546,
      "step": 24385
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7499683712299954,
      "learning_rate": 3.17771555092935e-06,
      "loss": 0.8182,
      "step": 24386
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29267908839654516,
      "learning_rate": 3.17699037969428e-06,
      "loss": 0.1733,
      "step": 24387
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41418836074319726,
      "learning_rate": 3.1762652755862234e-06,
      "loss": 0.1071,
      "step": 24388
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46955404507352294,
      "learning_rate": 3.1755402386123133e-06,
      "loss": 0.2817,
      "step": 24389
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31229747663555846,
      "learning_rate": 3.1748152687796875e-06,
      "loss": 0.2198,
      "step": 24390
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5822303044556465,
      "learning_rate": 3.1740903660954693e-06,
      "loss": 0.2481,
      "step": 24391
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39216465419973373,
      "learning_rate": 3.173365530566801e-06,
      "loss": 0.2336,
      "step": 24392
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7100389926273023,
      "learning_rate": 3.172640762200807e-06,
      "loss": 0.7615,
      "step": 24393
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0561705465063478,
      "learning_rate": 3.1719160610046195e-06,
      "loss": 0.4336,
      "step": 24394
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1604589534504886,
      "learning_rate": 3.1711914269853715e-06,
      "loss": 0.6158,
      "step": 24395
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2836827605575745,
      "learning_rate": 3.1704668601501853e-06,
      "loss": 0.1756,
      "step": 24396
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.33742805857383856,
      "learning_rate": 3.169742360506194e-06,
      "loss": 0.2827,
      "step": 24397
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.22096715287553903,
      "learning_rate": 3.169017928060525e-06,
      "loss": 0.0701,
      "step": 24398
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6178733716419467,
      "learning_rate": 3.168293562820308e-06,
      "loss": 0.3687,
      "step": 24399
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3690021361530975,
      "learning_rate": 3.1675692647926647e-06,
      "loss": 0.185,
      "step": 24400
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35027230159044703,
      "learning_rate": 3.1668450339847225e-06,
      "loss": 0.2138,
      "step": 24401
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5112425817110199,
      "learning_rate": 3.1661208704036105e-06,
      "loss": 0.0178,
      "step": 24402
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3310939386234601,
      "learning_rate": 3.165396774056443e-06,
      "loss": 0.2438,
      "step": 24403
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7871162189793803,
      "learning_rate": 3.1646727449503576e-06,
      "loss": 0.4459,
      "step": 24404
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3687573215903506,
      "learning_rate": 3.163948783092469e-06,
      "loss": 0.1809,
      "step": 24405
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5385125061099983,
      "learning_rate": 3.1632248884899e-06,
      "loss": 0.2837,
      "step": 24406
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6171160332617238,
      "learning_rate": 3.1625010611497776e-06,
      "loss": 0.2915,
      "step": 24407
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.32735851128359983,
      "learning_rate": 3.161777301079214e-06,
      "loss": 0.2773,
      "step": 24408
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.20285279375414814,
      "learning_rate": 3.1610536082853404e-06,
      "loss": 0.0708,
      "step": 24409
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3497958276037764,
      "learning_rate": 3.16032998277527e-06,
      "loss": 0.2531,
      "step": 24410
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4645559235477165,
      "learning_rate": 3.159606424556123e-06,
      "loss": 0.0982,
      "step": 24411
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5507149987049453,
      "learning_rate": 3.1588829336350225e-06,
      "loss": 0.8005,
      "step": 24412
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9468242562099238,
      "learning_rate": 3.1581595100190785e-06,
      "loss": 0.2842,
      "step": 24413
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3928302304690404,
      "learning_rate": 3.157436153715413e-06,
      "loss": 0.2261,
      "step": 24414
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.25834198391297636,
      "learning_rate": 3.156712864731143e-06,
      "loss": 0.2092,
      "step": 24415
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5909160563232708,
      "learning_rate": 3.1559896430733863e-06,
      "loss": 0.2746,
      "step": 24416
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9480338135610964,
      "learning_rate": 3.155266488749251e-06,
      "loss": 0.4437,
      "step": 24417
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.16924236437316367,
      "learning_rate": 3.1545434017658574e-06,
      "loss": 0.0712,
      "step": 24418
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29823506130815575,
      "learning_rate": 3.1538203821303213e-06,
      "loss": 0.2267,
      "step": 24419
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3129696269747367,
      "learning_rate": 3.1530974298497462e-06,
      "loss": 0.2097,
      "step": 24420
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5928320739446928,
      "learning_rate": 3.1523745449312594e-06,
      "loss": 0.7067,
      "step": 24421
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9683436285411459,
      "learning_rate": 3.1516517273819603e-06,
      "loss": 0.5155,
      "step": 24422
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.421333347857495,
      "learning_rate": 3.1509289772089645e-06,
      "loss": 0.269,
      "step": 24423
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.47374981263625415,
      "learning_rate": 3.1502062944193843e-06,
      "loss": 0.2448,
      "step": 24424
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9414359373764299,
      "learning_rate": 3.1494836790203308e-06,
      "loss": 0.4202,
      "step": 24425
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31154157336960375,
      "learning_rate": 3.1487611310189083e-06,
      "loss": 0.223,
      "step": 24426
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36481171214656444,
      "learning_rate": 3.148038650422228e-06,
      "loss": 0.2064,
      "step": 24427
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2954133869856553,
      "learning_rate": 3.147316237237401e-06,
      "loss": 0.1828,
      "step": 24428
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4212316764229658,
      "learning_rate": 3.1465938914715268e-06,
      "loss": 0.0903,
      "step": 24429
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5590130675746328,
      "learning_rate": 3.1458716131317213e-06,
      "loss": 0.8181,
      "step": 24430
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35970568776860357,
      "learning_rate": 3.1451494022250838e-06,
      "loss": 0.1946,
      "step": 24431
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.45737536943732665,
      "learning_rate": 3.144427258758722e-06,
      "loss": 0.2758,
      "step": 24432
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3287367515012161,
      "learning_rate": 3.1437051827397435e-06,
      "loss": 0.2249,
      "step": 24433
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7020038813561205,
      "learning_rate": 3.142983174175244e-06,
      "loss": 0.3689,
      "step": 24434
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7432783595799384,
      "learning_rate": 3.1422612330723377e-06,
      "loss": 0.1276,
      "step": 24435
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41378032035004425,
      "learning_rate": 3.1415393594381194e-06,
      "loss": 0.2232,
      "step": 24436
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2943981025249582,
      "learning_rate": 3.1408175532796925e-06,
      "loss": 0.1814,
      "step": 24437
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.33365187717338574,
      "learning_rate": 3.1400958146041637e-06,
      "loss": 0.2582,
      "step": 24438
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4122812752742572,
      "learning_rate": 3.1393741434186264e-06,
      "loss": 0.2666,
      "step": 24439
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9462767730874095,
      "learning_rate": 3.1386525397301835e-06,
      "loss": 0.7734,
      "step": 24440
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7604396758502155,
      "learning_rate": 3.1379310035459343e-06,
      "loss": 0.1674,
      "step": 24441
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38006905291856014,
      "learning_rate": 3.1372095348729812e-06,
      "loss": 0.2365,
      "step": 24442
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8595228245276066,
      "learning_rate": 3.136488133718415e-06,
      "loss": 0.3915,
      "step": 24443
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31129938264787005,
      "learning_rate": 3.1357668000893382e-06,
      "loss": 0.2214,
      "step": 24444
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8387177722323104,
      "learning_rate": 3.1350455339928488e-06,
      "loss": 0.3456,
      "step": 24445
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30377639081839475,
      "learning_rate": 3.134324335436033e-06,
      "loss": 0.1805,
      "step": 24446
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26886770058341014,
      "learning_rate": 3.1336032044260014e-06,
      "loss": 0.1696,
      "step": 24447
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.122247828452953,
      "learning_rate": 3.132882140969837e-06,
      "loss": 0.4551,
      "step": 24448
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8204822751760369,
      "learning_rate": 3.1321611450746382e-06,
      "loss": 0.4307,
      "step": 24449
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2619538743917734,
      "learning_rate": 3.1314402167474998e-06,
      "loss": 0.1819,
      "step": 24450
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36196083135346657,
      "learning_rate": 3.1307193559955106e-06,
      "loss": 0.2742,
      "step": 24451
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8982403700060079,
      "learning_rate": 3.1299985628257633e-06,
      "loss": 0.2813,
      "step": 24452
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3096413776206346,
      "learning_rate": 3.1292778372453526e-06,
      "loss": 0.5514,
      "step": 24453
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3537357276219914,
      "learning_rate": 3.1285571792613687e-06,
      "loss": 0.0882,
      "step": 24454
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30127662207462474,
      "learning_rate": 3.127836588880895e-06,
      "loss": 0.1833,
      "step": 24455
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.33632310491117506,
      "learning_rate": 3.127116066111032e-06,
      "loss": 0.2506,
      "step": 24456
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3867085510772651,
      "learning_rate": 3.12639561095886e-06,
      "loss": 0.2749,
      "step": 24457
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8303716746107539,
      "learning_rate": 3.1256752234314693e-06,
      "loss": 0.3585,
      "step": 24458
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5593484695643072,
      "learning_rate": 3.1249549035359505e-06,
      "loss": 0.1753,
      "step": 24459
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39075249555845926,
      "learning_rate": 3.124234651279382e-06,
      "loss": 0.3031,
      "step": 24460
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1336376749793138,
      "learning_rate": 3.123514466668861e-06,
      "loss": 0.2412,
      "step": 24461
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3469270528848365,
      "learning_rate": 3.1227943497114655e-06,
      "loss": 0.2874,
      "step": 24462
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8249019374810166,
      "learning_rate": 3.1220743004142805e-06,
      "loss": 0.0985,
      "step": 24463
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5110461526724882,
      "learning_rate": 3.1213543187843953e-06,
      "loss": 0.2976,
      "step": 24464
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27936100179796697,
      "learning_rate": 3.120634404828887e-06,
      "loss": 0.1533,
      "step": 24465
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9873189484773179,
      "learning_rate": 3.1199145585548405e-06,
      "loss": 0.5452,
      "step": 24466
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0441686771536494,
      "learning_rate": 3.119194779969339e-06,
      "loss": 0.2778,
      "step": 24467
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27522093596117514,
      "learning_rate": 3.1184750690794665e-06,
      "loss": 0.2026,
      "step": 24468
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3837038263473571,
      "learning_rate": 3.117755425892297e-06,
      "loss": 0.2879,
      "step": 24469
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3797517371051319,
      "learning_rate": 3.1170358504149146e-06,
      "loss": 0.1028,
      "step": 24470
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4388765038872147,
      "learning_rate": 3.1163163426544006e-06,
      "loss": 0.5742,
      "step": 24471
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41337193943954864,
      "learning_rate": 3.1155969026178257e-06,
      "loss": 0.0798,
      "step": 24472
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4091260618327436,
      "learning_rate": 3.1148775303122803e-06,
      "loss": 0.261,
      "step": 24473
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29863873008380304,
      "learning_rate": 3.114158225744831e-06,
      "loss": 0.2326,
      "step": 24474
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7365269692590974,
      "learning_rate": 3.1134389889225604e-06,
      "loss": 0.7631,
      "step": 24475
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6810508946375737,
      "learning_rate": 3.1127198198525453e-06,
      "loss": 0.3009,
      "step": 24476
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.842910314897971,
      "learning_rate": 3.1120007185418553e-06,
      "loss": 0.3016,
      "step": 24477
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23463344934929678,
      "learning_rate": 3.111281684997569e-06,
      "loss": 0.1648,
      "step": 24478
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9930502707206894,
      "learning_rate": 3.1105627192267608e-06,
      "loss": 0.6494,
      "step": 24479
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.31345024470631677,
      "learning_rate": 3.1098438212365067e-06,
      "loss": 0.214,
      "step": 24480
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34006892865800714,
      "learning_rate": 3.1091249910338717e-06,
      "loss": 0.0692,
      "step": 24481
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34642364161482714,
      "learning_rate": 3.1084062286259344e-06,
      "loss": 0.2366,
      "step": 24482
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9932075646954871,
      "learning_rate": 3.107687534019762e-06,
      "loss": 0.3122,
      "step": 24483
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2079152221931695,
      "learning_rate": 3.1069689072224296e-06,
      "loss": 0.3763,
      "step": 24484
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3386353005765271,
      "learning_rate": 3.106250348241007e-06,
      "loss": 0.2477,
      "step": 24485
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6018831247313025,
      "learning_rate": 3.1055318570825586e-06,
      "loss": 0.3001,
      "step": 24486
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35026708016674807,
      "learning_rate": 3.104813433754157e-06,
      "loss": 0.2001,
      "step": 24487
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2779332043330977,
      "learning_rate": 3.104095078262873e-06,
      "loss": 0.1738,
      "step": 24488
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1482890834441886,
      "learning_rate": 3.1033767906157642e-06,
      "loss": 0.136,
      "step": 24489
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.7123992442960607,
      "learning_rate": 3.10265857081991e-06,
      "loss": 0.6644,
      "step": 24490
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29765493376549823,
      "learning_rate": 3.1019404188823667e-06,
      "loss": 0.0675,
      "step": 24491
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3164331341602233,
      "learning_rate": 3.1012223348102043e-06,
      "loss": 0.276,
      "step": 24492
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5149917916140605,
      "learning_rate": 3.1005043186104854e-06,
      "loss": 0.2578,
      "step": 24493
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9580104688799408,
      "learning_rate": 3.0997863702902797e-06,
      "loss": 0.2981,
      "step": 24494
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8505916417821335,
      "learning_rate": 3.0990684898566426e-06,
      "loss": 0.5022,
      "step": 24495
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29196450328871193,
      "learning_rate": 3.0983506773166395e-06,
      "loss": 0.1884,
      "step": 24496
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2284436314542797,
      "learning_rate": 3.097632932677337e-06,
      "loss": 0.1841,
      "step": 24497
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4027355593012477,
      "learning_rate": 3.096915255945787e-06,
      "loss": 0.1918,
      "step": 24498
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.602921396648157,
      "learning_rate": 3.096197647129062e-06,
      "loss": 0.7529,
      "step": 24499
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38310475892864504,
      "learning_rate": 3.0954801062342134e-06,
      "loss": 0.1595,
      "step": 24500
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5528545004532022,
      "learning_rate": 3.0947626332683035e-06,
      "loss": 0.325,
      "step": 24501
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9203450721632689,
      "learning_rate": 3.0940452282383937e-06,
      "loss": 0.2723,
      "step": 24502
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5024492197300748,
      "learning_rate": 3.0933278911515365e-06,
      "loss": 0.3278,
      "step": 24503
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3328402104662885,
      "learning_rate": 3.092610622014791e-06,
      "loss": 0.2174,
      "step": 24504
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34585712827893716,
      "learning_rate": 3.091893420835216e-06,
      "loss": 0.249,
      "step": 24505
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1952509611416506,
      "learning_rate": 3.091176287619867e-06,
      "loss": 0.0893,
      "step": 24506
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.418281974967115,
      "learning_rate": 3.0904592223758014e-06,
      "loss": 0.0845,
      "step": 24507
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8556075449767634,
      "learning_rate": 3.0897422251100685e-06,
      "loss": 0.3707,
      "step": 24508
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.26990043183240414,
      "learning_rate": 3.089025295829725e-06,
      "loss": 0.1704,
      "step": 24509
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.48268255808514926,
      "learning_rate": 3.088308434541826e-06,
      "loss": 0.3269,
      "step": 24510
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42685315353599773,
      "learning_rate": 3.0875916412534246e-06,
      "loss": 0.2335,
      "step": 24511
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.155122046517471,
      "learning_rate": 3.086874915971568e-06,
      "loss": 0.5162,
      "step": 24512
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3136653861855696,
      "learning_rate": 3.0861582587033124e-06,
      "loss": 0.1501,
      "step": 24513
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5872885817140379,
      "learning_rate": 3.0854416694557075e-06,
      "loss": 0.3721,
      "step": 24514
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2396367449359967,
      "learning_rate": 3.0847251482357988e-06,
      "loss": 0.1714,
      "step": 24515
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.30866140389924807,
      "learning_rate": 3.0840086950506444e-06,
      "loss": 0.2107,
      "step": 24516
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8511321094742421,
      "learning_rate": 3.0832923099072842e-06,
      "loss": 0.1967,
      "step": 24517
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.762786889446828,
      "learning_rate": 3.082575992812772e-06,
      "loss": 0.3882,
      "step": 24518
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3168149877639117,
      "learning_rate": 3.081859743774155e-06,
      "loss": 0.1996,
      "step": 24519
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7738230761263576,
      "learning_rate": 3.0811435627984766e-06,
      "loss": 0.2912,
      "step": 24520
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36429250714446804,
      "learning_rate": 3.0804274498927834e-06,
      "loss": 0.3149,
      "step": 24521
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5067414266378392,
      "learning_rate": 3.079711405064121e-06,
      "loss": 0.2152,
      "step": 24522
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3807165276580494,
      "learning_rate": 3.078995428319539e-06,
      "loss": 0.2848,
      "step": 24523
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2093417228900516,
      "learning_rate": 3.078279519666071e-06,
      "loss": 0.0863,
      "step": 24524
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5940938288884896,
      "learning_rate": 3.0775636791107733e-06,
      "loss": 0.6086,
      "step": 24525
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9087278430677382,
      "learning_rate": 3.0768479066606783e-06,
      "loss": 0.2923,
      "step": 24526
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3797583591181743,
      "learning_rate": 3.076132202322831e-06,
      "loss": 0.25,
      "step": 24527
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.28025002057782133,
      "learning_rate": 3.075416566104277e-06,
      "loss": 0.2132,
      "step": 24528
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6406259253281683,
      "learning_rate": 3.0747009980120503e-06,
      "loss": 0.9104,
      "step": 24529
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9602781344773178,
      "learning_rate": 3.0739854980531935e-06,
      "loss": 0.4396,
      "step": 24530
    },
    {
      "epoch": 0.75,
      "grad_norm": 2.7430114628883127,
      "learning_rate": 3.073270066234747e-06,
      "loss": 0.7326,
      "step": 24531
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2892723876303989,
      "learning_rate": 3.072554702563748e-06,
      "loss": 0.1809,
      "step": 24532
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4015984859755007,
      "learning_rate": 3.071839407047239e-06,
      "loss": 0.2065,
      "step": 24533
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2261067562684206,
      "learning_rate": 3.0711241796922507e-06,
      "loss": 0.1855,
      "step": 24534
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8721731179428481,
      "learning_rate": 3.070409020505822e-06,
      "loss": 0.2921,
      "step": 24535
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6725428707991723,
      "learning_rate": 3.0696939294949903e-06,
      "loss": 0.2433,
      "step": 24536
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.33466354793850045,
      "learning_rate": 3.068978906666792e-06,
      "loss": 0.1947,
      "step": 24537
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2248335294083674,
      "learning_rate": 3.0682639520282575e-06,
      "loss": 0.3732,
      "step": 24538
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2884951961039464,
      "learning_rate": 3.0675490655864228e-06,
      "loss": 0.2191,
      "step": 24539
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8698744412787055,
      "learning_rate": 3.0668342473483248e-06,
      "loss": 0.5674,
      "step": 24540
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4182958219546213,
      "learning_rate": 3.066119497320986e-06,
      "loss": 0.1953,
      "step": 24541
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5389049185220393,
      "learning_rate": 3.0654048155114526e-06,
      "loss": 0.2897,
      "step": 24542
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.32840333312962144,
      "learning_rate": 3.0646902019267442e-06,
      "loss": 0.147,
      "step": 24543
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.456936994788981,
      "learning_rate": 3.0639756565738966e-06,
      "loss": 0.3114,
      "step": 24544
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3278799254716086,
      "learning_rate": 3.063261179459941e-06,
      "loss": 0.1744,
      "step": 24545
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29845372828372313,
      "learning_rate": 3.062546770591901e-06,
      "loss": 0.1767,
      "step": 24546
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.8519620868684215,
      "learning_rate": 3.0618324299768097e-06,
      "loss": 0.6925,
      "step": 24547
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9457617679775521,
      "learning_rate": 3.061118157621693e-06,
      "loss": 0.5253,
      "step": 24548
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.532055270792163,
      "learning_rate": 3.0604039535335827e-06,
      "loss": 0.7517,
      "step": 24549
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3073119125225463,
      "learning_rate": 3.0596898177194977e-06,
      "loss": 0.1512,
      "step": 24550
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35696027636571354,
      "learning_rate": 3.058975750186469e-06,
      "loss": 0.2567,
      "step": 24551
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.38870729562805345,
      "learning_rate": 3.0582617509415237e-06,
      "loss": 0.2412,
      "step": 24552
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7599325137115459,
      "learning_rate": 3.0575478199916764e-06,
      "loss": 0.4351,
      "step": 24553
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.14797175383625366,
      "learning_rate": 3.056833957343965e-06,
      "loss": 0.068,
      "step": 24554
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.42517819701516735,
      "learning_rate": 3.0561201630054026e-06,
      "loss": 0.295,
      "step": 24555
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34598214478978,
      "learning_rate": 3.055406436983015e-06,
      "loss": 0.0101,
      "step": 24556
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35435696012821455,
      "learning_rate": 3.0546927792838267e-06,
      "loss": 0.2862,
      "step": 24557
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3593406267391843,
      "learning_rate": 3.0539791899148495e-06,
      "loss": 0.3837,
      "step": 24558
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27802344320263983,
      "learning_rate": 3.0532656688831174e-06,
      "loss": 0.1782,
      "step": 24559
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.4182131511714235,
      "learning_rate": 3.052552216195641e-06,
      "loss": 0.5988,
      "step": 24560
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6927371148002768,
      "learning_rate": 3.0518388318594404e-06,
      "loss": 0.2773,
      "step": 24561
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5209623575076914,
      "learning_rate": 3.0511255158815366e-06,
      "loss": 0.326,
      "step": 24562
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2345435258996972,
      "learning_rate": 3.0504122682689496e-06,
      "loss": 0.1463,
      "step": 24563
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5930019541369772,
      "learning_rate": 3.0496990890286905e-06,
      "loss": 0.3589,
      "step": 24564
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3056021219151141,
      "learning_rate": 3.048985978167778e-06,
      "loss": 0.1481,
      "step": 24565
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2914133248067925,
      "learning_rate": 3.0482729356932315e-06,
      "loss": 0.7293,
      "step": 24566
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9417258914458841,
      "learning_rate": 3.0475599616120577e-06,
      "loss": 0.2876,
      "step": 24567
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0010573142293244,
      "learning_rate": 3.0468470559312834e-06,
      "loss": 0.4862,
      "step": 24568
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.25117267865186615,
      "learning_rate": 3.0461342186579114e-06,
      "loss": 0.1906,
      "step": 24569
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4319929325961843,
      "learning_rate": 3.04542144979896e-06,
      "loss": 0.2916,
      "step": 24570
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3520375321400016,
      "learning_rate": 3.0447087493614435e-06,
      "loss": 0.1318,
      "step": 24571
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.19752510425070158,
      "learning_rate": 3.0439961173523668e-06,
      "loss": 0.0696,
      "step": 24572
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3489727293967294,
      "learning_rate": 3.0432835537787465e-06,
      "loss": 0.2534,
      "step": 24573
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4405359366181764,
      "learning_rate": 3.042571058647591e-06,
      "loss": 0.2143,
      "step": 24574
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.466980694576444,
      "learning_rate": 3.0418586319659136e-06,
      "loss": 0.326,
      "step": 24575
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.632721032070198,
      "learning_rate": 3.041146273740718e-06,
      "loss": 0.2439,
      "step": 24576
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5477573493294242,
      "learning_rate": 3.0404339839790154e-06,
      "loss": 0.34,
      "step": 24577
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.46413025664955904,
      "learning_rate": 3.039721762687816e-06,
      "loss": 0.2274,
      "step": 24578
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1441911970248029,
      "learning_rate": 3.0390096098741173e-06,
      "loss": 0.452,
      "step": 24579
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3964609863637362,
      "learning_rate": 3.03829752554494e-06,
      "loss": 0.2057,
      "step": 24580
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.34982498780432886,
      "learning_rate": 3.0375855097072782e-06,
      "loss": 0.25,
      "step": 24581
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2814608597338135,
      "learning_rate": 3.036873562368141e-06,
      "loss": 0.1716,
      "step": 24582
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4458600639480188,
      "learning_rate": 3.0361616835345365e-06,
      "loss": 0.2314,
      "step": 24583
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0558260285327903,
      "learning_rate": 3.0354498732134584e-06,
      "loss": 0.5489,
      "step": 24584
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7014071482389928,
      "learning_rate": 3.034738131411923e-06,
      "loss": 0.4172,
      "step": 24585
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3273351741181527,
      "learning_rate": 3.0340264581369217e-06,
      "loss": 0.1892,
      "step": 24586
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3202641407931553,
      "learning_rate": 3.0333148533954604e-06,
      "loss": 0.2148,
      "step": 24587
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4897431899591586,
      "learning_rate": 3.0326033171945436e-06,
      "loss": 0.327,
      "step": 24588
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2096101747980403,
      "learning_rate": 3.0318918495411644e-06,
      "loss": 0.1745,
      "step": 24589
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2574484028239836,
      "learning_rate": 3.031180450442326e-06,
      "loss": 0.1662,
      "step": 24590
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3962417579145854,
      "learning_rate": 3.0304691199050275e-06,
      "loss": 0.1822,
      "step": 24591
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4905406421529656,
      "learning_rate": 3.0297578579362697e-06,
      "loss": 0.3122,
      "step": 24592
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.301753708801257,
      "learning_rate": 3.0290466645430417e-06,
      "loss": 0.2343,
      "step": 24593
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7380232315473005,
      "learning_rate": 3.028335539732352e-06,
      "loss": 0.4094,
      "step": 24594
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.68150404475214,
      "learning_rate": 3.027624483511189e-06,
      "loss": 0.2119,
      "step": 24595
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.35639280137516544,
      "learning_rate": 3.0269134958865497e-06,
      "loss": 0.277,
      "step": 24596
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7210067389147755,
      "learning_rate": 3.0262025768654334e-06,
      "loss": 0.1959,
      "step": 24597
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3374825773227685,
      "learning_rate": 3.025491726454829e-06,
      "loss": 0.2743,
      "step": 24598
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7532456578955278,
      "learning_rate": 3.0247809446617302e-06,
      "loss": 0.0251,
      "step": 24599
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.23683450242006351,
      "learning_rate": 3.0240702314931325e-06,
      "loss": 0.1301,
      "step": 24600
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2967809324309424,
      "learning_rate": 3.0233595869560294e-06,
      "loss": 0.2081,
      "step": 24601
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.7722145854042947,
      "learning_rate": 3.022649011057408e-06,
      "loss": 0.311,
      "step": 24602
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9280664824454024,
      "learning_rate": 3.0219385038042614e-06,
      "loss": 0.3997,
      "step": 24603
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3340233778881263,
      "learning_rate": 3.021228065203583e-06,
      "loss": 0.2024,
      "step": 24604
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2834692261392241,
      "learning_rate": 3.020517695262353e-06,
      "loss": 0.2315,
      "step": 24605
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9665014539678203,
      "learning_rate": 3.0198073939875726e-06,
      "loss": 0.3008,
      "step": 24606
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0896079616029155,
      "learning_rate": 3.019097161386221e-06,
      "loss": 0.4171,
      "step": 24607
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4345334082021704,
      "learning_rate": 3.018386997465288e-06,
      "loss": 0.1096,
      "step": 24608
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39664871588827,
      "learning_rate": 3.017676902231764e-06,
      "loss": 0.2375,
      "step": 24609
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.43280635075237955,
      "learning_rate": 3.0169668756926263e-06,
      "loss": 0.2316,
      "step": 24610
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3369869909907257,
      "learning_rate": 3.016256917854873e-06,
      "loss": 0.2822,
      "step": 24611
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.2137572103811207,
      "learning_rate": 3.0155470287254775e-06,
      "loss": 0.298,
      "step": 24612
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.27070934804986374,
      "learning_rate": 3.0148372083114297e-06,
      "loss": 0.1543,
      "step": 24613
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.41998228784139213,
      "learning_rate": 3.0141274566197144e-06,
      "loss": 0.3039,
      "step": 24614
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.25468748318948403,
      "learning_rate": 3.013417773657309e-06,
      "loss": 0.0645,
      "step": 24615
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.40522989187634356,
      "learning_rate": 3.012708159431199e-06,
      "loss": 0.2744,
      "step": 24616
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8109428036306626,
      "learning_rate": 3.0119986139483636e-06,
      "loss": 0.1279,
      "step": 24617
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0716411029257105,
      "learning_rate": 3.0112891372157903e-06,
      "loss": 0.4202,
      "step": 24618
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.21711749914441458,
      "learning_rate": 3.0105797292404494e-06,
      "loss": 0.1591,
      "step": 24619
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.0603795128512787,
      "learning_rate": 3.0098703900293257e-06,
      "loss": 0.4485,
      "step": 24620
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6635424430855349,
      "learning_rate": 3.0091611195894e-06,
      "loss": 0.2616,
      "step": 24621
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5122553789835275,
      "learning_rate": 3.0084519179276417e-06,
      "loss": 0.331,
      "step": 24622
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2831018979111034,
      "learning_rate": 3.0077427850510398e-06,
      "loss": 0.1787,
      "step": 24623
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4301445745278389,
      "learning_rate": 3.0070337209665625e-06,
      "loss": 0.2268,
      "step": 24624
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.6367061439936126,
      "learning_rate": 3.006324725681188e-06,
      "loss": 0.1647,
      "step": 24625
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1112989472918746,
      "learning_rate": 3.005615799201892e-06,
      "loss": 0.0789,
      "step": 24626
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.6655476635404317,
      "learning_rate": 3.0049069415356525e-06,
      "loss": 0.3734,
      "step": 24627
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29680426506422386,
      "learning_rate": 3.004198152689437e-06,
      "loss": 0.2185,
      "step": 24628
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.5718103853878101,
      "learning_rate": 3.0034894326702224e-06,
      "loss": 0.3223,
      "step": 24629
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.1750955308048534,
      "learning_rate": 3.002780781484983e-06,
      "loss": 0.3138,
      "step": 24630
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3885418275333393,
      "learning_rate": 3.0020721991406834e-06,
      "loss": 0.2381,
      "step": 24631
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.29009019428464533,
      "learning_rate": 3.0013636856443063e-06,
      "loss": 0.179,
      "step": 24632
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4659044380567118,
      "learning_rate": 3.0006552410028122e-06,
      "loss": 0.2155,
      "step": 24633
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.422659397616146,
      "learning_rate": 2.9999468652231757e-06,
      "loss": 0.2141,
      "step": 24634
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.45310706877202284,
      "learning_rate": 2.9992385583123683e-06,
      "loss": 0.2954,
      "step": 24635
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3054864968628893,
      "learning_rate": 2.9985303202773487e-06,
      "loss": 0.1459,
      "step": 24636
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.692245417849321,
      "learning_rate": 2.997822151125098e-06,
      "loss": 0.346,
      "step": 24637
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3627429333870356,
      "learning_rate": 2.997114050862575e-06,
      "loss": 0.2762,
      "step": 24638
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4009668155267199,
      "learning_rate": 2.996406019496747e-06,
      "loss": 0.2644,
      "step": 24639
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.39318041874201226,
      "learning_rate": 2.9956980570345852e-06,
      "loss": 0.2601,
      "step": 24640
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3142145115801504,
      "learning_rate": 2.994990163483047e-06,
      "loss": 0.1887,
      "step": 24641
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2465000383708035,
      "learning_rate": 2.9942823388490992e-06,
      "loss": 0.1257,
      "step": 24642
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3168470208982066,
      "learning_rate": 2.993574583139709e-06,
      "loss": 0.1289,
      "step": 24643
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.9566611898366266,
      "learning_rate": 2.99286689636184e-06,
      "loss": 0.3964,
      "step": 24644
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.8118165132431122,
      "learning_rate": 2.9921592785224484e-06,
      "loss": 0.2026,
      "step": 24645
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.36368972396659793,
      "learning_rate": 2.9914517296284995e-06,
      "loss": 0.2514,
      "step": 24646
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3785260010141881,
      "learning_rate": 2.990744249686959e-06,
      "loss": 0.2316,
      "step": 24647
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.3084077792873972,
      "learning_rate": 2.9900368387047752e-06,
      "loss": 0.6916,
      "step": 24648
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.5693931480258163,
      "learning_rate": 2.989329496688923e-06,
      "loss": 0.314,
      "step": 24649
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.3920625423040576,
      "learning_rate": 2.9886222236463504e-06,
      "loss": 0.2405,
      "step": 24650
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.1799774360124824,
      "learning_rate": 2.98791501958402e-06,
      "loss": 0.1112,
      "step": 24651
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.32459319362895206,
      "learning_rate": 2.9872078845088912e-06,
      "loss": 0.2277,
      "step": 24652
    },
    {
      "epoch": 0.75,
      "grad_norm": 1.9997379162834232,
      "learning_rate": 2.9865008184279155e-06,
      "loss": 0.7874,
      "step": 24653
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5481712466386743,
      "learning_rate": 2.9857938213480532e-06,
      "loss": 0.1851,
      "step": 24654
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37194679624995763,
      "learning_rate": 2.9850868932762587e-06,
      "loss": 0.2818,
      "step": 24655
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7471723047792506,
      "learning_rate": 2.9843800342194916e-06,
      "loss": 0.3114,
      "step": 24656
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.551431460914118,
      "learning_rate": 2.9836732441846983e-06,
      "loss": 0.8318,
      "step": 24657
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29973594047021823,
      "learning_rate": 2.9829665231788365e-06,
      "loss": 0.2421,
      "step": 24658
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3924421097007514,
      "learning_rate": 2.982259871208859e-06,
      "loss": 0.2607,
      "step": 24659
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18843493257260588,
      "learning_rate": 2.9815532882817176e-06,
      "loss": 0.1105,
      "step": 24660
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6717806042381524,
      "learning_rate": 2.980846774404368e-06,
      "loss": 0.6859,
      "step": 24661
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7183669297181743,
      "learning_rate": 2.9801403295837507e-06,
      "loss": 0.2841,
      "step": 24662
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5262931937758203,
      "learning_rate": 2.979433953826829e-06,
      "loss": 0.2967,
      "step": 24663
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23231682467970938,
      "learning_rate": 2.9787276471405435e-06,
      "loss": 0.215,
      "step": 24664
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6089323511579041,
      "learning_rate": 2.978021409531845e-06,
      "loss": 0.0332,
      "step": 24665
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3230782010870905,
      "learning_rate": 2.977315241007687e-06,
      "loss": 0.7169,
      "step": 24666
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8877215664925306,
      "learning_rate": 2.9766091415750075e-06,
      "loss": 0.1824,
      "step": 24667
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49960118683427823,
      "learning_rate": 2.9759031112407587e-06,
      "loss": 0.3144,
      "step": 24668
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.17683934266587065,
      "learning_rate": 2.975197150011887e-06,
      "loss": 0.1119,
      "step": 24669
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.33422614676214923,
      "learning_rate": 2.97449125789534e-06,
      "loss": 0.2735,
      "step": 24670
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7128582466300611,
      "learning_rate": 2.9737854348980566e-06,
      "loss": 0.2698,
      "step": 24671
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7536121282111777,
      "learning_rate": 2.9730796810269833e-06,
      "loss": 0.3042,
      "step": 24672
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3342356698104399,
      "learning_rate": 2.9723739962890684e-06,
      "loss": 0.1958,
      "step": 24673
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3533377467928562,
      "learning_rate": 2.971668380691244e-06,
      "loss": 0.5868,
      "step": 24674
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43580517533003926,
      "learning_rate": 2.970962834240465e-06,
      "loss": 0.2888,
      "step": 24675
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5159093206661455,
      "learning_rate": 2.9702573569436634e-06,
      "loss": 0.3078,
      "step": 24676
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29802187953469356,
      "learning_rate": 2.969551948807782e-06,
      "loss": 0.063,
      "step": 24677
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.24854238229788678,
      "learning_rate": 2.968846609839766e-06,
      "loss": 0.1532,
      "step": 24678
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5491191793696417,
      "learning_rate": 2.9681413400465475e-06,
      "loss": 0.2692,
      "step": 24679
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7771354439338972,
      "learning_rate": 2.967436139435068e-06,
      "loss": 0.2795,
      "step": 24680
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4470064426998795,
      "learning_rate": 2.9667310080122656e-06,
      "loss": 0.2953,
      "step": 24681
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2517984239126918,
      "learning_rate": 2.9660259457850814e-06,
      "loss": 0.2033,
      "step": 24682
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0519569458340858,
      "learning_rate": 2.9653209527604454e-06,
      "loss": 0.4157,
      "step": 24683
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0463263164880023,
      "learning_rate": 2.9646160289452965e-06,
      "loss": 0.4506,
      "step": 24684
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6291619746864106,
      "learning_rate": 2.963911174346569e-06,
      "loss": 0.6861,
      "step": 24685
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3455274510892243,
      "learning_rate": 2.9632063889711992e-06,
      "loss": 0.1604,
      "step": 24686
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3138788510999084,
      "learning_rate": 2.9625016728261237e-06,
      "loss": 0.2273,
      "step": 24687
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2985973350572343,
      "learning_rate": 2.9617970259182695e-06,
      "loss": 0.2209,
      "step": 24688
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.800189243219986,
      "learning_rate": 2.9610924482545724e-06,
      "loss": 0.3566,
      "step": 24689
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3067254362746821,
      "learning_rate": 2.9603879398419656e-06,
      "loss": 0.0839,
      "step": 24690
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2927244795098498,
      "learning_rate": 2.9596835006873737e-06,
      "loss": 0.1677,
      "step": 24691
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.768059997796937,
      "learning_rate": 2.9589791307977376e-06,
      "loss": 0.8397,
      "step": 24692
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42377185655737704,
      "learning_rate": 2.958274830179978e-06,
      "loss": 0.251,
      "step": 24693
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5060098355805702,
      "learning_rate": 2.9575705988410285e-06,
      "loss": 0.3309,
      "step": 24694
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32550692058146474,
      "learning_rate": 2.9568664367878163e-06,
      "loss": 0.1466,
      "step": 24695
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5209173275601783,
      "learning_rate": 2.9561623440272736e-06,
      "loss": 0.3054,
      "step": 24696
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6872041268508446,
      "learning_rate": 2.955458320566319e-06,
      "loss": 0.2814,
      "step": 24697
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29936116967380605,
      "learning_rate": 2.9547543664118837e-06,
      "loss": 0.1856,
      "step": 24698
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2659730467106029,
      "learning_rate": 2.954050481570897e-06,
      "loss": 0.208,
      "step": 24699
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.48347185126315684,
      "learning_rate": 2.953346666050274e-06,
      "loss": 0.3237,
      "step": 24700
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5103375491731145,
      "learning_rate": 2.9526429198569507e-06,
      "loss": 0.2267,
      "step": 24701
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3843068127619016,
      "learning_rate": 2.951939242997842e-06,
      "loss": 0.7202,
      "step": 24702
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0724139279686569,
      "learning_rate": 2.9512356354798742e-06,
      "loss": 0.0829,
      "step": 24703
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4727304596450997,
      "learning_rate": 2.9505320973099736e-06,
      "loss": 0.1686,
      "step": 24704
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3674443833714773,
      "learning_rate": 2.949828628495054e-06,
      "loss": 0.2704,
      "step": 24705
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.31252681383290715,
      "learning_rate": 2.94912522904204e-06,
      "loss": 0.2325,
      "step": 24706
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3884334025635805,
      "learning_rate": 2.9484218989578516e-06,
      "loss": 0.5212,
      "step": 24707
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.1711095232833849,
      "learning_rate": 2.947718638249413e-06,
      "loss": 0.0688,
      "step": 24708
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38555237202510845,
      "learning_rate": 2.9470154469236355e-06,
      "loss": 0.2411,
      "step": 24709
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4896195811842189,
      "learning_rate": 2.9463123249874404e-06,
      "loss": 0.2668,
      "step": 24710
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.7638623663009183,
      "learning_rate": 2.945609272447746e-06,
      "loss": 0.7827,
      "step": 24711
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2973043694060334,
      "learning_rate": 2.9449062893114686e-06,
      "loss": 0.2162,
      "step": 24712
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5900399303437655,
      "learning_rate": 2.9442033755855282e-06,
      "loss": 0.3026,
      "step": 24713
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3180491839439847,
      "learning_rate": 2.9435005312768326e-06,
      "loss": 0.2246,
      "step": 24714
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8658810411090113,
      "learning_rate": 2.9427977563923015e-06,
      "loss": 0.4878,
      "step": 24715
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3941465320880999,
      "learning_rate": 2.942095050938851e-06,
      "loss": 0.0793,
      "step": 24716
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2102516524515869,
      "learning_rate": 2.941392414923385e-06,
      "loss": 0.1645,
      "step": 24717
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3702241907384598,
      "learning_rate": 2.94068984835283e-06,
      "loss": 0.2664,
      "step": 24718
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.525951603508197,
      "learning_rate": 2.9399873512340883e-06,
      "loss": 0.2853,
      "step": 24719
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4032955001659158,
      "learning_rate": 2.9392849235740728e-06,
      "loss": 0.8311,
      "step": 24720
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8186404583684224,
      "learning_rate": 2.9385825653796998e-06,
      "loss": 0.2808,
      "step": 24721
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.595145049920699,
      "learning_rate": 2.9378802766578717e-06,
      "loss": 0.3131,
      "step": 24722
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.35564373671632954,
      "learning_rate": 2.9371780574155005e-06,
      "loss": 0.1989,
      "step": 24723
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32005431259969874,
      "learning_rate": 2.936475907659496e-06,
      "loss": 0.2843,
      "step": 24724
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1568567057908785,
      "learning_rate": 2.9357738273967684e-06,
      "loss": 0.1023,
      "step": 24725
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.28000389692579164,
      "learning_rate": 2.9350718166342162e-06,
      "loss": 0.1405,
      "step": 24726
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3636585896547796,
      "learning_rate": 2.9343698753787596e-06,
      "loss": 0.1677,
      "step": 24727
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5525125720650756,
      "learning_rate": 2.9336680036372954e-06,
      "loss": 0.3726,
      "step": 24728
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2855952118752049,
      "learning_rate": 2.932966201416726e-06,
      "loss": 0.2282,
      "step": 24729
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7868980975132129,
      "learning_rate": 2.9322644687239656e-06,
      "loss": 0.2703,
      "step": 24730
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7091663482377665,
      "learning_rate": 2.9315628055659096e-06,
      "loss": 0.4162,
      "step": 24731
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30960483230845326,
      "learning_rate": 2.9308612119494653e-06,
      "loss": 0.1927,
      "step": 24732
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3817191270648286,
      "learning_rate": 2.930159687881533e-06,
      "loss": 0.5395,
      "step": 24733
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8580961835243084,
      "learning_rate": 2.92945823336902e-06,
      "loss": 0.1251,
      "step": 24734
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2761468773647955,
      "learning_rate": 2.928756848418821e-06,
      "loss": 0.2327,
      "step": 24735
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22319342452805305,
      "learning_rate": 2.928055533037838e-06,
      "loss": 0.1246,
      "step": 24736
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5224034278856594,
      "learning_rate": 2.9273542872329718e-06,
      "loss": 0.3476,
      "step": 24737
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0478784213721535,
      "learning_rate": 2.9266531110111217e-06,
      "loss": 0.4465,
      "step": 24738
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7873606700835611,
      "learning_rate": 2.925952004379189e-06,
      "loss": 0.3812,
      "step": 24739
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37848038475418244,
      "learning_rate": 2.9252509673440642e-06,
      "loss": 0.2157,
      "step": 24740
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34272964401424155,
      "learning_rate": 2.924549999912649e-06,
      "loss": 0.2403,
      "step": 24741
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.48635226852477675,
      "learning_rate": 2.9238491020918424e-06,
      "loss": 0.21,
      "step": 24742
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3539254495457946,
      "learning_rate": 2.9231482738885296e-06,
      "loss": 0.161,
      "step": 24743
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.46002396006207796,
      "learning_rate": 2.9224475153096188e-06,
      "loss": 0.2305,
      "step": 24744
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2596611173034196,
      "learning_rate": 2.921746826361996e-06,
      "loss": 0.1739,
      "step": 24745
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5738650578642268,
      "learning_rate": 2.9210462070525568e-06,
      "loss": 0.7194,
      "step": 24746
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2987525609598946,
      "learning_rate": 2.920345657388197e-06,
      "loss": 0.2372,
      "step": 24747
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7876818876547552,
      "learning_rate": 2.919645177375803e-06,
      "loss": 0.3874,
      "step": 24748
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29897110184886405,
      "learning_rate": 2.9189447670222693e-06,
      "loss": 0.1201,
      "step": 24749
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6217116942685315,
      "learning_rate": 2.9182444263344877e-06,
      "loss": 0.336,
      "step": 24750
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.51561756799091,
      "learning_rate": 2.9175441553193494e-06,
      "loss": 0.1857,
      "step": 24751
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4102747057357115,
      "learning_rate": 2.91684395398374e-06,
      "loss": 0.4961,
      "step": 24752
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23904095705856407,
      "learning_rate": 2.9161438223345508e-06,
      "loss": 0.1899,
      "step": 24753
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1035771501744454,
      "learning_rate": 2.9154437603786724e-06,
      "loss": 0.3947,
      "step": 24754
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3119449976401231,
      "learning_rate": 2.914743768122983e-06,
      "loss": 0.1974,
      "step": 24755
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7243960115565315,
      "learning_rate": 2.9140438455743824e-06,
      "loss": 0.2971,
      "step": 24756
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1095028477845497,
      "learning_rate": 2.9133439927397465e-06,
      "loss": 0.4737,
      "step": 24757
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22259873070733152,
      "learning_rate": 2.9126442096259646e-06,
      "loss": 0.0941,
      "step": 24758
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.31674939383705797,
      "learning_rate": 2.9119444962399234e-06,
      "loss": 0.2399,
      "step": 24759
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3312529253688354,
      "learning_rate": 2.911244852588502e-06,
      "loss": 0.2256,
      "step": 24760
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4120299831038454,
      "learning_rate": 2.910545278678586e-06,
      "loss": 0.5102,
      "step": 24761
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2752232365094365,
      "learning_rate": 2.9098457745170584e-06,
      "loss": 0.0713,
      "step": 24762
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7664400967830688,
      "learning_rate": 2.909146340110801e-06,
      "loss": 0.2961,
      "step": 24763
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39368692597509314,
      "learning_rate": 2.908446975466694e-06,
      "loss": 0.2312,
      "step": 24764
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34221524938884007,
      "learning_rate": 2.907747680591624e-06,
      "loss": 0.2889,
      "step": 24765
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.067110196447197,
      "learning_rate": 2.907048455492463e-06,
      "loss": 0.3097,
      "step": 24766
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3400974537272912,
      "learning_rate": 2.9063493001760924e-06,
      "loss": 0.1878,
      "step": 24767
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2729189557571749,
      "learning_rate": 2.9056502146493935e-06,
      "loss": 0.1781,
      "step": 24768
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2823122631549828,
      "learning_rate": 2.904951198919237e-06,
      "loss": 0.1041,
      "step": 24769
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4679335390840833,
      "learning_rate": 2.904252252992512e-06,
      "loss": 0.7051,
      "step": 24770
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.33525989048750293,
      "learning_rate": 2.903553376876085e-06,
      "loss": 0.2189,
      "step": 24771
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6346497719111254,
      "learning_rate": 2.902854570576834e-06,
      "loss": 0.3267,
      "step": 24772
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3443443203238537,
      "learning_rate": 2.9021558341016376e-06,
      "loss": 0.1982,
      "step": 24773
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8908854635058756,
      "learning_rate": 2.9014571674573654e-06,
      "loss": 0.3658,
      "step": 24774
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.978665460759625,
      "learning_rate": 2.9007585706508932e-06,
      "loss": 0.3475,
      "step": 24775
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2947265214321948,
      "learning_rate": 2.9000600436890947e-06,
      "loss": 0.2296,
      "step": 24776
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3411126806297839,
      "learning_rate": 2.899361586578845e-06,
      "loss": 0.1546,
      "step": 24777
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2778391865537212,
      "learning_rate": 2.8986631993270076e-06,
      "loss": 0.2146,
      "step": 24778
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.593033428953111,
      "learning_rate": 2.8979648819404594e-06,
      "loss": 0.1153,
      "step": 24779
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.020164192846912,
      "learning_rate": 2.8972666344260725e-06,
      "loss": 0.4546,
      "step": 24780
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6555552722593223,
      "learning_rate": 2.896568456790708e-06,
      "loss": 0.1818,
      "step": 24781
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34969088945217547,
      "learning_rate": 2.8958703490412454e-06,
      "loss": 0.2716,
      "step": 24782
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29141703811538555,
      "learning_rate": 2.8951723111845455e-06,
      "loss": 0.2341,
      "step": 24783
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0060108696536367,
      "learning_rate": 2.894474343227478e-06,
      "loss": 0.5267,
      "step": 24784
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40431783646671693,
      "learning_rate": 2.8937764451769136e-06,
      "loss": 0.2495,
      "step": 24785
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.22948323287789965,
      "learning_rate": 2.8930786170397086e-06,
      "loss": 0.1227,
      "step": 24786
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6009319678042246,
      "learning_rate": 2.89238085882274e-06,
      "loss": 0.3088,
      "step": 24787
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.47604362218966184,
      "learning_rate": 2.891683170532864e-06,
      "loss": 0.2464,
      "step": 24788
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5181046053938645,
      "learning_rate": 2.890985552176949e-06,
      "loss": 0.3482,
      "step": 24789
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7578845137412161,
      "learning_rate": 2.89028800376186e-06,
      "loss": 0.2109,
      "step": 24790
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3845111840480307,
      "learning_rate": 2.8895905252944533e-06,
      "loss": 0.292,
      "step": 24791
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.768191667003215,
      "learning_rate": 2.888893116781595e-06,
      "loss": 0.3001,
      "step": 24792
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5706194289060111,
      "learning_rate": 2.8881957782301463e-06,
      "loss": 0.8264,
      "step": 24793
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30633185120236495,
      "learning_rate": 2.887498509646971e-06,
      "loss": 0.2396,
      "step": 24794
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38598514493590064,
      "learning_rate": 2.886801311038918e-06,
      "loss": 0.2368,
      "step": 24795
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2894651475087885,
      "learning_rate": 2.886104182412862e-06,
      "loss": 0.1385,
      "step": 24796
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.382239216736461,
      "learning_rate": 2.885407123775651e-06,
      "loss": 0.0801,
      "step": 24797
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6831396434191721,
      "learning_rate": 2.8847101351341443e-06,
      "loss": 0.3605,
      "step": 24798
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49218593687023265,
      "learning_rate": 2.8840132164952038e-06,
      "loss": 0.2114,
      "step": 24799
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.513376996247765,
      "learning_rate": 2.8833163678656806e-06,
      "loss": 0.2893,
      "step": 24800
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2573914864499599,
      "learning_rate": 2.8826195892524323e-06,
      "loss": 0.2056,
      "step": 24801
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.305793141381143,
      "learning_rate": 2.8819228806623144e-06,
      "loss": 0.7446,
      "step": 24802
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7755500483582126,
      "learning_rate": 2.881226242102184e-06,
      "loss": 0.1717,
      "step": 24803
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0056455264098234,
      "learning_rate": 2.88052967357889e-06,
      "loss": 0.4281,
      "step": 24804
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.189168061039905,
      "learning_rate": 2.8798331750992868e-06,
      "loss": 0.1436,
      "step": 24805
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.48688756234157543,
      "learning_rate": 2.8791367466702315e-06,
      "loss": 0.3209,
      "step": 24806
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42255150293388277,
      "learning_rate": 2.878440388298566e-06,
      "loss": 0.2472,
      "step": 24807
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0122299511897985,
      "learning_rate": 2.877744099991152e-06,
      "loss": 0.4382,
      "step": 24808
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29696271238976396,
      "learning_rate": 2.877047881754833e-06,
      "loss": 0.1788,
      "step": 24809
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1419155868420743,
      "learning_rate": 2.876351733596461e-06,
      "loss": 0.4396,
      "step": 24810
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3162750843133684,
      "learning_rate": 2.8756556555228874e-06,
      "loss": 0.7679,
      "step": 24811
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2880856452895973,
      "learning_rate": 2.8749596475409515e-06,
      "loss": 0.2231,
      "step": 24812
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6428740654978503,
      "learning_rate": 2.8742637096575133e-06,
      "loss": 0.2407,
      "step": 24813
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.18767429068657873,
      "learning_rate": 2.87356784187941e-06,
      "loss": 0.1328,
      "step": 24814
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7904579598126984,
      "learning_rate": 2.8728720442134907e-06,
      "loss": 0.341,
      "step": 24815
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7074065987531369,
      "learning_rate": 2.872176316666605e-06,
      "loss": 0.2705,
      "step": 24816
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4994032322027004,
      "learning_rate": 2.8714806592455915e-06,
      "loss": 0.5528,
      "step": 24817
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.25781026886395486,
      "learning_rate": 2.870785071957296e-06,
      "loss": 0.17,
      "step": 24818
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3388937019676285,
      "learning_rate": 2.870089554808564e-06,
      "loss": 0.2965,
      "step": 24819
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0895682587964297,
      "learning_rate": 2.869394107806238e-06,
      "loss": 0.4688,
      "step": 24820
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1381880984585382,
      "learning_rate": 2.868698730957158e-06,
      "loss": 0.4952,
      "step": 24821
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34684021212930355,
      "learning_rate": 2.868003424268165e-06,
      "loss": 0.1626,
      "step": 24822
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.26474390358908717,
      "learning_rate": 2.867308187746105e-06,
      "loss": 0.1433,
      "step": 24823
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3171817556093379,
      "learning_rate": 2.866613021397808e-06,
      "loss": 0.2319,
      "step": 24824
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5182320314890634,
      "learning_rate": 2.865917925230125e-06,
      "loss": 0.245,
      "step": 24825
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.906144178548245,
      "learning_rate": 2.8652228992498856e-06,
      "loss": 0.3586,
      "step": 24826
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3838440650438384,
      "learning_rate": 2.864527943463932e-06,
      "loss": 0.1847,
      "step": 24827
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5237944312693321,
      "learning_rate": 2.863833057879102e-06,
      "loss": 0.3471,
      "step": 24828
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0353997945013391,
      "learning_rate": 2.863138242502228e-06,
      "loss": 0.5218,
      "step": 24829
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.35903121251545345,
      "learning_rate": 2.862443497340148e-06,
      "loss": 0.3105,
      "step": 24830
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49067823052278714,
      "learning_rate": 2.8617488223996983e-06,
      "loss": 0.0691,
      "step": 24831
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32900200130354157,
      "learning_rate": 2.8610542176877152e-06,
      "loss": 0.255,
      "step": 24832
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4097169952256967,
      "learning_rate": 2.8603596832110236e-06,
      "loss": 0.1926,
      "step": 24833
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.47469843118778726,
      "learning_rate": 2.8596652189764686e-06,
      "loss": 0.2656,
      "step": 24834
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1823318663473572,
      "learning_rate": 2.8589708249908732e-06,
      "loss": 0.0986,
      "step": 24835
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2604521366705071,
      "learning_rate": 2.8582765012610738e-06,
      "loss": 0.1649,
      "step": 24836
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3345959481134257,
      "learning_rate": 2.857582247793902e-06,
      "loss": 0.2812,
      "step": 24837
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9764244177723279,
      "learning_rate": 2.8568880645961804e-06,
      "loss": 0.5133,
      "step": 24838
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6639880112744658,
      "learning_rate": 2.8561939516747507e-06,
      "loss": 0.7438,
      "step": 24839
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.46400037617017065,
      "learning_rate": 2.855499909036432e-06,
      "loss": 0.1579,
      "step": 24840
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3610653206241783,
      "learning_rate": 2.8548059366880564e-06,
      "loss": 0.2629,
      "step": 24841
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3116223025972426,
      "learning_rate": 2.8541120346364548e-06,
      "loss": 0.2516,
      "step": 24842
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.509276260858789,
      "learning_rate": 2.853418202888447e-06,
      "loss": 0.5723,
      "step": 24843
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.15145123915513312,
      "learning_rate": 2.8527244414508615e-06,
      "loss": 0.0695,
      "step": 24844
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4183138978997636,
      "learning_rate": 2.8520307503305257e-06,
      "loss": 0.2489,
      "step": 24845
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4687368695790303,
      "learning_rate": 2.851337129534266e-06,
      "loss": 0.2565,
      "step": 24846
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5103337718070233,
      "learning_rate": 2.8506435790689003e-06,
      "loss": 0.7018,
      "step": 24847
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3106042036388863,
      "learning_rate": 2.8499500989412563e-06,
      "loss": 0.2089,
      "step": 24848
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5003586985376008,
      "learning_rate": 2.8492566891581587e-06,
      "loss": 0.1748,
      "step": 24849
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39229837546143115,
      "learning_rate": 2.848563349726421e-06,
      "loss": 0.2492,
      "step": 24850
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7085144471203569,
      "learning_rate": 2.847870080652876e-06,
      "loss": 0.2783,
      "step": 24851
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38459956053696537,
      "learning_rate": 2.8471768819443345e-06,
      "loss": 0.1822,
      "step": 24852
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2502314095059832,
      "learning_rate": 2.8464837536076216e-06,
      "loss": 0.14,
      "step": 24853
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5043816591796427,
      "learning_rate": 2.8457906956495584e-06,
      "loss": 0.3124,
      "step": 24854
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.33341061699142566,
      "learning_rate": 2.845097708076957e-06,
      "loss": 0.2169,
      "step": 24855
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5147680491206703,
      "learning_rate": 2.8444047908966386e-06,
      "loss": 0.7777,
      "step": 24856
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.782097427559655,
      "learning_rate": 2.84371194411542e-06,
      "loss": 0.2608,
      "step": 24857
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9816965731203163,
      "learning_rate": 2.843019167740122e-06,
      "loss": 0.4086,
      "step": 24858
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.273741828749054,
      "learning_rate": 2.8423264617775526e-06,
      "loss": 0.1719,
      "step": 24859
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43917448257628167,
      "learning_rate": 2.8416338262345313e-06,
      "loss": 0.2918,
      "step": 24860
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.25850513168914535,
      "learning_rate": 2.8409412611178742e-06,
      "loss": 0.164,
      "step": 24861
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.25512870395117726,
      "learning_rate": 2.840248766434386e-06,
      "loss": 0.0715,
      "step": 24862
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5631619954419966,
      "learning_rate": 2.8395563421908934e-06,
      "loss": 0.3065,
      "step": 24863
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3126151893643653,
      "learning_rate": 2.838863988394195e-06,
      "loss": 0.2097,
      "step": 24864
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4345976781311249,
      "learning_rate": 2.838171705051116e-06,
      "loss": 0.3103,
      "step": 24865
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4722343745666462,
      "learning_rate": 2.8374794921684567e-06,
      "loss": 0.2391,
      "step": 24866
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8149835891173184,
      "learning_rate": 2.8367873497530308e-06,
      "loss": 0.3014,
      "step": 24867
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3279826369461952,
      "learning_rate": 2.8360952778116515e-06,
      "loss": 0.2054,
      "step": 24868
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0634229374942443,
      "learning_rate": 2.8354032763511197e-06,
      "loss": 0.38,
      "step": 24869
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1907764507364624,
      "learning_rate": 2.8347113453782495e-06,
      "loss": 0.0853,
      "step": 24870
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2385146902304375,
      "learning_rate": 2.8340194848998457e-06,
      "loss": 0.1947,
      "step": 24871
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2672050909552368,
      "learning_rate": 2.8333276949227195e-06,
      "loss": 0.1839,
      "step": 24872
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5101448024990067,
      "learning_rate": 2.8326359754536712e-06,
      "loss": 0.3298,
      "step": 24873
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.011910291498048,
      "learning_rate": 2.831944326499507e-06,
      "loss": 0.4487,
      "step": 24874
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9848075223662153,
      "learning_rate": 2.831252748067038e-06,
      "loss": 0.2812,
      "step": 24875
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7963707951567549,
      "learning_rate": 2.830561240163057e-06,
      "loss": 0.3806,
      "step": 24876
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.34070067111715563,
      "learning_rate": 2.829869802794378e-06,
      "loss": 0.1924,
      "step": 24877
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5074160423879762,
      "learning_rate": 2.8291784359677976e-06,
      "loss": 0.3493,
      "step": 24878
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.41018170477835253,
      "learning_rate": 2.828487139690118e-06,
      "loss": 0.2098,
      "step": 24879
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2779943175689716,
      "learning_rate": 2.8277959139681465e-06,
      "loss": 0.1897,
      "step": 24880
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.28838352437316583,
      "learning_rate": 2.8271047588086743e-06,
      "loss": 0.112,
      "step": 24881
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.36787475846121764,
      "learning_rate": 2.8264136742185066e-06,
      "loss": 0.2399,
      "step": 24882
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1348971485108834,
      "learning_rate": 2.8257226602044407e-06,
      "loss": 0.3241,
      "step": 24883
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3870488789654682,
      "learning_rate": 2.8250317167732787e-06,
      "loss": 0.2842,
      "step": 24884
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0458364301790517,
      "learning_rate": 2.8243408439318144e-06,
      "loss": 0.235,
      "step": 24885
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4067869459679432,
      "learning_rate": 2.823650041686844e-06,
      "loss": 0.2507,
      "step": 24886
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4550299149212481,
      "learning_rate": 2.82295931004517e-06,
      "loss": 0.1844,
      "step": 24887
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.49208153450714676,
      "learning_rate": 2.8222686490135777e-06,
      "loss": 0.0887,
      "step": 24888
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.45072413822409474,
      "learning_rate": 2.821578058598874e-06,
      "loss": 0.2815,
      "step": 24889
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.29619243631915715,
      "learning_rate": 2.820887538807844e-06,
      "loss": 0.1875,
      "step": 24890
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38600017789638313,
      "learning_rate": 2.8201970896472853e-06,
      "loss": 0.3166,
      "step": 24891
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5564064975700713,
      "learning_rate": 2.8195067111239927e-06,
      "loss": 0.3309,
      "step": 24892
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2123681444071992,
      "learning_rate": 2.8188164032447497e-06,
      "loss": 0.5047,
      "step": 24893
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.25515043541743615,
      "learning_rate": 2.8181261660163596e-06,
      "loss": 0.0981,
      "step": 24894
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.40216888557770564,
      "learning_rate": 2.8174359994456047e-06,
      "loss": 0.2406,
      "step": 24895
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2597850420999493,
      "learning_rate": 2.8167459035392773e-06,
      "loss": 0.2168,
      "step": 24896
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.6470260594668416,
      "learning_rate": 2.816055878304169e-06,
      "loss": 0.5654,
      "step": 24897
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3955178851569958,
      "learning_rate": 2.8153659237470686e-06,
      "loss": 0.2164,
      "step": 24898
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5946517161229357,
      "learning_rate": 2.8146760398747597e-06,
      "loss": 0.3043,
      "step": 24899
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.46192705770224857,
      "learning_rate": 2.813986226694031e-06,
      "loss": 0.2276,
      "step": 24900
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5741917112979567,
      "learning_rate": 2.813296484211675e-06,
      "loss": 0.2896,
      "step": 24901
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3907905723820761,
      "learning_rate": 2.812606812434466e-06,
      "loss": 0.3034,
      "step": 24902
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.14581330567083522,
      "learning_rate": 2.8119172113692017e-06,
      "loss": 0.0834,
      "step": 24903
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3373735660292423,
      "learning_rate": 2.811227681022659e-06,
      "loss": 0.2429,
      "step": 24904
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.27612923453171,
      "learning_rate": 2.810538221401624e-06,
      "loss": 0.1348,
      "step": 24905
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2364904036430695,
      "learning_rate": 2.8098488325128813e-06,
      "loss": 0.4837,
      "step": 24906
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2909359498540335,
      "learning_rate": 2.80915951436321e-06,
      "loss": 0.2242,
      "step": 24907
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6127254854758367,
      "learning_rate": 2.8084702669593934e-06,
      "loss": 0.3236,
      "step": 24908
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3473366576461778,
      "learning_rate": 2.8077810903082114e-06,
      "loss": 0.2046,
      "step": 24909
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9545236804859457,
      "learning_rate": 2.80709198441645e-06,
      "loss": 0.3938,
      "step": 24910
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43669208682822197,
      "learning_rate": 2.806402949290881e-06,
      "loss": 0.1935,
      "step": 24911
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.43232212842717943,
      "learning_rate": 2.8057139849382863e-06,
      "loss": 0.2275,
      "step": 24912
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3089570344702398,
      "learning_rate": 2.805025091365444e-06,
      "loss": 0.1646,
      "step": 24913
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.26397985232441723,
      "learning_rate": 2.804336268579134e-06,
      "loss": 0.2112,
      "step": 24914
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4491877384531107,
      "learning_rate": 2.8036475165861342e-06,
      "loss": 0.4719,
      "step": 24915
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.4388495507535988,
      "learning_rate": 2.8029588353932147e-06,
      "loss": 0.2282,
      "step": 24916
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5521724448799508,
      "learning_rate": 2.8022702250071533e-06,
      "loss": 0.3403,
      "step": 24917
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3085947549503826,
      "learning_rate": 2.80158168543473e-06,
      "loss": 0.2055,
      "step": 24918
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4252364955098835,
      "learning_rate": 2.8008932166827087e-06,
      "loss": 0.2882,
      "step": 24919
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39476531831806055,
      "learning_rate": 2.8002048187578745e-06,
      "loss": 0.2245,
      "step": 24920
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4497223629854968,
      "learning_rate": 2.7995164916669913e-06,
      "loss": 0.2338,
      "step": 24921
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.21228022792520337,
      "learning_rate": 2.798828235416834e-06,
      "loss": 0.114,
      "step": 24922
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5048981884346254,
      "learning_rate": 2.7981400500141765e-06,
      "loss": 0.2951,
      "step": 24923
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1348318516561369,
      "learning_rate": 2.797451935465785e-06,
      "loss": 0.05,
      "step": 24924
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.42528923188886425,
      "learning_rate": 2.7967638917784314e-06,
      "loss": 0.2602,
      "step": 24925
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37189236073352056,
      "learning_rate": 2.7960759189588835e-06,
      "loss": 0.2286,
      "step": 24926
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3332605806898139,
      "learning_rate": 2.7953880170139146e-06,
      "loss": 0.2084,
      "step": 24927
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.9751515636983757,
      "learning_rate": 2.7947001859502865e-06,
      "loss": 0.4894,
      "step": 24928
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.006085573526134,
      "learning_rate": 2.794012425774769e-06,
      "loss": 0.4242,
      "step": 24929
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.248495858832882,
      "learning_rate": 2.7933247364941306e-06,
      "loss": 0.1887,
      "step": 24930
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2834282692578768,
      "learning_rate": 2.792637118115129e-06,
      "loss": 0.1722,
      "step": 24931
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5001606638602019,
      "learning_rate": 2.791949570644541e-06,
      "loss": 0.3291,
      "step": 24932
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1539879797409873,
      "learning_rate": 2.791262094089122e-06,
      "loss": 0.1716,
      "step": 24933
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7180588962012432,
      "learning_rate": 2.7905746884556386e-06,
      "loss": 0.4345,
      "step": 24934
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5628751167209648,
      "learning_rate": 2.789887353750853e-06,
      "loss": 0.2231,
      "step": 24935
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4699200032370184,
      "learning_rate": 2.7892000899815317e-06,
      "loss": 0.3024,
      "step": 24936
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3313738948528083,
      "learning_rate": 2.7885128971544295e-06,
      "loss": 0.2221,
      "step": 24937
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.48881113279605953,
      "learning_rate": 2.7878257752763104e-06,
      "loss": 0.3706,
      "step": 24938
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.30693439803739714,
      "learning_rate": 2.787138724353934e-06,
      "loss": 0.078,
      "step": 24939
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.15962546086780516,
      "learning_rate": 2.7864517443940607e-06,
      "loss": 0.068,
      "step": 24940
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5309364070635938,
      "learning_rate": 2.7857648354034516e-06,
      "loss": 0.2635,
      "step": 24941
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.801160360544451,
      "learning_rate": 2.7850779973888588e-06,
      "loss": 0.2695,
      "step": 24942
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3685298229852301,
      "learning_rate": 2.784391230357042e-06,
      "loss": 0.2749,
      "step": 24943
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5575501726926796,
      "learning_rate": 2.783704534314763e-06,
      "loss": 0.2223,
      "step": 24944
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3572307769080307,
      "learning_rate": 2.783017909268766e-06,
      "loss": 0.2557,
      "step": 24945
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.1666159543667136,
      "learning_rate": 2.78233135522582e-06,
      "loss": 0.5506,
      "step": 24946
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.8582195588624388,
      "learning_rate": 2.7816448721926694e-06,
      "loss": 0.8075,
      "step": 24947
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4418085482793611,
      "learning_rate": 2.780958460176073e-06,
      "loss": 0.2224,
      "step": 24948
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4493884618698212,
      "learning_rate": 2.7802721191827843e-06,
      "loss": 0.2481,
      "step": 24949
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2040304249845613,
      "learning_rate": 2.7795858492195516e-06,
      "loss": 0.1409,
      "step": 24950
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.2260297345237683,
      "learning_rate": 2.7788996502931297e-06,
      "loss": 0.4956,
      "step": 24951
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.7384703462164828,
      "learning_rate": 2.778213522410268e-06,
      "loss": 0.2787,
      "step": 24952
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6830177607831996,
      "learning_rate": 2.777527465577722e-06,
      "loss": 0.2443,
      "step": 24953
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.325283810600585,
      "learning_rate": 2.7768414798022334e-06,
      "loss": 0.2445,
      "step": 24954
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.39494560196494044,
      "learning_rate": 2.7761555650905557e-06,
      "loss": 0.2312,
      "step": 24955
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4648691097262366,
      "learning_rate": 2.7754697214494396e-06,
      "loss": 0.3481,
      "step": 24956
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3072823982685064,
      "learning_rate": 2.7747839488856232e-06,
      "loss": 0.0709,
      "step": 24957
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2717162886353756,
      "learning_rate": 2.7740982474058664e-06,
      "loss": 0.148,
      "step": 24958
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32164743712199606,
      "learning_rate": 2.773412617016905e-06,
      "loss": 0.1893,
      "step": 24959
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.916264268591612,
      "learning_rate": 2.7727270577254883e-06,
      "loss": 0.3842,
      "step": 24960
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3145413874770909,
      "learning_rate": 2.772041569538364e-06,
      "loss": 0.2157,
      "step": 24961
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.8952888191512163,
      "learning_rate": 2.771356152462269e-06,
      "loss": 0.3767,
      "step": 24962
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.38695599944018627,
      "learning_rate": 2.770670806503952e-06,
      "loss": 0.1675,
      "step": 24963
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.5669344202926974,
      "learning_rate": 2.769985531670154e-06,
      "loss": 0.3993,
      "step": 24964
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.070753911831223,
      "learning_rate": 2.769300327967617e-06,
      "loss": 0.4507,
      "step": 24965
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.4848660299635565,
      "learning_rate": 2.7686151954030825e-06,
      "loss": 0.3092,
      "step": 24966
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.37544330433758355,
      "learning_rate": 2.7679301339832953e-06,
      "loss": 0.214,
      "step": 24967
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.23708644606557777,
      "learning_rate": 2.767245143714987e-06,
      "loss": 0.1549,
      "step": 24968
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6191025233374609,
      "learning_rate": 2.766560224604902e-06,
      "loss": 0.3481,
      "step": 24969
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3555767096907962,
      "learning_rate": 2.765875376659779e-06,
      "loss": 0.13,
      "step": 24970
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.5086583532169822,
      "learning_rate": 2.7651905998863494e-06,
      "loss": 0.7077,
      "step": 24971
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.32050829962447286,
      "learning_rate": 2.7645058942913615e-06,
      "loss": 0.1695,
      "step": 24972
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2761554181756986,
      "learning_rate": 2.7638212598815418e-06,
      "loss": 0.2462,
      "step": 24973
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.0653441140159527,
      "learning_rate": 2.7631366966636295e-06,
      "loss": 0.4607,
      "step": 24974
    },
    {
      "epoch": 0.76,
      "grad_norm": 1.3990880372496548,
      "learning_rate": 2.7624522046443626e-06,
      "loss": 0.5811,
      "step": 24975
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.2187501412871764,
      "learning_rate": 2.76176778383047e-06,
      "loss": 0.0741,
      "step": 24976
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.35525448495967243,
      "learning_rate": 2.7610834342286874e-06,
      "loss": 0.2614,
      "step": 24977
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.6933710227960064,
      "learning_rate": 2.760399155845748e-06,
      "loss": 0.2706,
      "step": 24978
    },
    {
      "epoch": 0.76,
      "grad_norm": 0.3891258129872372,
      "learning_rate": 2.759714948688386e-06,
      "loss": 0.2655,
      "step": 24979
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42928792421535966,
      "learning_rate": 2.759030812763328e-06,
      "loss": 0.127,
      "step": 24980
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.276023880672723,
      "learning_rate": 2.7583467480773084e-06,
      "loss": 0.1744,
      "step": 24981
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5665622020990575,
      "learning_rate": 2.7576627546370582e-06,
      "loss": 0.7073,
      "step": 24982
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7997932975221128,
      "learning_rate": 2.7569788324492996e-06,
      "loss": 0.4235,
      "step": 24983
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4600196462879542,
      "learning_rate": 2.7562949815207716e-06,
      "loss": 0.3117,
      "step": 24984
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40263120178749157,
      "learning_rate": 2.755611201858195e-06,
      "loss": 0.1968,
      "step": 24985
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5683203911712572,
      "learning_rate": 2.754927493468299e-06,
      "loss": 0.3433,
      "step": 24986
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5069032321354605,
      "learning_rate": 2.7542438563578135e-06,
      "loss": 0.2487,
      "step": 24987
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27114388556590235,
      "learning_rate": 2.753560290533458e-06,
      "loss": 0.1588,
      "step": 24988
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3991485740046136,
      "learning_rate": 2.7528767960019597e-06,
      "loss": 0.2301,
      "step": 24989
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9979371056459139,
      "learning_rate": 2.7521933727700455e-06,
      "loss": 0.4551,
      "step": 24990
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2490220136735843,
      "learning_rate": 2.7515100208444365e-06,
      "loss": 0.2011,
      "step": 24991
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3803774262488493,
      "learning_rate": 2.7508267402318613e-06,
      "loss": 0.7815,
      "step": 24992
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6318784774483965,
      "learning_rate": 2.7501435309390333e-06,
      "loss": 0.2777,
      "step": 24993
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6465768172780157,
      "learning_rate": 2.7494603929726805e-06,
      "loss": 0.2028,
      "step": 24994
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3432367891547732,
      "learning_rate": 2.7487773263395213e-06,
      "loss": 0.2588,
      "step": 24995
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3907375615715332,
      "learning_rate": 2.748094331046279e-06,
      "loss": 0.22,
      "step": 24996
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5420728440991299,
      "learning_rate": 2.747411407099666e-06,
      "loss": 0.3374,
      "step": 24997
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.16811171668445704,
      "learning_rate": 2.7467285545064117e-06,
      "loss": 0.0835,
      "step": 24998
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6435911630621078,
      "learning_rate": 2.746045773273225e-06,
      "loss": 0.3611,
      "step": 24999
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3320753427241682,
      "learning_rate": 2.7453630634068273e-06,
      "loss": 0.2133,
      "step": 25000
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2688607350911352,
      "learning_rate": 2.744680424913938e-06,
      "loss": 0.5044,
      "step": 25001
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4370188609492678,
      "learning_rate": 2.743997857801267e-06,
      "loss": 0.2331,
      "step": 25002
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4417613064418272,
      "learning_rate": 2.7433153620755326e-06,
      "loss": 0.3226,
      "step": 25003
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2774539966625662,
      "learning_rate": 2.742632937743449e-06,
      "loss": 0.163,
      "step": 25004
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.320452533460507,
      "learning_rate": 2.7419505848117333e-06,
      "loss": 0.3914,
      "step": 25005
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22211402215829243,
      "learning_rate": 2.7412683032870936e-06,
      "loss": 0.069,
      "step": 25006
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22392811402873622,
      "learning_rate": 2.740586093176245e-06,
      "loss": 0.1185,
      "step": 25007
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41226370455933914,
      "learning_rate": 2.7399039544859017e-06,
      "loss": 0.2701,
      "step": 25008
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.32476760501013313,
      "learning_rate": 2.739221887222766e-06,
      "loss": 0.2218,
      "step": 25009
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.49560338907061235,
      "learning_rate": 2.73853989139356e-06,
      "loss": 0.3283,
      "step": 25010
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6151702617686771,
      "learning_rate": 2.7378579670049867e-06,
      "loss": 0.2911,
      "step": 25011
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7093811423236227,
      "learning_rate": 2.7371761140637543e-06,
      "loss": 0.302,
      "step": 25012
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3687108482581852,
      "learning_rate": 2.7364943325765768e-06,
      "loss": 0.1674,
      "step": 25013
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5074206510579204,
      "learning_rate": 2.735812622550156e-06,
      "loss": 0.3098,
      "step": 25014
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.305474400408855,
      "learning_rate": 2.7351309839911998e-06,
      "loss": 0.2227,
      "step": 25015
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3034690120879003,
      "learning_rate": 2.734449416906415e-06,
      "loss": 0.1827,
      "step": 25016
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4387996351701536,
      "learning_rate": 2.7337679213025093e-06,
      "loss": 0.1434,
      "step": 25017
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3518128711090198,
      "learning_rate": 2.733086497186187e-06,
      "loss": 0.2708,
      "step": 25018
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6682390662539868,
      "learning_rate": 2.732405144564149e-06,
      "loss": 0.3293,
      "step": 25019
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3136482405640136,
      "learning_rate": 2.7317238634430998e-06,
      "loss": 0.2332,
      "step": 25020
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7649573281097126,
      "learning_rate": 2.731042653829743e-06,
      "loss": 0.7007,
      "step": 25021
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3520587352843418,
      "learning_rate": 2.7303615157307827e-06,
      "loss": 0.1498,
      "step": 25022
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5202136754916824,
      "learning_rate": 2.729680449152916e-06,
      "loss": 0.3043,
      "step": 25023
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.20553161215696814,
      "learning_rate": 2.7289994541028454e-06,
      "loss": 0.0663,
      "step": 25024
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4779915435289775,
      "learning_rate": 2.728318530587273e-06,
      "loss": 0.2679,
      "step": 25025
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3214750306055363,
      "learning_rate": 2.727637678612889e-06,
      "loss": 0.2007,
      "step": 25026
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.29490474959095925,
      "learning_rate": 2.726956898186405e-06,
      "loss": 0.2651,
      "step": 25027
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8743252520717238,
      "learning_rate": 2.7262761893145097e-06,
      "loss": 0.3297,
      "step": 25028
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.73115997473787,
      "learning_rate": 2.7255955520039025e-06,
      "loss": 0.3785,
      "step": 25029
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8589242865428733,
      "learning_rate": 2.7249149862612824e-06,
      "loss": 0.2127,
      "step": 25030
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3994112984498337,
      "learning_rate": 2.7242344920933395e-06,
      "loss": 0.263,
      "step": 25031
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3211512559071727,
      "learning_rate": 2.723554069506772e-06,
      "loss": 0.2075,
      "step": 25032
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5086601171574932,
      "learning_rate": 2.722873718508273e-06,
      "loss": 0.2418,
      "step": 25033
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.523483261007602,
      "learning_rate": 2.722193439104541e-06,
      "loss": 0.2689,
      "step": 25034
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33439930039339905,
      "learning_rate": 2.7215132313022586e-06,
      "loss": 0.1484,
      "step": 25035
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5453464493799706,
      "learning_rate": 2.7208330951081286e-06,
      "loss": 0.3066,
      "step": 25036
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5689647367204876,
      "learning_rate": 2.720153030528836e-06,
      "loss": 0.25,
      "step": 25037
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38301438678252114,
      "learning_rate": 2.7194730375710733e-06,
      "loss": 0.3006,
      "step": 25038
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.32120428888979907,
      "learning_rate": 2.7187931162415324e-06,
      "loss": 0.1034,
      "step": 25039
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0029338719241916,
      "learning_rate": 2.7181132665468956e-06,
      "loss": 0.4138,
      "step": 25040
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3498460409864711,
      "learning_rate": 2.7174334884938625e-06,
      "loss": 0.2023,
      "step": 25041
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6252344211753433,
      "learning_rate": 2.716753782089112e-06,
      "loss": 0.7724,
      "step": 25042
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2532946163841741,
      "learning_rate": 2.716074147339335e-06,
      "loss": 0.1778,
      "step": 25043
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5102193930352739,
      "learning_rate": 2.7153945842512187e-06,
      "loss": 0.3749,
      "step": 25044
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.258932091399397,
      "learning_rate": 2.714715092831446e-06,
      "loss": 0.1612,
      "step": 25045
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6462951700613657,
      "learning_rate": 2.714035673086702e-06,
      "loss": 0.2748,
      "step": 25046
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4332756409810726,
      "learning_rate": 2.713356325023673e-06,
      "loss": 0.2648,
      "step": 25047
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.423456860961346,
      "learning_rate": 2.712677048649045e-06,
      "loss": 0.1412,
      "step": 25048
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8575558280387442,
      "learning_rate": 2.711997843969495e-06,
      "loss": 0.3079,
      "step": 25049
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2889070580542612,
      "learning_rate": 2.7113187109917084e-06,
      "loss": 0.2151,
      "step": 25050
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4976280184763576,
      "learning_rate": 2.7106396497223697e-06,
      "loss": 0.3305,
      "step": 25051
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9204677780013639,
      "learning_rate": 2.7099606601681503e-06,
      "loss": 0.299,
      "step": 25052
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7244099447505752,
      "learning_rate": 2.709281742335743e-06,
      "loss": 0.408,
      "step": 25053
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30624244524427513,
      "learning_rate": 2.708602896231818e-06,
      "loss": 0.1853,
      "step": 25054
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0213103620853388,
      "learning_rate": 2.7079241218630568e-06,
      "loss": 0.3934,
      "step": 25055
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33797131727458857,
      "learning_rate": 2.7072454192361407e-06,
      "loss": 0.2274,
      "step": 25056
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33732398149657367,
      "learning_rate": 2.706566788357741e-06,
      "loss": 0.1825,
      "step": 25057
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3676833596277016,
      "learning_rate": 2.7058882292345356e-06,
      "loss": 0.1558,
      "step": 25058
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4452960458758211,
      "learning_rate": 2.7052097418732038e-06,
      "loss": 0.1955,
      "step": 25059
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3044530240487122,
      "learning_rate": 2.704531326280421e-06,
      "loss": 0.5885,
      "step": 25060
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4046843380627173,
      "learning_rate": 2.7038529824628556e-06,
      "loss": 0.2221,
      "step": 25061
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.45013920431911236,
      "learning_rate": 2.7031747104271864e-06,
      "loss": 0.2806,
      "step": 25062
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.35476090190009013,
      "learning_rate": 2.7024965101800883e-06,
      "loss": 0.1659,
      "step": 25063
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.558309645827568,
      "learning_rate": 2.701818381728225e-06,
      "loss": 0.29,
      "step": 25064
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7810116095336129,
      "learning_rate": 2.7011403250782797e-06,
      "loss": 0.5003,
      "step": 25065
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26979995468296025,
      "learning_rate": 2.7004623402369113e-06,
      "loss": 0.1544,
      "step": 25066
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3546760002396715,
      "learning_rate": 2.699784427210802e-06,
      "loss": 0.0624,
      "step": 25067
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3611248279039129,
      "learning_rate": 2.6991065860066133e-06,
      "loss": 0.2701,
      "step": 25068
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31840143979218044,
      "learning_rate": 2.6984288166310156e-06,
      "loss": 0.2087,
      "step": 25069
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7922883172169972,
      "learning_rate": 2.6977511190906813e-06,
      "loss": 0.359,
      "step": 25070
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7733764123303016,
      "learning_rate": 2.6970734933922706e-06,
      "loss": 0.2553,
      "step": 25071
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25872116364428616,
      "learning_rate": 2.696395939542454e-06,
      "loss": 0.1686,
      "step": 25072
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3627814768395312,
      "learning_rate": 2.695718457547898e-06,
      "loss": 0.8324,
      "step": 25073
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3065085940599801,
      "learning_rate": 2.69504104741527e-06,
      "loss": 0.2349,
      "step": 25074
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4235218314257279,
      "learning_rate": 2.6943637091512288e-06,
      "loss": 0.2135,
      "step": 25075
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27799332202658356,
      "learning_rate": 2.693686442762441e-06,
      "loss": 0.0784,
      "step": 25076
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3889209493405798,
      "learning_rate": 2.6930092482555727e-06,
      "loss": 0.2917,
      "step": 25077
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9133989223395277,
      "learning_rate": 2.692332125637278e-06,
      "loss": 0.2386,
      "step": 25078
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.008558528850705,
      "learning_rate": 2.6916550749142313e-06,
      "loss": 0.4763,
      "step": 25079
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3412033805786166,
      "learning_rate": 2.6909780960930833e-06,
      "loss": 0.2428,
      "step": 25080
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42743965025941993,
      "learning_rate": 2.6903011891804975e-06,
      "loss": 0.2739,
      "step": 25081
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4274278580284951,
      "learning_rate": 2.6896243541831367e-06,
      "loss": 0.2535,
      "step": 25082
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4002292328652783,
      "learning_rate": 2.6889475911076534e-06,
      "loss": 0.669,
      "step": 25083
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.21157175251991178,
      "learning_rate": 2.688270899960711e-06,
      "loss": 0.077,
      "step": 25084
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.29892975378496706,
      "learning_rate": 2.6875942807489642e-06,
      "loss": 0.1488,
      "step": 25085
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3469597071128799,
      "learning_rate": 2.686917733479074e-06,
      "loss": 0.2551,
      "step": 25086
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4042396271839648,
      "learning_rate": 2.68624125815769e-06,
      "loss": 0.2072,
      "step": 25087
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9491132534307963,
      "learning_rate": 2.6855648547914726e-06,
      "loss": 0.3746,
      "step": 25088
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7196342019996711,
      "learning_rate": 2.684888523387077e-06,
      "loss": 0.2073,
      "step": 25089
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44211273021990616,
      "learning_rate": 2.684212263951149e-06,
      "loss": 0.2862,
      "step": 25090
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40310672495768696,
      "learning_rate": 2.683536076490354e-06,
      "loss": 0.2287,
      "step": 25091
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.34017571941204966,
      "learning_rate": 2.682859961011336e-06,
      "loss": 0.2949,
      "step": 25092
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3542462267664202,
      "learning_rate": 2.6821839175207486e-06,
      "loss": 0.0796,
      "step": 25093
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8806092038554442,
      "learning_rate": 2.681507946025247e-06,
      "loss": 0.4152,
      "step": 25094
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3488126156106965,
      "learning_rate": 2.680832046531472e-06,
      "loss": 0.1785,
      "step": 25095
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0906682820453968,
      "learning_rate": 2.680156219046086e-06,
      "loss": 0.348,
      "step": 25096
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.45474350964999954,
      "learning_rate": 2.6794804635757286e-06,
      "loss": 0.2491,
      "step": 25097
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5621905714720699,
      "learning_rate": 2.6788047801270525e-06,
      "loss": 0.0811,
      "step": 25098
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6117989261225477,
      "learning_rate": 2.678129168706702e-06,
      "loss": 0.3332,
      "step": 25099
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.34463518682443317,
      "learning_rate": 2.677453629321329e-06,
      "loss": 0.2168,
      "step": 25100
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6006843681287404,
      "learning_rate": 2.6767781619775745e-06,
      "loss": 0.694,
      "step": 25101
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3876927740741379,
      "learning_rate": 2.676102766682086e-06,
      "loss": 0.0862,
      "step": 25102
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3155758522155271,
      "learning_rate": 2.6754274434415116e-06,
      "loss": 0.2197,
      "step": 25103
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25939643270490836,
      "learning_rate": 2.674752192262486e-06,
      "loss": 0.1951,
      "step": 25104
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9118423634656482,
      "learning_rate": 2.6740770131516645e-06,
      "loss": 0.3396,
      "step": 25105
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2488273960701692,
      "learning_rate": 2.67340190611568e-06,
      "loss": 0.2949,
      "step": 25106
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5341235195741025,
      "learning_rate": 2.67272687116118e-06,
      "loss": 0.7437,
      "step": 25107
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.346022724447752,
      "learning_rate": 2.672051908294806e-06,
      "loss": 0.1608,
      "step": 25108
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5561254109289424,
      "learning_rate": 2.6713770175231935e-06,
      "loss": 0.3516,
      "step": 25109
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3203514622540656,
      "learning_rate": 2.6707021988529857e-06,
      "loss": 0.2288,
      "step": 25110
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5722143441427758,
      "learning_rate": 2.670027452290821e-06,
      "loss": 0.1929,
      "step": 25111
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.47678267624154735,
      "learning_rate": 2.6693527778433415e-06,
      "loss": 0.2081,
      "step": 25112
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.29736066335457934,
      "learning_rate": 2.6686781755171785e-06,
      "loss": 0.1731,
      "step": 25113
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1688645337425274,
      "learning_rate": 2.6680036453189715e-06,
      "loss": 0.3757,
      "step": 25114
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2795836233568261,
      "learning_rate": 2.667329187255361e-06,
      "loss": 0.1973,
      "step": 25115
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.0455118284104505,
      "learning_rate": 2.6666548013329717e-06,
      "loss": 0.4449,
      "step": 25116
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31418862583266627,
      "learning_rate": 2.665980487558452e-06,
      "loss": 0.1469,
      "step": 25117
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5787137155178872,
      "learning_rate": 2.665306245938426e-06,
      "loss": 0.3499,
      "step": 25118
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1220913995393362,
      "learning_rate": 2.664632076479531e-06,
      "loss": 0.4435,
      "step": 25119
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.956976104752178,
      "learning_rate": 2.6639579791884017e-06,
      "loss": 0.442,
      "step": 25120
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3113643965742778,
      "learning_rate": 2.663283954071663e-06,
      "loss": 0.2026,
      "step": 25121
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4535096903878242,
      "learning_rate": 2.662610001135956e-06,
      "loss": 0.2615,
      "step": 25122
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5897402833002807,
      "learning_rate": 2.6619361203879026e-06,
      "loss": 0.2679,
      "step": 25123
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.15777129050003008,
      "learning_rate": 2.6612623118341373e-06,
      "loss": 0.0697,
      "step": 25124
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7772827244561076,
      "learning_rate": 2.6605885754812898e-06,
      "loss": 0.7916,
      "step": 25125
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3791371468490266,
      "learning_rate": 2.6599149113359855e-06,
      "loss": 0.1579,
      "step": 25126
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.38213914963848306,
      "learning_rate": 2.659241319404853e-06,
      "loss": 0.2896,
      "step": 25127
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3183353338489228,
      "learning_rate": 2.658567799694519e-06,
      "loss": 0.2131,
      "step": 25128
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.9757611917122888,
      "learning_rate": 2.657894352211614e-06,
      "loss": 0.7534,
      "step": 25129
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6272452696338938,
      "learning_rate": 2.6572209769627576e-06,
      "loss": 0.2685,
      "step": 25130
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3944496820496899,
      "learning_rate": 2.656547673954577e-06,
      "loss": 0.2292,
      "step": 25131
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3165094480378108,
      "learning_rate": 2.6558744431937e-06,
      "loss": 0.1655,
      "step": 25132
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40799003384573196,
      "learning_rate": 2.6552012846867404e-06,
      "loss": 0.1974,
      "step": 25133
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.313716707162584,
      "learning_rate": 2.6545281984403336e-06,
      "loss": 0.2391,
      "step": 25134
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6346625722336988,
      "learning_rate": 2.6538551844610927e-06,
      "loss": 0.3114,
      "step": 25135
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.32371876772703556,
      "learning_rate": 2.6531822427556407e-06,
      "loss": 0.2192,
      "step": 25136
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5971423124086948,
      "learning_rate": 2.6525093733305995e-06,
      "loss": 0.0227,
      "step": 25137
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6818791653886906,
      "learning_rate": 2.6518365761925923e-06,
      "loss": 0.4279,
      "step": 25138
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2987094132846623,
      "learning_rate": 2.6511638513482317e-06,
      "loss": 0.2146,
      "step": 25139
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3636712640094823,
      "learning_rate": 2.650491198804139e-06,
      "loss": 0.2824,
      "step": 25140
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8387648988461948,
      "learning_rate": 2.649818618566935e-06,
      "loss": 0.2299,
      "step": 25141
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27197004270049924,
      "learning_rate": 2.6491461106432283e-06,
      "loss": 0.1471,
      "step": 25142
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.2510204516211758,
      "learning_rate": 2.6484736750396467e-06,
      "loss": 0.3487,
      "step": 25143
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.611086457789957,
      "learning_rate": 2.6478013117627964e-06,
      "loss": 0.2664,
      "step": 25144
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.350517336133067,
      "learning_rate": 2.6471290208192967e-06,
      "loss": 0.2105,
      "step": 25145
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.32971661833035704,
      "learning_rate": 2.6464568022157645e-06,
      "loss": 0.2958,
      "step": 25146
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.863033588298068,
      "learning_rate": 2.645784655958803e-06,
      "loss": 0.2849,
      "step": 25147
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6463978002069788,
      "learning_rate": 2.6451125820550393e-06,
      "loss": 0.3446,
      "step": 25148
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3632727613349751,
      "learning_rate": 2.644440580511074e-06,
      "loss": 0.1665,
      "step": 25149
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5411927554503972,
      "learning_rate": 2.6437686513335225e-06,
      "loss": 0.2779,
      "step": 25150
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2563622147907639,
      "learning_rate": 2.6430967945289985e-06,
      "loss": 0.1829,
      "step": 25151
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.22184757329933663,
      "learning_rate": 2.6424250101041073e-06,
      "loss": 0.0971,
      "step": 25152
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7876879108012065,
      "learning_rate": 2.641753298065458e-06,
      "loss": 0.3957,
      "step": 25153
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2808133798348357,
      "learning_rate": 2.641081658419662e-06,
      "loss": 0.1865,
      "step": 25154
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.050860779979623,
      "learning_rate": 2.640410091173329e-06,
      "loss": 0.5227,
      "step": 25155
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.989915221884477,
      "learning_rate": 2.639738596333059e-06,
      "loss": 0.2642,
      "step": 25156
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3644372060055659,
      "learning_rate": 2.639067173905464e-06,
      "loss": 0.2777,
      "step": 25157
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31315626058189855,
      "learning_rate": 2.638395823897151e-06,
      "loss": 0.1501,
      "step": 25158
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5251898019733621,
      "learning_rate": 2.637724546314715e-06,
      "loss": 0.2925,
      "step": 25159
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.612521907080594,
      "learning_rate": 2.6370533411647748e-06,
      "loss": 0.0272,
      "step": 25160
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5137905689718159,
      "learning_rate": 2.6363822084539225e-06,
      "loss": 0.3174,
      "step": 25161
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3654434896155209,
      "learning_rate": 2.6357111481887653e-06,
      "loss": 0.2207,
      "step": 25162
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36332842612404825,
      "learning_rate": 2.635040160375909e-06,
      "loss": 0.2567,
      "step": 25163
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.26139621916131783,
      "learning_rate": 2.6343692450219473e-06,
      "loss": 0.1526,
      "step": 25164
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3547385066561533,
      "learning_rate": 2.6336984021334846e-06,
      "loss": 0.2894,
      "step": 25165
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1315750333162424,
      "learning_rate": 2.63302763171712e-06,
      "loss": 0.3822,
      "step": 25166
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2800044304935316,
      "learning_rate": 2.632356933779454e-06,
      "loss": 0.1707,
      "step": 25167
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4264511290732702,
      "learning_rate": 2.631686308327086e-06,
      "loss": 0.4301,
      "step": 25168
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3211797191648234,
      "learning_rate": 2.631015755366615e-06,
      "loss": 0.2303,
      "step": 25169
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4228647718662098,
      "learning_rate": 2.630345274904632e-06,
      "loss": 0.2502,
      "step": 25170
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3081360100582171,
      "learning_rate": 2.6296748669477368e-06,
      "loss": 0.1531,
      "step": 25171
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5324068335007908,
      "learning_rate": 2.629004531502529e-06,
      "loss": 0.3442,
      "step": 25172
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.732304478156813,
      "learning_rate": 2.6283342685755954e-06,
      "loss": 0.2553,
      "step": 25173
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.542481248659162,
      "learning_rate": 2.6276640781735386e-06,
      "loss": 0.2431,
      "step": 25174
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3156654740288968,
      "learning_rate": 2.626993960302947e-06,
      "loss": 0.2217,
      "step": 25175
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9388015432867689,
      "learning_rate": 2.6263239149704146e-06,
      "loss": 0.464,
      "step": 25176
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2930714448369511,
      "learning_rate": 2.6256539421825357e-06,
      "loss": 0.1804,
      "step": 25177
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.044650222562766,
      "learning_rate": 2.624984041945898e-06,
      "loss": 0.1676,
      "step": 25178
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6588551287065294,
      "learning_rate": 2.624314214267093e-06,
      "loss": 0.6992,
      "step": 25179
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3540038000926851,
      "learning_rate": 2.623644459152712e-06,
      "loss": 0.1949,
      "step": 25180
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3431782332208261,
      "learning_rate": 2.6229747766093484e-06,
      "loss": 0.2399,
      "step": 25181
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4135600503831603,
      "learning_rate": 2.6223051666435817e-06,
      "loss": 0.2554,
      "step": 25182
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44249544108520844,
      "learning_rate": 2.621635629262005e-06,
      "loss": 0.239,
      "step": 25183
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30243432665838216,
      "learning_rate": 2.620966164471207e-06,
      "loss": 0.0711,
      "step": 25184
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.654506351945739,
      "learning_rate": 2.620296772277766e-06,
      "loss": 0.322,
      "step": 25185
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3709221328447927,
      "learning_rate": 2.6196274526882803e-06,
      "loss": 0.2099,
      "step": 25186
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3542198876114982,
      "learning_rate": 2.6189582057093254e-06,
      "loss": 0.3225,
      "step": 25187
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9968072740729477,
      "learning_rate": 2.6182890313474873e-06,
      "loss": 0.4275,
      "step": 25188
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8738730884596432,
      "learning_rate": 2.617619929609354e-06,
      "loss": 0.4196,
      "step": 25189
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.25831836312005746,
      "learning_rate": 2.6169509005015027e-06,
      "loss": 0.169,
      "step": 25190
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.659839353225275,
      "learning_rate": 2.6162819440305177e-06,
      "loss": 0.2642,
      "step": 25191
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2694841291629514,
      "learning_rate": 2.6156130602029795e-06,
      "loss": 0.1731,
      "step": 25192
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31235471111018015,
      "learning_rate": 2.6149442490254708e-06,
      "loss": 0.1982,
      "step": 25193
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5606993147210924,
      "learning_rate": 2.6142755105045724e-06,
      "loss": 0.1911,
      "step": 25194
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33220617833968635,
      "learning_rate": 2.6136068446468597e-06,
      "loss": 0.1859,
      "step": 25195
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.12440901756351,
      "learning_rate": 2.612938251458913e-06,
      "loss": 0.6602,
      "step": 25196
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7470614103264284,
      "learning_rate": 2.612269730947311e-06,
      "loss": 0.3324,
      "step": 25197
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2933911903673651,
      "learning_rate": 2.6116012831186323e-06,
      "loss": 0.2468,
      "step": 25198
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.42496229637163524,
      "learning_rate": 2.6109329079794453e-06,
      "loss": 0.1801,
      "step": 25199
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4831276799326618,
      "learning_rate": 2.6102646055363388e-06,
      "loss": 0.3455,
      "step": 25200
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31097196077857886,
      "learning_rate": 2.60959637579588e-06,
      "loss": 0.1168,
      "step": 25201
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5288178631352812,
      "learning_rate": 2.608928218764638e-06,
      "loss": 0.2303,
      "step": 25202
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4476565712485988,
      "learning_rate": 2.6082601344491978e-06,
      "loss": 0.0658,
      "step": 25203
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.31269374014424295,
      "learning_rate": 2.607592122856124e-06,
      "loss": 0.1874,
      "step": 25204
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33271622511442944,
      "learning_rate": 2.6069241839919913e-06,
      "loss": 0.2593,
      "step": 25205
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.019437704533941,
      "learning_rate": 2.6062563178633703e-06,
      "loss": 0.3304,
      "step": 25206
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7972754803498628,
      "learning_rate": 2.6055885244768363e-06,
      "loss": 0.4003,
      "step": 25207
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3349568779039001,
      "learning_rate": 2.6049208038389527e-06,
      "loss": 0.1755,
      "step": 25208
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5832570414565184,
      "learning_rate": 2.604253155956291e-06,
      "loss": 0.3506,
      "step": 25209
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4518000649540514,
      "learning_rate": 2.603585580835423e-06,
      "loss": 0.2004,
      "step": 25210
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3176755490065798,
      "learning_rate": 2.602918078482909e-06,
      "loss": 0.2025,
      "step": 25211
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2689980973119779,
      "learning_rate": 2.6022506489053255e-06,
      "loss": 0.0745,
      "step": 25212
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.37687332688535924,
      "learning_rate": 2.6015832921092323e-06,
      "loss": 0.2758,
      "step": 25213
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3910769084854342,
      "learning_rate": 2.6009160081011963e-06,
      "loss": 0.1508,
      "step": 25214
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8755456967626252,
      "learning_rate": 2.6002487968877863e-06,
      "loss": 0.4158,
      "step": 25215
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3372717912429946,
      "learning_rate": 2.5995816584755605e-06,
      "loss": 0.2293,
      "step": 25216
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27188808373245493,
      "learning_rate": 2.5989145928710868e-06,
      "loss": 0.1624,
      "step": 25217
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.855875644351369,
      "learning_rate": 2.5982476000809254e-06,
      "loss": 0.8298,
      "step": 25218
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1390769053959438,
      "learning_rate": 2.59758068011164e-06,
      "loss": 0.2853,
      "step": 25219
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30833743019898446,
      "learning_rate": 2.596913832969794e-06,
      "loss": 0.1695,
      "step": 25220
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36646168397567824,
      "learning_rate": 2.5962470586619427e-06,
      "loss": 0.0683,
      "step": 25221
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6056698842804334,
      "learning_rate": 2.595580357194649e-06,
      "loss": 0.3269,
      "step": 25222
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.29907024342588906,
      "learning_rate": 2.594913728574473e-06,
      "loss": 0.2137,
      "step": 25223
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9032922655589348,
      "learning_rate": 2.5942471728079742e-06,
      "loss": 0.3725,
      "step": 25224
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6341583952805588,
      "learning_rate": 2.5935806899017057e-06,
      "loss": 0.2563,
      "step": 25225
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9325444230933578,
      "learning_rate": 2.5929142798622287e-06,
      "loss": 0.47,
      "step": 25226
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33079119496522674,
      "learning_rate": 2.5922479426961e-06,
      "loss": 0.2175,
      "step": 25227
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.49991897556814185,
      "learning_rate": 2.591581678409868e-06,
      "loss": 0.3145,
      "step": 25228
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27804352024276136,
      "learning_rate": 2.5909154870100983e-06,
      "loss": 0.1388,
      "step": 25229
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30981786534221006,
      "learning_rate": 2.590249368503338e-06,
      "loss": 0.0613,
      "step": 25230
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3642608831639805,
      "learning_rate": 2.589583322896142e-06,
      "loss": 0.2415,
      "step": 25231
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.1622034243839474,
      "learning_rate": 2.588917350195067e-06,
      "loss": 0.2585,
      "step": 25232
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9599398233511734,
      "learning_rate": 2.588251450406658e-06,
      "loss": 0.4384,
      "step": 25233
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3257057135725151,
      "learning_rate": 2.587585623537471e-06,
      "loss": 0.2284,
      "step": 25234
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6660580821615822,
      "learning_rate": 2.5869198695940557e-06,
      "loss": 0.3878,
      "step": 25235
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3305529401799516,
      "learning_rate": 2.586254188582964e-06,
      "loss": 0.2226,
      "step": 25236
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7041193212997987,
      "learning_rate": 2.585588580510737e-06,
      "loss": 0.4572,
      "step": 25237
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.17913932489287446,
      "learning_rate": 2.584923045383936e-06,
      "loss": 0.0689,
      "step": 25238
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4821119492064932,
      "learning_rate": 2.584257583209099e-06,
      "loss": 0.2843,
      "step": 25239
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2991512901028735,
      "learning_rate": 2.5835921939927744e-06,
      "loss": 0.1702,
      "step": 25240
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5946315134816716,
      "learning_rate": 2.582926877741515e-06,
      "loss": 0.3185,
      "step": 25241
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7756514047748795,
      "learning_rate": 2.582261634461858e-06,
      "loss": 0.2979,
      "step": 25242
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9593717839379434,
      "learning_rate": 2.581596464160351e-06,
      "loss": 0.5106,
      "step": 25243
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39348785839138717,
      "learning_rate": 2.58093136684354e-06,
      "loss": 0.2499,
      "step": 25244
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4841414054464599,
      "learning_rate": 2.5802663425179663e-06,
      "loss": 0.2223,
      "step": 25245
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4967986016247869,
      "learning_rate": 2.5796013911901774e-06,
      "loss": 0.3364,
      "step": 25246
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3141077904302627,
      "learning_rate": 2.5789365128667086e-06,
      "loss": 0.1443,
      "step": 25247
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4397923052891437,
      "learning_rate": 2.578271707554105e-06,
      "loss": 0.2101,
      "step": 25248
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39490400781371265,
      "learning_rate": 2.5776069752589063e-06,
      "loss": 0.1734,
      "step": 25249
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7397442789358186,
      "learning_rate": 2.5769423159876537e-06,
      "loss": 0.3522,
      "step": 25250
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.3724259874715814,
      "learning_rate": 2.5762777297468834e-06,
      "loss": 0.2691,
      "step": 25251
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.35203389284858283,
      "learning_rate": 2.5756132165431358e-06,
      "loss": 0.2757,
      "step": 25252
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.39032953849951496,
      "learning_rate": 2.5749487763829504e-06,
      "loss": 0.1396,
      "step": 25253
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.332076227085996,
      "learning_rate": 2.574284409272857e-06,
      "loss": 0.2506,
      "step": 25254
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8109347192625109,
      "learning_rate": 2.5736201152194027e-06,
      "loss": 0.2843,
      "step": 25255
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3944715145283462,
      "learning_rate": 2.5729558942291144e-06,
      "loss": 0.1393,
      "step": 25256
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.04723695593127,
      "learning_rate": 2.5722917463085295e-06,
      "loss": 0.5163,
      "step": 25257
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2669418800769811,
      "learning_rate": 2.5716276714641853e-06,
      "loss": 0.1646,
      "step": 25258
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.34627453237161926,
      "learning_rate": 2.57096366970261e-06,
      "loss": 0.2864,
      "step": 25259
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4226444099609465,
      "learning_rate": 2.5702997410303377e-06,
      "loss": 0.1388,
      "step": 25260
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.5409234740261435,
      "learning_rate": 2.5696358854539016e-06,
      "loss": 0.7391,
      "step": 25261
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33333328720138405,
      "learning_rate": 2.5689721029798364e-06,
      "loss": 0.1499,
      "step": 25262
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.376006585405234,
      "learning_rate": 2.5683083936146646e-06,
      "loss": 0.2909,
      "step": 25263
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3036060278633475,
      "learning_rate": 2.567644757364921e-06,
      "loss": 0.2184,
      "step": 25264
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.4547107867687716,
      "learning_rate": 2.566981194237137e-06,
      "loss": 0.5795,
      "step": 25265
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7091856878022675,
      "learning_rate": 2.5663177042378305e-06,
      "loss": 0.2785,
      "step": 25266
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44867748447765804,
      "learning_rate": 2.5656542873735434e-06,
      "loss": 0.2965,
      "step": 25267
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.24905442584073578,
      "learning_rate": 2.5649909436507926e-06,
      "loss": 0.1578,
      "step": 25268
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44463460397271093,
      "learning_rate": 2.564327673076107e-06,
      "loss": 0.1007,
      "step": 25269
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.32884801832179567,
      "learning_rate": 2.5636644756560123e-06,
      "loss": 0.2826,
      "step": 25270
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3312423562986718,
      "learning_rate": 2.563001351397034e-06,
      "loss": 0.1495,
      "step": 25271
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.40055828366072754,
      "learning_rate": 2.562338300305699e-06,
      "loss": 0.2941,
      "step": 25272
    },
    {
      "epoch": 0.77,
      "grad_norm": 2.4101795333784133,
      "learning_rate": 2.5616753223885226e-06,
      "loss": 0.2403,
      "step": 25273
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8260949169862014,
      "learning_rate": 2.561012417652032e-06,
      "loss": 0.3838,
      "step": 25274
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.44182846332141107,
      "learning_rate": 2.5603495861027506e-06,
      "loss": 0.2485,
      "step": 25275
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.41242334453864765,
      "learning_rate": 2.5596868277472e-06,
      "loss": 0.2611,
      "step": 25276
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3155714537010231,
      "learning_rate": 2.559024142591895e-06,
      "loss": 0.2064,
      "step": 25277
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.28598710181906684,
      "learning_rate": 2.5583615306433596e-06,
      "loss": 0.1453,
      "step": 25278
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9665656037774982,
      "learning_rate": 2.5576989919081154e-06,
      "loss": 0.5227,
      "step": 25279
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.7022094285103693,
      "learning_rate": 2.557036526392671e-06,
      "loss": 0.9103,
      "step": 25280
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2697452283710513,
      "learning_rate": 2.5563741341035555e-06,
      "loss": 0.1729,
      "step": 25281
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.36638848645263156,
      "learning_rate": 2.555711815047278e-06,
      "loss": 0.2156,
      "step": 25282
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.743396565832624,
      "learning_rate": 2.555049569230358e-06,
      "loss": 0.4417,
      "step": 25283
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.7963902094357665,
      "learning_rate": 2.5543873966593123e-06,
      "loss": 0.2674,
      "step": 25284
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6314940923226544,
      "learning_rate": 2.55372529734065e-06,
      "loss": 0.2238,
      "step": 25285
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.23272222115659458,
      "learning_rate": 2.553063271280889e-06,
      "loss": 0.1524,
      "step": 25286
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.33463173453383627,
      "learning_rate": 2.5524013184865415e-06,
      "loss": 0.2534,
      "step": 25287
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4406777873857771,
      "learning_rate": 2.5517394389641236e-06,
      "loss": 0.2494,
      "step": 25288
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.8227630081071002,
      "learning_rate": 2.5510776327201413e-06,
      "loss": 0.4254,
      "step": 25289
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3149989826385455,
      "learning_rate": 2.550415899761108e-06,
      "loss": 0.1976,
      "step": 25290
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9758558843602222,
      "learning_rate": 2.549754240093537e-06,
      "loss": 0.4214,
      "step": 25291
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9673696581340058,
      "learning_rate": 2.54909265372393e-06,
      "loss": 0.2638,
      "step": 25292
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.48294244881071935,
      "learning_rate": 2.5484311406588056e-06,
      "loss": 0.2747,
      "step": 25293
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.27674349550557764,
      "learning_rate": 2.5477697009046665e-06,
      "loss": 0.183,
      "step": 25294
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.5796187002167945,
      "learning_rate": 2.54710833446802e-06,
      "loss": 0.211,
      "step": 25295
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.43708127469972524,
      "learning_rate": 2.5464470413553775e-06,
      "loss": 0.202,
      "step": 25296
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.30567035201730425,
      "learning_rate": 2.5457858215732354e-06,
      "loss": 0.1544,
      "step": 25297
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.9454636698057903,
      "learning_rate": 2.545124675128111e-06,
      "loss": 0.4521,
      "step": 25298
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.305024694432985,
      "learning_rate": 2.5444636020265e-06,
      "loss": 0.1828,
      "step": 25299
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4945117153899549,
      "learning_rate": 2.5438026022749096e-06,
      "loss": 0.3114,
      "step": 25300
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.6530499924008617,
      "learning_rate": 2.5431416758798444e-06,
      "loss": 0.2443,
      "step": 25301
    },
    {
      "epoch": 0.77,
      "grad_norm": 1.6692847733132252,
      "learning_rate": 2.5424808228478036e-06,
      "loss": 0.7017,
      "step": 25302
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3874523987702772,
      "learning_rate": 2.54182004318529e-06,
      "loss": 0.0637,
      "step": 25303
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.3559348080196987,
      "learning_rate": 2.5411593368988042e-06,
      "loss": 0.2576,
      "step": 25304
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.2713050638674676,
      "learning_rate": 2.5404987039948505e-06,
      "loss": 0.1499,
      "step": 25305
    },
    {
      "epoch": 0.77,
      "grad_norm": 0.4859192251016952,
      "learning_rate": 2.5398381444799192e-06,
      "loss": 0.3347,
      "step": 25306
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.005374239498616,
      "learning_rate": 2.539177658360521e-06,
      "loss": 0.4531,
      "step": 25307
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3111537116602895,
      "learning_rate": 2.538517245643145e-06,
      "loss": 0.1768,
      "step": 25308
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40710986818874206,
      "learning_rate": 2.5378569063342905e-06,
      "loss": 0.2598,
      "step": 25309
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.640587275707632,
      "learning_rate": 2.537196640440459e-06,
      "loss": 0.2425,
      "step": 25310
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5014632905388459,
      "learning_rate": 2.53653644796814e-06,
      "loss": 0.3172,
      "step": 25311
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3234861066843826,
      "learning_rate": 2.53587632892383e-06,
      "loss": 0.1621,
      "step": 25312
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3703962906341178,
      "learning_rate": 2.535216283314026e-06,
      "loss": 0.2834,
      "step": 25313
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23577833896025546,
      "learning_rate": 2.534556311145222e-06,
      "loss": 0.1026,
      "step": 25314
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5518078038164402,
      "learning_rate": 2.5338964124239075e-06,
      "loss": 0.9093,
      "step": 25315
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6320641331429654,
      "learning_rate": 2.5332365871565766e-06,
      "loss": 0.2969,
      "step": 25316
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4219486949772845,
      "learning_rate": 2.532576835349724e-06,
      "loss": 0.2602,
      "step": 25317
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2776910122691747,
      "learning_rate": 2.5319171570098312e-06,
      "loss": 0.2291,
      "step": 25318
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1386198340981284,
      "learning_rate": 2.5312575521434014e-06,
      "loss": 0.4807,
      "step": 25319
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.47938442277708093,
      "learning_rate": 2.5305980207569137e-06,
      "loss": 0.1041,
      "step": 25320
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29917743967955834,
      "learning_rate": 2.5299385628568607e-06,
      "loss": 0.1466,
      "step": 25321
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5480227054508308,
      "learning_rate": 2.529279178449734e-06,
      "loss": 0.2945,
      "step": 25322
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4054850606132487,
      "learning_rate": 2.528619867542009e-06,
      "loss": 0.1984,
      "step": 25323
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.47404577966422246,
      "learning_rate": 2.527960630140188e-06,
      "loss": 0.3058,
      "step": 25324
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6028477011169131,
      "learning_rate": 2.5273014662507446e-06,
      "loss": 0.2407,
      "step": 25325
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.381066058106349,
      "learning_rate": 2.5266423758801695e-06,
      "loss": 0.2447,
      "step": 25326
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7539806806632147,
      "learning_rate": 2.525983359034948e-06,
      "loss": 0.283,
      "step": 25327
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28721883686540656,
      "learning_rate": 2.5253244157215585e-06,
      "loss": 0.1758,
      "step": 25328
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3255220807912915,
      "learning_rate": 2.5246655459464865e-06,
      "loss": 0.2223,
      "step": 25329
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8797170647272499,
      "learning_rate": 2.5240067497162156e-06,
      "loss": 0.3254,
      "step": 25330
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33780268231176364,
      "learning_rate": 2.523348027037228e-06,
      "loss": 0.191,
      "step": 25331
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3087650336106949,
      "learning_rate": 2.5226893779160012e-06,
      "loss": 0.419,
      "step": 25332
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8498533302398985,
      "learning_rate": 2.5220308023590156e-06,
      "loss": 0.3464,
      "step": 25333
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7579879706760793,
      "learning_rate": 2.5213723003727542e-06,
      "loss": 0.2768,
      "step": 25334
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34579819046010724,
      "learning_rate": 2.520713871963687e-06,
      "loss": 0.2599,
      "step": 25335
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3543217491328091,
      "learning_rate": 2.5200555171383045e-06,
      "loss": 0.2144,
      "step": 25336
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2707002679875293,
      "learning_rate": 2.5193972359030737e-06,
      "loss": 0.1824,
      "step": 25337
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5265989712697594,
      "learning_rate": 2.518739028264474e-06,
      "loss": 0.0387,
      "step": 25338
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8919827082970709,
      "learning_rate": 2.5180808942289827e-06,
      "loss": 0.382,
      "step": 25339
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3346287795925028,
      "learning_rate": 2.517422833803075e-06,
      "loss": 0.1898,
      "step": 25340
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3539741382849725,
      "learning_rate": 2.5167648469932206e-06,
      "loss": 0.2768,
      "step": 25341
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9678480522765122,
      "learning_rate": 2.5161069338058964e-06,
      "loss": 0.4454,
      "step": 25342
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9754707213772701,
      "learning_rate": 2.5154490942475785e-06,
      "loss": 0.3695,
      "step": 25343
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4052106577812164,
      "learning_rate": 2.5147913283247282e-06,
      "loss": 0.1815,
      "step": 25344
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5408835827958469,
      "learning_rate": 2.5141336360438295e-06,
      "loss": 0.3384,
      "step": 25345
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2953681611157844,
      "learning_rate": 2.513476017411345e-06,
      "loss": 0.0783,
      "step": 25346
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.25212695187576384,
      "learning_rate": 2.512818472433747e-06,
      "loss": 0.2095,
      "step": 25347
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4585161338488101,
      "learning_rate": 2.5121610011175056e-06,
      "loss": 0.0604,
      "step": 25348
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31882898491907646,
      "learning_rate": 2.5115036034690844e-06,
      "loss": 0.1905,
      "step": 25349
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4410642184264844,
      "learning_rate": 2.5108462794949595e-06,
      "loss": 0.5579,
      "step": 25350
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8988317209769591,
      "learning_rate": 2.5101890292015905e-06,
      "loss": 0.2963,
      "step": 25351
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4703585896091735,
      "learning_rate": 2.509531852595446e-06,
      "loss": 0.321,
      "step": 25352
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3074914316873189,
      "learning_rate": 2.5088747496829957e-06,
      "loss": 0.1486,
      "step": 25353
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3797439859642113,
      "learning_rate": 2.508217720470697e-06,
      "loss": 0.2749,
      "step": 25354
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8819227436710263,
      "learning_rate": 2.507560764965017e-06,
      "loss": 0.4314,
      "step": 25355
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.27451579974661405,
      "learning_rate": 2.5069038831724202e-06,
      "loss": 0.1524,
      "step": 25356
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6419664163306241,
      "learning_rate": 2.506247075099372e-06,
      "loss": 0.0754,
      "step": 25357
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.46213231725905846,
      "learning_rate": 2.5055903407523265e-06,
      "loss": 0.2739,
      "step": 25358
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33138375839036227,
      "learning_rate": 2.50493368013775e-06,
      "loss": 0.2218,
      "step": 25359
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4563329525549839,
      "learning_rate": 2.504277093262105e-06,
      "loss": 0.3179,
      "step": 25360
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8002993360769003,
      "learning_rate": 2.5036205801318425e-06,
      "loss": 0.2893,
      "step": 25361
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33012388180450725,
      "learning_rate": 2.502964140753433e-06,
      "loss": 0.1124,
      "step": 25362
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3893766734747235,
      "learning_rate": 2.502307775133327e-06,
      "loss": 0.2961,
      "step": 25363
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30965409744017264,
      "learning_rate": 2.5016514832779835e-06,
      "loss": 0.1583,
      "step": 25364
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5414714441809159,
      "learning_rate": 2.500995265193863e-06,
      "loss": 0.3641,
      "step": 25365
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3260464232324807,
      "learning_rate": 2.500339120887416e-06,
      "loss": 0.0779,
      "step": 25366
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33845687274674646,
      "learning_rate": 2.4996830503650993e-06,
      "loss": 0.2549,
      "step": 25367
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8571110338810799,
      "learning_rate": 2.4990270536333705e-06,
      "loss": 0.2813,
      "step": 25368
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9477881076733303,
      "learning_rate": 2.4983711306986834e-06,
      "loss": 0.4065,
      "step": 25369
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4585559160250957,
      "learning_rate": 2.4977152815674834e-06,
      "loss": 0.2655,
      "step": 25370
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3600847977763299,
      "learning_rate": 2.4970595062462355e-06,
      "loss": 0.2571,
      "step": 25371
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3521745052334415,
      "learning_rate": 2.496403804741382e-06,
      "loss": 0.2169,
      "step": 25372
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4694846705898224,
      "learning_rate": 2.4957481770593773e-06,
      "loss": 0.6121,
      "step": 25373
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.17597603012681237,
      "learning_rate": 2.495092623206673e-06,
      "loss": 0.0674,
      "step": 25374
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6451227667425341,
      "learning_rate": 2.494437143189712e-06,
      "loss": 0.1649,
      "step": 25375
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3424443185139975,
      "learning_rate": 2.493781737014953e-06,
      "loss": 0.2465,
      "step": 25376
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4400973298265197,
      "learning_rate": 2.4931264046888372e-06,
      "loss": 0.2345,
      "step": 25377
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5308849214642998,
      "learning_rate": 2.492471146217814e-06,
      "loss": 0.3352,
      "step": 25378
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1131835944151383,
      "learning_rate": 2.4918159616083325e-06,
      "loss": 0.5311,
      "step": 25379
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5270343288404055,
      "learning_rate": 2.491160850866833e-06,
      "loss": 0.2543,
      "step": 25380
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34832599365947176,
      "learning_rate": 2.490505813999764e-06,
      "loss": 0.2073,
      "step": 25381
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4288805379727183,
      "learning_rate": 2.489850851013569e-06,
      "loss": 0.5678,
      "step": 25382
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1953163533065033,
      "learning_rate": 2.4891959619146968e-06,
      "loss": 0.1678,
      "step": 25383
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0139960882581238,
      "learning_rate": 2.4885411467095823e-06,
      "loss": 0.5365,
      "step": 25384
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35315017844209223,
      "learning_rate": 2.4878864054046713e-06,
      "loss": 0.1689,
      "step": 25385
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5082750452415832,
      "learning_rate": 2.4872317380064093e-06,
      "loss": 0.3524,
      "step": 25386
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6770896944804626,
      "learning_rate": 2.4865771445212274e-06,
      "loss": 0.2883,
      "step": 25387
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28940581531762055,
      "learning_rate": 2.485922624955578e-06,
      "loss": 0.2249,
      "step": 25388
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8659940890244155,
      "learning_rate": 2.485268179315891e-06,
      "loss": 0.463,
      "step": 25389
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3190164357830335,
      "learning_rate": 2.484613807608608e-06,
      "loss": 0.2063,
      "step": 25390
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2949651318403483,
      "learning_rate": 2.48395950984017e-06,
      "loss": 0.4533,
      "step": 25391
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3437919491560477,
      "learning_rate": 2.483305286017009e-06,
      "loss": 0.0835,
      "step": 25392
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40300640004779154,
      "learning_rate": 2.482651136145564e-06,
      "loss": 0.2484,
      "step": 25393
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3292330838446772,
      "learning_rate": 2.481997060232271e-06,
      "loss": 0.1818,
      "step": 25394
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2801648811540637,
      "learning_rate": 2.4813430582835676e-06,
      "loss": 0.2672,
      "step": 25395
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8433914856854209,
      "learning_rate": 2.480689130305881e-06,
      "loss": 0.2454,
      "step": 25396
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4027344897498146,
      "learning_rate": 2.48003527630565e-06,
      "loss": 0.8652,
      "step": 25397
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2997635785193443,
      "learning_rate": 2.479381496289306e-06,
      "loss": 0.1524,
      "step": 25398
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4766239985760616,
      "learning_rate": 2.478727790263281e-06,
      "loss": 0.277,
      "step": 25399
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5394428551744872,
      "learning_rate": 2.4780741582340107e-06,
      "loss": 0.2098,
      "step": 25400
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30010618627832125,
      "learning_rate": 2.477420600207918e-06,
      "loss": 0.2162,
      "step": 25401
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6795715966500449,
      "learning_rate": 2.4767671161914374e-06,
      "loss": 0.3867,
      "step": 25402
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2682229574387187,
      "learning_rate": 2.4761137061910003e-06,
      "loss": 0.174,
      "step": 25403
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8777393532511477,
      "learning_rate": 2.4754603702130265e-06,
      "loss": 0.4419,
      "step": 25404
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.25230292101616086,
      "learning_rate": 2.4748071082639545e-06,
      "loss": 0.0842,
      "step": 25405
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31646708240235777,
      "learning_rate": 2.474153920350204e-06,
      "loss": 0.2742,
      "step": 25406
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3703148833736602,
      "learning_rate": 2.473500806478203e-06,
      "loss": 0.1485,
      "step": 25407
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5587718211973849,
      "learning_rate": 2.4728477666543784e-06,
      "loss": 0.2997,
      "step": 25408
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4488671679286265,
      "learning_rate": 2.472194800885157e-06,
      "loss": 0.1925,
      "step": 25409
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8551140954244445,
      "learning_rate": 2.4715419091769565e-06,
      "loss": 0.4182,
      "step": 25410
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5034171821189961,
      "learning_rate": 2.4708890915362047e-06,
      "loss": 0.2428,
      "step": 25411
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9145157359486112,
      "learning_rate": 2.4702363479693258e-06,
      "loss": 0.3808,
      "step": 25412
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.25919637586601946,
      "learning_rate": 2.469583678482734e-06,
      "loss": 0.2042,
      "step": 25413
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1636272322360134,
      "learning_rate": 2.468931083082862e-06,
      "loss": 0.0868,
      "step": 25414
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2811450561995912,
      "learning_rate": 2.4682785617761196e-06,
      "loss": 0.7858,
      "step": 25415
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7299417362059474,
      "learning_rate": 2.4676261145689317e-06,
      "loss": 0.11,
      "step": 25416
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35192516860161005,
      "learning_rate": 2.46697374146772e-06,
      "loss": 0.2475,
      "step": 25417
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.44430881280392565,
      "learning_rate": 2.466321442478896e-06,
      "loss": 0.2046,
      "step": 25418
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.45182634598564303,
      "learning_rate": 2.4656692176088803e-06,
      "loss": 0.3103,
      "step": 25419
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6661001921842472,
      "learning_rate": 2.4650170668640905e-06,
      "loss": 0.2884,
      "step": 25420
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6373972604022599,
      "learning_rate": 2.464364990250945e-06,
      "loss": 0.2874,
      "step": 25421
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2618753328699323,
      "learning_rate": 2.463712987775854e-06,
      "loss": 0.1663,
      "step": 25422
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40714340228646395,
      "learning_rate": 2.463061059445233e-06,
      "loss": 0.2146,
      "step": 25423
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28574980015867424,
      "learning_rate": 2.462409205265498e-06,
      "loss": 0.2315,
      "step": 25424
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2050138539388067,
      "learning_rate": 2.461757425243061e-06,
      "loss": 0.6336,
      "step": 25425
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2741221950456294,
      "learning_rate": 2.461105719384338e-06,
      "loss": 0.1705,
      "step": 25426
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5627215106899888,
      "learning_rate": 2.460454087695735e-06,
      "loss": 0.1731,
      "step": 25427
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6297401228522451,
      "learning_rate": 2.459802530183665e-06,
      "loss": 0.3731,
      "step": 25428
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42262381016941886,
      "learning_rate": 2.459151046854542e-06,
      "loss": 0.2279,
      "step": 25429
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40816605521643046,
      "learning_rate": 2.4584996377147663e-06,
      "loss": 0.2494,
      "step": 25430
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.32493916264166683,
      "learning_rate": 2.457848302770759e-06,
      "loss": 0.2128,
      "step": 25431
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3091718499502941,
      "learning_rate": 2.457197042028918e-06,
      "loss": 0.1693,
      "step": 25432
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0606983757881983,
      "learning_rate": 2.4565458554956546e-06,
      "loss": 0.6001,
      "step": 25433
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9821132739105424,
      "learning_rate": 2.455894743177378e-06,
      "loss": 0.3636,
      "step": 25434
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40420504367462795,
      "learning_rate": 2.4552437050804887e-06,
      "loss": 0.1892,
      "step": 25435
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4917795031202395,
      "learning_rate": 2.4545927412113944e-06,
      "loss": 0.2855,
      "step": 25436
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3293666961999451,
      "learning_rate": 2.4539418515765e-06,
      "loss": 0.2326,
      "step": 25437
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0534594492915346,
      "learning_rate": 2.4532910361822116e-06,
      "loss": 0.4346,
      "step": 25438
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6688909280852897,
      "learning_rate": 2.4526402950349225e-06,
      "loss": 0.085,
      "step": 25439
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38569306897035455,
      "learning_rate": 2.451989628141047e-06,
      "loss": 0.2005,
      "step": 25440
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39334978886328287,
      "learning_rate": 2.451339035506979e-06,
      "loss": 0.2081,
      "step": 25441
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2938914932692104,
      "learning_rate": 2.4506885171391216e-06,
      "loss": 0.228,
      "step": 25442
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.512272357575976,
      "learning_rate": 2.4500380730438765e-06,
      "loss": 0.2837,
      "step": 25443
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35033267655544614,
      "learning_rate": 2.449387703227639e-06,
      "loss": 0.1485,
      "step": 25444
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4838951373412543,
      "learning_rate": 2.4487374076968106e-06,
      "loss": 0.3107,
      "step": 25445
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.823070229457015,
      "learning_rate": 2.4480871864577872e-06,
      "loss": 0.2507,
      "step": 25446
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5127661325064274,
      "learning_rate": 2.4474370395169665e-06,
      "loss": 0.612,
      "step": 25447
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24013138384934576,
      "learning_rate": 2.44678696688075e-06,
      "loss": 0.18,
      "step": 25448
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3870009965259402,
      "learning_rate": 2.4461369685555236e-06,
      "loss": 0.2789,
      "step": 25449
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1869240711005213,
      "learning_rate": 2.4454870445476887e-06,
      "loss": 0.3206,
      "step": 25450
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4106517920529265,
      "learning_rate": 2.4448371948636372e-06,
      "loss": 0.8205,
      "step": 25451
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4400729019690477,
      "learning_rate": 2.444187419509767e-06,
      "loss": 0.181,
      "step": 25452
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29014075715951315,
      "learning_rate": 2.4435377184924635e-06,
      "loss": 0.1772,
      "step": 25453
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0201848104968723,
      "learning_rate": 2.442888091818123e-06,
      "loss": 0.4031,
      "step": 25454
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24138379084198117,
      "learning_rate": 2.442238539493139e-06,
      "loss": 0.1971,
      "step": 25455
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2318909101593327,
      "learning_rate": 2.4415890615238925e-06,
      "loss": 0.4061,
      "step": 25456
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21687248249438307,
      "learning_rate": 2.4409396579167865e-06,
      "loss": 0.0656,
      "step": 25457
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5787612037620659,
      "learning_rate": 2.4402903286781996e-06,
      "loss": 0.3225,
      "step": 25458
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4348915233560611,
      "learning_rate": 2.439641073814525e-06,
      "loss": 0.2287,
      "step": 25459
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.32399532970678857,
      "learning_rate": 2.438991893332152e-06,
      "loss": 0.2719,
      "step": 25460
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2755260869992227,
      "learning_rate": 2.438342787237461e-06,
      "loss": 0.1312,
      "step": 25461
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0467792843354289,
      "learning_rate": 2.4376937555368417e-06,
      "loss": 0.459,
      "step": 25462
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37065592372173234,
      "learning_rate": 2.4370447982366807e-06,
      "loss": 0.2342,
      "step": 25463
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4574329819761095,
      "learning_rate": 2.4363959153433636e-06,
      "loss": 0.2092,
      "step": 25464
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3148812936327642,
      "learning_rate": 2.4357471068632687e-06,
      "loss": 0.2169,
      "step": 25465
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4420785540897253,
      "learning_rate": 2.4350983728027846e-06,
      "loss": 0.0749,
      "step": 25466
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34028165626374707,
      "learning_rate": 2.4344497131682933e-06,
      "loss": 0.2592,
      "step": 25467
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8099361109490533,
      "learning_rate": 2.433801127966169e-06,
      "loss": 0.4128,
      "step": 25468
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8935015240478377,
      "learning_rate": 2.4331526172028053e-06,
      "loss": 0.4955,
      "step": 25469
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6863488166490783,
      "learning_rate": 2.432504180884573e-06,
      "loss": 0.2691,
      "step": 25470
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5595296019337981,
      "learning_rate": 2.4318558190178553e-06,
      "loss": 0.3095,
      "step": 25471
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.23064569112082192,
      "learning_rate": 2.4312075316090324e-06,
      "loss": 0.2173,
      "step": 25472
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2510444618057481,
      "learning_rate": 2.430559318664475e-06,
      "loss": 0.149,
      "step": 25473
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.472259853588178,
      "learning_rate": 2.4299111801905707e-06,
      "loss": 0.1021,
      "step": 25474
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.220789409411568,
      "learning_rate": 2.4292631161936887e-06,
      "loss": 0.3438,
      "step": 25475
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29697429179353446,
      "learning_rate": 2.4286151266802072e-06,
      "loss": 0.1741,
      "step": 25476
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.5761445903140214,
      "learning_rate": 2.427967211656501e-06,
      "loss": 0.7597,
      "step": 25477
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.303255811684956,
      "learning_rate": 2.4273193711289477e-06,
      "loss": 0.2043,
      "step": 25478
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8000827763420161,
      "learning_rate": 2.4266716051039163e-06,
      "loss": 0.283,
      "step": 25479
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3920634237257763,
      "learning_rate": 2.4260239135877807e-06,
      "loss": 0.243,
      "step": 25480
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4373808054090017,
      "learning_rate": 2.4253762965869177e-06,
      "loss": 0.2031,
      "step": 25481
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2623646900210011,
      "learning_rate": 2.424728754107689e-06,
      "loss": 0.1703,
      "step": 25482
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4298305893441078,
      "learning_rate": 2.4240812861564765e-06,
      "loss": 0.2162,
      "step": 25483
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4357840440111011,
      "learning_rate": 2.423433892739643e-06,
      "loss": 0.2867,
      "step": 25484
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3884244446131416,
      "learning_rate": 2.422786573863559e-06,
      "loss": 0.173,
      "step": 25485
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8600771608739635,
      "learning_rate": 2.4221393295345984e-06,
      "loss": 0.346,
      "step": 25486
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9142289989996496,
      "learning_rate": 2.4214921597591203e-06,
      "loss": 0.2949,
      "step": 25487
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9205360196353524,
      "learning_rate": 2.4208450645434967e-06,
      "loss": 0.4142,
      "step": 25488
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.35079952128602465,
      "learning_rate": 2.420198043894093e-06,
      "loss": 0.179,
      "step": 25489
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36673898102617747,
      "learning_rate": 2.4195510978172776e-06,
      "loss": 0.2748,
      "step": 25490
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3478921941463775,
      "learning_rate": 2.4189042263194095e-06,
      "loss": 0.1746,
      "step": 25491
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33894006381481095,
      "learning_rate": 2.4182574294068573e-06,
      "loss": 0.0872,
      "step": 25492
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4051118445583295,
      "learning_rate": 2.4176107070859856e-06,
      "loss": 0.5785,
      "step": 25493
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3222600322974162,
      "learning_rate": 2.4169640593631494e-06,
      "loss": 0.1678,
      "step": 25494
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2373630613886826,
      "learning_rate": 2.4163174862447203e-06,
      "loss": 0.4711,
      "step": 25495
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4968474467401819,
      "learning_rate": 2.4156709877370532e-06,
      "loss": 0.2406,
      "step": 25496
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1682217681112756,
      "learning_rate": 2.4150245638465108e-06,
      "loss": 0.435,
      "step": 25497
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34735444289666534,
      "learning_rate": 2.414378214579455e-06,
      "loss": 0.1524,
      "step": 25498
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38326169351451556,
      "learning_rate": 2.4137319399422367e-06,
      "loss": 0.2578,
      "step": 25499
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24260882057277494,
      "learning_rate": 2.4130857399412254e-06,
      "loss": 0.0787,
      "step": 25500
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9162320432902498,
      "learning_rate": 2.41243961458277e-06,
      "loss": 0.817,
      "step": 25501
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3376338619191412,
      "learning_rate": 2.4117935638732304e-06,
      "loss": 0.2205,
      "step": 25502
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3983528831639794,
      "learning_rate": 2.411147587818966e-06,
      "loss": 0.2411,
      "step": 25503
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8346374777869527,
      "learning_rate": 2.4105016864263253e-06,
      "loss": 0.2601,
      "step": 25504
    },
    {
      "epoch": 0.78,
      "grad_norm": 2.0604267154342972,
      "learning_rate": 2.4098558597016665e-06,
      "loss": 0.2959,
      "step": 25505
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4502532859415895,
      "learning_rate": 2.409210107651343e-06,
      "loss": 0.742,
      "step": 25506
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2759217898914776,
      "learning_rate": 2.4085644302817113e-06,
      "loss": 0.1887,
      "step": 25507
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.39726443145032964,
      "learning_rate": 2.407918827599114e-06,
      "loss": 0.3048,
      "step": 25508
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1456875677148968,
      "learning_rate": 2.4072732996099167e-06,
      "loss": 0.0983,
      "step": 25509
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.26298794951841065,
      "learning_rate": 2.4066278463204584e-06,
      "loss": 0.1634,
      "step": 25510
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8274442764682426,
      "learning_rate": 2.4059824677370956e-06,
      "loss": 0.1063,
      "step": 25511
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4738032505381807,
      "learning_rate": 2.4053371638661783e-06,
      "loss": 0.3037,
      "step": 25512
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8727101104360155,
      "learning_rate": 2.4046919347140494e-06,
      "loss": 0.2372,
      "step": 25513
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.38865208347264046,
      "learning_rate": 2.4040467802870616e-06,
      "loss": 0.2744,
      "step": 25514
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4195598643877614,
      "learning_rate": 2.40340170059156e-06,
      "loss": 0.3572,
      "step": 25515
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8433274596466492,
      "learning_rate": 2.4027566956338956e-06,
      "loss": 0.3294,
      "step": 25516
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30751602658379606,
      "learning_rate": 2.4021117654204075e-06,
      "loss": 0.1899,
      "step": 25517
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.21343053529465283,
      "learning_rate": 2.4014669099574428e-06,
      "loss": 0.0699,
      "step": 25518
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28075645041261776,
      "learning_rate": 2.400822129251351e-06,
      "loss": 0.2178,
      "step": 25519
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.9697014909703234,
      "learning_rate": 2.4001774233084664e-06,
      "loss": 0.1892,
      "step": 25520
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5583377528188008,
      "learning_rate": 2.399532792135142e-06,
      "loss": 0.3277,
      "step": 25521
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.46797835565831974,
      "learning_rate": 2.3988882357377117e-06,
      "loss": 0.2358,
      "step": 25522
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0721056898100316,
      "learning_rate": 2.3982437541225202e-06,
      "loss": 0.4424,
      "step": 25523
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.992670873815851,
      "learning_rate": 2.39759934729591e-06,
      "loss": 0.4574,
      "step": 25524
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4061667593545212,
      "learning_rate": 2.396955015264215e-06,
      "loss": 0.2958,
      "step": 25525
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2633499850710895,
      "learning_rate": 2.396310758033783e-06,
      "loss": 0.1952,
      "step": 25526
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.458913844209817,
      "learning_rate": 2.3956665756109443e-06,
      "loss": 0.5343,
      "step": 25527
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.1793265132076464,
      "learning_rate": 2.395022468002042e-06,
      "loss": 0.0717,
      "step": 25528
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.6689422528950559,
      "learning_rate": 2.3943784352134113e-06,
      "loss": 0.3673,
      "step": 25529
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34201838434275006,
      "learning_rate": 2.393734477251387e-06,
      "loss": 0.187,
      "step": 25530
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.329080901261093,
      "learning_rate": 2.3930905941223048e-06,
      "loss": 0.2661,
      "step": 25531
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4040632363202905,
      "learning_rate": 2.3924467858325008e-06,
      "loss": 0.1889,
      "step": 25532
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1661920900634253,
      "learning_rate": 2.391803052388312e-06,
      "loss": 0.5188,
      "step": 25533
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8670583064933023,
      "learning_rate": 2.391159393796064e-06,
      "loss": 0.4447,
      "step": 25534
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29940388742171414,
      "learning_rate": 2.390515810062095e-06,
      "loss": 0.1821,
      "step": 25535
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4222856819258698,
      "learning_rate": 2.3898723011927372e-06,
      "loss": 0.5642,
      "step": 25536
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2346628483319212,
      "learning_rate": 2.389228867194314e-06,
      "loss": 0.1782,
      "step": 25537
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4652538761062256,
      "learning_rate": 2.388585508073168e-06,
      "loss": 0.2417,
      "step": 25538
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.32839078635502794,
      "learning_rate": 2.3879422238356188e-06,
      "loss": 0.1532,
      "step": 25539
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5178643036052257,
      "learning_rate": 2.387299014487998e-06,
      "loss": 0.3278,
      "step": 25540
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8633554819472664,
      "learning_rate": 2.386655880036636e-06,
      "loss": 0.2618,
      "step": 25541
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2776754716413052,
      "learning_rate": 2.3860128204878597e-06,
      "loss": 0.6722,
      "step": 25542
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.29338276640287475,
      "learning_rate": 2.3853698358479927e-06,
      "loss": 0.2086,
      "step": 25543
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3918420466956336,
      "learning_rate": 2.384726926123363e-06,
      "loss": 0.2439,
      "step": 25544
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5232006437873661,
      "learning_rate": 2.3840840913202976e-06,
      "loss": 0.2017,
      "step": 25545
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1916036386243543,
      "learning_rate": 2.3834413314451146e-06,
      "loss": 0.2378,
      "step": 25546
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9630749044838908,
      "learning_rate": 2.3827986465041462e-06,
      "loss": 0.4075,
      "step": 25547
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3072804247173831,
      "learning_rate": 2.382156036503709e-06,
      "loss": 0.1194,
      "step": 25548
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31000576849190836,
      "learning_rate": 2.3815135014501266e-06,
      "loss": 0.2524,
      "step": 25549
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24105621717681672,
      "learning_rate": 2.380871041349725e-06,
      "loss": 0.154,
      "step": 25550
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6516332612840887,
      "learning_rate": 2.380228656208815e-06,
      "loss": 0.7898,
      "step": 25551
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9548508093391115,
      "learning_rate": 2.3795863460337287e-06,
      "loss": 0.4205,
      "step": 25552
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42139676412958316,
      "learning_rate": 2.3789441108307766e-06,
      "loss": 0.2335,
      "step": 25553
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.47010371560459707,
      "learning_rate": 2.378301950606279e-06,
      "loss": 0.2106,
      "step": 25554
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4885155743639307,
      "learning_rate": 2.3776598653665583e-06,
      "loss": 0.3057,
      "step": 25555
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.40437111336728604,
      "learning_rate": 2.3770178551179247e-06,
      "loss": 0.258,
      "step": 25556
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7218695765308965,
      "learning_rate": 2.376375919866698e-06,
      "loss": 0.2874,
      "step": 25557
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19643405543727868,
      "learning_rate": 2.3757340596191937e-06,
      "loss": 0.1427,
      "step": 25558
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9284198783101374,
      "learning_rate": 2.3750922743817294e-06,
      "loss": 0.1832,
      "step": 25559
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.4319593050391604,
      "learning_rate": 2.3744505641606132e-06,
      "loss": 0.7556,
      "step": 25560
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2733078619283266,
      "learning_rate": 2.3738089289621615e-06,
      "loss": 0.2045,
      "step": 25561
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3633251535782227,
      "learning_rate": 2.3731673687926892e-06,
      "loss": 0.2752,
      "step": 25562
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.569123104136843,
      "learning_rate": 2.372525883658501e-06,
      "loss": 0.0976,
      "step": 25563
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7586353167077128,
      "learning_rate": 2.3718844735659175e-06,
      "loss": 0.3759,
      "step": 25564
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7399920237723893,
      "learning_rate": 2.371243138521242e-06,
      "loss": 0.2908,
      "step": 25565
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33968409520023857,
      "learning_rate": 2.3706018785307873e-06,
      "loss": 0.1724,
      "step": 25566
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3478084918055086,
      "learning_rate": 2.3699606936008646e-06,
      "loss": 0.1978,
      "step": 25567
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.28338278592013083,
      "learning_rate": 2.3693195837377758e-06,
      "loss": 0.2504,
      "step": 25568
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9211006177945037,
      "learning_rate": 2.3686785489478324e-06,
      "loss": 0.5548,
      "step": 25569
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6122067022639945,
      "learning_rate": 2.36803758923734e-06,
      "loss": 0.5953,
      "step": 25570
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3241281500112138,
      "learning_rate": 2.367396704612608e-06,
      "loss": 0.1523,
      "step": 25571
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7200633664968091,
      "learning_rate": 2.3667558950799363e-06,
      "loss": 0.2481,
      "step": 25572
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.33104167333475654,
      "learning_rate": 2.366115160645631e-06,
      "loss": 0.2795,
      "step": 25573
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0463379680491622,
      "learning_rate": 2.3654745013159973e-06,
      "loss": 0.2153,
      "step": 25574
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8177992630133264,
      "learning_rate": 2.3648339170973367e-06,
      "loss": 0.3317,
      "step": 25575
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.22445932427250945,
      "learning_rate": 2.3641934079959562e-06,
      "loss": 0.1503,
      "step": 25576
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4097972380836533,
      "learning_rate": 2.3635529740181473e-06,
      "loss": 0.223,
      "step": 25577
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.1643689714739585,
      "learning_rate": 2.362912615170222e-06,
      "loss": 0.4557,
      "step": 25578
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3601631866352209,
      "learning_rate": 2.3622723314584726e-06,
      "loss": 0.261,
      "step": 25579
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.37351707114287447,
      "learning_rate": 2.3616321228892027e-06,
      "loss": 0.162,
      "step": 25580
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.45332217574999584,
      "learning_rate": 2.3609919894687107e-06,
      "loss": 0.306,
      "step": 25581
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7516669693723196,
      "learning_rate": 2.3603519312032906e-06,
      "loss": 0.2484,
      "step": 25582
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.2408517689091456,
      "learning_rate": 2.3597119480992424e-06,
      "loss": 0.5533,
      "step": 25583
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3708292323900135,
      "learning_rate": 2.3590720401628618e-06,
      "loss": 0.1812,
      "step": 25584
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.24800129915908461,
      "learning_rate": 2.3584322074004473e-06,
      "loss": 0.2036,
      "step": 25585
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.49636608927597403,
      "learning_rate": 2.357792449818288e-06,
      "loss": 0.2076,
      "step": 25586
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8786926825116712,
      "learning_rate": 2.3571527674226812e-06,
      "loss": 0.572,
      "step": 25587
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.45150851157557415,
      "learning_rate": 2.356513160219924e-06,
      "loss": 0.2504,
      "step": 25588
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.26613555984834314,
      "learning_rate": 2.3558736282163e-06,
      "loss": 0.1588,
      "step": 25589
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3995573934279912,
      "learning_rate": 2.3552341714181106e-06,
      "loss": 0.5214,
      "step": 25590
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30940543351534494,
      "learning_rate": 2.3545947898316414e-06,
      "loss": 0.2379,
      "step": 25591
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6622710016223938,
      "learning_rate": 2.353955483463183e-06,
      "loss": 0.5604,
      "step": 25592
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.8514064009368247,
      "learning_rate": 2.35331625231903e-06,
      "loss": 0.118,
      "step": 25593
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.4677560446598785,
      "learning_rate": 2.3526770964054656e-06,
      "loss": 0.3155,
      "step": 25594
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2863863303849035,
      "learning_rate": 2.352038015728779e-06,
      "loss": 0.1657,
      "step": 25595
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.49506227949119447,
      "learning_rate": 2.35139901029526e-06,
      "loss": 0.3562,
      "step": 25596
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31481715454843695,
      "learning_rate": 2.3507600801111964e-06,
      "loss": 0.194,
      "step": 25597
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5231895987288344,
      "learning_rate": 2.3501212251828697e-06,
      "loss": 0.1662,
      "step": 25598
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.36832605867711804,
      "learning_rate": 2.3494824455165676e-06,
      "loss": 0.258,
      "step": 25599
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7818262612788387,
      "learning_rate": 2.3488437411185736e-06,
      "loss": 0.2229,
      "step": 25600
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.3921006161331986,
      "learning_rate": 2.3482051119951733e-06,
      "loss": 0.5386,
      "step": 25601
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3742087986936071,
      "learning_rate": 2.347566558152652e-06,
      "loss": 0.1848,
      "step": 25602
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3634320025549186,
      "learning_rate": 2.3469280795972847e-06,
      "loss": 0.2946,
      "step": 25603
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.42059708749359165,
      "learning_rate": 2.3462896763353583e-06,
      "loss": 0.2246,
      "step": 25604
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.581986267192324,
      "learning_rate": 2.345651348373155e-06,
      "loss": 0.6868,
      "step": 25605
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7695326559471493,
      "learning_rate": 2.345013095716947e-06,
      "loss": 0.2638,
      "step": 25606
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.539448340870541,
      "learning_rate": 2.3443749183730234e-06,
      "loss": 0.2757,
      "step": 25607
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.19574628548809855,
      "learning_rate": 2.343736816347657e-06,
      "loss": 0.1394,
      "step": 25608
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.469672143266607,
      "learning_rate": 2.343098789647127e-06,
      "loss": 0.3129,
      "step": 25609
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5179620767848793,
      "learning_rate": 2.3424608382777104e-06,
      "loss": 0.2247,
      "step": 25610
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.9301124947649644,
      "learning_rate": 2.3418229622456855e-06,
      "loss": 0.0992,
      "step": 25611
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.355517771669378,
      "learning_rate": 2.3411851615573244e-06,
      "loss": 0.2443,
      "step": 25612
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.0220139671903603,
      "learning_rate": 2.340547436218903e-06,
      "loss": 0.4455,
      "step": 25613
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5130408089526777,
      "learning_rate": 2.3399097862367004e-06,
      "loss": 0.378,
      "step": 25614
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.49674088064803157,
      "learning_rate": 2.3392722116169787e-06,
      "loss": 0.2668,
      "step": 25615
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5966747281784986,
      "learning_rate": 2.3386347123660238e-06,
      "loss": 0.3124,
      "step": 25616
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.31719855729196117,
      "learning_rate": 2.3379972884900994e-06,
      "loss": 0.195,
      "step": 25617
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2703245896464904,
      "learning_rate": 2.3373599399954783e-06,
      "loss": 0.1694,
      "step": 25618
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.234365389988915,
      "learning_rate": 2.3367226668884337e-06,
      "loss": 0.1122,
      "step": 25619
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.30251335977001764,
      "learning_rate": 2.336085469175231e-06,
      "loss": 0.2628,
      "step": 25620
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.3633471939474246,
      "learning_rate": 2.33544834686214e-06,
      "loss": 0.1697,
      "step": 25621
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5161091846388026,
      "learning_rate": 2.3348112999554317e-06,
      "loss": 0.2892,
      "step": 25622
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7072127673036875,
      "learning_rate": 2.334174328461374e-06,
      "loss": 0.2947,
      "step": 25623
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.7633129996750053,
      "learning_rate": 2.3335374323862293e-06,
      "loss": 0.2826,
      "step": 25624
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.885074049453275,
      "learning_rate": 2.3329006117362664e-06,
      "loss": 0.3683,
      "step": 25625
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.225792878872063,
      "learning_rate": 2.332263866517749e-06,
      "loss": 0.1468,
      "step": 25626
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.2456898059453318,
      "learning_rate": 2.331627196736943e-06,
      "loss": 0.2168,
      "step": 25627
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.074426148410506,
      "learning_rate": 2.330990602400115e-06,
      "loss": 0.112,
      "step": 25628
    },
    {
      "epoch": 0.78,
      "grad_norm": 1.6035051814304302,
      "learning_rate": 2.330354083513522e-06,
      "loss": 0.7875,
      "step": 25629
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.34153092594885354,
      "learning_rate": 2.329717640083429e-06,
      "loss": 0.1821,
      "step": 25630
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5701502375084618,
      "learning_rate": 2.3290812721161004e-06,
      "loss": 0.3344,
      "step": 25631
    },
    {
      "epoch": 0.78,
      "grad_norm": 0.5307414623349773,
      "learning_rate": 2.328444979617789e-06,
      "loss": 0.2449,
      "step": 25632
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.636112476571446,
      "learning_rate": 2.3278087625947656e-06,
      "loss": 0.3334,
      "step": 25633
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34512588517681114,
      "learning_rate": 2.327172621053281e-06,
      "loss": 0.1482,
      "step": 25634
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3728345983585023,
      "learning_rate": 2.3265365549995966e-06,
      "loss": 0.2776,
      "step": 25635
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1939539494288118,
      "learning_rate": 2.3259005644399747e-06,
      "loss": 0.0698,
      "step": 25636
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.46401976337271594,
      "learning_rate": 2.325264649380664e-06,
      "loss": 0.0941,
      "step": 25637
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3252242427831008,
      "learning_rate": 2.3246288098279255e-06,
      "loss": 0.2573,
      "step": 25638
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2898335757593667,
      "learning_rate": 2.3239930457880144e-06,
      "loss": 0.1745,
      "step": 25639
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.599922222607197,
      "learning_rate": 2.3233573572671875e-06,
      "loss": 0.7737,
      "step": 25640
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6767977205168438,
      "learning_rate": 2.322721744271692e-06,
      "loss": 0.28,
      "step": 25641
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.039716598015689,
      "learning_rate": 2.322086206807792e-06,
      "loss": 0.5333,
      "step": 25642
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5272011530533772,
      "learning_rate": 2.3214507448817314e-06,
      "loss": 0.0838,
      "step": 25643
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5346116131808804,
      "learning_rate": 2.3208153584997637e-06,
      "loss": 0.3305,
      "step": 25644
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.20066696239109028,
      "learning_rate": 2.320180047668146e-06,
      "loss": 0.1777,
      "step": 25645
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4415607345906088,
      "learning_rate": 2.3195448123931198e-06,
      "loss": 0.2226,
      "step": 25646
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9021122834559371,
      "learning_rate": 2.3189096526809398e-06,
      "loss": 0.2966,
      "step": 25647
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9002538500546731,
      "learning_rate": 2.3182745685378528e-06,
      "loss": 0.4362,
      "step": 25648
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.32521336864281364,
      "learning_rate": 2.317639559970112e-06,
      "loss": 0.2129,
      "step": 25649
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33466621608628044,
      "learning_rate": 2.3170046269839584e-06,
      "loss": 0.2426,
      "step": 25650
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4552928272498864,
      "learning_rate": 2.3163697695856414e-06,
      "loss": 0.8252,
      "step": 25651
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3645508047710265,
      "learning_rate": 2.315734987781406e-06,
      "loss": 0.0599,
      "step": 25652
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3462733275833011,
      "learning_rate": 2.3151002815774983e-06,
      "loss": 0.2503,
      "step": 25653
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2979366377819615,
      "learning_rate": 2.3144656509801668e-06,
      "loss": 0.0765,
      "step": 25654
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5337392040476944,
      "learning_rate": 2.3138310959956466e-06,
      "loss": 0.3186,
      "step": 25655
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3887117756859599,
      "learning_rate": 2.3131966166301855e-06,
      "loss": 0.2055,
      "step": 25656
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5286116229361425,
      "learning_rate": 2.3125622128900294e-06,
      "loss": 0.2795,
      "step": 25657
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3576467634000419,
      "learning_rate": 2.3119278847814084e-06,
      "loss": 0.2356,
      "step": 25658
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9479398879671387,
      "learning_rate": 2.311293632310578e-06,
      "loss": 0.4489,
      "step": 25659
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0786716725544145,
      "learning_rate": 2.310659455483767e-06,
      "loss": 0.5346,
      "step": 25660
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5459718328918813,
      "learning_rate": 2.310025354307218e-06,
      "loss": 0.3262,
      "step": 25661
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2731933797694497,
      "learning_rate": 2.3093913287871737e-06,
      "loss": 0.1688,
      "step": 25662
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43443675020135053,
      "learning_rate": 2.3087573789298647e-06,
      "loss": 0.2006,
      "step": 25663
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24233954901540006,
      "learning_rate": 2.3081235047415317e-06,
      "loss": 0.1345,
      "step": 25664
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7232253607398668,
      "learning_rate": 2.3074897062284106e-06,
      "loss": 0.2669,
      "step": 25665
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38015736924763227,
      "learning_rate": 2.3068559833967396e-06,
      "loss": 0.2339,
      "step": 25666
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.48371818554756624,
      "learning_rate": 2.3062223362527468e-06,
      "loss": 0.2339,
      "step": 25667
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.45651995176520954,
      "learning_rate": 2.305588764802672e-06,
      "loss": 0.2968,
      "step": 25668
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4282055239182977,
      "learning_rate": 2.3049552690527488e-06,
      "loss": 0.2958,
      "step": 25669
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8885196661849467,
      "learning_rate": 2.3043218490092022e-06,
      "loss": 0.3275,
      "step": 25670
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36324717869215684,
      "learning_rate": 2.3036885046782755e-06,
      "loss": 0.1638,
      "step": 25671
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5594170470194115,
      "learning_rate": 2.3030552360661895e-06,
      "loss": 0.3097,
      "step": 25672
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.19036013073014585,
      "learning_rate": 2.30242204317918e-06,
      "loss": 0.1248,
      "step": 25673
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.45461323555750177,
      "learning_rate": 2.301788926023478e-06,
      "loss": 0.321,
      "step": 25674
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6841281292759074,
      "learning_rate": 2.301155884605306e-06,
      "loss": 0.1906,
      "step": 25675
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3197267690344173,
      "learning_rate": 2.3005229189308966e-06,
      "loss": 0.2132,
      "step": 25676
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3781762922289365,
      "learning_rate": 2.299890029006475e-06,
      "loss": 0.5219,
      "step": 25677
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4912939925774553,
      "learning_rate": 2.299257214838271e-06,
      "loss": 0.2984,
      "step": 25678
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5384735782946994,
      "learning_rate": 2.298624476432506e-06,
      "loss": 0.3468,
      "step": 25679
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29222736835139623,
      "learning_rate": 2.2979918137954115e-06,
      "loss": 0.1736,
      "step": 25680
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4986036203777097,
      "learning_rate": 2.2973592269332058e-06,
      "loss": 0.3189,
      "step": 25681
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3601163675315474,
      "learning_rate": 2.2967267158521134e-06,
      "loss": 0.1376,
      "step": 25682
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7933213511249512,
      "learning_rate": 2.2960942805583607e-06,
      "loss": 0.3867,
      "step": 25683
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24880591815239786,
      "learning_rate": 2.295461921058163e-06,
      "loss": 0.1244,
      "step": 25684
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3956042516067955,
      "learning_rate": 2.2948296373577506e-06,
      "loss": 0.2789,
      "step": 25685
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3129862838844873,
      "learning_rate": 2.2941974294633375e-06,
      "loss": 0.2398,
      "step": 25686
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.517581003945434,
      "learning_rate": 2.293565297381145e-06,
      "loss": 0.8003,
      "step": 25687
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2299990965657677,
      "learning_rate": 2.2929332411173965e-06,
      "loss": 0.1983,
      "step": 25688
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2920180533754502,
      "learning_rate": 2.2923012606783035e-06,
      "loss": 0.1806,
      "step": 25689
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.708094991183018,
      "learning_rate": 2.2916693560700863e-06,
      "loss": 0.696,
      "step": 25690
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.37688895352396595,
      "learning_rate": 2.2910375272989637e-06,
      "loss": 0.1715,
      "step": 25691
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33849708709115645,
      "learning_rate": 2.2904057743711526e-06,
      "loss": 0.2683,
      "step": 25692
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2894040586809804,
      "learning_rate": 2.289774097292863e-06,
      "loss": 0.1191,
      "step": 25693
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36887812343578186,
      "learning_rate": 2.2891424960703133e-06,
      "loss": 0.2346,
      "step": 25694
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9675470334606742,
      "learning_rate": 2.2885109707097196e-06,
      "loss": 0.4064,
      "step": 25695
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5859996262486764,
      "learning_rate": 2.2878795212172867e-06,
      "loss": 0.7471,
      "step": 25696
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27399179072857344,
      "learning_rate": 2.2872481475992383e-06,
      "loss": 0.1999,
      "step": 25697
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.903173060259095,
      "learning_rate": 2.286616849861778e-06,
      "loss": 0.3808,
      "step": 25698
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3476209543756201,
      "learning_rate": 2.285985628011119e-06,
      "loss": 0.2046,
      "step": 25699
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7595848782793894,
      "learning_rate": 2.2853544820534745e-06,
      "loss": 0.348,
      "step": 25700
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39734431207665216,
      "learning_rate": 2.284723411995046e-06,
      "loss": 0.1763,
      "step": 25701
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.18648906851871214,
      "learning_rate": 2.284092417842051e-06,
      "loss": 0.0743,
      "step": 25702
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3361267766308524,
      "learning_rate": 2.283461499600692e-06,
      "loss": 0.2452,
      "step": 25703
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3222515259025932,
      "learning_rate": 2.2828306572771785e-06,
      "loss": 0.2493,
      "step": 25704
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2827154586966,
      "learning_rate": 2.2821998908777177e-06,
      "loss": 0.6412,
      "step": 25705
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.49516462026422986,
      "learning_rate": 2.281569200408512e-06,
      "loss": 0.0148,
      "step": 25706
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6163778505119546,
      "learning_rate": 2.2809385858757683e-06,
      "loss": 0.2526,
      "step": 25707
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33189465003793334,
      "learning_rate": 2.2803080472856907e-06,
      "loss": 0.2175,
      "step": 25708
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.45261487895181907,
      "learning_rate": 2.2796775846444853e-06,
      "loss": 0.3294,
      "step": 25709
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4129232131301643,
      "learning_rate": 2.2790471979583463e-06,
      "loss": 0.2073,
      "step": 25710
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9354190188271527,
      "learning_rate": 2.2784168872334868e-06,
      "loss": 0.4194,
      "step": 25711
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23524410339840465,
      "learning_rate": 2.2777866524761006e-06,
      "loss": 0.1478,
      "step": 25712
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4752071301730698,
      "learning_rate": 2.2771564936923897e-06,
      "loss": 0.8898,
      "step": 25713
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.22152858527681032,
      "learning_rate": 2.2765264108885575e-06,
      "loss": 0.0947,
      "step": 25714
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3487381184022912,
      "learning_rate": 2.2758964040707965e-06,
      "loss": 0.2958,
      "step": 25715
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36843975891683933,
      "learning_rate": 2.275266473245309e-06,
      "loss": 0.1402,
      "step": 25716
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4868159107279971,
      "learning_rate": 2.274636618418291e-06,
      "loss": 0.2124,
      "step": 25717
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9984728236537029,
      "learning_rate": 2.2740068395959437e-06,
      "loss": 0.3924,
      "step": 25718
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0292157086470748,
      "learning_rate": 2.273377136784456e-06,
      "loss": 0.2379,
      "step": 25719
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4374073351840214,
      "learning_rate": 2.2727475099900266e-06,
      "loss": 0.2551,
      "step": 25720
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3518296834823811,
      "learning_rate": 2.2721179592188524e-06,
      "loss": 0.1608,
      "step": 25721
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.28380225576435664,
      "learning_rate": 2.271488484477119e-06,
      "loss": 0.2467,
      "step": 25722
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2327884793935823,
      "learning_rate": 2.27085908577103e-06,
      "loss": 0.1376,
      "step": 25723
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9634988824547881,
      "learning_rate": 2.2702297631067706e-06,
      "loss": 0.3241,
      "step": 25724
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7002719227464197,
      "learning_rate": 2.2696005164905333e-06,
      "loss": 0.1911,
      "step": 25725
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36504739814673043,
      "learning_rate": 2.2689713459285136e-06,
      "loss": 0.2714,
      "step": 25726
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4297987556837444,
      "learning_rate": 2.2683422514268906e-06,
      "loss": 0.2566,
      "step": 25727
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.51121391610819,
      "learning_rate": 2.267713232991867e-06,
      "loss": 0.3116,
      "step": 25728
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6876687221977301,
      "learning_rate": 2.267084290629622e-06,
      "loss": 0.034,
      "step": 25729
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2867011368873859,
      "learning_rate": 2.2664554243463466e-06,
      "loss": 0.175,
      "step": 25730
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.2570983863759293,
      "learning_rate": 2.2658266341482306e-06,
      "loss": 0.7435,
      "step": 25731
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3064994323679165,
      "learning_rate": 2.265197920041453e-06,
      "loss": 0.1282,
      "step": 25732
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5049716048700299,
      "learning_rate": 2.2645692820322042e-06,
      "loss": 0.3191,
      "step": 25733
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3499742466211563,
      "learning_rate": 2.2639407201266674e-06,
      "loss": 0.2019,
      "step": 25734
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36962562045461683,
      "learning_rate": 2.263312234331031e-06,
      "loss": 0.2785,
      "step": 25735
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8243187184801636,
      "learning_rate": 2.2626838246514705e-06,
      "loss": 0.2337,
      "step": 25736
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.0111409764431,
      "learning_rate": 2.262055491094174e-06,
      "loss": 0.832,
      "step": 25737
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4614439163011191,
      "learning_rate": 2.2614272336653233e-06,
      "loss": 0.2207,
      "step": 25738
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.44051143870230947,
      "learning_rate": 2.2607990523710923e-06,
      "loss": 0.2578,
      "step": 25739
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.30183867915121293,
      "learning_rate": 2.2601709472176724e-06,
      "loss": 0.2181,
      "step": 25740
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4878347900710767,
      "learning_rate": 2.2595429182112362e-06,
      "loss": 0.7372,
      "step": 25741
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.28215628556291594,
      "learning_rate": 2.258914965357962e-06,
      "loss": 0.1484,
      "step": 25742
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34225267280723587,
      "learning_rate": 2.2582870886640296e-06,
      "loss": 0.1832,
      "step": 25743
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.49441056734623573,
      "learning_rate": 2.25765928813562e-06,
      "loss": 0.3266,
      "step": 25744
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.43045509078191996,
      "learning_rate": 2.257031563778902e-06,
      "loss": 0.2045,
      "step": 25745
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5398567992974607,
      "learning_rate": 2.256403915600056e-06,
      "loss": 0.3203,
      "step": 25746
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7847652850423542,
      "learning_rate": 2.2557763436052583e-06,
      "loss": 0.1553,
      "step": 25747
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.45992574793348706,
      "learning_rate": 2.2551488478006758e-06,
      "loss": 0.26,
      "step": 25748
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4409264423935598,
      "learning_rate": 2.2545214281924933e-06,
      "loss": 0.2335,
      "step": 25749
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7521343132896723,
      "learning_rate": 2.2538940847868753e-06,
      "loss": 0.3419,
      "step": 25750
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31870750640034423,
      "learning_rate": 2.253266817589995e-06,
      "loss": 0.2374,
      "step": 25751
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7264307501434156,
      "learning_rate": 2.2526396266080276e-06,
      "loss": 0.3579,
      "step": 25752
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3037905332714527,
      "learning_rate": 2.252012511847136e-06,
      "loss": 0.1812,
      "step": 25753
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29438800874667787,
      "learning_rate": 2.2513854733135e-06,
      "loss": 0.1691,
      "step": 25754
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4281509065237492,
      "learning_rate": 2.2507585110132803e-06,
      "loss": 0.1459,
      "step": 25755
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42567620008625373,
      "learning_rate": 2.2501316249526484e-06,
      "loss": 0.2347,
      "step": 25756
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3385320147649335,
      "learning_rate": 2.249504815137774e-06,
      "loss": 0.2149,
      "step": 25757
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35823222722691617,
      "learning_rate": 2.248878081574819e-06,
      "loss": 0.23,
      "step": 25758
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6695286199571705,
      "learning_rate": 2.2482514242699516e-06,
      "loss": 0.3618,
      "step": 25759
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7262085242355752,
      "learning_rate": 2.247624843229338e-06,
      "loss": 0.275,
      "step": 25760
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8508839088149096,
      "learning_rate": 2.2469983384591443e-06,
      "loss": 0.3584,
      "step": 25761
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.24159014393660988,
      "learning_rate": 2.246371909965529e-06,
      "loss": 0.1539,
      "step": 25762
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2633070359895007,
      "learning_rate": 2.2457455577546593e-06,
      "loss": 0.2347,
      "step": 25763
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9149081082695965,
      "learning_rate": 2.2451192818326985e-06,
      "loss": 0.0419,
      "step": 25764
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5353626145458847,
      "learning_rate": 2.2444930822057997e-06,
      "loss": 0.6862,
      "step": 25765
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31065007347096063,
      "learning_rate": 2.2438669588801354e-06,
      "loss": 0.1574,
      "step": 25766
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5400477867735306,
      "learning_rate": 2.243240911861858e-06,
      "loss": 0.3358,
      "step": 25767
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7876459228609787,
      "learning_rate": 2.2426149411571287e-06,
      "loss": 0.272,
      "step": 25768
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3330390139590562,
      "learning_rate": 2.2419890467721084e-06,
      "loss": 0.2372,
      "step": 25769
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.826028194159019,
      "learning_rate": 2.2413632287129504e-06,
      "loss": 0.2951,
      "step": 25770
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3266326999265464,
      "learning_rate": 2.2407374869858136e-06,
      "loss": 0.188,
      "step": 25771
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2931087438583026,
      "learning_rate": 2.2401118215968554e-06,
      "loss": 0.1865,
      "step": 25772
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.33989037599277,
      "learning_rate": 2.2394862325522315e-06,
      "loss": 0.1184,
      "step": 25773
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.45644859193418075,
      "learning_rate": 2.2388607198580945e-06,
      "loss": 0.311,
      "step": 25774
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3302226886100823,
      "learning_rate": 2.238235283520599e-06,
      "loss": 0.1504,
      "step": 25775
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3534956446322598,
      "learning_rate": 2.2376099235459005e-06,
      "loss": 0.2619,
      "step": 25776
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7260746374942225,
      "learning_rate": 2.236984639940145e-06,
      "loss": 0.2807,
      "step": 25777
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0189617295949356,
      "learning_rate": 2.2363594327094952e-06,
      "loss": 0.4632,
      "step": 25778
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39058532656260786,
      "learning_rate": 2.2357343018600885e-06,
      "loss": 0.0663,
      "step": 25779
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4447861319370315,
      "learning_rate": 2.2351092473980896e-06,
      "loss": 0.2494,
      "step": 25780
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2110343736366137,
      "learning_rate": 2.2344842693296377e-06,
      "loss": 0.1904,
      "step": 25781
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42781044396059936,
      "learning_rate": 2.2338593676608843e-06,
      "loss": 0.095,
      "step": 25782
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.241157138902784,
      "learning_rate": 2.2332345423979816e-06,
      "loss": 0.5181,
      "step": 25783
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6017071729264578,
      "learning_rate": 2.2326097935470705e-06,
      "loss": 0.1765,
      "step": 25784
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.41398745050504837,
      "learning_rate": 2.2319851211142983e-06,
      "loss": 0.29,
      "step": 25785
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9451225230756595,
      "learning_rate": 2.2313605251058137e-06,
      "loss": 0.2414,
      "step": 25786
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5062329245517333,
      "learning_rate": 2.2307360055277627e-06,
      "loss": 0.3475,
      "step": 25787
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9912193854565315,
      "learning_rate": 2.230111562386286e-06,
      "loss": 0.4241,
      "step": 25788
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4202334884786882,
      "learning_rate": 2.2294871956875263e-06,
      "loss": 0.2437,
      "step": 25789
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23647356565110172,
      "learning_rate": 2.2288629054376322e-06,
      "loss": 0.1235,
      "step": 25790
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6989406092054935,
      "learning_rate": 2.2282386916427366e-06,
      "loss": 0.5165,
      "step": 25791
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.47709796041567554,
      "learning_rate": 2.227614554308991e-06,
      "loss": 0.2528,
      "step": 25792
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34821376483045213,
      "learning_rate": 2.226990493442529e-06,
      "loss": 0.214,
      "step": 25793
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.38260104743832113,
      "learning_rate": 2.2263665090494912e-06,
      "loss": 0.2309,
      "step": 25794
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8427357397238021,
      "learning_rate": 2.2257426011360208e-06,
      "loss": 0.2777,
      "step": 25795
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.288438555533832,
      "learning_rate": 2.22511876970825e-06,
      "loss": 0.7883,
      "step": 25796
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33547354861428813,
      "learning_rate": 2.2244950147723187e-06,
      "loss": 0.0611,
      "step": 25797
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3495972291895027,
      "learning_rate": 2.2238713363343635e-06,
      "loss": 0.2386,
      "step": 25798
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2960852597177196,
      "learning_rate": 2.2232477344005245e-06,
      "loss": 0.2219,
      "step": 25799
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2633550293468415,
      "learning_rate": 2.22262420897693e-06,
      "loss": 0.1469,
      "step": 25800
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7651235536717362,
      "learning_rate": 2.222000760069718e-06,
      "loss": 0.2428,
      "step": 25801
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6793310002027968,
      "learning_rate": 2.221377387685024e-06,
      "loss": 0.3044,
      "step": 25802
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34396481143799346,
      "learning_rate": 2.220754091828973e-06,
      "loss": 0.1906,
      "step": 25803
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8670336438540547,
      "learning_rate": 2.2201308725077085e-06,
      "loss": 0.4886,
      "step": 25804
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3369150896457242,
      "learning_rate": 2.219507729727355e-06,
      "loss": 0.2462,
      "step": 25805
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9243123981063757,
      "learning_rate": 2.218884663494043e-06,
      "loss": 0.4028,
      "step": 25806
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3516098518447829,
      "learning_rate": 2.2182616738139074e-06,
      "loss": 0.1642,
      "step": 25807
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3423830244387424,
      "learning_rate": 2.217638760693069e-06,
      "loss": 0.1451,
      "step": 25808
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.46506593999328494,
      "learning_rate": 2.2170159241376664e-06,
      "loss": 0.2589,
      "step": 25809
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4184158893256131,
      "learning_rate": 2.2163931641538185e-06,
      "loss": 0.2378,
      "step": 25810
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4107272240666682,
      "learning_rate": 2.2157704807476566e-06,
      "loss": 0.2832,
      "step": 25811
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31001422960481106,
      "learning_rate": 2.215147873925306e-06,
      "loss": 0.1898,
      "step": 25812
    },
    {
      "epoch": 0.79,
      "grad_norm": 2.077107864890846,
      "learning_rate": 2.214525343692895e-06,
      "loss": 0.6868,
      "step": 25813
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0324578853126953,
      "learning_rate": 2.213902890056543e-06,
      "loss": 0.4597,
      "step": 25814
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5321659243387007,
      "learning_rate": 2.2132805130223756e-06,
      "loss": 0.6208,
      "step": 25815
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3804224533132936,
      "learning_rate": 2.212658212596519e-06,
      "loss": 0.1523,
      "step": 25816
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.26726151476773696,
      "learning_rate": 2.2120359887850884e-06,
      "loss": 0.2453,
      "step": 25817
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.1683322210428721,
      "learning_rate": 2.2114138415942167e-06,
      "loss": 0.0693,
      "step": 25818
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7110990534789781,
      "learning_rate": 2.210791771030014e-06,
      "loss": 0.3705,
      "step": 25819
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4582751789121116,
      "learning_rate": 2.2101697770986065e-06,
      "loss": 0.1983,
      "step": 25820
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33700015046431925,
      "learning_rate": 2.209547859806114e-06,
      "loss": 0.2078,
      "step": 25821
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7180590069407178,
      "learning_rate": 2.208926019158649e-06,
      "loss": 0.8243,
      "step": 25822
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3036764324726795,
      "learning_rate": 2.2083042551623347e-06,
      "loss": 0.2441,
      "step": 25823
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.8093634419534599,
      "learning_rate": 2.2076825678232873e-06,
      "loss": 0.6192,
      "step": 25824
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27896574776382416,
      "learning_rate": 2.2070609571476253e-06,
      "loss": 0.1753,
      "step": 25825
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3890491510395724,
      "learning_rate": 2.206439423141459e-06,
      "loss": 0.498,
      "step": 25826
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4414613947305857,
      "learning_rate": 2.2058179658109057e-06,
      "loss": 0.1627,
      "step": 25827
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.336690704562188,
      "learning_rate": 2.205196585162079e-06,
      "loss": 0.2732,
      "step": 25828
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29303485137922786,
      "learning_rate": 2.2045752812010946e-06,
      "loss": 0.1223,
      "step": 25829
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5202495433193468,
      "learning_rate": 2.2039540539340665e-06,
      "loss": 0.2986,
      "step": 25830
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4643199056507726,
      "learning_rate": 2.2033329033671004e-06,
      "loss": 0.2392,
      "step": 25831
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.1578858247139632,
      "learning_rate": 2.20271182950631e-06,
      "loss": 0.5176,
      "step": 25832
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4463823700845413,
      "learning_rate": 2.2020908323578107e-06,
      "loss": 0.1955,
      "step": 25833
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2303236110771308,
      "learning_rate": 2.2014699119277004e-06,
      "loss": 0.0644,
      "step": 25834
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.28893040591395036,
      "learning_rate": 2.2008490682221016e-06,
      "loss": 0.2406,
      "step": 25835
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8025400985074203,
      "learning_rate": 2.2002283012471127e-06,
      "loss": 0.2604,
      "step": 25836
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4869175960796568,
      "learning_rate": 2.199607611008845e-06,
      "loss": 0.2717,
      "step": 25837
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2931702382984036,
      "learning_rate": 2.198986997513406e-06,
      "loss": 0.0739,
      "step": 25838
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36206715584477345,
      "learning_rate": 2.1983664607668976e-06,
      "loss": 0.2601,
      "step": 25839
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3075760713618741,
      "learning_rate": 2.197746000775427e-06,
      "loss": 0.2467,
      "step": 25840
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4906288944395034,
      "learning_rate": 2.1971256175450985e-06,
      "loss": 0.7982,
      "step": 25841
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.409042095055067,
      "learning_rate": 2.1965053110820177e-06,
      "loss": 0.1572,
      "step": 25842
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6922660942117818,
      "learning_rate": 2.1958850813922817e-06,
      "loss": 0.3371,
      "step": 25843
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3728634624425068,
      "learning_rate": 2.1952649284820005e-06,
      "loss": 0.2231,
      "step": 25844
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8735893757468454,
      "learning_rate": 2.1946448523572707e-06,
      "loss": 0.3829,
      "step": 25845
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3262002099208707,
      "learning_rate": 2.1940248530241882e-06,
      "loss": 0.2331,
      "step": 25846
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.12863485492850033,
      "learning_rate": 2.193404930488863e-06,
      "loss": 0.0665,
      "step": 25847
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.37627215450355983,
      "learning_rate": 2.192785084757386e-06,
      "loss": 0.2763,
      "step": 25848
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.977125140037033,
      "learning_rate": 2.192165315835858e-06,
      "loss": 0.4225,
      "step": 25849
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6118759798426392,
      "learning_rate": 2.1915456237303767e-06,
      "loss": 0.7505,
      "step": 25850
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5014872490120827,
      "learning_rate": 2.190926008447042e-06,
      "loss": 0.2055,
      "step": 25851
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4255959117936405,
      "learning_rate": 2.1903064699919443e-06,
      "loss": 0.2935,
      "step": 25852
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.30325371605224977,
      "learning_rate": 2.1896870083711807e-06,
      "loss": 0.1904,
      "step": 25853
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7154162635179605,
      "learning_rate": 2.189067623590847e-06,
      "loss": 0.3607,
      "step": 25854
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8676829868128569,
      "learning_rate": 2.188448315657036e-06,
      "loss": 0.1874,
      "step": 25855
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3801019631287636,
      "learning_rate": 2.187829084575842e-06,
      "loss": 0.2189,
      "step": 25856
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3654329613729585,
      "learning_rate": 2.1872099303533544e-06,
      "loss": 0.1446,
      "step": 25857
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5186894825409972,
      "learning_rate": 2.186590852995667e-06,
      "loss": 0.3466,
      "step": 25858
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2303934439507336,
      "learning_rate": 2.1859718525088714e-06,
      "loss": 0.1901,
      "step": 25859
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9356609559299615,
      "learning_rate": 2.1853529288990503e-06,
      "loss": 0.2429,
      "step": 25860
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6438256959287706,
      "learning_rate": 2.184734082172305e-06,
      "loss": 0.3005,
      "step": 25861
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.32789567152416943,
      "learning_rate": 2.1841153123347136e-06,
      "loss": 0.2077,
      "step": 25862
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9925123602955157,
      "learning_rate": 2.183496619392369e-06,
      "loss": 0.4317,
      "step": 25863
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31475603356770954,
      "learning_rate": 2.1828780033513576e-06,
      "loss": 0.2267,
      "step": 25864
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.47359931623541857,
      "learning_rate": 2.182259464217763e-06,
      "loss": 0.2252,
      "step": 25865
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3289299530089412,
      "learning_rate": 2.181641001997672e-06,
      "loss": 0.1669,
      "step": 25866
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5751011697644165,
      "learning_rate": 2.181022616697168e-06,
      "loss": 0.3787,
      "step": 25867
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2154687741806474,
      "learning_rate": 2.1804043083223404e-06,
      "loss": 0.0968,
      "step": 25868
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8534766445432154,
      "learning_rate": 2.1797860768792657e-06,
      "loss": 0.4204,
      "step": 25869
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.25344103543382746,
      "learning_rate": 2.1791679223740283e-06,
      "loss": 0.1818,
      "step": 25870
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4035927643323686,
      "learning_rate": 2.178549844812713e-06,
      "loss": 0.2754,
      "step": 25871
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8388860135950659,
      "learning_rate": 2.177931844201392e-06,
      "loss": 0.2631,
      "step": 25872
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7042247138148796,
      "learning_rate": 2.177313920546158e-06,
      "loss": 0.1003,
      "step": 25873
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5318955440321702,
      "learning_rate": 2.1766960738530794e-06,
      "loss": 0.5528,
      "step": 25874
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.27172507256577166,
      "learning_rate": 2.17607830412824e-06,
      "loss": 0.1706,
      "step": 25875
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6522456918473205,
      "learning_rate": 2.175460611377719e-06,
      "loss": 0.7549,
      "step": 25876
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2427776812543338,
      "learning_rate": 2.174842995607589e-06,
      "loss": 0.1827,
      "step": 25877
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.42450237044902783,
      "learning_rate": 2.1742254568239275e-06,
      "loss": 0.2456,
      "step": 25878
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5710391349768323,
      "learning_rate": 2.1736079950328116e-06,
      "loss": 0.1999,
      "step": 25879
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6113749851933734,
      "learning_rate": 2.1729906102403153e-06,
      "loss": 0.3364,
      "step": 25880
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5306318124429796,
      "learning_rate": 2.172373302452513e-06,
      "loss": 0.1958,
      "step": 25881
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36678957536864804,
      "learning_rate": 2.1717560716754805e-06,
      "loss": 0.2802,
      "step": 25882
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4031577414505837,
      "learning_rate": 2.171138917915284e-06,
      "loss": 0.1376,
      "step": 25883
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8354606268733387,
      "learning_rate": 2.1705218411780003e-06,
      "loss": 0.3688,
      "step": 25884
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.357877242880515,
      "learning_rate": 2.1699048414697012e-06,
      "loss": 0.2326,
      "step": 25885
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.39213796713474663,
      "learning_rate": 2.1692879187964487e-06,
      "loss": 0.1852,
      "step": 25886
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3915745874827156,
      "learning_rate": 2.1686710731643236e-06,
      "loss": 0.2974,
      "step": 25887
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6520807518120734,
      "learning_rate": 2.1680543045793878e-06,
      "loss": 0.2036,
      "step": 25888
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3370224454777969,
      "learning_rate": 2.1674376130477093e-06,
      "loss": 0.239,
      "step": 25889
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4663007388954252,
      "learning_rate": 2.1668209985753606e-06,
      "loss": 0.0987,
      "step": 25890
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5867076181382087,
      "learning_rate": 2.1662044611684017e-06,
      "loss": 0.5971,
      "step": 25891
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.0336594725270323,
      "learning_rate": 2.1655880008329e-06,
      "loss": 0.4304,
      "step": 25892
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.622754264151884,
      "learning_rate": 2.1649716175749226e-06,
      "loss": 0.3233,
      "step": 25893
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2482019438990151,
      "learning_rate": 2.1643553114005345e-06,
      "loss": 0.2209,
      "step": 25894
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8902105571238916,
      "learning_rate": 2.1637390823157945e-06,
      "loss": 0.4578,
      "step": 25895
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7429752019921384,
      "learning_rate": 2.1631229303267676e-06,
      "loss": 0.257,
      "step": 25896
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.32850863483615067,
      "learning_rate": 2.162506855439519e-06,
      "loss": 0.1875,
      "step": 25897
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2767344197554469,
      "learning_rate": 2.161890857660102e-06,
      "loss": 0.1685,
      "step": 25898
    },
    {
      "epoch": 0.79,
      "grad_norm": 3.400099091244107,
      "learning_rate": 2.161274936994586e-06,
      "loss": 0.5085,
      "step": 25899
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33863925724344024,
      "learning_rate": 2.1606590934490245e-06,
      "loss": 0.2212,
      "step": 25900
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3962326747339908,
      "learning_rate": 2.1600433270294787e-06,
      "loss": 0.3081,
      "step": 25901
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3831918867743874,
      "learning_rate": 2.1594276377420087e-06,
      "loss": 0.2171,
      "step": 25902
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4373145634461293,
      "learning_rate": 2.158812025592667e-06,
      "loss": 0.2134,
      "step": 25903
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8142193880111207,
      "learning_rate": 2.158196490587512e-06,
      "loss": 0.3845,
      "step": 25904
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3247097638923248,
      "learning_rate": 2.1575810327326006e-06,
      "loss": 0.2392,
      "step": 25905
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8833668198945624,
      "learning_rate": 2.156965652033988e-06,
      "loss": 0.3574,
      "step": 25906
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3551391739384547,
      "learning_rate": 2.15635034849773e-06,
      "loss": 0.1596,
      "step": 25907
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.25491720979024324,
      "learning_rate": 2.155735122129875e-06,
      "loss": 0.1702,
      "step": 25908
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7638426204981414,
      "learning_rate": 2.1551199729364804e-06,
      "loss": 0.1861,
      "step": 25909
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.349355868976592,
      "learning_rate": 2.1545049009235953e-06,
      "loss": 0.7963,
      "step": 25910
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3397206564123433,
      "learning_rate": 2.153889906097276e-06,
      "loss": 0.1859,
      "step": 25911
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35349811327893427,
      "learning_rate": 2.153274988463564e-06,
      "loss": 0.2513,
      "step": 25912
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.431385606251864,
      "learning_rate": 2.1526601480285205e-06,
      "loss": 0.256,
      "step": 25913
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8895899168119847,
      "learning_rate": 2.1520453847981848e-06,
      "loss": 0.2448,
      "step": 25914
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4941729152660239,
      "learning_rate": 2.1514306987786105e-06,
      "loss": 0.2254,
      "step": 25915
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.28531769468281315,
      "learning_rate": 2.1508160899758447e-06,
      "loss": 0.1706,
      "step": 25916
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5666997306976636,
      "learning_rate": 2.1502015583959313e-06,
      "loss": 0.2959,
      "step": 25917
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.34032499352763446,
      "learning_rate": 2.149587104044917e-06,
      "loss": 0.2254,
      "step": 25918
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5950255149356367,
      "learning_rate": 2.1489727269288483e-06,
      "loss": 0.8815,
      "step": 25919
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3121203673913329,
      "learning_rate": 2.1483584270537715e-06,
      "loss": 0.1527,
      "step": 25920
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.36423916407567997,
      "learning_rate": 2.1477442044257256e-06,
      "loss": 0.2488,
      "step": 25921
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8127279821902613,
      "learning_rate": 2.147130059050756e-06,
      "loss": 0.303,
      "step": 25922
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4632049361410917,
      "learning_rate": 2.1465159909349074e-06,
      "loss": 0.3025,
      "step": 25923
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.4621832853787419,
      "learning_rate": 2.1459020000842147e-06,
      "loss": 0.1996,
      "step": 25924
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.35728391872675047,
      "learning_rate": 2.1452880865047256e-06,
      "loss": 0.2115,
      "step": 25925
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2125195191659931,
      "learning_rate": 2.144674250202475e-06,
      "loss": 0.1232,
      "step": 25926
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5451537764679371,
      "learning_rate": 2.144060491183504e-06,
      "loss": 0.0407,
      "step": 25927
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8467777113005291,
      "learning_rate": 2.143446809453854e-06,
      "loss": 0.4718,
      "step": 25928
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.26054919118915515,
      "learning_rate": 2.1428332050195564e-06,
      "loss": 0.1936,
      "step": 25929
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3781190801769853,
      "learning_rate": 2.142219677886651e-06,
      "loss": 0.2634,
      "step": 25930
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6654930174081833,
      "learning_rate": 2.1416062280611737e-06,
      "loss": 0.2668,
      "step": 25931
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.4741774661352725,
      "learning_rate": 2.1409928555491598e-06,
      "loss": 0.8958,
      "step": 25932
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2191867470990029,
      "learning_rate": 2.140379560356648e-06,
      "loss": 0.0684,
      "step": 25933
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5057476680199025,
      "learning_rate": 2.139766342489664e-06,
      "loss": 0.318,
      "step": 25934
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2675828251517331,
      "learning_rate": 2.139153201954247e-06,
      "loss": 0.1292,
      "step": 25935
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3584783439926971,
      "learning_rate": 2.1385401387564263e-06,
      "loss": 0.2777,
      "step": 25936
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.6752502519314221,
      "learning_rate": 2.137927152902237e-06,
      "loss": 0.2672,
      "step": 25937
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8474914542175512,
      "learning_rate": 2.1373142443977047e-06,
      "loss": 0.446,
      "step": 25938
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2695856020929057,
      "learning_rate": 2.136701413248862e-06,
      "loss": 0.1692,
      "step": 25939
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8969237793394552,
      "learning_rate": 2.136088659461741e-06,
      "loss": 0.3399,
      "step": 25940
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3723606023326397,
      "learning_rate": 2.135475983042362e-06,
      "loss": 0.2679,
      "step": 25941
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5329277340948019,
      "learning_rate": 2.1348633839967637e-06,
      "loss": 0.0153,
      "step": 25942
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.5860074961406455,
      "learning_rate": 2.1342508623309645e-06,
      "loss": 0.2595,
      "step": 25943
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.2054770082826263,
      "learning_rate": 2.1336384180509927e-06,
      "loss": 0.1413,
      "step": 25944
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.3866503388448395,
      "learning_rate": 2.133026051162879e-06,
      "loss": 0.6415,
      "step": 25945
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9787275761145436,
      "learning_rate": 2.1324137616726394e-06,
      "loss": 0.2701,
      "step": 25946
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.33108104419090434,
      "learning_rate": 2.131801549586302e-06,
      "loss": 0.2834,
      "step": 25947
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.31661010372683024,
      "learning_rate": 2.13118941490989e-06,
      "loss": 0.2013,
      "step": 25948
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.9690408440881013,
      "learning_rate": 2.1305773576494294e-06,
      "loss": 0.5137,
      "step": 25949
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.365601120678064,
      "learning_rate": 2.129965377810932e-06,
      "loss": 0.2094,
      "step": 25950
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.5052010098623496,
      "learning_rate": 2.1293534754004295e-06,
      "loss": 0.5771,
      "step": 25951
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.29761897482673616,
      "learning_rate": 2.1287416504239355e-06,
      "loss": 0.1432,
      "step": 25952
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.23908058982012875,
      "learning_rate": 2.1281299028874714e-06,
      "loss": 0.1676,
      "step": 25953
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3573181119891504,
      "learning_rate": 2.127518232797058e-06,
      "loss": 0.2476,
      "step": 25954
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.8735497334345295,
      "learning_rate": 2.1269066401587046e-06,
      "loss": 0.2747,
      "step": 25955
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.7460491503670412,
      "learning_rate": 2.1262951249784403e-06,
      "loss": 0.3349,
      "step": 25956
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.325310008900197,
      "learning_rate": 2.125683687262272e-06,
      "loss": 0.1918,
      "step": 25957
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.6201682444308145,
      "learning_rate": 2.1250723270162197e-06,
      "loss": 0.8617,
      "step": 25958
    },
    {
      "epoch": 0.79,
      "grad_norm": 0.3214021636985872,
      "learning_rate": 2.124461044246299e-06,
      "loss": 0.2344,
      "step": 25959
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4177494218126494,
      "learning_rate": 2.1238498389585195e-06,
      "loss": 0.5437,
      "step": 25960
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31640113923187124,
      "learning_rate": 2.123238711158896e-06,
      "loss": 0.0614,
      "step": 25961
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39504194964776823,
      "learning_rate": 2.122627660853442e-06,
      "loss": 0.2792,
      "step": 25962
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3526575939014981,
      "learning_rate": 2.122016688048173e-06,
      "loss": 0.1386,
      "step": 25963
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.523920101059705,
      "learning_rate": 2.1214057927490926e-06,
      "loss": 0.3238,
      "step": 25964
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43218428746856347,
      "learning_rate": 2.120794974962215e-06,
      "loss": 0.214,
      "step": 25965
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27346626930952445,
      "learning_rate": 2.120184234693551e-06,
      "loss": 0.1675,
      "step": 25966
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3799904754707144,
      "learning_rate": 2.1195735719491016e-06,
      "loss": 0.7629,
      "step": 25967
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4968781900979184,
      "learning_rate": 2.118962986734886e-06,
      "loss": 0.2123,
      "step": 25968
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5814038235146766,
      "learning_rate": 2.1183524790569045e-06,
      "loss": 0.5938,
      "step": 25969
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4696424939782056,
      "learning_rate": 2.1177420489211633e-06,
      "loss": 0.0682,
      "step": 25970
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.29900379666847565,
      "learning_rate": 2.117131696333672e-06,
      "loss": 0.2374,
      "step": 25971
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2856069691780695,
      "learning_rate": 2.1165214213004304e-06,
      "loss": 0.19,
      "step": 25972
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0070170303960237,
      "learning_rate": 2.115911223827445e-06,
      "loss": 0.461,
      "step": 25973
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3316497589940707,
      "learning_rate": 2.1153011039207195e-06,
      "loss": 0.1661,
      "step": 25974
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42272837553819886,
      "learning_rate": 2.1146910615862594e-06,
      "loss": 0.2625,
      "step": 25975
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.480552695381025,
      "learning_rate": 2.1140810968300595e-06,
      "loss": 0.2406,
      "step": 25976
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4900829292085993,
      "learning_rate": 2.1134712096581245e-06,
      "loss": 0.3206,
      "step": 25977
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40792916573856836,
      "learning_rate": 2.112861400076458e-06,
      "loss": 0.2025,
      "step": 25978
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3330323892814909,
      "learning_rate": 2.112251668091051e-06,
      "loss": 0.1548,
      "step": 25979
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5564392962886987,
      "learning_rate": 2.111642013707913e-06,
      "loss": 0.3208,
      "step": 25980
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7496490471028749,
      "learning_rate": 2.11103243693303e-06,
      "loss": 0.2608,
      "step": 25981
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3025177003198844,
      "learning_rate": 2.110422937772413e-06,
      "loss": 0.156,
      "step": 25982
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2744844674364044,
      "learning_rate": 2.1098135162320464e-06,
      "loss": 0.2169,
      "step": 25983
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34112170676707154,
      "learning_rate": 2.1092041723179312e-06,
      "loss": 0.2494,
      "step": 25984
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9651114821424285,
      "learning_rate": 2.1085949060360654e-06,
      "loss": 0.4373,
      "step": 25985
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3192189372394034,
      "learning_rate": 2.107985717392437e-06,
      "loss": 0.4314,
      "step": 25986
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9467439437991794,
      "learning_rate": 2.1073766063930402e-06,
      "loss": 0.2166,
      "step": 25987
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4370085514461282,
      "learning_rate": 2.1067675730438707e-06,
      "loss": 0.2548,
      "step": 25988
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.32621691536595654,
      "learning_rate": 2.1061586173509206e-06,
      "loss": 0.2027,
      "step": 25989
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.486154427460015,
      "learning_rate": 2.1055497393201786e-06,
      "loss": 0.3092,
      "step": 25990
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33778263636053846,
      "learning_rate": 2.1049409389576347e-06,
      "loss": 0.1205,
      "step": 25991
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1267097641021155,
      "learning_rate": 2.1043322162692827e-06,
      "loss": 0.4546,
      "step": 25992
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41912937734774197,
      "learning_rate": 2.1037235712611016e-06,
      "loss": 0.242,
      "step": 25993
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.25859612916787306,
      "learning_rate": 2.1031150039390937e-06,
      "loss": 0.1489,
      "step": 25994
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4208922250524665,
      "learning_rate": 2.102506514309235e-06,
      "loss": 0.2727,
      "step": 25995
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.0658392530077196,
      "learning_rate": 2.1018981023775152e-06,
      "loss": 0.2354,
      "step": 25996
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6434739835636083,
      "learning_rate": 2.1012897681499247e-06,
      "loss": 0.3628,
      "step": 25997
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3142271493236958,
      "learning_rate": 2.100681511632441e-06,
      "loss": 0.2005,
      "step": 25998
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7985888712858632,
      "learning_rate": 2.100073332831052e-06,
      "loss": 0.3458,
      "step": 25999
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.42447114679026016,
      "learning_rate": 2.0994652317517415e-06,
      "loss": 0.0163,
      "step": 26000
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.327391711287272,
      "learning_rate": 2.0988572084004934e-06,
      "loss": 0.2809,
      "step": 26001
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3017797313786625,
      "learning_rate": 2.0982492627832864e-06,
      "loss": 0.1653,
      "step": 26002
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5563814899755584,
      "learning_rate": 2.0976413949061025e-06,
      "loss": 0.3836,
      "step": 26003
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1785029942573129,
      "learning_rate": 2.097033604774925e-06,
      "loss": 0.0699,
      "step": 26004
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8139137894168245,
      "learning_rate": 2.096425892395726e-06,
      "loss": 0.2634,
      "step": 26005
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31687393716438805,
      "learning_rate": 2.0958182577744955e-06,
      "loss": 0.2816,
      "step": 26006
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3407516977701276,
      "learning_rate": 2.095210700917203e-06,
      "loss": 0.1904,
      "step": 26007
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.098379542980773,
      "learning_rate": 2.0946032218298297e-06,
      "loss": 0.4556,
      "step": 26008
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.378855692230727,
      "learning_rate": 2.093995820518352e-06,
      "loss": 0.1255,
      "step": 26009
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3280703907390694,
      "learning_rate": 2.0933884969887406e-06,
      "loss": 0.7794,
      "step": 26010
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3228701562057046,
      "learning_rate": 2.0927812512469804e-06,
      "loss": 0.1611,
      "step": 26011
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5615967430634007,
      "learning_rate": 2.0921740832990367e-06,
      "loss": 0.3309,
      "step": 26012
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19052766417581024,
      "learning_rate": 2.0915669931508862e-06,
      "loss": 0.1556,
      "step": 26013
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9691152675305257,
      "learning_rate": 2.0909599808085023e-06,
      "loss": 0.3678,
      "step": 26014
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5913622347796627,
      "learning_rate": 2.0903530462778586e-06,
      "loss": 0.2279,
      "step": 26015
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4409699932664384,
      "learning_rate": 2.0897461895649217e-06,
      "loss": 0.2946,
      "step": 26016
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5174104399698334,
      "learning_rate": 2.0891394106756655e-06,
      "loss": 0.2009,
      "step": 26017
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31847735088247753,
      "learning_rate": 2.0885327096160614e-06,
      "loss": 0.2136,
      "step": 26018
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.3627665011992947,
      "learning_rate": 2.087926086392069e-06,
      "loss": 0.8425,
      "step": 26019
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27887260157598637,
      "learning_rate": 2.0873195410096693e-06,
      "loss": 0.1131,
      "step": 26020
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40379831425376295,
      "learning_rate": 2.086713073474821e-06,
      "loss": 0.3008,
      "step": 26021
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3044128471070783,
      "learning_rate": 2.086106683793494e-06,
      "loss": 0.1138,
      "step": 26022
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9699328274362152,
      "learning_rate": 2.0855003719716547e-06,
      "loss": 0.4123,
      "step": 26023
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36843682392553123,
      "learning_rate": 2.0848941380152655e-06,
      "loss": 0.2148,
      "step": 26024
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3500129258330632,
      "learning_rate": 2.084287981930292e-06,
      "loss": 0.2477,
      "step": 26025
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3887465693268748,
      "learning_rate": 2.0836819037226976e-06,
      "loss": 0.1861,
      "step": 26026
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8220104305568054,
      "learning_rate": 2.083075903398448e-06,
      "loss": 0.7156,
      "step": 26027
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9109068094560094,
      "learning_rate": 2.0824699809634995e-06,
      "loss": 0.2501,
      "step": 26028
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.636763460112023,
      "learning_rate": 2.081864136423818e-06,
      "loss": 0.3214,
      "step": 26029
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27233166264584113,
      "learning_rate": 2.081258369785364e-06,
      "loss": 0.2167,
      "step": 26030
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.26550108282643536,
      "learning_rate": 2.0806526810540907e-06,
      "loss": 0.1633,
      "step": 26031
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7860284038238411,
      "learning_rate": 2.0800470702359676e-06,
      "loss": 0.379,
      "step": 26032
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38836984620384635,
      "learning_rate": 2.079441537336945e-06,
      "loss": 0.1334,
      "step": 26033
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6969385712943096,
      "learning_rate": 2.0788360823629825e-06,
      "loss": 0.2843,
      "step": 26034
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6739281335990273,
      "learning_rate": 2.07823070532004e-06,
      "loss": 0.0973,
      "step": 26035
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4729920027096655,
      "learning_rate": 2.0776254062140646e-06,
      "loss": 0.2881,
      "step": 26036
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4277461187735527,
      "learning_rate": 2.077020185051023e-06,
      "loss": 0.2284,
      "step": 26037
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6029275583944573,
      "learning_rate": 2.0764150418368613e-06,
      "loss": 0.3,
      "step": 26038
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33656107171421706,
      "learning_rate": 2.0758099765775353e-06,
      "loss": 0.189,
      "step": 26039
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8719352277616634,
      "learning_rate": 2.0752049892790006e-06,
      "loss": 0.4419,
      "step": 26040
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4050069268519523,
      "learning_rate": 2.0746000799472043e-06,
      "loss": 0.1594,
      "step": 26041
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.328381773092957,
      "learning_rate": 2.0739952485880997e-06,
      "loss": 0.2575,
      "step": 26042
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3949920731944656,
      "learning_rate": 2.073390495207639e-06,
      "loss": 0.1637,
      "step": 26043
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3125314727184288,
      "learning_rate": 2.0727858198117724e-06,
      "loss": 0.1474,
      "step": 26044
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1582374173985097,
      "learning_rate": 2.072181222406445e-06,
      "loss": 0.367,
      "step": 26045
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9771777822164519,
      "learning_rate": 2.0715767029976076e-06,
      "loss": 0.5163,
      "step": 26046
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.901751203421462,
      "learning_rate": 2.0709722615912097e-06,
      "loss": 0.455,
      "step": 26047
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.26142110608413,
      "learning_rate": 2.0703678981931908e-06,
      "loss": 0.195,
      "step": 26048
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.47511815733783397,
      "learning_rate": 2.0697636128095066e-06,
      "loss": 0.3229,
      "step": 26049
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8635792863985043,
      "learning_rate": 2.069159405446095e-06,
      "loss": 0.2696,
      "step": 26050
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.49764546393592296,
      "learning_rate": 2.0685552761089034e-06,
      "loss": 0.2324,
      "step": 26051
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.20469719449046397,
      "learning_rate": 2.0679512248038737e-06,
      "loss": 0.1208,
      "step": 26052
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5932871380070132,
      "learning_rate": 2.067347251536953e-06,
      "loss": 0.3096,
      "step": 26053
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4190873665885241,
      "learning_rate": 2.066743356314077e-06,
      "loss": 0.2057,
      "step": 26054
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.475913295500671,
      "learning_rate": 2.066139539141191e-06,
      "loss": 0.3087,
      "step": 26055
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34763456606693477,
      "learning_rate": 2.0655358000242377e-06,
      "loss": 0.1544,
      "step": 26056
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3908506259595602,
      "learning_rate": 2.064932138969148e-06,
      "loss": 0.2349,
      "step": 26057
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2183017464845003,
      "learning_rate": 2.0643285559818728e-06,
      "loss": 0.4271,
      "step": 26058
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9317743417555122,
      "learning_rate": 2.0637250510683405e-06,
      "loss": 0.2778,
      "step": 26059
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2746789418873827,
      "learning_rate": 2.0631216242344933e-06,
      "loss": 0.2439,
      "step": 26060
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27153648620106274,
      "learning_rate": 2.0625182754862707e-06,
      "loss": 0.1668,
      "step": 26061
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5135673208109216,
      "learning_rate": 2.0619150048295975e-06,
      "loss": 0.2912,
      "step": 26062
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.126596127743737,
      "learning_rate": 2.0613118122704235e-06,
      "loss": 0.0521,
      "step": 26063
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.258684317113664,
      "learning_rate": 2.060708697814674e-06,
      "loss": 0.4519,
      "step": 26064
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.794262087673188,
      "learning_rate": 2.0601056614682835e-06,
      "loss": 0.2132,
      "step": 26065
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.584521298849825,
      "learning_rate": 2.0595027032371884e-06,
      "loss": 0.3264,
      "step": 26066
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2698316313810818,
      "learning_rate": 2.0588998231273173e-06,
      "loss": 0.2187,
      "step": 26067
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4489506715986888,
      "learning_rate": 2.058297021144602e-06,
      "loss": 0.5851,
      "step": 26068
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.264746460371512,
      "learning_rate": 2.057694297294973e-06,
      "loss": 0.0785,
      "step": 26069
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27350348547858255,
      "learning_rate": 2.0570916515843643e-06,
      "loss": 0.0691,
      "step": 26070
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38785282317328423,
      "learning_rate": 2.056489084018699e-06,
      "loss": 0.2633,
      "step": 26071
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3045601307496061,
      "learning_rate": 2.0558865946039073e-06,
      "loss": 0.2145,
      "step": 26072
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.201078195807244,
      "learning_rate": 2.05528418334592e-06,
      "loss": 0.4713,
      "step": 26073
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8366784661462942,
      "learning_rate": 2.054681850250656e-06,
      "loss": 0.184,
      "step": 26074
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3502846905391934,
      "learning_rate": 2.0540795953240512e-06,
      "loss": 0.2587,
      "step": 26075
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7739137116275018,
      "learning_rate": 2.053477418572024e-06,
      "loss": 0.3159,
      "step": 26076
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6014600129286145,
      "learning_rate": 2.0528753200005003e-06,
      "loss": 0.8321,
      "step": 26077
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35100654939658005,
      "learning_rate": 2.052273299615406e-06,
      "loss": 0.2211,
      "step": 26078
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35165168161156535,
      "learning_rate": 2.051671357422661e-06,
      "loss": 0.2107,
      "step": 26079
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.25943279599373475,
      "learning_rate": 2.0510694934281874e-06,
      "loss": 0.1573,
      "step": 26080
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.1243139672366755,
      "learning_rate": 2.0504677076379077e-06,
      "loss": 0.7597,
      "step": 26081
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5774047720713438,
      "learning_rate": 2.049866000057743e-06,
      "loss": 0.3053,
      "step": 26082
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6284264459743822,
      "learning_rate": 2.049264370693612e-06,
      "loss": 0.2104,
      "step": 26083
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3874113047424482,
      "learning_rate": 2.048662819551438e-06,
      "loss": 0.2645,
      "step": 26084
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4740568035526732,
      "learning_rate": 2.0480613466371315e-06,
      "loss": 0.2453,
      "step": 26085
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.7288845233888366,
      "learning_rate": 2.0474599519566153e-06,
      "loss": 0.7148,
      "step": 26086
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4844534478802926,
      "learning_rate": 2.046858635515806e-06,
      "loss": 0.0877,
      "step": 26087
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38922782614659934,
      "learning_rate": 2.0462573973206136e-06,
      "loss": 0.2363,
      "step": 26088
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27047574136005975,
      "learning_rate": 2.045656237376964e-06,
      "loss": 0.1343,
      "step": 26089
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33742189461192224,
      "learning_rate": 2.0450551556907628e-06,
      "loss": 0.2836,
      "step": 26090
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5575207816806884,
      "learning_rate": 2.044454152267926e-06,
      "loss": 0.2971,
      "step": 26091
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8221147777105051,
      "learning_rate": 2.0438532271143697e-06,
      "loss": 0.4148,
      "step": 26092
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39928789996059366,
      "learning_rate": 2.0432523802360016e-06,
      "loss": 0.1638,
      "step": 26093
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5568999877909632,
      "learning_rate": 2.042651611638733e-06,
      "loss": 0.3996,
      "step": 26094
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.41389499748176506,
      "learning_rate": 2.0420509213284777e-06,
      "loss": 0.2066,
      "step": 26095
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5115870254351907,
      "learning_rate": 2.041450309311146e-06,
      "loss": 0.3253,
      "step": 26096
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.583972936712184,
      "learning_rate": 2.0408497755926417e-06,
      "loss": 0.0705,
      "step": 26097
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23311232716331973,
      "learning_rate": 2.0402493201788774e-06,
      "loss": 0.1502,
      "step": 26098
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5451086097322957,
      "learning_rate": 2.0396489430757617e-06,
      "loss": 0.2533,
      "step": 26099
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2682210991473,
      "learning_rate": 2.039048644289192e-06,
      "loss": 0.2751,
      "step": 26100
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4634910668613018,
      "learning_rate": 2.038448423825089e-06,
      "loss": 0.3365,
      "step": 26101
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.274561394586207,
      "learning_rate": 2.0378482816893462e-06,
      "loss": 0.1671,
      "step": 26102
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4863855072411378,
      "learning_rate": 2.037248217887872e-06,
      "loss": 0.3566,
      "step": 26103
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2315221360002953,
      "learning_rate": 2.0366482324265723e-06,
      "loss": 0.1882,
      "step": 26104
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5492279456937668,
      "learning_rate": 2.036048325311345e-06,
      "loss": 0.5273,
      "step": 26105
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.15104292685556825,
      "learning_rate": 2.0354484965480957e-06,
      "loss": 0.07,
      "step": 26106
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.27235788205068573,
      "learning_rate": 2.0348487461427233e-06,
      "loss": 0.2096,
      "step": 26107
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3388839164481462,
      "learning_rate": 2.034249074101131e-06,
      "loss": 0.2399,
      "step": 26108
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.860010710937281,
      "learning_rate": 2.0336494804292194e-06,
      "loss": 0.4,
      "step": 26109
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9048880001942665,
      "learning_rate": 2.0330499651328827e-06,
      "loss": 0.3237,
      "step": 26110
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.26983889221327195,
      "learning_rate": 2.032450528218023e-06,
      "loss": 0.1744,
      "step": 26111
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5447031917944518,
      "learning_rate": 2.031851169690536e-06,
      "loss": 0.787,
      "step": 26112
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39885577295610214,
      "learning_rate": 2.0312518895563217e-06,
      "loss": 0.1971,
      "step": 26113
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5089265104961147,
      "learning_rate": 2.0306526878212674e-06,
      "loss": 0.3186,
      "step": 26114
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5199969590541492,
      "learning_rate": 2.030053564491281e-06,
      "loss": 0.097,
      "step": 26115
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.540340109130379,
      "learning_rate": 2.029454519572249e-06,
      "loss": 0.3012,
      "step": 26116
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6491922223923763,
      "learning_rate": 2.0288555530700604e-06,
      "loss": 0.2462,
      "step": 26117
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34186368765014546,
      "learning_rate": 2.02825666499062e-06,
      "loss": 0.192,
      "step": 26118
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.29737508473318636,
      "learning_rate": 2.0276578553398108e-06,
      "loss": 0.1996,
      "step": 26119
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9591271714548593,
      "learning_rate": 2.027059124123527e-06,
      "loss": 0.4377,
      "step": 26120
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3331263286735058,
      "learning_rate": 2.0264604713476586e-06,
      "loss": 0.2249,
      "step": 26121
    },
    {
      "epoch": 0.8,
      "grad_norm": 3.386597914422804,
      "learning_rate": 2.0258618970180987e-06,
      "loss": 0.7062,
      "step": 26122
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3242936056873555,
      "learning_rate": 2.0252634011407303e-06,
      "loss": 0.0813,
      "step": 26123
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5788983698378337,
      "learning_rate": 2.024664983721444e-06,
      "loss": 0.181,
      "step": 26124
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3479170338957209,
      "learning_rate": 2.0240666447661315e-06,
      "loss": 0.249,
      "step": 26125
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3172097408073096,
      "learning_rate": 2.02346838428067e-06,
      "loss": 0.2381,
      "step": 26126
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.280305046831163,
      "learning_rate": 2.0228702022709568e-06,
      "loss": 0.1858,
      "step": 26127
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9917924410390163,
      "learning_rate": 2.0222720987428677e-06,
      "loss": 0.4445,
      "step": 26128
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7205575773356426,
      "learning_rate": 2.0216740737022923e-06,
      "loss": 0.3192,
      "step": 26129
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3545465710325889,
      "learning_rate": 2.0210761271551137e-06,
      "loss": 0.2148,
      "step": 26130
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3542640974771734,
      "learning_rate": 2.0204782591072123e-06,
      "loss": 0.2615,
      "step": 26131
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4401978391388461,
      "learning_rate": 2.01988046956447e-06,
      "loss": 0.0123,
      "step": 26132
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6475473090957087,
      "learning_rate": 2.0192827585327712e-06,
      "loss": 0.32,
      "step": 26133
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3700559152791961,
      "learning_rate": 2.018685126017993e-06,
      "loss": 0.1952,
      "step": 26134
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7297103750789757,
      "learning_rate": 2.018087572026021e-06,
      "loss": 0.4311,
      "step": 26135
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3323837995586366,
      "learning_rate": 2.0174900965627253e-06,
      "loss": 0.0726,
      "step": 26136
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.29563505183497096,
      "learning_rate": 2.0168926996339898e-06,
      "loss": 0.2275,
      "step": 26137
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3983323918324952,
      "learning_rate": 2.016295381245691e-06,
      "loss": 0.2516,
      "step": 26138
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4318416058681841,
      "learning_rate": 2.015698141403707e-06,
      "loss": 0.2105,
      "step": 26139
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5259474949525466,
      "learning_rate": 2.01510098011391e-06,
      "loss": 0.2626,
      "step": 26140
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7174382289291341,
      "learning_rate": 2.014503897382176e-06,
      "loss": 0.2605,
      "step": 26141
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6453702082727468,
      "learning_rate": 2.0139068932143834e-06,
      "loss": 0.3402,
      "step": 26142
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38226524730932526,
      "learning_rate": 2.013309967616397e-06,
      "loss": 0.1592,
      "step": 26143
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30360850025555336,
      "learning_rate": 2.012713120594101e-06,
      "loss": 0.2663,
      "step": 26144
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.1730204858802755,
      "learning_rate": 2.012116352153358e-06,
      "loss": 0.0624,
      "step": 26145
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.511725490075589,
      "learning_rate": 2.0115196623000433e-06,
      "loss": 0.8164,
      "step": 26146
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44061616503461043,
      "learning_rate": 2.010923051040029e-06,
      "loss": 0.1384,
      "step": 26147
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39843955243468737,
      "learning_rate": 2.010326518379181e-06,
      "loss": 0.2661,
      "step": 26148
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2518523311481126,
      "learning_rate": 2.0097300643233686e-06,
      "loss": 0.172,
      "step": 26149
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7423378433394706,
      "learning_rate": 2.0091336888784606e-06,
      "loss": 0.2627,
      "step": 26150
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0065877379670587,
      "learning_rate": 2.008537392050328e-06,
      "loss": 0.4954,
      "step": 26151
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.28536174502130435,
      "learning_rate": 2.0079411738448286e-06,
      "loss": 0.1711,
      "step": 26152
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.933511597107289,
      "learning_rate": 2.007345034267839e-06,
      "loss": 0.3316,
      "step": 26153
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4117839660375836,
      "learning_rate": 2.0067489733252165e-06,
      "loss": 0.225,
      "step": 26154
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5034246486685718,
      "learning_rate": 2.0061529910228273e-06,
      "loss": 0.3337,
      "step": 26155
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3958606343900475,
      "learning_rate": 2.0055570873665376e-06,
      "loss": 0.1516,
      "step": 26156
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3747802087489422,
      "learning_rate": 2.0049612623622063e-06,
      "loss": 0.2733,
      "step": 26157
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.39048476792906356,
      "learning_rate": 2.004365516015696e-06,
      "loss": 0.1156,
      "step": 26158
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8969348035633552,
      "learning_rate": 2.0037698483328684e-06,
      "loss": 0.3998,
      "step": 26159
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4075079098901749,
      "learning_rate": 2.0031742593195847e-06,
      "loss": 0.2381,
      "step": 26160
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4291578468803289,
      "learning_rate": 2.002578748981706e-06,
      "loss": 0.2828,
      "step": 26161
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.35856171711091117,
      "learning_rate": 2.001983317325087e-06,
      "loss": 0.2163,
      "step": 26162
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1810735777083747,
      "learning_rate": 2.001387964355588e-06,
      "loss": 0.0784,
      "step": 26163
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4358677752664537,
      "learning_rate": 2.0007926900790662e-06,
      "loss": 0.8274,
      "step": 26164
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31933160160115787,
      "learning_rate": 2.0001974945013803e-06,
      "loss": 0.1656,
      "step": 26165
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5391818618679458,
      "learning_rate": 1.99960237762838e-06,
      "loss": 0.2893,
      "step": 26166
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.301638625413925,
      "learning_rate": 1.999007339465926e-06,
      "loss": 0.1647,
      "step": 26167
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4895394150527822,
      "learning_rate": 1.9984123800198716e-06,
      "loss": 0.2896,
      "step": 26168
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3509726615653611,
      "learning_rate": 1.997817499296064e-06,
      "loss": 0.1193,
      "step": 26169
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4971008952378358,
      "learning_rate": 1.9972226973003674e-06,
      "loss": 0.2509,
      "step": 26170
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5101180509796273,
      "learning_rate": 1.996627974038623e-06,
      "loss": 0.1954,
      "step": 26171
    },
    {
      "epoch": 0.8,
      "grad_norm": 2.606968108526325,
      "learning_rate": 1.996033329516687e-06,
      "loss": 0.6439,
      "step": 26172
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3060236611753422,
      "learning_rate": 1.9954387637404115e-06,
      "loss": 0.2419,
      "step": 26173
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8719980134148614,
      "learning_rate": 1.9948442767156395e-06,
      "loss": 0.4806,
      "step": 26174
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2998604935739193,
      "learning_rate": 1.994249868448225e-06,
      "loss": 0.1849,
      "step": 26175
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36234635113878305,
      "learning_rate": 1.993655538944014e-06,
      "loss": 0.1554,
      "step": 26176
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40712714909388276,
      "learning_rate": 1.993061288208856e-06,
      "loss": 0.2542,
      "step": 26177
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5434135448007947,
      "learning_rate": 1.992467116248594e-06,
      "loss": 0.228,
      "step": 26178
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37806273005095287,
      "learning_rate": 1.991873023069075e-06,
      "loss": 0.2493,
      "step": 26179
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36431768302899087,
      "learning_rate": 1.991279008676147e-06,
      "loss": 0.2046,
      "step": 26180
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2060945594916601,
      "learning_rate": 1.9906850730756454e-06,
      "loss": 0.5256,
      "step": 26181
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9947753135192399,
      "learning_rate": 1.9900912162734253e-06,
      "loss": 0.4536,
      "step": 26182
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.9798734846127506,
      "learning_rate": 1.9894974382753217e-06,
      "loss": 0.4132,
      "step": 26183
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31888115298740993,
      "learning_rate": 1.988903739087177e-06,
      "loss": 0.1901,
      "step": 26184
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3334805172286106,
      "learning_rate": 1.9883101187148333e-06,
      "loss": 0.2763,
      "step": 26185
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.94590970747771,
      "learning_rate": 1.9877165771641316e-06,
      "loss": 0.2861,
      "step": 26186
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2827986818668502,
      "learning_rate": 1.987123114440912e-06,
      "loss": 0.1801,
      "step": 26187
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3265202352635833,
      "learning_rate": 1.98652973055101e-06,
      "loss": 0.16,
      "step": 26188
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4761277991373507,
      "learning_rate": 1.9859364255002655e-06,
      "loss": 0.2011,
      "step": 26189
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3836021324903127,
      "learning_rate": 1.985343199294515e-06,
      "loss": 0.6175,
      "step": 26190
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30648532717711363,
      "learning_rate": 1.9847500519395978e-06,
      "loss": 0.2361,
      "step": 26191
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8164059140637621,
      "learning_rate": 1.9841569834413443e-06,
      "loss": 0.3718,
      "step": 26192
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30070651471542953,
      "learning_rate": 1.983563993805593e-06,
      "loss": 0.1853,
      "step": 26193
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7675499084002233,
      "learning_rate": 1.982971083038179e-06,
      "loss": 0.417,
      "step": 26194
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4999944288703461,
      "learning_rate": 1.9823782511449276e-06,
      "loss": 0.1489,
      "step": 26195
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.28104338794242034,
      "learning_rate": 1.981785498131682e-06,
      "loss": 0.2282,
      "step": 26196
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.30622514402831325,
      "learning_rate": 1.9811928240042665e-06,
      "loss": 0.1429,
      "step": 26197
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5088025553215921,
      "learning_rate": 1.9806002287685157e-06,
      "loss": 0.2766,
      "step": 26198
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1429392854483262,
      "learning_rate": 1.9800077124302607e-06,
      "loss": 0.1667,
      "step": 26199
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.328896504699651,
      "learning_rate": 1.9794152749953253e-06,
      "loss": 0.7014,
      "step": 26200
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.713219902770386,
      "learning_rate": 1.9788229164695427e-06,
      "loss": 0.2288,
      "step": 26201
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33180935743698126,
      "learning_rate": 1.978230636858739e-06,
      "loss": 0.1871,
      "step": 26202
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.316204051626634,
      "learning_rate": 1.977638436168745e-06,
      "loss": 0.2931,
      "step": 26203
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9101942051797978,
      "learning_rate": 1.977046314405381e-06,
      "loss": 0.2521,
      "step": 26204
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4445194777228751,
      "learning_rate": 1.9764542715744737e-06,
      "loss": 0.2504,
      "step": 26205
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.18240547773983826,
      "learning_rate": 1.975862307681854e-06,
      "loss": 0.0691,
      "step": 26206
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37383278088002636,
      "learning_rate": 1.9752704227333353e-06,
      "loss": 0.2544,
      "step": 26207
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.43961142471097486,
      "learning_rate": 1.9746786167347522e-06,
      "loss": 0.2538,
      "step": 26208
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4855310300200932,
      "learning_rate": 1.9740868896919176e-06,
      "loss": 0.3217,
      "step": 26209
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7197303180018838,
      "learning_rate": 1.973495241610658e-06,
      "loss": 0.2936,
      "step": 26210
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3977608414421115,
      "learning_rate": 1.9729036724967967e-06,
      "loss": 0.2657,
      "step": 26211
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5143009877817071,
      "learning_rate": 1.972312182356143e-06,
      "loss": 0.2307,
      "step": 26212
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3142836424469464,
      "learning_rate": 1.9717207711945306e-06,
      "loss": 0.5343,
      "step": 26213
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3045750811598807,
      "learning_rate": 1.971129439017768e-06,
      "loss": 0.2034,
      "step": 26214
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2218522383529314,
      "learning_rate": 1.9705381858316753e-06,
      "loss": 0.1193,
      "step": 26215
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3728418340889743,
      "learning_rate": 1.9699470116420725e-06,
      "loss": 0.2122,
      "step": 26216
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.1893330452423014,
      "learning_rate": 1.9693559164547715e-06,
      "loss": 0.1904,
      "step": 26217
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8096109102232779,
      "learning_rate": 1.9687649002755884e-06,
      "loss": 0.4192,
      "step": 26218
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5538229420804321,
      "learning_rate": 1.9681739631103393e-06,
      "loss": 0.2416,
      "step": 26219
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3372081692073704,
      "learning_rate": 1.967583104964841e-06,
      "loss": 0.252,
      "step": 26220
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.29406730823378896,
      "learning_rate": 1.9669923258448963e-06,
      "loss": 0.2086,
      "step": 26221
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3821995984048712,
      "learning_rate": 1.96640162575633e-06,
      "loss": 0.5552,
      "step": 26222
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.19127559889313517,
      "learning_rate": 1.965811004704946e-06,
      "loss": 0.071,
      "step": 26223
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7046671242181738,
      "learning_rate": 1.9652204626965564e-06,
      "loss": 0.3523,
      "step": 26224
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34614488173983754,
      "learning_rate": 1.9646299997369744e-06,
      "loss": 0.2005,
      "step": 26225
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.560198810839204,
      "learning_rate": 1.964039615832004e-06,
      "loss": 0.8046,
      "step": 26226
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.319730075031035,
      "learning_rate": 1.9634493109874562e-06,
      "loss": 0.2432,
      "step": 26227
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7393084904940053,
      "learning_rate": 1.962859085209138e-06,
      "loss": 0.2754,
      "step": 26228
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.40678164668590994,
      "learning_rate": 1.9622689385028593e-06,
      "loss": 0.2641,
      "step": 26229
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4346182443408144,
      "learning_rate": 1.9616788708744218e-06,
      "loss": 0.2374,
      "step": 26230
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.180486619162067,
      "learning_rate": 1.9610888823296316e-06,
      "loss": 0.3237,
      "step": 26231
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2972136801179184,
      "learning_rate": 1.9604989728742974e-06,
      "loss": 0.2191,
      "step": 26232
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33970720449729197,
      "learning_rate": 1.9599091425142138e-06,
      "loss": 0.173,
      "step": 26233
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.23440973694915937,
      "learning_rate": 1.9593193912551943e-06,
      "loss": 0.156,
      "step": 26234
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9284578740474221,
      "learning_rate": 1.958729719103035e-06,
      "loss": 0.476,
      "step": 26235
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8113061883408149,
      "learning_rate": 1.958140126063538e-06,
      "loss": 0.2744,
      "step": 26236
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8208997692648272,
      "learning_rate": 1.957550612142507e-06,
      "loss": 0.5023,
      "step": 26237
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.28737449158530354,
      "learning_rate": 1.956961177345734e-06,
      "loss": 0.1941,
      "step": 26238
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.36371968502138985,
      "learning_rate": 1.9563718216790297e-06,
      "loss": 0.2815,
      "step": 26239
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6697613292194494,
      "learning_rate": 1.9557825451481826e-06,
      "loss": 0.0387,
      "step": 26240
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2092993397340805,
      "learning_rate": 1.9551933477589945e-06,
      "loss": 0.0693,
      "step": 26241
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3066916372413782,
      "learning_rate": 1.954604229517264e-06,
      "loss": 0.1806,
      "step": 26242
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.38027915429478865,
      "learning_rate": 1.954015190428781e-06,
      "loss": 0.202,
      "step": 26243
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0149786311704825,
      "learning_rate": 1.9534262304993444e-06,
      "loss": 0.534,
      "step": 26244
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3446112240815518,
      "learning_rate": 1.9528373497347473e-06,
      "loss": 0.2274,
      "step": 26245
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.3856960951190853,
      "learning_rate": 1.9522485481407884e-06,
      "loss": 0.7931,
      "step": 26246
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2937625300688848,
      "learning_rate": 1.951659825723253e-06,
      "loss": 0.1784,
      "step": 26247
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.5315652906900037,
      "learning_rate": 1.951071182487936e-06,
      "loss": 0.7319,
      "step": 26248
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.4215921906925513,
      "learning_rate": 1.950482618440632e-06,
      "loss": 0.1002,
      "step": 26249
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.33882960678760454,
      "learning_rate": 1.9498941335871247e-06,
      "loss": 0.2618,
      "step": 26250
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2742906742203112,
      "learning_rate": 1.9493057279332116e-06,
      "loss": 0.0757,
      "step": 26251
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5764009715785712,
      "learning_rate": 1.9487174014846756e-06,
      "loss": 0.327,
      "step": 26252
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3133016907660768,
      "learning_rate": 1.948129154247307e-06,
      "loss": 0.1784,
      "step": 26253
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.8097186998156523,
      "learning_rate": 1.9475409862268924e-06,
      "loss": 0.2958,
      "step": 26254
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.49145452716993127,
      "learning_rate": 1.946952897429223e-06,
      "loss": 0.3513,
      "step": 26255
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.31555330566039996,
      "learning_rate": 1.946364887860077e-06,
      "loss": 0.1765,
      "step": 26256
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.37607095938305074,
      "learning_rate": 1.945776957525245e-06,
      "loss": 0.2762,
      "step": 26257
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2307159640830447,
      "learning_rate": 1.9451891064305105e-06,
      "loss": 0.1543,
      "step": 26258
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.8808456753083769,
      "learning_rate": 1.9446013345816515e-06,
      "loss": 0.5208,
      "step": 26259
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5891068697849124,
      "learning_rate": 1.94401364198446e-06,
      "loss": 0.179,
      "step": 26260
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.34035879258698004,
      "learning_rate": 1.9434260286447105e-06,
      "loss": 0.2463,
      "step": 26261
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3213375460802513,
      "learning_rate": 1.9428384945681867e-06,
      "loss": 0.2271,
      "step": 26262
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4857033285292483,
      "learning_rate": 1.942251039760672e-06,
      "loss": 0.2213,
      "step": 26263
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.0777368353938586,
      "learning_rate": 1.9416636642279373e-06,
      "loss": 0.522,
      "step": 26264
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.621729513781381,
      "learning_rate": 1.9410763679757716e-06,
      "loss": 0.3377,
      "step": 26265
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.2650511137069347,
      "learning_rate": 1.9404891510099467e-06,
      "loss": 0.151,
      "step": 26266
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.28059044925992,
      "learning_rate": 1.939902013336241e-06,
      "loss": 0.0592,
      "step": 26267
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3464652788801185,
      "learning_rate": 1.939314954960434e-06,
      "loss": 0.2649,
      "step": 26268
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.6231706941737104,
      "learning_rate": 1.938727975888297e-06,
      "loss": 0.2156,
      "step": 26269
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3779150781663414,
      "learning_rate": 1.9381410761256057e-06,
      "loss": 0.2581,
      "step": 26270
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7459324966270174,
      "learning_rate": 1.937554255678136e-06,
      "loss": 0.267,
      "step": 26271
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.46881100126808906,
      "learning_rate": 1.936967514551662e-06,
      "loss": 0.2732,
      "step": 26272
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.333636661216085,
      "learning_rate": 1.9363808527519533e-06,
      "loss": 0.2402,
      "step": 26273
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.44115855423996064,
      "learning_rate": 1.935794270284782e-06,
      "loss": 0.2516,
      "step": 26274
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.4621696060224997,
      "learning_rate": 1.935207767155923e-06,
      "loss": 0.2113,
      "step": 26275
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.5067139503019601,
      "learning_rate": 1.934621343371139e-06,
      "loss": 0.257,
      "step": 26276
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.961787972437784,
      "learning_rate": 1.9340349989362083e-06,
      "loss": 0.2402,
      "step": 26277
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.660442655053171,
      "learning_rate": 1.933448733856893e-06,
      "loss": 0.3766,
      "step": 26278
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3895315367992063,
      "learning_rate": 1.932862548138964e-06,
      "loss": 0.1608,
      "step": 26279
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3949646397523681,
      "learning_rate": 1.93227644178819e-06,
      "loss": 0.2774,
      "step": 26280
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.3298365233264424,
      "learning_rate": 1.931690414810332e-06,
      "loss": 0.16,
      "step": 26281
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.9722577185265605,
      "learning_rate": 1.9311044672111588e-06,
      "loss": 0.504,
      "step": 26282
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.7918627751468659,
      "learning_rate": 1.9305185989964346e-06,
      "loss": 0.3391,
      "step": 26283
    },
    {
      "epoch": 0.8,
      "grad_norm": 0.22695302703174203,
      "learning_rate": 1.929932810171927e-06,
      "loss": 0.1493,
      "step": 26284
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.6004905609755467,
      "learning_rate": 1.9293471007433905e-06,
      "loss": 0.4791,
      "step": 26285
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45916129615870566,
      "learning_rate": 1.928761470716598e-06,
      "loss": 0.2417,
      "step": 26286
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9120419461961407,
      "learning_rate": 1.928175920097303e-06,
      "loss": 0.3613,
      "step": 26287
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38851372769243936,
      "learning_rate": 1.927590448891269e-06,
      "loss": 0.1816,
      "step": 26288
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6217273471983614,
      "learning_rate": 1.9270050571042587e-06,
      "loss": 0.3139,
      "step": 26289
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2299769354561632,
      "learning_rate": 1.926419744742024e-06,
      "loss": 0.097,
      "step": 26290
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4739822041990798,
      "learning_rate": 1.925834511810335e-06,
      "loss": 0.7346,
      "step": 26291
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2681032478939371,
      "learning_rate": 1.925249358314939e-06,
      "loss": 0.1954,
      "step": 26292
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36936202573589594,
      "learning_rate": 1.924664284261597e-06,
      "loss": 0.2743,
      "step": 26293
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23271797700829772,
      "learning_rate": 1.924079289656067e-06,
      "loss": 0.0637,
      "step": 26294
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7044968377392979,
      "learning_rate": 1.9234943745041002e-06,
      "loss": 0.2824,
      "step": 26295
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8758916125014272,
      "learning_rate": 1.922909538811453e-06,
      "loss": 0.4597,
      "step": 26296
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26705175842182344,
      "learning_rate": 1.92232478258388e-06,
      "loss": 0.1609,
      "step": 26297
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.48995776296777455,
      "learning_rate": 1.9217401058271356e-06,
      "loss": 0.3339,
      "step": 26298
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4855729073115225,
      "learning_rate": 1.921155508546968e-06,
      "loss": 0.2276,
      "step": 26299
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.353425988574574,
      "learning_rate": 1.920570990749131e-06,
      "loss": 0.8088,
      "step": 26300
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16095837822441708,
      "learning_rate": 1.919986552439377e-06,
      "loss": 0.0867,
      "step": 26301
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.544513881798611,
      "learning_rate": 1.9194021936234484e-06,
      "loss": 0.3336,
      "step": 26302
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.29825694065009467,
      "learning_rate": 1.9188179143071063e-06,
      "loss": 0.1579,
      "step": 26303
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3629896596505508,
      "learning_rate": 1.9182337144960892e-06,
      "loss": 0.2916,
      "step": 26304
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.685748829016738,
      "learning_rate": 1.9176495941961493e-06,
      "loss": 0.2143,
      "step": 26305
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9731568784037912,
      "learning_rate": 1.9170655534130333e-06,
      "loss": 0.4287,
      "step": 26306
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3589485677260259,
      "learning_rate": 1.9164815921524836e-06,
      "loss": 0.1941,
      "step": 26307
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0079153266485363,
      "learning_rate": 1.915897710420249e-06,
      "loss": 0.42,
      "step": 26308
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3550542433806683,
      "learning_rate": 1.915313908222072e-06,
      "loss": 0.3001,
      "step": 26309
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4400187772570128,
      "learning_rate": 1.914730185563699e-06,
      "loss": 0.1448,
      "step": 26310
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5126914356839692,
      "learning_rate": 1.9141465424508685e-06,
      "loss": 0.2423,
      "step": 26311
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4397648422919488,
      "learning_rate": 1.913562978889326e-06,
      "loss": 0.1737,
      "step": 26312
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4295250690608974,
      "learning_rate": 1.9129794948848103e-06,
      "loss": 0.2602,
      "step": 26313
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0472828381851302,
      "learning_rate": 1.912396090443064e-06,
      "loss": 0.052,
      "step": 26314
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41687287393765754,
      "learning_rate": 1.9118127655698284e-06,
      "loss": 0.3003,
      "step": 26315
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26422618501642176,
      "learning_rate": 1.9112295202708387e-06,
      "loss": 0.1931,
      "step": 26316
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3016332049402277,
      "learning_rate": 1.910646354551833e-06,
      "loss": 0.4633,
      "step": 26317
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0505217541641203,
      "learning_rate": 1.910063268418554e-06,
      "loss": 0.5181,
      "step": 26318
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9313068819544175,
      "learning_rate": 1.909480261876728e-06,
      "loss": 0.4583,
      "step": 26319
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2706032380980308,
      "learning_rate": 1.9088973349321038e-06,
      "loss": 0.166,
      "step": 26320
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8269733772470974,
      "learning_rate": 1.908314487590407e-06,
      "loss": 0.2742,
      "step": 26321
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31821149486225075,
      "learning_rate": 1.9077317198573743e-06,
      "loss": 0.2662,
      "step": 26322
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.18142719175578587,
      "learning_rate": 1.9071490317387398e-06,
      "loss": 0.0669,
      "step": 26323
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.426056421001028,
      "learning_rate": 1.9065664232402392e-06,
      "loss": 0.2715,
      "step": 26324
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.48396776268697417,
      "learning_rate": 1.905983894367598e-06,
      "loss": 0.194,
      "step": 26325
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5256177071304968,
      "learning_rate": 1.9054014451265502e-06,
      "loss": 0.6872,
      "step": 26326
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3049974571634895,
      "learning_rate": 1.9048190755228303e-06,
      "loss": 0.2302,
      "step": 26327
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9804569015762954,
      "learning_rate": 1.904236785562158e-06,
      "loss": 0.406,
      "step": 26328
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.42488580889476657,
      "learning_rate": 1.9036545752502734e-06,
      "loss": 0.1735,
      "step": 26329
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5046848511207187,
      "learning_rate": 1.903072444592896e-06,
      "loss": 0.3124,
      "step": 26330
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.409528084234228,
      "learning_rate": 1.9024903935957562e-06,
      "loss": 0.1703,
      "step": 26331
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5473912747587155,
      "learning_rate": 1.9019084222645835e-06,
      "loss": 0.2322,
      "step": 26332
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34313676675530386,
      "learning_rate": 1.9013265306050977e-06,
      "loss": 0.1824,
      "step": 26333
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2564065057041257,
      "learning_rate": 1.9007447186230265e-06,
      "loss": 0.1913,
      "step": 26334
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2958717233094124,
      "learning_rate": 1.9001629863240933e-06,
      "loss": 0.4863,
      "step": 26335
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7598519118192827,
      "learning_rate": 1.8995813337140246e-06,
      "loss": 0.3423,
      "step": 26336
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0385406307735525,
      "learning_rate": 1.8989997607985377e-06,
      "loss": 0.5469,
      "step": 26337
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2837731981105653,
      "learning_rate": 1.8984182675833562e-06,
      "loss": 0.1631,
      "step": 26338
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4211006981509754,
      "learning_rate": 1.8978368540742021e-06,
      "loss": 0.3008,
      "step": 26339
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2760863070582737,
      "learning_rate": 1.897255520276795e-06,
      "loss": 0.1511,
      "step": 26340
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.48217762304927286,
      "learning_rate": 1.8966742661968573e-06,
      "loss": 0.2093,
      "step": 26341
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4456808299730378,
      "learning_rate": 1.896093091840101e-06,
      "loss": 0.0715,
      "step": 26342
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37729479059695054,
      "learning_rate": 1.895511997212247e-06,
      "loss": 0.2638,
      "step": 26343
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9988908565537505,
      "learning_rate": 1.8949309823190165e-06,
      "loss": 0.4227,
      "step": 26344
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4954332464000244,
      "learning_rate": 1.8943500471661158e-06,
      "loss": 0.3684,
      "step": 26345
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5499349236894552,
      "learning_rate": 1.8937691917592726e-06,
      "loss": 0.2594,
      "step": 26346
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26077964210814003,
      "learning_rate": 1.8931884161041925e-06,
      "loss": 0.1641,
      "step": 26347
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9053403128647015,
      "learning_rate": 1.8926077202065917e-06,
      "loss": 0.4773,
      "step": 26348
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2674905118089086,
      "learning_rate": 1.8920271040721872e-06,
      "loss": 0.0809,
      "step": 26349
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3258394407796335,
      "learning_rate": 1.891446567706684e-06,
      "loss": 0.1741,
      "step": 26350
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.275416638213593,
      "learning_rate": 1.890866111115799e-06,
      "loss": 0.202,
      "step": 26351
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.411842229161497,
      "learning_rate": 1.8902857343052405e-06,
      "loss": 0.3024,
      "step": 26352
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7552136371838514,
      "learning_rate": 1.8897054372807222e-06,
      "loss": 0.2388,
      "step": 26353
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8318198386502735,
      "learning_rate": 1.889125220047946e-06,
      "loss": 0.4548,
      "step": 26354
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5850520689055718,
      "learning_rate": 1.8885450826126296e-06,
      "loss": 0.2329,
      "step": 26355
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0186561671234533,
      "learning_rate": 1.8879650249804725e-06,
      "loss": 0.4782,
      "step": 26356
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2773434669649921,
      "learning_rate": 1.8873850471571854e-06,
      "loss": 0.2201,
      "step": 26357
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5297184933671595,
      "learning_rate": 1.8868051491484752e-06,
      "loss": 0.3587,
      "step": 26358
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19566852672088506,
      "learning_rate": 1.8862253309600443e-06,
      "loss": 0.0708,
      "step": 26359
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8037369967999719,
      "learning_rate": 1.885645592597597e-06,
      "loss": 0.1137,
      "step": 26360
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.346332420013884,
      "learning_rate": 1.8850659340668388e-06,
      "loss": 0.2468,
      "step": 26361
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8372083868981359,
      "learning_rate": 1.8844863553734728e-06,
      "loss": 0.3116,
      "step": 26362
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3542250764169441,
      "learning_rate": 1.883906856523202e-06,
      "loss": 0.2868,
      "step": 26363
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7596403685024056,
      "learning_rate": 1.8833274375217248e-06,
      "loss": 0.2736,
      "step": 26364
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6270313606906129,
      "learning_rate": 1.8827480983747415e-06,
      "loss": 0.3175,
      "step": 26365
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38951090460786186,
      "learning_rate": 1.8821688390879545e-06,
      "loss": 0.2099,
      "step": 26366
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.3580004928417952,
      "learning_rate": 1.8815896596670636e-06,
      "loss": 0.4494,
      "step": 26367
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.14152110090616676,
      "learning_rate": 1.881010560117762e-06,
      "loss": 0.0681,
      "step": 26368
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3023775238092083,
      "learning_rate": 1.8804315404457508e-06,
      "loss": 0.2448,
      "step": 26369
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.29455756018769774,
      "learning_rate": 1.8798526006567275e-06,
      "loss": 0.1712,
      "step": 26370
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0209662525678815,
      "learning_rate": 1.8792737407563822e-06,
      "loss": 0.4447,
      "step": 26371
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0209663576435153,
      "learning_rate": 1.8786949607504179e-06,
      "loss": 0.2685,
      "step": 26372
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9258408772424733,
      "learning_rate": 1.8781162606445236e-06,
      "loss": 0.363,
      "step": 26373
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3952092465876611,
      "learning_rate": 1.8775376404443924e-06,
      "loss": 0.2658,
      "step": 26374
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31695882861091706,
      "learning_rate": 1.876959100155722e-06,
      "loss": 0.2096,
      "step": 26375
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.49654025058564005,
      "learning_rate": 1.8763806397841976e-06,
      "loss": 0.2987,
      "step": 26376
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3866424962767777,
      "learning_rate": 1.8758022593355152e-06,
      "loss": 0.0898,
      "step": 26377
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4025605620861833,
      "learning_rate": 1.8752239588153621e-06,
      "loss": 0.2077,
      "step": 26378
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3904953711591353,
      "learning_rate": 1.8746457382294315e-06,
      "loss": 0.1637,
      "step": 26379
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6400015245680402,
      "learning_rate": 1.8740675975834078e-06,
      "loss": 0.3308,
      "step": 26380
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5148208416175971,
      "learning_rate": 1.8734895368829798e-06,
      "loss": 0.2483,
      "step": 26381
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1165799933643488,
      "learning_rate": 1.8729115561338396e-06,
      "loss": 0.4413,
      "step": 26382
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4210506223887862,
      "learning_rate": 1.8723336553416637e-06,
      "loss": 0.1416,
      "step": 26383
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37262894022333815,
      "learning_rate": 1.8717558345121488e-06,
      "loss": 0.2428,
      "step": 26384
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5719972244561662,
      "learning_rate": 1.8711780936509715e-06,
      "loss": 0.1404,
      "step": 26385
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23491571580063833,
      "learning_rate": 1.8706004327638204e-06,
      "loss": 0.1806,
      "step": 26386
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5051585375684762,
      "learning_rate": 1.8700228518563756e-06,
      "loss": 0.2488,
      "step": 26387
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.28574265024966655,
      "learning_rate": 1.8694453509343225e-06,
      "loss": 0.1805,
      "step": 26388
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9878267527185774,
      "learning_rate": 1.868867930003343e-06,
      "loss": 0.3518,
      "step": 26389
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.166008339890989,
      "learning_rate": 1.8682905890691139e-06,
      "loss": 0.3127,
      "step": 26390
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.7268323983372642,
      "learning_rate": 1.8677133281373172e-06,
      "loss": 0.7468,
      "step": 26391
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5371714318909554,
      "learning_rate": 1.867136147213634e-06,
      "loss": 0.1249,
      "step": 26392
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33874689025534993,
      "learning_rate": 1.866559046303743e-06,
      "loss": 0.2556,
      "step": 26393
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4789206831031682,
      "learning_rate": 1.8659820254133177e-06,
      "loss": 0.2065,
      "step": 26394
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.666941001908082,
      "learning_rate": 1.8654050845480375e-06,
      "loss": 0.7067,
      "step": 26395
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5606857872953829,
      "learning_rate": 1.8648282237135818e-06,
      "loss": 0.1629,
      "step": 26396
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4707545819630274,
      "learning_rate": 1.8642514429156178e-06,
      "loss": 0.2549,
      "step": 26397
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.48088824677924885,
      "learning_rate": 1.86367474215983e-06,
      "loss": 0.2437,
      "step": 26398
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26531461920365357,
      "learning_rate": 1.8630981214518852e-06,
      "loss": 0.1616,
      "step": 26399
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8388065400291895,
      "learning_rate": 1.8625215807974583e-06,
      "loss": 0.7931,
      "step": 26400
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3353537666378092,
      "learning_rate": 1.8619451202022244e-06,
      "loss": 0.1645,
      "step": 26401
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.540318171893978,
      "learning_rate": 1.8613687396718482e-06,
      "loss": 0.2616,
      "step": 26402
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.47378482546218975,
      "learning_rate": 1.8607924392120058e-06,
      "loss": 0.026,
      "step": 26403
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4905603936335421,
      "learning_rate": 1.8602162188283656e-06,
      "loss": 0.291,
      "step": 26404
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37241860655982334,
      "learning_rate": 1.8596400785265977e-06,
      "loss": 0.1951,
      "step": 26405
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.49088763160813265,
      "learning_rate": 1.8590640183123676e-06,
      "loss": 0.2993,
      "step": 26406
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5121081022739838,
      "learning_rate": 1.858488038191345e-06,
      "loss": 0.2406,
      "step": 26407
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4686419686059022,
      "learning_rate": 1.8579121381691978e-06,
      "loss": 0.2842,
      "step": 26408
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2650293814647769,
      "learning_rate": 1.8573363182515857e-06,
      "loss": 0.1559,
      "step": 26409
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4533859463814828,
      "learning_rate": 1.8567605784441823e-06,
      "loss": 0.2762,
      "step": 26410
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3273049835559685,
      "learning_rate": 1.856184918752646e-06,
      "loss": 0.1897,
      "step": 26411
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43908570639711614,
      "learning_rate": 1.8556093391826412e-06,
      "loss": 0.1911,
      "step": 26412
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8890085189304112,
      "learning_rate": 1.8550338397398349e-06,
      "loss": 0.4523,
      "step": 26413
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6756041571480816,
      "learning_rate": 1.854458420429881e-06,
      "loss": 0.281,
      "step": 26414
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3942020203663631,
      "learning_rate": 1.8538830812584495e-06,
      "loss": 0.2262,
      "step": 26415
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2749208664834115,
      "learning_rate": 1.853307822231195e-06,
      "loss": 0.1646,
      "step": 26416
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3227429494098855,
      "learning_rate": 1.8527326433537783e-06,
      "loss": 0.2915,
      "step": 26417
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.051545536639279,
      "learning_rate": 1.8521575446318618e-06,
      "loss": 0.4332,
      "step": 26418
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3689314875596827,
      "learning_rate": 1.8515825260710973e-06,
      "loss": 0.2212,
      "step": 26419
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3311928540725265,
      "learning_rate": 1.8510075876771449e-06,
      "loss": 0.186,
      "step": 26420
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5152900426528189,
      "learning_rate": 1.8504327294556613e-06,
      "loss": 0.4092,
      "step": 26421
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4090753475293763,
      "learning_rate": 1.849857951412305e-06,
      "loss": 0.2363,
      "step": 26422
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5276750841816481,
      "learning_rate": 1.8492832535527228e-06,
      "loss": 0.3467,
      "step": 26423
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33333604390653043,
      "learning_rate": 1.848708635882578e-06,
      "loss": 0.1792,
      "step": 26424
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4346305187418714,
      "learning_rate": 1.848134098407518e-06,
      "loss": 0.1686,
      "step": 26425
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6438910921374403,
      "learning_rate": 1.8475596411331975e-06,
      "loss": 0.7899,
      "step": 26426
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2674585897393497,
      "learning_rate": 1.8469852640652696e-06,
      "loss": 0.1393,
      "step": 26427
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3129391869948152,
      "learning_rate": 1.8464109672093811e-06,
      "loss": 0.2571,
      "step": 26428
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2378229985904845,
      "learning_rate": 1.845836750571185e-06,
      "loss": 0.1556,
      "step": 26429
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.9396676677142455,
      "learning_rate": 1.845262614156329e-06,
      "loss": 0.7614,
      "step": 26430
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6828760353138574,
      "learning_rate": 1.8446885579704653e-06,
      "loss": 0.2495,
      "step": 26431
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6948896010893412,
      "learning_rate": 1.8441145820192374e-06,
      "loss": 0.393,
      "step": 26432
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2962867860635446,
      "learning_rate": 1.8435406863082927e-06,
      "loss": 0.1395,
      "step": 26433
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5665819453439398,
      "learning_rate": 1.8429668708432813e-06,
      "loss": 0.3146,
      "step": 26434
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3159896143591584,
      "learning_rate": 1.842393135629842e-06,
      "loss": 0.2369,
      "step": 26435
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5123710705484934,
      "learning_rate": 1.8418194806736267e-06,
      "loss": 0.6924,
      "step": 26436
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.17012611756735377,
      "learning_rate": 1.8412459059802744e-06,
      "loss": 0.0692,
      "step": 26437
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3033247528397584,
      "learning_rate": 1.8406724115554287e-06,
      "loss": 0.1726,
      "step": 26438
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0853283979462027,
      "learning_rate": 1.8400989974047357e-06,
      "loss": 0.4786,
      "step": 26439
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.32820978281519225,
      "learning_rate": 1.8395256635338287e-06,
      "loss": 0.2373,
      "step": 26440
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9692058895436467,
      "learning_rate": 1.838952409948358e-06,
      "loss": 0.4326,
      "step": 26441
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5586868858079649,
      "learning_rate": 1.8383792366539566e-06,
      "loss": 0.0963,
      "step": 26442
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.448042460249331,
      "learning_rate": 1.8378061436562656e-06,
      "loss": 0.2838,
      "step": 26443
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9809684786511494,
      "learning_rate": 1.837233130960926e-06,
      "loss": 0.3285,
      "step": 26444
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5349802781638655,
      "learning_rate": 1.83666019857357e-06,
      "loss": 0.7045,
      "step": 26445
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2862025325893355,
      "learning_rate": 1.8360873464998375e-06,
      "loss": 0.2044,
      "step": 26446
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3666083963508028,
      "learning_rate": 1.835514574745364e-06,
      "loss": 0.2458,
      "step": 26447
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36498070462325427,
      "learning_rate": 1.8349418833157862e-06,
      "loss": 0.142,
      "step": 26448
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1410113820919763,
      "learning_rate": 1.834369272216735e-06,
      "loss": 0.4612,
      "step": 26449
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0114425799116604,
      "learning_rate": 1.833796741453846e-06,
      "loss": 0.2182,
      "step": 26450
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33382011174413034,
      "learning_rate": 1.8332242910327535e-06,
      "loss": 0.1527,
      "step": 26451
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.387734259151891,
      "learning_rate": 1.8326519209590832e-06,
      "loss": 0.2727,
      "step": 26452
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41718027051432943,
      "learning_rate": 1.832079631238476e-06,
      "loss": 0.2252,
      "step": 26453
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.438207152056981,
      "learning_rate": 1.8315074218765538e-06,
      "loss": 0.7945,
      "step": 26454
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3146472096120922,
      "learning_rate": 1.8309352928789502e-06,
      "loss": 0.0765,
      "step": 26455
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3499401546674614,
      "learning_rate": 1.830363244251292e-06,
      "loss": 0.2616,
      "step": 26456
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1176434366966213,
      "learning_rate": 1.829791275999212e-06,
      "loss": 0.273,
      "step": 26457
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5728350568948848,
      "learning_rate": 1.8292193881283304e-06,
      "loss": 0.3046,
      "step": 26458
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3189944481706918,
      "learning_rate": 1.8286475806442772e-06,
      "loss": 0.1573,
      "step": 26459
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7414246267441446,
      "learning_rate": 1.828075853552681e-06,
      "loss": 0.3178,
      "step": 26460
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34893390542339964,
      "learning_rate": 1.8275042068591575e-06,
      "loss": 0.2037,
      "step": 26461
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4794843217896418,
      "learning_rate": 1.8269326405693422e-06,
      "loss": 0.7378,
      "step": 26462
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43532033597277414,
      "learning_rate": 1.8263611546888505e-06,
      "loss": 0.2598,
      "step": 26463
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4377044660892503,
      "learning_rate": 1.8257897492233079e-06,
      "loss": 0.2859,
      "step": 26464
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3826019260433938,
      "learning_rate": 1.8252184241783367e-06,
      "loss": 0.1729,
      "step": 26465
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5142313908962283,
      "learning_rate": 1.824647179559552e-06,
      "loss": 0.2392,
      "step": 26466
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.28572584816959157,
      "learning_rate": 1.8240760153725835e-06,
      "loss": 0.1703,
      "step": 26467
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0146309881963864,
      "learning_rate": 1.8235049316230425e-06,
      "loss": 0.1084,
      "step": 26468
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8529039385596248,
      "learning_rate": 1.8229339283165504e-06,
      "loss": 0.3983,
      "step": 26469
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.28226670963851785,
      "learning_rate": 1.8223630054587283e-06,
      "loss": 0.2087,
      "step": 26470
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.46938277027548436,
      "learning_rate": 1.821792163055186e-06,
      "loss": 0.3552,
      "step": 26471
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.054732482204629,
      "learning_rate": 1.8212214011115448e-06,
      "loss": 0.4297,
      "step": 26472
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7665560537112364,
      "learning_rate": 1.8206507196334179e-06,
      "loss": 0.3401,
      "step": 26473
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.35642470036268686,
      "learning_rate": 1.8200801186264228e-06,
      "loss": 0.1895,
      "step": 26474
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5200219406903546,
      "learning_rate": 1.8195095980961685e-06,
      "loss": 0.3284,
      "step": 26475
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.42689812304728153,
      "learning_rate": 1.8189391580482707e-06,
      "loss": 0.2189,
      "step": 26476
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.25382089711041284,
      "learning_rate": 1.8183687984883436e-06,
      "loss": 0.195,
      "step": 26477
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43048767002712063,
      "learning_rate": 1.8177985194219916e-06,
      "loss": 0.0708,
      "step": 26478
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3423344357905106,
      "learning_rate": 1.8172283208548358e-06,
      "loss": 0.1968,
      "step": 26479
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.8823310204663857,
      "learning_rate": 1.8166582027924774e-06,
      "loss": 0.8144,
      "step": 26480
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45528894132017367,
      "learning_rate": 1.816088165240527e-06,
      "loss": 0.2526,
      "step": 26481
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4328446871537733,
      "learning_rate": 1.8155182082045985e-06,
      "loss": 0.3138,
      "step": 26482
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2711246225118451,
      "learning_rate": 1.8149483316902917e-06,
      "loss": 0.1691,
      "step": 26483
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2656189504897883,
      "learning_rate": 1.8143785357032162e-06,
      "loss": 0.4287,
      "step": 26484
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3600631255445974,
      "learning_rate": 1.8138088202489778e-06,
      "loss": 0.1339,
      "step": 26485
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3983460611381959,
      "learning_rate": 1.8132391853331842e-06,
      "loss": 0.1986,
      "step": 26486
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3090729656393081,
      "learning_rate": 1.8126696309614334e-06,
      "loss": 0.1708,
      "step": 26487
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.52230276408434,
      "learning_rate": 1.8121001571393372e-06,
      "loss": 0.2891,
      "step": 26488
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3139980746062858,
      "learning_rate": 1.811530763872491e-06,
      "loss": 0.2129,
      "step": 26489
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8339747226442678,
      "learning_rate": 1.8109614511664997e-06,
      "loss": 0.4608,
      "step": 26490
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.747600272720566,
      "learning_rate": 1.8103922190269673e-06,
      "loss": 0.2556,
      "step": 26491
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.30810980307985103,
      "learning_rate": 1.8098230674594852e-06,
      "loss": 0.0679,
      "step": 26492
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41422902566926156,
      "learning_rate": 1.8092539964696643e-06,
      "loss": 0.2607,
      "step": 26493
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3329301587726074,
      "learning_rate": 1.8086850060630944e-06,
      "loss": 0.207,
      "step": 26494
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.24954766940026366,
      "learning_rate": 1.808116096245378e-06,
      "loss": 0.1446,
      "step": 26495
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4056024731612095,
      "learning_rate": 1.8075472670221117e-06,
      "loss": 0.0591,
      "step": 26496
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.36821257475337876,
      "learning_rate": 1.8069785183988896e-06,
      "loss": 0.2541,
      "step": 26497
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9454376752021406,
      "learning_rate": 1.8064098503813087e-06,
      "loss": 0.3692,
      "step": 26498
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8110482544937478,
      "learning_rate": 1.8058412629749645e-06,
      "loss": 0.3891,
      "step": 26499
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.29773609723661576,
      "learning_rate": 1.805272756185451e-06,
      "loss": 0.2309,
      "step": 26500
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.594040271331954,
      "learning_rate": 1.8047043300183587e-06,
      "loss": 0.3089,
      "step": 26501
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3627492084687229,
      "learning_rate": 1.8041359844792827e-06,
      "loss": 0.2049,
      "step": 26502
    },
    {
      "epoch": 0.81,
      "grad_norm": 2.0295187387165647,
      "learning_rate": 1.8035677195738156e-06,
      "loss": 0.6988,
      "step": 26503
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.16311364103439976,
      "learning_rate": 1.8029995353075403e-06,
      "loss": 0.0665,
      "step": 26504
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3072218062411869,
      "learning_rate": 1.802431431686059e-06,
      "loss": 0.2059,
      "step": 26505
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3682592870532662,
      "learning_rate": 1.8018634087149512e-06,
      "loss": 0.2476,
      "step": 26506
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1286889040232293,
      "learning_rate": 1.8012954663998094e-06,
      "loss": 0.4439,
      "step": 26507
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8840828871793136,
      "learning_rate": 1.800727604746223e-06,
      "loss": 0.3841,
      "step": 26508
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8210456816025831,
      "learning_rate": 1.800159823759774e-06,
      "loss": 0.2758,
      "step": 26509
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.44600084350410363,
      "learning_rate": 1.79959212344605e-06,
      "loss": 0.2861,
      "step": 26510
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5291083646590473,
      "learning_rate": 1.799024503810638e-06,
      "loss": 0.1869,
      "step": 26511
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3748916773782288,
      "learning_rate": 1.7984569648591233e-06,
      "loss": 0.2888,
      "step": 26512
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.26704570701747066,
      "learning_rate": 1.7978895065970848e-06,
      "loss": 0.0759,
      "step": 26513
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8728891974343161,
      "learning_rate": 1.797322129030108e-06,
      "loss": 0.4377,
      "step": 26514
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38422316904585757,
      "learning_rate": 1.7967548321637763e-06,
      "loss": 0.1721,
      "step": 26515
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5445997476554,
      "learning_rate": 1.7961876160036685e-06,
      "loss": 0.3263,
      "step": 26516
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6611756700603939,
      "learning_rate": 1.79562048055537e-06,
      "loss": 0.2617,
      "step": 26517
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4408579467126689,
      "learning_rate": 1.7950534258244535e-06,
      "loss": 0.2505,
      "step": 26518
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9187423678884432,
      "learning_rate": 1.7944864518165017e-06,
      "loss": 0.3022,
      "step": 26519
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2986971748192435,
      "learning_rate": 1.793919558537095e-06,
      "loss": 0.1738,
      "step": 26520
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2768390393646467,
      "learning_rate": 1.7933527459918022e-06,
      "loss": 0.4573,
      "step": 26521
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19592218123663616,
      "learning_rate": 1.7927860141862097e-06,
      "loss": 0.0727,
      "step": 26522
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3279016702096346,
      "learning_rate": 1.7922193631258877e-06,
      "loss": 0.2596,
      "step": 26523
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3078810001008973,
      "learning_rate": 1.7916527928164119e-06,
      "loss": 0.1564,
      "step": 26524
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5730143726669007,
      "learning_rate": 1.7910863032633564e-06,
      "loss": 0.3486,
      "step": 26525
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.739146027820441,
      "learning_rate": 1.790519894472298e-06,
      "loss": 0.2901,
      "step": 26526
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.5409060379591484,
      "learning_rate": 1.7899535664488033e-06,
      "loss": 0.857,
      "step": 26527
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4090915569974754,
      "learning_rate": 1.7893873191984478e-06,
      "loss": 0.0691,
      "step": 26528
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38963941615116104,
      "learning_rate": 1.788821152726803e-06,
      "loss": 0.2816,
      "step": 26529
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.35580577879435227,
      "learning_rate": 1.7882550670394328e-06,
      "loss": 0.2411,
      "step": 26530
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2312863569217361,
      "learning_rate": 1.7876890621419163e-06,
      "loss": 0.0656,
      "step": 26531
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.45364059341086543,
      "learning_rate": 1.7871231380398147e-06,
      "loss": 0.261,
      "step": 26532
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.29601849958105525,
      "learning_rate": 1.7865572947386977e-06,
      "loss": 0.1701,
      "step": 26533
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9694337363672649,
      "learning_rate": 1.7859915322441358e-06,
      "loss": 0.4702,
      "step": 26534
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.41332501549845785,
      "learning_rate": 1.7854258505616884e-06,
      "loss": 0.2478,
      "step": 26535
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5546040950995534,
      "learning_rate": 1.7848602496969247e-06,
      "loss": 0.3772,
      "step": 26536
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3201365544270361,
      "learning_rate": 1.7842947296554092e-06,
      "loss": 0.1163,
      "step": 26537
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5487544957217685,
      "learning_rate": 1.7837292904427085e-06,
      "loss": 0.3264,
      "step": 26538
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5477627583891116,
      "learning_rate": 1.783163932064379e-06,
      "loss": 0.1853,
      "step": 26539
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.494764281087705,
      "learning_rate": 1.7825986545259876e-06,
      "loss": 0.1985,
      "step": 26540
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3007126649232084,
      "learning_rate": 1.7820334578330933e-06,
      "loss": 0.2117,
      "step": 26541
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.013907673726759,
      "learning_rate": 1.7814683419912592e-06,
      "loss": 0.4355,
      "step": 26542
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4119960624196342,
      "learning_rate": 1.7809033070060455e-06,
      "loss": 0.2255,
      "step": 26543
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.42534142787624957,
      "learning_rate": 1.7803383528830076e-06,
      "loss": 0.1774,
      "step": 26544
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.1958468264462017,
      "learning_rate": 1.7797734796277045e-06,
      "loss": 0.5754,
      "step": 26545
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.42949075962343086,
      "learning_rate": 1.7792086872456981e-06,
      "loss": 0.0632,
      "step": 26546
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33675614059725445,
      "learning_rate": 1.7786439757425366e-06,
      "loss": 0.2494,
      "step": 26547
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33596015771375765,
      "learning_rate": 1.7780793451237855e-06,
      "loss": 0.2047,
      "step": 26548
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9901538351582675,
      "learning_rate": 1.7775147953949934e-06,
      "loss": 0.5047,
      "step": 26549
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8104505085895747,
      "learning_rate": 1.7769503265617149e-06,
      "loss": 0.2644,
      "step": 26550
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5526107772122016,
      "learning_rate": 1.776385938629509e-06,
      "loss": 0.276,
      "step": 26551
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.23822494337555677,
      "learning_rate": 1.7758216316039201e-06,
      "loss": 0.155,
      "step": 26552
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.38984169745240443,
      "learning_rate": 1.775257405490506e-06,
      "loss": 0.2751,
      "step": 26553
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0776970191527993,
      "learning_rate": 1.774693260294814e-06,
      "loss": 0.4585,
      "step": 26554
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3734802401654386,
      "learning_rate": 1.7741291960223995e-06,
      "loss": 0.2218,
      "step": 26555
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3006134993328173,
      "learning_rate": 1.7735652126788038e-06,
      "loss": 0.1753,
      "step": 26556
    },
    {
      "epoch": 0.81,
      "grad_norm": 3.212726488088821,
      "learning_rate": 1.7730013102695842e-06,
      "loss": 0.0952,
      "step": 26557
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.9218549319768116,
      "learning_rate": 1.7724374888002837e-06,
      "loss": 0.3736,
      "step": 26558
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34549947221601063,
      "learning_rate": 1.7718737482764504e-06,
      "loss": 0.2399,
      "step": 26559
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3882091277119274,
      "learning_rate": 1.7713100887036316e-06,
      "loss": 0.2274,
      "step": 26560
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5306324346737638,
      "learning_rate": 1.770746510087371e-06,
      "loss": 0.2377,
      "step": 26561
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5274171673140177,
      "learning_rate": 1.770183012433213e-06,
      "loss": 0.2389,
      "step": 26562
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2575049227703627,
      "learning_rate": 1.7696195957467021e-06,
      "loss": 0.1482,
      "step": 26563
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4397968182723739,
      "learning_rate": 1.769056260033385e-06,
      "loss": 0.3059,
      "step": 26564
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3777802524826073,
      "learning_rate": 1.768493005298798e-06,
      "loss": 0.1688,
      "step": 26565
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3669651204816655,
      "learning_rate": 1.7679298315484862e-06,
      "loss": 0.2706,
      "step": 26566
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.708886407666132,
      "learning_rate": 1.7673667387879889e-06,
      "loss": 0.2602,
      "step": 26567
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7345800838086499,
      "learning_rate": 1.7668037270228456e-06,
      "loss": 0.3633,
      "step": 26568
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3550176945458982,
      "learning_rate": 1.7662407962585993e-06,
      "loss": 0.0625,
      "step": 26569
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3206589968763262,
      "learning_rate": 1.7656779465007834e-06,
      "loss": 0.1847,
      "step": 26570
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.34055377643772095,
      "learning_rate": 1.765115177754937e-06,
      "loss": 0.2997,
      "step": 26571
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2422556853247664,
      "learning_rate": 1.7645524900265998e-06,
      "loss": 0.139,
      "step": 26572
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4007860005583868,
      "learning_rate": 1.7639898833213e-06,
      "loss": 0.1866,
      "step": 26573
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2761765455946824,
      "learning_rate": 1.7634273576445827e-06,
      "loss": 0.1675,
      "step": 26574
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.002751317380543,
      "learning_rate": 1.7628649130019738e-06,
      "loss": 0.3793,
      "step": 26575
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4831749965404655,
      "learning_rate": 1.7623025493990108e-06,
      "loss": 0.2319,
      "step": 26576
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.47862584032290784,
      "learning_rate": 1.7617402668412286e-06,
      "loss": 0.322,
      "step": 26577
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6346404874734306,
      "learning_rate": 1.7611780653341548e-06,
      "loss": 0.0857,
      "step": 26578
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.43961808627275967,
      "learning_rate": 1.7606159448833204e-06,
      "loss": 0.2506,
      "step": 26579
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0484127456863537,
      "learning_rate": 1.7600539054942588e-06,
      "loss": 0.5105,
      "step": 26580
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6047383343422443,
      "learning_rate": 1.7594919471725003e-06,
      "loss": 0.7842,
      "step": 26581
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.19942989242315715,
      "learning_rate": 1.7589300699235689e-06,
      "loss": 0.1257,
      "step": 26582
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2778575689990607,
      "learning_rate": 1.7583682737529951e-06,
      "loss": 0.1745,
      "step": 26583
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.4470250596590764,
      "learning_rate": 1.7578065586663085e-06,
      "loss": 0.2684,
      "step": 26584
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6801670120482947,
      "learning_rate": 1.7572449246690281e-06,
      "loss": 0.2691,
      "step": 26585
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0008651262457509,
      "learning_rate": 1.7566833717666887e-06,
      "loss": 0.4568,
      "step": 26586
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3857534603031777,
      "learning_rate": 1.7561218999648088e-06,
      "loss": 0.1617,
      "step": 26587
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5678179115746018,
      "learning_rate": 1.7555605092689143e-06,
      "loss": 0.3313,
      "step": 26588
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3198113631184921,
      "learning_rate": 1.7549991996845317e-06,
      "loss": 0.2294,
      "step": 26589
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.4222706379867966,
      "learning_rate": 1.7544379712171766e-06,
      "loss": 0.7983,
      "step": 26590
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.2132849726500379,
      "learning_rate": 1.7538768238723736e-06,
      "loss": 0.0706,
      "step": 26591
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.506173187175206,
      "learning_rate": 1.7533157576556436e-06,
      "loss": 0.2587,
      "step": 26592
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.51836903139616,
      "learning_rate": 1.7527547725725068e-06,
      "loss": 0.2537,
      "step": 26593
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8157891561778323,
      "learning_rate": 1.7521938686284823e-06,
      "loss": 0.364,
      "step": 26594
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.32795765148806155,
      "learning_rate": 1.751633045829092e-06,
      "loss": 0.2076,
      "step": 26595
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.6042322572694016,
      "learning_rate": 1.751072304179846e-06,
      "loss": 0.0681,
      "step": 26596
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3442348147959954,
      "learning_rate": 1.7505116436862657e-06,
      "loss": 0.2612,
      "step": 26597
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.2741882850742805,
      "learning_rate": 1.7499510643538687e-06,
      "loss": 0.343,
      "step": 26598
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.6991791732874788,
      "learning_rate": 1.7493905661881616e-06,
      "loss": 0.7528,
      "step": 26599
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.469478235117602,
      "learning_rate": 1.7488301491946713e-06,
      "loss": 0.2389,
      "step": 26600
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.37369085022912685,
      "learning_rate": 1.7482698133789023e-06,
      "loss": 0.2498,
      "step": 26601
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3512639339238891,
      "learning_rate": 1.74770955874637e-06,
      "loss": 0.215,
      "step": 26602
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.3212918243690652,
      "learning_rate": 1.7471493853025888e-06,
      "loss": 0.185,
      "step": 26603
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7427775708613463,
      "learning_rate": 1.7465892930530649e-06,
      "loss": 0.0243,
      "step": 26604
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.8669857564822735,
      "learning_rate": 1.7460292820033109e-06,
      "loss": 0.425,
      "step": 26605
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.31530510286670804,
      "learning_rate": 1.7454693521588374e-06,
      "loss": 0.1936,
      "step": 26606
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.33724865746481486,
      "learning_rate": 1.744909503525154e-06,
      "loss": 0.3069,
      "step": 26607
    },
    {
      "epoch": 0.81,
      "grad_norm": 1.0083895291054552,
      "learning_rate": 1.7443497361077654e-06,
      "loss": 0.4476,
      "step": 26608
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.7606941745145428,
      "learning_rate": 1.7437900499121807e-06,
      "loss": 0.234,
      "step": 26609
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.39760648301248624,
      "learning_rate": 1.7432304449439085e-06,
      "loss": 0.2438,
      "step": 26610
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.5066562701915421,
      "learning_rate": 1.742670921208447e-06,
      "loss": 0.2337,
      "step": 26611
    },
    {
      "epoch": 0.81,
      "grad_norm": 0.27514148480598566,
      "learning_rate": 1.7421114787113113e-06,
      "loss": 0.1651,
      "step": 26612
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37079133960178046,
      "learning_rate": 1.7415521174579975e-06,
      "loss": 0.2145,
      "step": 26613
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9821967993299445,
      "learning_rate": 1.7409928374540108e-06,
      "loss": 0.407,
      "step": 26614
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3193277682805219,
      "learning_rate": 1.7404336387048571e-06,
      "loss": 0.1887,
      "step": 26615
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.432810937414502,
      "learning_rate": 1.7398745212160305e-06,
      "loss": 0.7404,
      "step": 26616
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8970608497270306,
      "learning_rate": 1.7393154849930405e-06,
      "loss": 0.3048,
      "step": 26617
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33448338804447514,
      "learning_rate": 1.7387565300413812e-06,
      "loss": 0.2699,
      "step": 26618
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.31121086269910553,
      "learning_rate": 1.7381976563665526e-06,
      "loss": 0.1453,
      "step": 26619
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4910827182407994,
      "learning_rate": 1.737638863974056e-06,
      "loss": 0.3017,
      "step": 26620
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3772608884336198,
      "learning_rate": 1.737080152869386e-06,
      "loss": 0.1374,
      "step": 26621
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.461062022479954,
      "learning_rate": 1.736521523058039e-06,
      "loss": 0.1058,
      "step": 26622
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6234944482819236,
      "learning_rate": 1.735962974545513e-06,
      "loss": 0.7459,
      "step": 26623
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3069030299015995,
      "learning_rate": 1.735404507337305e-06,
      "loss": 0.1782,
      "step": 26624
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34233022406969765,
      "learning_rate": 1.7348461214389011e-06,
      "loss": 0.2882,
      "step": 26625
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8347402031263098,
      "learning_rate": 1.7342878168558064e-06,
      "loss": 0.2946,
      "step": 26626
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0893395907491075,
      "learning_rate": 1.733729593593505e-06,
      "loss": 0.4237,
      "step": 26627
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5251957979844093,
      "learning_rate": 1.7331714516574927e-06,
      "loss": 0.165,
      "step": 26628
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4009927172157733,
      "learning_rate": 1.7326133910532616e-06,
      "loss": 0.2996,
      "step": 26629
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.25417119944627176,
      "learning_rate": 1.7320554117862976e-06,
      "loss": 0.1878,
      "step": 26630
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4997472056800904,
      "learning_rate": 1.7314975138620926e-06,
      "loss": 0.2496,
      "step": 26631
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3911981762745724,
      "learning_rate": 1.730939697286137e-06,
      "loss": 0.186,
      "step": 26632
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4234706046309333,
      "learning_rate": 1.7303819620639194e-06,
      "loss": 0.2098,
      "step": 26633
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4433945498949857,
      "learning_rate": 1.7298243082009226e-06,
      "loss": 0.2529,
      "step": 26634
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2332960730778602,
      "learning_rate": 1.7292667357026348e-06,
      "loss": 0.2643,
      "step": 26635
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36361926524266536,
      "learning_rate": 1.7287092445745458e-06,
      "loss": 0.2949,
      "step": 26636
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37281520169622123,
      "learning_rate": 1.7281518348221315e-06,
      "loss": 0.1819,
      "step": 26637
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6403146633570329,
      "learning_rate": 1.727594506450887e-06,
      "loss": 0.2863,
      "step": 26638
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.940833595317309,
      "learning_rate": 1.7270372594662864e-06,
      "loss": 0.143,
      "step": 26639
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.25676415527764657,
      "learning_rate": 1.7264800938738157e-06,
      "loss": 0.1292,
      "step": 26640
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.48028296368763757,
      "learning_rate": 1.725923009678958e-06,
      "loss": 0.1934,
      "step": 26641
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3522318090348694,
      "learning_rate": 1.725366006887188e-06,
      "loss": 0.2339,
      "step": 26642
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.30428155064146123,
      "learning_rate": 1.7248090855039946e-06,
      "loss": 0.2265,
      "step": 26643
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9931942973896437,
      "learning_rate": 1.7242522455348499e-06,
      "loss": 0.4298,
      "step": 26644
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9252552312300523,
      "learning_rate": 1.723695486985235e-06,
      "loss": 0.2581,
      "step": 26645
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.806938158043403,
      "learning_rate": 1.7231388098606294e-06,
      "loss": 0.3431,
      "step": 26646
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3394244272343353,
      "learning_rate": 1.722582214166505e-06,
      "loss": 0.1795,
      "step": 26647
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5224787138246346,
      "learning_rate": 1.7220256999083407e-06,
      "loss": 0.3333,
      "step": 26648
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.21261533149236048,
      "learning_rate": 1.721469267091611e-06,
      "loss": 0.1373,
      "step": 26649
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6991219264600324,
      "learning_rate": 1.720912915721794e-06,
      "loss": 0.2264,
      "step": 26650
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40258013748631405,
      "learning_rate": 1.7203566458043563e-06,
      "loss": 0.2572,
      "step": 26651
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7582943777092473,
      "learning_rate": 1.7198004573447747e-06,
      "loss": 0.2425,
      "step": 26652
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2079857634280289,
      "learning_rate": 1.7192443503485235e-06,
      "loss": 0.4351,
      "step": 26653
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32863896671060394,
      "learning_rate": 1.7186883248210663e-06,
      "loss": 0.2272,
      "step": 26654
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8228921488624971,
      "learning_rate": 1.7181323807678829e-06,
      "loss": 0.3976,
      "step": 26655
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.31001390423583025,
      "learning_rate": 1.717576518194436e-06,
      "loss": 0.1825,
      "step": 26656
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4516484257378495,
      "learning_rate": 1.7170207371061953e-06,
      "loss": 0.5441,
      "step": 26657
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1566462971237887,
      "learning_rate": 1.7164650375086311e-06,
      "loss": 0.0667,
      "step": 26658
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2560597795908066,
      "learning_rate": 1.7159094194072123e-06,
      "loss": 0.6231,
      "step": 26659
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2980432001405607,
      "learning_rate": 1.7153538828074002e-06,
      "loss": 0.1904,
      "step": 26660
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4253194080651882,
      "learning_rate": 1.7147984277146612e-06,
      "loss": 0.2892,
      "step": 26661
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5501755022399897,
      "learning_rate": 1.7142430541344658e-06,
      "loss": 0.2579,
      "step": 26662
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0708513311726955,
      "learning_rate": 1.713687762072267e-06,
      "loss": 0.2863,
      "step": 26663
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.895074303355715,
      "learning_rate": 1.7131325515335407e-06,
      "loss": 0.3808,
      "step": 26664
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34217370190022056,
      "learning_rate": 1.7125774225237401e-06,
      "loss": 0.1945,
      "step": 26665
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7323035779018876,
      "learning_rate": 1.712022375048329e-06,
      "loss": 0.6581,
      "step": 26666
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2275969032293239,
      "learning_rate": 1.7114674091127726e-06,
      "loss": 0.1646,
      "step": 26667
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6495897107097124,
      "learning_rate": 1.7109125247225211e-06,
      "loss": 0.2936,
      "step": 26668
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2951670465879828,
      "learning_rate": 1.7103577218830458e-06,
      "loss": 0.1721,
      "step": 26669
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5614529582954204,
      "learning_rate": 1.7098030005997957e-06,
      "loss": 0.4859,
      "step": 26670
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.8589212147665541,
      "learning_rate": 1.7092483608782317e-06,
      "loss": 0.3177,
      "step": 26671
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3218928066913649,
      "learning_rate": 1.7086938027238121e-06,
      "loss": 0.2678,
      "step": 26672
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8123320547335278,
      "learning_rate": 1.7081393261419898e-06,
      "loss": 0.1115,
      "step": 26673
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4730944890790176,
      "learning_rate": 1.7075849311382197e-06,
      "loss": 0.3055,
      "step": 26674
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5932326153498066,
      "learning_rate": 1.7070306177179586e-06,
      "loss": 0.2007,
      "step": 26675
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37687624322975394,
      "learning_rate": 1.7064763858866617e-06,
      "loss": 0.1134,
      "step": 26676
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6271136452026725,
      "learning_rate": 1.7059222356497763e-06,
      "loss": 0.3129,
      "step": 26677
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.309312272641121,
      "learning_rate": 1.7053681670127576e-06,
      "loss": 0.1771,
      "step": 26678
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5658070498638037,
      "learning_rate": 1.704814179981057e-06,
      "loss": 0.2986,
      "step": 26679
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9353866757598137,
      "learning_rate": 1.70426027456012e-06,
      "loss": 0.3692,
      "step": 26680
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1811263401141001,
      "learning_rate": 1.7037064507554058e-06,
      "loss": 0.574,
      "step": 26681
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5906421542373955,
      "learning_rate": 1.7031527085723542e-06,
      "loss": 0.0705,
      "step": 26682
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3487216111001488,
      "learning_rate": 1.7025990480164157e-06,
      "loss": 0.262,
      "step": 26683
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32939856820669816,
      "learning_rate": 1.7020454690930411e-06,
      "loss": 0.2315,
      "step": 26684
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7846754727172967,
      "learning_rate": 1.7014919718076717e-06,
      "loss": 0.5073,
      "step": 26685
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40149000540195673,
      "learning_rate": 1.7009385561657544e-06,
      "loss": 0.1778,
      "step": 26686
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6168545428448413,
      "learning_rate": 1.7003852221727335e-06,
      "loss": 0.2623,
      "step": 26687
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37080834503689475,
      "learning_rate": 1.699831969834057e-06,
      "loss": 0.2119,
      "step": 26688
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.183579264289712,
      "learning_rate": 1.6992787991551619e-06,
      "loss": 0.4464,
      "step": 26689
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33710475663592876,
      "learning_rate": 1.6987257101414933e-06,
      "loss": 0.2868,
      "step": 26690
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.22529689147178436,
      "learning_rate": 1.6981727027984952e-06,
      "loss": 0.0711,
      "step": 26691
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33975047642661543,
      "learning_rate": 1.6976197771316005e-06,
      "loss": 0.2428,
      "step": 26692
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4117515673406844,
      "learning_rate": 1.6970669331462598e-06,
      "loss": 0.2066,
      "step": 26693
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9166575178800205,
      "learning_rate": 1.6965141708479015e-06,
      "loss": 0.3648,
      "step": 26694
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40417672340735056,
      "learning_rate": 1.695961490241974e-06,
      "loss": 0.229,
      "step": 26695
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38471661957154224,
      "learning_rate": 1.6954088913339072e-06,
      "loss": 0.2431,
      "step": 26696
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4852872122966337,
      "learning_rate": 1.6948563741291401e-06,
      "loss": 0.2243,
      "step": 26697
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4653967917479854,
      "learning_rate": 1.6943039386331117e-06,
      "loss": 0.2811,
      "step": 26698
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24514486027167057,
      "learning_rate": 1.6937515848512509e-06,
      "loss": 0.1199,
      "step": 26699
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0067509843008757,
      "learning_rate": 1.6931993127889957e-06,
      "loss": 0.4458,
      "step": 26700
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3659116178956383,
      "learning_rate": 1.6926471224517783e-06,
      "loss": 0.1607,
      "step": 26701
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28179678269542996,
      "learning_rate": 1.6920950138450353e-06,
      "loss": 0.2177,
      "step": 26702
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7287807470105717,
      "learning_rate": 1.6915429869741918e-06,
      "loss": 0.3841,
      "step": 26703
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6995868310646162,
      "learning_rate": 1.6909910418446828e-06,
      "loss": 0.24,
      "step": 26704
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9081221018865038,
      "learning_rate": 1.6904391784619412e-06,
      "loss": 0.461,
      "step": 26705
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3046771535633535,
      "learning_rate": 1.6898873968313867e-06,
      "loss": 0.1968,
      "step": 26706
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.380344032688824,
      "learning_rate": 1.6893356969584607e-06,
      "loss": 0.7932,
      "step": 26707
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24201376983726663,
      "learning_rate": 1.688784078848581e-06,
      "loss": 0.1779,
      "step": 26708
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4763431571678133,
      "learning_rate": 1.6882325425071789e-06,
      "loss": 0.2452,
      "step": 26709
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3492292738391243,
      "learning_rate": 1.6876810879396821e-06,
      "loss": 0.1514,
      "step": 26710
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.536619292459475,
      "learning_rate": 1.687129715151512e-06,
      "loss": 0.325,
      "step": 26711
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.716478328422176,
      "learning_rate": 1.6865784241480954e-06,
      "loss": 0.2603,
      "step": 26712
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5043918691521713,
      "learning_rate": 1.6860272149348545e-06,
      "loss": 0.3157,
      "step": 26713
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3229772481876654,
      "learning_rate": 1.6854760875172171e-06,
      "loss": 0.1627,
      "step": 26714
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32418679287314794,
      "learning_rate": 1.684925041900598e-06,
      "loss": 0.2087,
      "step": 26715
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.805016798569163,
      "learning_rate": 1.6843740780904238e-06,
      "loss": 0.7962,
      "step": 26716
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.29023311480818936,
      "learning_rate": 1.6838231960921148e-06,
      "loss": 0.1234,
      "step": 26717
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6482193901507172,
      "learning_rate": 1.6832723959110853e-06,
      "loss": 0.2049,
      "step": 26718
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28125478936148335,
      "learning_rate": 1.6827216775527633e-06,
      "loss": 0.1732,
      "step": 26719
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4987575700384949,
      "learning_rate": 1.68217104102256e-06,
      "loss": 0.3102,
      "step": 26720
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44685805899352393,
      "learning_rate": 1.6816204863258956e-06,
      "loss": 0.2368,
      "step": 26721
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8751150496960688,
      "learning_rate": 1.6810700134681867e-06,
      "loss": 0.4474,
      "step": 26722
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3775262334301097,
      "learning_rate": 1.6805196224548447e-06,
      "loss": 0.115,
      "step": 26723
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5675862404263032,
      "learning_rate": 1.6799693132912931e-06,
      "loss": 0.3423,
      "step": 26724
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41654281664023346,
      "learning_rate": 1.6794190859829373e-06,
      "loss": 0.2401,
      "step": 26725
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36141797806349557,
      "learning_rate": 1.6788689405351953e-06,
      "loss": 0.2734,
      "step": 26726
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.15600163934096542,
      "learning_rate": 1.6783188769534787e-06,
      "loss": 0.0683,
      "step": 26727
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3777095918084654,
      "learning_rate": 1.6777688952432014e-06,
      "loss": 0.1595,
      "step": 26728
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4603131239615157,
      "learning_rate": 1.6772189954097706e-06,
      "loss": 0.3026,
      "step": 26729
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6591510560599055,
      "learning_rate": 1.6766691774585964e-06,
      "loss": 0.2128,
      "step": 26730
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3300084432856335,
      "learning_rate": 1.6761194413950933e-06,
      "loss": 0.2581,
      "step": 26731
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3261293165692938,
      "learning_rate": 1.6755697872246601e-06,
      "loss": 0.0659,
      "step": 26732
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3447355522883161,
      "learning_rate": 1.675020214952716e-06,
      "loss": 0.2594,
      "step": 26733
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9203735856686666,
      "learning_rate": 1.6744707245846591e-06,
      "loss": 0.5544,
      "step": 26734
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3934014200433724,
      "learning_rate": 1.6739213161258994e-06,
      "loss": 0.6468,
      "step": 26735
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2868405356179388,
      "learning_rate": 1.6733719895818435e-06,
      "loss": 0.1166,
      "step": 26736
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.360033984413277,
      "learning_rate": 1.6728227449578916e-06,
      "loss": 0.2693,
      "step": 26737
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4054801055568497,
      "learning_rate": 1.6722735822594493e-06,
      "loss": 0.2076,
      "step": 26738
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7129022606990905,
      "learning_rate": 1.6717245014919204e-06,
      "loss": 0.3955,
      "step": 26739
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3652479245272588,
      "learning_rate": 1.671175502660708e-06,
      "loss": 0.0764,
      "step": 26740
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.47842546421998183,
      "learning_rate": 1.670626585771209e-06,
      "loss": 0.1929,
      "step": 26741
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.359325189984102,
      "learning_rate": 1.670077750828827e-06,
      "loss": 0.2367,
      "step": 26742
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9362443020459471,
      "learning_rate": 1.6695289978389606e-06,
      "loss": 0.4149,
      "step": 26743
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3471604282272237,
      "learning_rate": 1.6689803268070081e-06,
      "loss": 0.2814,
      "step": 26744
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3983767491783497,
      "learning_rate": 1.6684317377383718e-06,
      "loss": 0.1582,
      "step": 26745
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42913321826510864,
      "learning_rate": 1.667883230638443e-06,
      "loss": 0.2264,
      "step": 26746
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5140178274125287,
      "learning_rate": 1.6673348055126204e-06,
      "loss": 0.215,
      "step": 26747
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4589396822353951,
      "learning_rate": 1.6667864623663033e-06,
      "loss": 0.242,
      "step": 26748
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3041210950738419,
      "learning_rate": 1.6662382012048772e-06,
      "loss": 0.2112,
      "step": 26749
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9015719455051687,
      "learning_rate": 1.6656900220337469e-06,
      "loss": 0.4025,
      "step": 26750
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36620933112973447,
      "learning_rate": 1.6651419248582978e-06,
      "loss": 0.1684,
      "step": 26751
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6234805313571613,
      "learning_rate": 1.6645939096839248e-06,
      "loss": 0.3638,
      "step": 26752
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8856714182783344,
      "learning_rate": 1.6640459765160222e-06,
      "loss": 0.2309,
      "step": 26753
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32805998175724027,
      "learning_rate": 1.6634981253599758e-06,
      "loss": 0.2362,
      "step": 26754
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0314749952861653,
      "learning_rate": 1.662950356221178e-06,
      "loss": 0.5262,
      "step": 26755
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33382027418203913,
      "learning_rate": 1.6624026691050177e-06,
      "loss": 0.1978,
      "step": 26756
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3283462881790193,
      "learning_rate": 1.6618550640168862e-06,
      "loss": 0.1405,
      "step": 26757
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6129671546264424,
      "learning_rate": 1.6613075409621637e-06,
      "loss": 0.227,
      "step": 26758
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.4562648491753725,
      "learning_rate": 1.6607600999462447e-06,
      "loss": 0.62,
      "step": 26759
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28639244331390695,
      "learning_rate": 1.6602127409745127e-06,
      "loss": 0.156,
      "step": 26760
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5288290891676038,
      "learning_rate": 1.659665464052348e-06,
      "loss": 0.38,
      "step": 26761
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4252659382119422,
      "learning_rate": 1.6591182691851426e-06,
      "loss": 0.2094,
      "step": 26762
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7984843186280365,
      "learning_rate": 1.6585711563782736e-06,
      "loss": 0.3653,
      "step": 26763
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5049895737356025,
      "learning_rate": 1.6580241256371265e-06,
      "loss": 0.2028,
      "step": 26764
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44650937708050603,
      "learning_rate": 1.6574771769670827e-06,
      "loss": 0.2719,
      "step": 26765
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1609137609725852,
      "learning_rate": 1.6569303103735267e-06,
      "loss": 0.0681,
      "step": 26766
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3246169844759169,
      "learning_rate": 1.6563835258618334e-06,
      "loss": 0.222,
      "step": 26767
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.1993514727167187,
      "learning_rate": 1.6558368234373834e-06,
      "loss": 0.483,
      "step": 26768
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.27754880460489834,
      "learning_rate": 1.6552902031055562e-06,
      "loss": 0.1683,
      "step": 26769
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3474014181731921,
      "learning_rate": 1.6547436648717297e-06,
      "loss": 0.7485,
      "step": 26770
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7842053682579039,
      "learning_rate": 1.6541972087412839e-06,
      "loss": 0.2539,
      "step": 26771
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4447086372939733,
      "learning_rate": 1.6536508347195902e-06,
      "loss": 0.3062,
      "step": 26772
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.306960969818175,
      "learning_rate": 1.6531045428120251e-06,
      "loss": 0.1715,
      "step": 26773
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.589718970544211,
      "learning_rate": 1.6525583330239657e-06,
      "loss": 0.3187,
      "step": 26774
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3390206666223692,
      "learning_rate": 1.6520122053607811e-06,
      "loss": 0.1401,
      "step": 26775
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5453628892724968,
      "learning_rate": 1.6514661598278503e-06,
      "loss": 0.2359,
      "step": 26776
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.48043616958408564,
      "learning_rate": 1.650920196430541e-06,
      "loss": 0.2508,
      "step": 26777
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9909599702832336,
      "learning_rate": 1.6503743151742258e-06,
      "loss": 0.3796,
      "step": 26778
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28604879148863377,
      "learning_rate": 1.6498285160642792e-06,
      "loss": 0.2217,
      "step": 26779
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42792715575134843,
      "learning_rate": 1.6492827991060633e-06,
      "loss": 0.2513,
      "step": 26780
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9771669628133838,
      "learning_rate": 1.648737164304951e-06,
      "loss": 0.4438,
      "step": 26781
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4589672713942384,
      "learning_rate": 1.648191611666311e-06,
      "loss": 0.0696,
      "step": 26782
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3394932533352812,
      "learning_rate": 1.6476461411955114e-06,
      "loss": 0.2301,
      "step": 26783
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17139268319275183,
      "learning_rate": 1.6471007528979155e-06,
      "loss": 0.0689,
      "step": 26784
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3357997211403887,
      "learning_rate": 1.6465554467788913e-06,
      "loss": 0.2534,
      "step": 26785
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.116140412011839,
      "learning_rate": 1.6460102228438058e-06,
      "loss": 0.4341,
      "step": 26786
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6481023731183148,
      "learning_rate": 1.6454650810980155e-06,
      "loss": 0.366,
      "step": 26787
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38358643814999127,
      "learning_rate": 1.644920021546893e-06,
      "loss": 0.2317,
      "step": 26788
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6453363070942743,
      "learning_rate": 1.644375044195795e-06,
      "loss": 0.3353,
      "step": 26789
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.534443960678125,
      "learning_rate": 1.643830149050084e-06,
      "loss": 0.1136,
      "step": 26790
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3224771986588755,
      "learning_rate": 1.6432853361151236e-06,
      "loss": 0.267,
      "step": 26791
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3179465034899848,
      "learning_rate": 1.6427406053962702e-06,
      "loss": 0.1818,
      "step": 26792
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4387782699263427,
      "learning_rate": 1.6421959568988844e-06,
      "loss": 0.0887,
      "step": 26793
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.442863274770173,
      "learning_rate": 1.641651390628325e-06,
      "loss": 0.1933,
      "step": 26794
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8271757944164432,
      "learning_rate": 1.6411069065899488e-06,
      "loss": 0.3353,
      "step": 26795
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.42806359046068865,
      "learning_rate": 1.6405625047891138e-06,
      "loss": 0.2617,
      "step": 26796
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2669639387894593,
      "learning_rate": 1.6400181852311781e-06,
      "loss": 0.2123,
      "step": 26797
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7397634611215667,
      "learning_rate": 1.6394739479214916e-06,
      "loss": 0.3831,
      "step": 26798
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8042717868203897,
      "learning_rate": 1.6389297928654114e-06,
      "loss": 0.3222,
      "step": 26799
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9557058162002626,
      "learning_rate": 1.6383857200682928e-06,
      "loss": 0.4327,
      "step": 26800
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4288814511906831,
      "learning_rate": 1.637841729535483e-06,
      "loss": 0.172,
      "step": 26801
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.51253138862576,
      "learning_rate": 1.6372978212723423e-06,
      "loss": 0.2591,
      "step": 26802
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.20138301116355334,
      "learning_rate": 1.636753995284215e-06,
      "loss": 0.1593,
      "step": 26803
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8657889344959182,
      "learning_rate": 1.6362102515764533e-06,
      "loss": 0.3773,
      "step": 26804
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.735480012210479,
      "learning_rate": 1.63566659015441e-06,
      "loss": 0.1957,
      "step": 26805
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3698292621381556,
      "learning_rate": 1.6351230110234295e-06,
      "loss": 0.207,
      "step": 26806
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9802476945667703,
      "learning_rate": 1.6345795141888598e-06,
      "loss": 0.4818,
      "step": 26807
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.31916490088000493,
      "learning_rate": 1.63403609965605e-06,
      "loss": 0.2322,
      "step": 26808
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6323579004439368,
      "learning_rate": 1.633492767430348e-06,
      "loss": 0.8496,
      "step": 26809
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.27545037745975604,
      "learning_rate": 1.632949517517095e-06,
      "loss": 0.1669,
      "step": 26810
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.476390616096454,
      "learning_rate": 1.632406349921638e-06,
      "loss": 0.5,
      "step": 26811
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.15769268522334726,
      "learning_rate": 1.6318632646493227e-06,
      "loss": 0.0833,
      "step": 26812
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.231866053185639,
      "learning_rate": 1.6313202617054847e-06,
      "loss": 0.3949,
      "step": 26813
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38555182105442803,
      "learning_rate": 1.6307773410954775e-06,
      "loss": 0.1991,
      "step": 26814
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.29634353426752824,
      "learning_rate": 1.6302345028246337e-06,
      "loss": 0.2461,
      "step": 26815
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8021252296968585,
      "learning_rate": 1.6296917468982964e-06,
      "loss": 0.3116,
      "step": 26816
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6383475026238734,
      "learning_rate": 1.6291490733218086e-06,
      "loss": 0.6692,
      "step": 26817
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2346762841522039,
      "learning_rate": 1.6286064821005042e-06,
      "loss": 0.049,
      "step": 26818
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2886431087688076,
      "learning_rate": 1.628063973239724e-06,
      "loss": 0.1732,
      "step": 26819
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5672511395085083,
      "learning_rate": 1.6275215467448057e-06,
      "loss": 0.2968,
      "step": 26820
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2537303611166491,
      "learning_rate": 1.626979202621084e-06,
      "loss": 0.1411,
      "step": 26821
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4222462245954535,
      "learning_rate": 1.6264369408739e-06,
      "loss": 0.2346,
      "step": 26822
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3708499972470842,
      "learning_rate": 1.6258947615085807e-06,
      "loss": 0.1868,
      "step": 26823
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.49314243045007317,
      "learning_rate": 1.6253526645304651e-06,
      "loss": 0.3047,
      "step": 26824
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.2044873145354207,
      "learning_rate": 1.6248106499448857e-06,
      "loss": 0.5337,
      "step": 26825
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.47609631127162966,
      "learning_rate": 1.624268717757177e-06,
      "loss": 0.3426,
      "step": 26826
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4540348807410842,
      "learning_rate": 1.6237268679726636e-06,
      "loss": 0.1929,
      "step": 26827
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.50674290807412,
      "learning_rate": 1.623185100596687e-06,
      "loss": 0.2911,
      "step": 26828
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4551594527553436,
      "learning_rate": 1.6226434156345683e-06,
      "loss": 0.1921,
      "step": 26829
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.681878132435932,
      "learning_rate": 1.6221018130916411e-06,
      "loss": 0.4957,
      "step": 26830
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38433681207266135,
      "learning_rate": 1.6215602929732343e-06,
      "loss": 0.1883,
      "step": 26831
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.353029824837592,
      "learning_rate": 1.6210188552846729e-06,
      "loss": 0.1981,
      "step": 26832
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.363965574027951,
      "learning_rate": 1.6204775000312844e-06,
      "loss": 0.2562,
      "step": 26833
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.38221034310270297,
      "learning_rate": 1.619936227218396e-06,
      "loss": 0.2465,
      "step": 26834
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4660047164250201,
      "learning_rate": 1.6193950368513345e-06,
      "loss": 0.2764,
      "step": 26835
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.624718227006554,
      "learning_rate": 1.6188539289354187e-06,
      "loss": 0.1243,
      "step": 26836
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3991484905616866,
      "learning_rate": 1.618312903475977e-06,
      "loss": 0.2583,
      "step": 26837
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35373913547670366,
      "learning_rate": 1.6177719604783338e-06,
      "loss": 0.1986,
      "step": 26838
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4554641828764651,
      "learning_rate": 1.6172310999478026e-06,
      "loss": 0.3213,
      "step": 26839
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7311442042732421,
      "learning_rate": 1.6166903218897146e-06,
      "loss": 0.2456,
      "step": 26840
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5099511014968197,
      "learning_rate": 1.6161496263093833e-06,
      "loss": 0.2157,
      "step": 26841
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3121179685980378,
      "learning_rate": 1.6156090132121316e-06,
      "loss": 0.1874,
      "step": 26842
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.430440075301147,
      "learning_rate": 1.6150684826032791e-06,
      "loss": 0.8257,
      "step": 26843
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2683464644677809,
      "learning_rate": 1.6145280344881398e-06,
      "loss": 0.1584,
      "step": 26844
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44943293787700334,
      "learning_rate": 1.6139876688720324e-06,
      "loss": 0.2903,
      "step": 26845
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37515073510108915,
      "learning_rate": 1.613447385760274e-06,
      "loss": 0.1613,
      "step": 26846
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4594334515192353,
      "learning_rate": 1.612907185158179e-06,
      "loss": 0.2161,
      "step": 26847
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9916326525940891,
      "learning_rate": 1.6123670670710655e-06,
      "loss": 0.4647,
      "step": 26848
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.838158336747263,
      "learning_rate": 1.6118270315042428e-06,
      "loss": 0.2517,
      "step": 26849
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3172854253953707,
      "learning_rate": 1.6112870784630253e-06,
      "loss": 0.2747,
      "step": 26850
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3053263266605509,
      "learning_rate": 1.6107472079527265e-06,
      "loss": 0.1728,
      "step": 26851
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6066112840003635,
      "learning_rate": 1.6102074199786589e-06,
      "loss": 0.7822,
      "step": 26852
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.17825550197247952,
      "learning_rate": 1.6096677145461292e-06,
      "loss": 0.1032,
      "step": 26853
    },
    {
      "epoch": 0.82,
      "grad_norm": 2.2079598700771155,
      "learning_rate": 1.609128091660449e-06,
      "loss": 0.5813,
      "step": 26854
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32122213144109985,
      "learning_rate": 1.6085885513269306e-06,
      "loss": 0.1543,
      "step": 26855
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4796030350350796,
      "learning_rate": 1.6080490935508742e-06,
      "loss": 0.2951,
      "step": 26856
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33097024294837163,
      "learning_rate": 1.6075097183375966e-06,
      "loss": 0.2187,
      "step": 26857
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8481322529033776,
      "learning_rate": 1.606970425692398e-06,
      "loss": 0.3775,
      "step": 26858
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.41415887302554844,
      "learning_rate": 1.6064312156205853e-06,
      "loss": 0.1361,
      "step": 26859
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.31628437084566663,
      "learning_rate": 1.6058920881274664e-06,
      "loss": 0.1881,
      "step": 26860
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.3693275883546239,
      "learning_rate": 1.6053530432183417e-06,
      "loss": 0.7777,
      "step": 26861
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.235725093371099,
      "learning_rate": 1.6048140808985147e-06,
      "loss": 0.1842,
      "step": 26862
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4682059733840669,
      "learning_rate": 1.6042752011732908e-06,
      "loss": 0.2257,
      "step": 26863
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3247047396103301,
      "learning_rate": 1.6037364040479708e-06,
      "loss": 0.0693,
      "step": 26864
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40995351194535873,
      "learning_rate": 1.6031976895278512e-06,
      "loss": 0.2774,
      "step": 26865
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6900195021807426,
      "learning_rate": 1.60265905761824e-06,
      "loss": 0.2606,
      "step": 26866
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.9973455490466445,
      "learning_rate": 1.6021205083244297e-06,
      "loss": 0.6432,
      "step": 26867
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.28533802520884927,
      "learning_rate": 1.6015820416517214e-06,
      "loss": 0.2093,
      "step": 26868
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3593677635197421,
      "learning_rate": 1.6010436576054145e-06,
      "loss": 0.254,
      "step": 26869
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0420010390352417,
      "learning_rate": 1.6005053561908013e-06,
      "loss": 0.5765,
      "step": 26870
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.559616175492297,
      "learning_rate": 1.5999671374131798e-06,
      "loss": 0.5011,
      "step": 26871
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3272111603027943,
      "learning_rate": 1.5994290012778458e-06,
      "loss": 0.1196,
      "step": 26872
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35884193856200036,
      "learning_rate": 1.5988909477900938e-06,
      "loss": 0.1795,
      "step": 26873
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34494601647146433,
      "learning_rate": 1.5983529769552197e-06,
      "loss": 0.276,
      "step": 26874
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.406814145758869,
      "learning_rate": 1.5978150887785104e-06,
      "loss": 0.2345,
      "step": 26875
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.087504236817277,
      "learning_rate": 1.5972772832652617e-06,
      "loss": 0.3905,
      "step": 26876
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4645288012474988,
      "learning_rate": 1.5967395604207626e-06,
      "loss": 0.1998,
      "step": 26877
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.37991169118694906,
      "learning_rate": 1.5962019202503087e-06,
      "loss": 0.2779,
      "step": 26878
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.0187318917456591,
      "learning_rate": 1.5956643627591816e-06,
      "loss": 0.4458,
      "step": 26879
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3343052184095898,
      "learning_rate": 1.595126887952675e-06,
      "loss": 0.2852,
      "step": 26880
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5625334044857285,
      "learning_rate": 1.5945894958360785e-06,
      "loss": 0.1728,
      "step": 26881
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5390574564665345,
      "learning_rate": 1.5940521864146708e-06,
      "loss": 0.2998,
      "step": 26882
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.35175044602028843,
      "learning_rate": 1.5935149596937482e-06,
      "loss": 0.2011,
      "step": 26883
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.44325210674929355,
      "learning_rate": 1.5929778156785891e-06,
      "loss": 0.2498,
      "step": 26884
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.46633073616195914,
      "learning_rate": 1.5924407543744803e-06,
      "loss": 0.2047,
      "step": 26885
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.39461109436681036,
      "learning_rate": 1.5919037757867083e-06,
      "loss": 0.2063,
      "step": 26886
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3883126650747257,
      "learning_rate": 1.5913668799205518e-06,
      "loss": 0.2772,
      "step": 26887
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.45151982546702374,
      "learning_rate": 1.5908300667812948e-06,
      "loss": 0.2586,
      "step": 26888
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4864217861139431,
      "learning_rate": 1.590293336374218e-06,
      "loss": 0.2029,
      "step": 26889
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8979013314640473,
      "learning_rate": 1.5897566887046056e-06,
      "loss": 0.2789,
      "step": 26890
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6305080069392367,
      "learning_rate": 1.5892201237777304e-06,
      "loss": 0.3114,
      "step": 26891
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2461591074631677,
      "learning_rate": 1.588683641598877e-06,
      "loss": 0.1915,
      "step": 26892
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.31843423438486673,
      "learning_rate": 1.5881472421733224e-06,
      "loss": 0.2198,
      "step": 26893
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.728762883537581,
      "learning_rate": 1.58761092550634e-06,
      "loss": 0.1347,
      "step": 26894
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.6100210966545416,
      "learning_rate": 1.5870746916032132e-06,
      "loss": 0.7106,
      "step": 26895
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3522951730838699,
      "learning_rate": 1.586538540469208e-06,
      "loss": 0.1707,
      "step": 26896
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.5266645653733815,
      "learning_rate": 1.5860024721096112e-06,
      "loss": 0.3204,
      "step": 26897
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.33798899271277416,
      "learning_rate": 1.5854664865296877e-06,
      "loss": 0.2237,
      "step": 26898
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6147847656350437,
      "learning_rate": 1.584930583734714e-06,
      "loss": 0.2712,
      "step": 26899
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6228403042975078,
      "learning_rate": 1.5843947637299639e-06,
      "loss": 0.3196,
      "step": 26900
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.2765998627638315,
      "learning_rate": 1.5838590265207055e-06,
      "loss": 0.168,
      "step": 26901
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4580188718284779,
      "learning_rate": 1.583323372112211e-06,
      "loss": 0.2365,
      "step": 26902
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4420767321503632,
      "learning_rate": 1.5827878005097507e-06,
      "loss": 0.1946,
      "step": 26903
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4843994338604135,
      "learning_rate": 1.5822523117185963e-06,
      "loss": 0.3369,
      "step": 26904
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.27152075638068485,
      "learning_rate": 1.5817169057440107e-06,
      "loss": 0.1783,
      "step": 26905
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7206427921024836,
      "learning_rate": 1.5811815825912647e-06,
      "loss": 0.7845,
      "step": 26906
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7909700689322046,
      "learning_rate": 1.580646342265626e-06,
      "loss": 0.2867,
      "step": 26907
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.7634687822146814,
      "learning_rate": 1.580111184772355e-06,
      "loss": 0.3842,
      "step": 26908
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34112563728242057,
      "learning_rate": 1.579576110116725e-06,
      "loss": 0.1627,
      "step": 26909
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4885204817248927,
      "learning_rate": 1.5790411183039933e-06,
      "loss": 0.2799,
      "step": 26910
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24950535814182678,
      "learning_rate": 1.5785062093394253e-06,
      "loss": 0.1635,
      "step": 26911
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.24513833976450589,
      "learning_rate": 1.5779713832282883e-06,
      "loss": 0.0949,
      "step": 26912
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5415440423256284,
      "learning_rate": 1.5774366399758368e-06,
      "loss": 0.8463,
      "step": 26913
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.340136260527511,
      "learning_rate": 1.5769019795873352e-06,
      "loss": 0.1123,
      "step": 26914
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.43980210799306246,
      "learning_rate": 1.576367402068043e-06,
      "loss": 0.2641,
      "step": 26915
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.32219328969778727,
      "learning_rate": 1.5758329074232226e-06,
      "loss": 0.2427,
      "step": 26916
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.735563721998058,
      "learning_rate": 1.5752984956581286e-06,
      "loss": 0.3833,
      "step": 26917
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3094149695159208,
      "learning_rate": 1.574764166778019e-06,
      "loss": 0.0643,
      "step": 26918
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.36493265741103925,
      "learning_rate": 1.574229920788154e-06,
      "loss": 0.2617,
      "step": 26919
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.23579595956409788,
      "learning_rate": 1.5736957576937827e-06,
      "loss": 0.0706,
      "step": 26920
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4819905121319713,
      "learning_rate": 1.5731616775001713e-06,
      "loss": 0.2894,
      "step": 26921
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4741345085073635,
      "learning_rate": 1.572627680212565e-06,
      "loss": 0.271,
      "step": 26922
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.6080677281724736,
      "learning_rate": 1.57209376583622e-06,
      "loss": 0.3206,
      "step": 26923
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3908213164341773,
      "learning_rate": 1.571559934376392e-06,
      "loss": 0.2275,
      "step": 26924
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8753181658684276,
      "learning_rate": 1.5710261858383268e-06,
      "loss": 0.275,
      "step": 26925
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8997145162762409,
      "learning_rate": 1.5704925202272835e-06,
      "loss": 0.3992,
      "step": 26926
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3119154787511764,
      "learning_rate": 1.569958937548507e-06,
      "loss": 0.1989,
      "step": 26927
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3774640104137552,
      "learning_rate": 1.5694254378072472e-06,
      "loss": 0.2492,
      "step": 26928
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.5432511571271983,
      "learning_rate": 1.5688920210087544e-06,
      "loss": 0.0917,
      "step": 26929
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.3159150265832772,
      "learning_rate": 1.568358687158279e-06,
      "loss": 0.1831,
      "step": 26930
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9300785781154358,
      "learning_rate": 1.5678254362610623e-06,
      "loss": 0.366,
      "step": 26931
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4468348671685472,
      "learning_rate": 1.5672922683223536e-06,
      "loss": 0.2402,
      "step": 26932
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.4868108063796223,
      "learning_rate": 1.5667591833474004e-06,
      "loss": 0.2159,
      "step": 26933
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.34723756397989736,
      "learning_rate": 1.5662261813414404e-06,
      "loss": 0.2841,
      "step": 26934
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.9500770973206795,
      "learning_rate": 1.5656932623097276e-06,
      "loss": 0.2727,
      "step": 26935
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.8954495777146225,
      "learning_rate": 1.565160426257497e-06,
      "loss": 0.4052,
      "step": 26936
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.40947558289456537,
      "learning_rate": 1.5646276731899945e-06,
      "loss": 0.1698,
      "step": 26937
    },
    {
      "epoch": 0.82,
      "grad_norm": 0.1898836704750175,
      "learning_rate": 1.5640950031124614e-06,
      "loss": 0.1134,
      "step": 26938
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3561468041457318,
      "learning_rate": 1.5635624160301366e-06,
      "loss": 0.265,
      "step": 26939
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9522636504801197,
      "learning_rate": 1.5630299119482594e-06,
      "loss": 0.5027,
      "step": 26940
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7645990104969719,
      "learning_rate": 1.5624974908720703e-06,
      "loss": 0.355,
      "step": 26941
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3079089727061563,
      "learning_rate": 1.5619651528068104e-06,
      "loss": 0.2044,
      "step": 26942
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7549842594611824,
      "learning_rate": 1.561432897757711e-06,
      "loss": 0.4033,
      "step": 26943
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4916584201602767,
      "learning_rate": 1.5609007257300113e-06,
      "loss": 0.1216,
      "step": 26944
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.357890387138622,
      "learning_rate": 1.5603686367289494e-06,
      "loss": 0.2844,
      "step": 26945
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.29614297178951055,
      "learning_rate": 1.5598366307597534e-06,
      "loss": 0.1484,
      "step": 26946
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.58485508113778,
      "learning_rate": 1.5593047078276658e-06,
      "loss": 0.3321,
      "step": 26947
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1458344414664928,
      "learning_rate": 1.5587728679379144e-06,
      "loss": 0.0825,
      "step": 26948
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9063722473374759,
      "learning_rate": 1.558241111095733e-06,
      "loss": 0.4128,
      "step": 26949
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5190434372661022,
      "learning_rate": 1.557709437306355e-06,
      "loss": 0.1795,
      "step": 26950
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.27910516590563605,
      "learning_rate": 1.5571778465750043e-06,
      "loss": 0.2049,
      "step": 26951
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48951605868249126,
      "learning_rate": 1.556646338906922e-06,
      "loss": 0.3007,
      "step": 26952
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9179415630332967,
      "learning_rate": 1.5561149143073284e-06,
      "loss": 0.42,
      "step": 26953
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1694351600765678,
      "learning_rate": 1.5555835727814539e-06,
      "loss": 0.4241,
      "step": 26954
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2889027962883315,
      "learning_rate": 1.5550523143345297e-06,
      "loss": 0.1758,
      "step": 26955
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4660987187297045,
      "learning_rate": 1.5545211389717784e-06,
      "loss": 0.4215,
      "step": 26956
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24401570288979432,
      "learning_rate": 1.5539900466984258e-06,
      "loss": 0.1689,
      "step": 26957
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7801681127144375,
      "learning_rate": 1.553459037519698e-06,
      "loss": 0.4173,
      "step": 26958
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6131379995974121,
      "learning_rate": 1.5529281114408224e-06,
      "loss": 0.1902,
      "step": 26959
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5954444071260669,
      "learning_rate": 1.5523972684670175e-06,
      "loss": 0.3402,
      "step": 26960
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42861698253335717,
      "learning_rate": 1.5518665086035067e-06,
      "loss": 0.247,
      "step": 26961
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4174516707324093,
      "learning_rate": 1.5513358318555161e-06,
      "loss": 0.205,
      "step": 26962
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2750341357248881,
      "learning_rate": 1.550805238228259e-06,
      "loss": 0.1979,
      "step": 26963
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3778213102814624,
      "learning_rate": 1.5502747277269647e-06,
      "loss": 0.1719,
      "step": 26964
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5145295502810163,
      "learning_rate": 1.5497443003568458e-06,
      "loss": 0.2979,
      "step": 26965
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3476256378803885,
      "learning_rate": 1.5492139561231223e-06,
      "loss": 0.152,
      "step": 26966
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7425547788475831,
      "learning_rate": 1.5486836950310136e-06,
      "loss": 0.3552,
      "step": 26967
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41159383889250634,
      "learning_rate": 1.5481535170857366e-06,
      "loss": 0.1383,
      "step": 26968
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3398130473407757,
      "learning_rate": 1.5476234222925047e-06,
      "loss": 0.2473,
      "step": 26969
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3185234571456291,
      "learning_rate": 1.547093410656535e-06,
      "loss": 0.2343,
      "step": 26970
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5228869884279455,
      "learning_rate": 1.5465634821830432e-06,
      "loss": 0.2463,
      "step": 26971
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7851477373170078,
      "learning_rate": 1.5460336368772377e-06,
      "loss": 0.034,
      "step": 26972
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5995393313144889,
      "learning_rate": 1.5455038747443397e-06,
      "loss": 0.2312,
      "step": 26973
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3876066772528521,
      "learning_rate": 1.5449741957895536e-06,
      "loss": 0.2013,
      "step": 26974
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.46742862139553143,
      "learning_rate": 1.5444446000180935e-06,
      "loss": 0.3164,
      "step": 26975
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4355149551018897,
      "learning_rate": 1.5439150874351716e-06,
      "loss": 0.246,
      "step": 26976
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48081799445160717,
      "learning_rate": 1.5433856580459915e-06,
      "loss": 0.1738,
      "step": 26977
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3778371857524089,
      "learning_rate": 1.5428563118557704e-06,
      "loss": 0.266,
      "step": 26978
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2142907775807616,
      "learning_rate": 1.5423270488697096e-06,
      "loss": 0.4404,
      "step": 26979
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2841847126922951,
      "learning_rate": 1.541797869093018e-06,
      "loss": 0.1574,
      "step": 26980
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35082890491039287,
      "learning_rate": 1.541268772530904e-06,
      "loss": 0.2021,
      "step": 26981
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5260227778977404,
      "learning_rate": 1.5407397591885697e-06,
      "loss": 0.2448,
      "step": 26982
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4406919189627471,
      "learning_rate": 1.5402108290712203e-06,
      "loss": 0.1933,
      "step": 26983
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7505350418060652,
      "learning_rate": 1.5396819821840604e-06,
      "loss": 0.3947,
      "step": 26984
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8356695681032392,
      "learning_rate": 1.5391532185322956e-06,
      "loss": 0.2746,
      "step": 26985
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44404118199427284,
      "learning_rate": 1.5386245381211228e-06,
      "loss": 0.3059,
      "step": 26986
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41968315208768053,
      "learning_rate": 1.538095940955746e-06,
      "loss": 0.1708,
      "step": 26987
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3609519733666013,
      "learning_rate": 1.5375674270413687e-06,
      "loss": 0.2828,
      "step": 26988
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.1915852845706421,
      "learning_rate": 1.5370389963831823e-06,
      "loss": 0.0863,
      "step": 26989
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.945231153357941,
      "learning_rate": 1.5365106489863956e-06,
      "loss": 0.1345,
      "step": 26990
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9338090576954996,
      "learning_rate": 1.5359823848561994e-06,
      "loss": 0.3832,
      "step": 26991
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.33445510774292186,
      "learning_rate": 1.5354542039977937e-06,
      "loss": 0.1892,
      "step": 26992
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3508671588851543,
      "learning_rate": 1.5349261064163778e-06,
      "loss": 0.274,
      "step": 26993
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8127947080645243,
      "learning_rate": 1.5343980921171408e-06,
      "loss": 0.2672,
      "step": 26994
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.6091015843075416,
      "learning_rate": 1.533870161105282e-06,
      "loss": 0.7984,
      "step": 26995
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.286839758508019,
      "learning_rate": 1.5333423133859936e-06,
      "loss": 0.1741,
      "step": 26996
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5219037683111145,
      "learning_rate": 1.532814548964472e-06,
      "loss": 0.7207,
      "step": 26997
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18695951797038482,
      "learning_rate": 1.5322868678459024e-06,
      "loss": 0.1493,
      "step": 26998
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.9400848908137331,
      "learning_rate": 1.5317592700354855e-06,
      "loss": 0.7288,
      "step": 26999
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5997604937886802,
      "learning_rate": 1.5312317555384048e-06,
      "loss": 0.0857,
      "step": 27000
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39432215982569674,
      "learning_rate": 1.5307043243598518e-06,
      "loss": 0.2653,
      "step": 27001
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8165404001985839,
      "learning_rate": 1.53017697650502e-06,
      "loss": 0.2722,
      "step": 27002
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3847748871120948,
      "learning_rate": 1.5296497119790887e-06,
      "loss": 0.2185,
      "step": 27003
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35439809577721587,
      "learning_rate": 1.529122530787256e-06,
      "loss": 0.2786,
      "step": 27004
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2967540205248657,
      "learning_rate": 1.5285954329347009e-06,
      "loss": 0.165,
      "step": 27005
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5102105509155637,
      "learning_rate": 1.5280684184266103e-06,
      "loss": 0.7799,
      "step": 27006
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.45377583729483045,
      "learning_rate": 1.5275414872681738e-06,
      "loss": 0.0816,
      "step": 27007
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44020751880709796,
      "learning_rate": 1.527014639464568e-06,
      "loss": 0.2438,
      "step": 27008
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4303837236530104,
      "learning_rate": 1.526487875020981e-06,
      "loss": 0.1855,
      "step": 27009
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5080184373007138,
      "learning_rate": 1.5259611939425957e-06,
      "loss": 0.2808,
      "step": 27010
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3219915778881481,
      "learning_rate": 1.525434596234593e-06,
      "loss": 0.2369,
      "step": 27011
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7269462501375312,
      "learning_rate": 1.5249080819021521e-06,
      "loss": 0.773,
      "step": 27012
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4742444240725955,
      "learning_rate": 1.5243816509504539e-06,
      "loss": 0.253,
      "step": 27013
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9551045362681793,
      "learning_rate": 1.5238553033846793e-06,
      "loss": 0.4718,
      "step": 27014
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34527532577435116,
      "learning_rate": 1.523329039210002e-06,
      "loss": 0.2019,
      "step": 27015
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2225437600895834,
      "learning_rate": 1.5228028584316057e-06,
      "loss": 0.1293,
      "step": 27016
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5913524804474812,
      "learning_rate": 1.522276761054663e-06,
      "loss": 0.3107,
      "step": 27017
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6997177725913919,
      "learning_rate": 1.5217507470843517e-06,
      "loss": 0.2029,
      "step": 27018
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35180391398623073,
      "learning_rate": 1.5212248165258482e-06,
      "loss": 0.2397,
      "step": 27019
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8148255756698827,
      "learning_rate": 1.5206989693843222e-06,
      "loss": 0.2536,
      "step": 27020
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4087217774108556,
      "learning_rate": 1.5201732056649488e-06,
      "loss": 0.7495,
      "step": 27021
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3178792445922645,
      "learning_rate": 1.5196475253729026e-06,
      "loss": 0.2272,
      "step": 27022
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5803325587084034,
      "learning_rate": 1.5191219285133552e-06,
      "loss": 0.2857,
      "step": 27023
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.31714301618667173,
      "learning_rate": 1.5185964150914778e-06,
      "loss": 0.1967,
      "step": 27024
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5111719880854328,
      "learning_rate": 1.5180709851124388e-06,
      "loss": 0.2196,
      "step": 27025
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8013546454262197,
      "learning_rate": 1.5175456385814068e-06,
      "loss": 0.2735,
      "step": 27026
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5967954328221398,
      "learning_rate": 1.5170203755035528e-06,
      "loss": 0.2873,
      "step": 27027
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23567051247809706,
      "learning_rate": 1.5164951958840458e-06,
      "loss": 0.1529,
      "step": 27028
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5156248002796016,
      "learning_rate": 1.5159700997280458e-06,
      "loss": 0.2515,
      "step": 27029
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.797079582624248,
      "learning_rate": 1.5154450870407288e-06,
      "loss": 0.1141,
      "step": 27030
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.956271827816335,
      "learning_rate": 1.5149201578272544e-06,
      "loss": 0.5579,
      "step": 27031
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38502206605930206,
      "learning_rate": 1.5143953120927823e-06,
      "loss": 0.2401,
      "step": 27032
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5175864303691141,
      "learning_rate": 1.5138705498424865e-06,
      "loss": 0.193,
      "step": 27033
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9427428458695769,
      "learning_rate": 1.5133458710815219e-06,
      "loss": 0.365,
      "step": 27034
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3758218871746636,
      "learning_rate": 1.512821275815053e-06,
      "loss": 0.2355,
      "step": 27035
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44356298272524924,
      "learning_rate": 1.512296764048241e-06,
      "loss": 0.2301,
      "step": 27036
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.31824393721837907,
      "learning_rate": 1.511772335786249e-06,
      "loss": 0.1872,
      "step": 27037
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5499064809996389,
      "learning_rate": 1.5112479910342315e-06,
      "loss": 0.2199,
      "step": 27038
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0541480545280852,
      "learning_rate": 1.5107237297973487e-06,
      "loss": 0.2635,
      "step": 27039
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3842160251052982,
      "learning_rate": 1.5101995520807623e-06,
      "loss": 0.2763,
      "step": 27040
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23761689375046913,
      "learning_rate": 1.509675457889621e-06,
      "loss": 0.0675,
      "step": 27041
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39577619723826174,
      "learning_rate": 1.5091514472290913e-06,
      "loss": 0.286,
      "step": 27042
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8614405120619137,
      "learning_rate": 1.508627520104321e-06,
      "loss": 0.2284,
      "step": 27043
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6811708431937504,
      "learning_rate": 1.508103676520467e-06,
      "loss": 0.2852,
      "step": 27044
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4954748608592922,
      "learning_rate": 1.5075799164826866e-06,
      "loss": 0.272,
      "step": 27045
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2619370889708246,
      "learning_rate": 1.5070562399961253e-06,
      "loss": 0.1631,
      "step": 27046
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2871834674627001,
      "learning_rate": 1.5065326470659403e-06,
      "loss": 0.2103,
      "step": 27047
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1850256495215414,
      "learning_rate": 1.5060091376972818e-06,
      "loss": 0.1275,
      "step": 27048
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5483598137111703,
      "learning_rate": 1.5054857118953005e-06,
      "loss": 0.727,
      "step": 27049
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37011169593838644,
      "learning_rate": 1.5049623696651483e-06,
      "loss": 0.1528,
      "step": 27050
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44726640453169975,
      "learning_rate": 1.5044391110119682e-06,
      "loss": 0.2715,
      "step": 27051
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5810204756260998,
      "learning_rate": 1.5039159359409128e-06,
      "loss": 0.2476,
      "step": 27052
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0167680673137427,
      "learning_rate": 1.5033928444571267e-06,
      "loss": 0.4586,
      "step": 27053
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1848131788774623,
      "learning_rate": 1.50286983656576e-06,
      "loss": 0.2444,
      "step": 27054
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.38842662374251974,
      "learning_rate": 1.5023469122719547e-06,
      "loss": 0.2326,
      "step": 27055
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.20292964230642574,
      "learning_rate": 1.501824071580855e-06,
      "loss": 0.1119,
      "step": 27056
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7732474402609315,
      "learning_rate": 1.5013013144976097e-06,
      "loss": 0.2624,
      "step": 27057
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35421022449819956,
      "learning_rate": 1.5007786410273539e-06,
      "loss": 0.293,
      "step": 27058
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.32350894593657037,
      "learning_rate": 1.5002560511752395e-06,
      "loss": 0.161,
      "step": 27059
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5413944952953471,
      "learning_rate": 1.499733544946399e-06,
      "loss": 0.3124,
      "step": 27060
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1346301777948729,
      "learning_rate": 1.4992111223459781e-06,
      "loss": 0.2578,
      "step": 27061
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8927682093865559,
      "learning_rate": 1.4986887833791176e-06,
      "loss": 0.4327,
      "step": 27062
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3976285324255247,
      "learning_rate": 1.4981665280509516e-06,
      "loss": 0.1688,
      "step": 27063
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4442856965891425,
      "learning_rate": 1.4976443563666198e-06,
      "loss": 0.2495,
      "step": 27064
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2568404824947556,
      "learning_rate": 1.4971222683312614e-06,
      "loss": 0.1862,
      "step": 27065
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48944705102809805,
      "learning_rate": 1.4966002639500144e-06,
      "loss": 0.2448,
      "step": 27066
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.119982308419062,
      "learning_rate": 1.4960783432280067e-06,
      "loss": 0.5648,
      "step": 27067
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7756591145369496,
      "learning_rate": 1.4955565061703835e-06,
      "loss": 0.3448,
      "step": 27068
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3039161764987546,
      "learning_rate": 1.4950347527822717e-06,
      "loss": 0.1717,
      "step": 27069
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3772453902250843,
      "learning_rate": 1.494513083068806e-06,
      "loss": 0.2415,
      "step": 27070
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9566005548702334,
      "learning_rate": 1.4939914970351222e-06,
      "loss": 0.4224,
      "step": 27071
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7964315028880965,
      "learning_rate": 1.4934699946863463e-06,
      "loss": 0.0215,
      "step": 27072
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4100336638514654,
      "learning_rate": 1.492948576027612e-06,
      "loss": 0.2604,
      "step": 27073
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22748091844876536,
      "learning_rate": 1.4924272410640473e-06,
      "loss": 0.1181,
      "step": 27074
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5696520073786842,
      "learning_rate": 1.4919059898007838e-06,
      "loss": 0.7243,
      "step": 27075
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2981737973831646,
      "learning_rate": 1.4913848222429517e-06,
      "loss": 0.2095,
      "step": 27076
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7500710418041375,
      "learning_rate": 1.4908637383956715e-06,
      "loss": 0.4053,
      "step": 27077
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36934999814157377,
      "learning_rate": 1.4903427382640746e-06,
      "loss": 0.2068,
      "step": 27078
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8001256616439661,
      "learning_rate": 1.4898218218532868e-06,
      "loss": 0.4329,
      "step": 27079
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7538097788572959,
      "learning_rate": 1.4893009891684329e-06,
      "loss": 0.0391,
      "step": 27080
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.49670036876036333,
      "learning_rate": 1.4887802402146345e-06,
      "loss": 0.2853,
      "step": 27081
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2650067720258177,
      "learning_rate": 1.4882595749970163e-06,
      "loss": 0.1766,
      "step": 27082
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.22291759509976994,
      "learning_rate": 1.4877389935207042e-06,
      "loss": 0.1118,
      "step": 27083
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4280426468733463,
      "learning_rate": 1.487218495790812e-06,
      "loss": 0.6073,
      "step": 27084
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6993371069534188,
      "learning_rate": 1.4866980818124698e-06,
      "loss": 0.3192,
      "step": 27085
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.553884615373672,
      "learning_rate": 1.4861777515907904e-06,
      "loss": 0.2866,
      "step": 27086
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3338645776072748,
      "learning_rate": 1.4856575051308964e-06,
      "loss": 0.2085,
      "step": 27087
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4848415049382222,
      "learning_rate": 1.485137342437908e-06,
      "loss": 0.2697,
      "step": 27088
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.50738619458583,
      "learning_rate": 1.4846172635169366e-06,
      "loss": 0.2293,
      "step": 27089
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.8519604483660645,
      "learning_rate": 1.4840972683731036e-06,
      "loss": 0.6821,
      "step": 27090
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3502485791341083,
      "learning_rate": 1.483577357011523e-06,
      "loss": 0.1621,
      "step": 27091
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5660652200990828,
      "learning_rate": 1.483057529437314e-06,
      "loss": 0.3316,
      "step": 27092
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18561544244427003,
      "learning_rate": 1.4825377856555844e-06,
      "loss": 0.095,
      "step": 27093
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.32026022361340184,
      "learning_rate": 1.4820181256714506e-06,
      "loss": 0.2612,
      "step": 27094
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5763554735483575,
      "learning_rate": 1.481498549490028e-06,
      "loss": 0.2361,
      "step": 27095
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3002618125725782,
      "learning_rate": 1.4809790571164207e-06,
      "loss": 0.18,
      "step": 27096
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8993485328023382,
      "learning_rate": 1.4804596485557487e-06,
      "loss": 0.4843,
      "step": 27097
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1468502758905041,
      "learning_rate": 1.4799403238131161e-06,
      "loss": 0.1102,
      "step": 27098
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.49765490640093246,
      "learning_rate": 1.479421082893634e-06,
      "loss": 0.304,
      "step": 27099
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.327047752487199,
      "learning_rate": 1.4789019258024117e-06,
      "loss": 0.1803,
      "step": 27100
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4053241028950829,
      "learning_rate": 1.4783828525445553e-06,
      "loss": 0.2472,
      "step": 27101
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.40344179842278594,
      "learning_rate": 1.4778638631251751e-06,
      "loss": 0.1528,
      "step": 27102
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9699775642084294,
      "learning_rate": 1.4773449575493715e-06,
      "loss": 0.5076,
      "step": 27103
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5795340146984774,
      "learning_rate": 1.4768261358222524e-06,
      "loss": 0.225,
      "step": 27104
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4052813738611112,
      "learning_rate": 1.476307397948923e-06,
      "loss": 0.2791,
      "step": 27105
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3164728465966508,
      "learning_rate": 1.4757887439344887e-06,
      "loss": 0.2474,
      "step": 27106
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4267555722000304,
      "learning_rate": 1.4752701737840459e-06,
      "loss": 0.2157,
      "step": 27107
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5790806960955963,
      "learning_rate": 1.4747516875027012e-06,
      "loss": 0.1191,
      "step": 27108
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3427795231861475,
      "learning_rate": 1.4742332850955565e-06,
      "loss": 0.1463,
      "step": 27109
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.30731496996964336,
      "learning_rate": 1.4737149665677052e-06,
      "loss": 0.2017,
      "step": 27110
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8403839798625403,
      "learning_rate": 1.4731967319242568e-06,
      "loss": 0.2534,
      "step": 27111
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41808779545201924,
      "learning_rate": 1.4726785811703026e-06,
      "loss": 0.2772,
      "step": 27112
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6318430070484349,
      "learning_rate": 1.472160514310942e-06,
      "loss": 0.2452,
      "step": 27113
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36209865903279675,
      "learning_rate": 1.471642531351274e-06,
      "loss": 0.2472,
      "step": 27114
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1879170083546373,
      "learning_rate": 1.4711246322963912e-06,
      "loss": 0.4651,
      "step": 27115
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4399686038543138,
      "learning_rate": 1.470606817151391e-06,
      "loss": 0.2306,
      "step": 27116
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3344064014534479,
      "learning_rate": 1.4700890859213669e-06,
      "loss": 0.2031,
      "step": 27117
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3918033636244507,
      "learning_rate": 1.4695714386114157e-06,
      "loss": 0.2283,
      "step": 27118
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.46141632410663924,
      "learning_rate": 1.4690538752266248e-06,
      "loss": 0.1767,
      "step": 27119
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0726787341983646,
      "learning_rate": 1.468536395772089e-06,
      "loss": 0.4142,
      "step": 27120
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8654896154481787,
      "learning_rate": 1.4680190002529017e-06,
      "loss": 0.2748,
      "step": 27121
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4221452629771398,
      "learning_rate": 1.4675016886741467e-06,
      "loss": 0.1816,
      "step": 27122
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4665987513700356,
      "learning_rate": 1.4669844610409222e-06,
      "loss": 0.2481,
      "step": 27123
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2896980799932119,
      "learning_rate": 1.46646731735831e-06,
      "loss": 0.2277,
      "step": 27124
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2506050548389372,
      "learning_rate": 1.4659502576314e-06,
      "loss": 0.1461,
      "step": 27125
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6690143004969641,
      "learning_rate": 1.4654332818652817e-06,
      "loss": 0.0283,
      "step": 27126
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0916169450074305,
      "learning_rate": 1.4649163900650342e-06,
      "loss": 0.4229,
      "step": 27127
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4133065700669701,
      "learning_rate": 1.4643995822357526e-06,
      "loss": 0.1893,
      "step": 27128
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6013558317875104,
      "learning_rate": 1.4638828583825137e-06,
      "loss": 0.3054,
      "step": 27129
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5053766287101029,
      "learning_rate": 1.4633662185104048e-06,
      "loss": 0.2451,
      "step": 27130
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.826993866894897,
      "learning_rate": 1.4628496626245103e-06,
      "loss": 0.5384,
      "step": 27131
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35003274955613695,
      "learning_rate": 1.4623331907299066e-06,
      "loss": 0.1738,
      "step": 27132
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3101084607412247,
      "learning_rate": 1.4618168028316793e-06,
      "loss": 0.21,
      "step": 27133
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.32237233282585415,
      "learning_rate": 1.4613004989349077e-06,
      "loss": 0.0765,
      "step": 27134
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3103322503840129,
      "learning_rate": 1.4607842790446724e-06,
      "loss": 0.2058,
      "step": 27135
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7278144408107438,
      "learning_rate": 1.460268143166048e-06,
      "loss": 0.3318,
      "step": 27136
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3525760538922047,
      "learning_rate": 1.4597520913041197e-06,
      "loss": 0.217,
      "step": 27137
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9617305211496341,
      "learning_rate": 1.4592361234639574e-06,
      "loss": 0.4308,
      "step": 27138
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9542316983990833,
      "learning_rate": 1.4587202396506417e-06,
      "loss": 0.2741,
      "step": 27139
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3059475792086566,
      "learning_rate": 1.4582044398692485e-06,
      "loss": 0.8041,
      "step": 27140
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.27389744038523456,
      "learning_rate": 1.4576887241248482e-06,
      "loss": 0.1629,
      "step": 27141
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4997789412489725,
      "learning_rate": 1.457173092422517e-06,
      "loss": 0.3359,
      "step": 27142
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23475943715299646,
      "learning_rate": 1.456657544767327e-06,
      "loss": 0.1339,
      "step": 27143
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7410653195828947,
      "learning_rate": 1.4561420811643545e-06,
      "loss": 0.7203,
      "step": 27144
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6042692381549786,
      "learning_rate": 1.4556267016186653e-06,
      "loss": 0.1851,
      "step": 27145
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6159289113361741,
      "learning_rate": 1.4551114061353323e-06,
      "loss": 0.3583,
      "step": 27146
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37955578560911707,
      "learning_rate": 1.4545961947194266e-06,
      "loss": 0.2392,
      "step": 27147
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.43106616901443673,
      "learning_rate": 1.4540810673760109e-06,
      "loss": 0.2119,
      "step": 27148
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8576289231376024,
      "learning_rate": 1.4535660241101623e-06,
      "loss": 0.4193,
      "step": 27149
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.35003554471205917,
      "learning_rate": 1.4530510649269415e-06,
      "loss": 0.1561,
      "step": 27150
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5433642276737013,
      "learning_rate": 1.4525361898314161e-06,
      "loss": 0.314,
      "step": 27151
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23022859662916734,
      "learning_rate": 1.4520213988286546e-06,
      "loss": 0.066,
      "step": 27152
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3501751725307666,
      "learning_rate": 1.4515066919237142e-06,
      "loss": 0.2888,
      "step": 27153
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6381499453007694,
      "learning_rate": 1.4509920691216683e-06,
      "loss": 0.214,
      "step": 27154
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3739510550083597,
      "learning_rate": 1.4504775304275741e-06,
      "loss": 0.2571,
      "step": 27155
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9196662401139369,
      "learning_rate": 1.4499630758464945e-06,
      "loss": 0.2504,
      "step": 27156
    },
    {
      "epoch": 0.83,
      "grad_norm": 2.0772990951425845,
      "learning_rate": 1.4494487053834938e-06,
      "loss": 0.7736,
      "step": 27157
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0261240272930592,
      "learning_rate": 1.4489344190436284e-06,
      "loss": 0.5126,
      "step": 27158
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3938166763225213,
      "learning_rate": 1.4484202168319594e-06,
      "loss": 0.2438,
      "step": 27159
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2809891606740478,
      "learning_rate": 1.4479060987535453e-06,
      "loss": 0.2085,
      "step": 27160
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.40779697616287414,
      "learning_rate": 1.4473920648134486e-06,
      "loss": 0.0773,
      "step": 27161
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5123649589190866,
      "learning_rate": 1.44687811501672e-06,
      "loss": 0.2676,
      "step": 27162
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6948599263089985,
      "learning_rate": 1.4463642493684194e-06,
      "loss": 0.2348,
      "step": 27163
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36855155760060426,
      "learning_rate": 1.4458504678736042e-06,
      "loss": 0.2661,
      "step": 27164
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3961056966951043,
      "learning_rate": 1.4453367705373212e-06,
      "loss": 0.2006,
      "step": 27165
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.490683505884177,
      "learning_rate": 1.4448231573646354e-06,
      "loss": 0.3066,
      "step": 27166
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8848885038609415,
      "learning_rate": 1.4443096283605917e-06,
      "loss": 0.5368,
      "step": 27167
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5493664261260138,
      "learning_rate": 1.443796183530245e-06,
      "loss": 0.2855,
      "step": 27168
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.505002663452921,
      "learning_rate": 1.4432828228786478e-06,
      "loss": 0.1818,
      "step": 27169
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5120567887206394,
      "learning_rate": 1.4427695464108504e-06,
      "loss": 0.2054,
      "step": 27170
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36110523939368133,
      "learning_rate": 1.4422563541319012e-06,
      "loss": 0.2327,
      "step": 27171
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6619419048191499,
      "learning_rate": 1.4417432460468484e-06,
      "loss": 0.3319,
      "step": 27172
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42213291422623245,
      "learning_rate": 1.4412302221607455e-06,
      "loss": 0.172,
      "step": 27173
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.27786577283016783,
      "learning_rate": 1.4407172824786298e-06,
      "loss": 0.1395,
      "step": 27174
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4516699158320208,
      "learning_rate": 1.4402044270055594e-06,
      "loss": 0.7171,
      "step": 27175
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2793183614864749,
      "learning_rate": 1.4396916557465713e-06,
      "loss": 0.2198,
      "step": 27176
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9640769924395822,
      "learning_rate": 1.4391789687067137e-06,
      "loss": 0.4321,
      "step": 27177
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34035532831584614,
      "learning_rate": 1.4386663658910327e-06,
      "loss": 0.1892,
      "step": 27178
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0180660020206505,
      "learning_rate": 1.4381538473045643e-06,
      "loss": 0.49,
      "step": 27179
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6771295685042025,
      "learning_rate": 1.4376414129523607e-06,
      "loss": 0.2672,
      "step": 27180
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3226745348829861,
      "learning_rate": 1.4371290628394551e-06,
      "loss": 0.1641,
      "step": 27181
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2921061054143619,
      "learning_rate": 1.4366167969708921e-06,
      "loss": 0.1761,
      "step": 27182
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.47917660561194675,
      "learning_rate": 1.4361046153517133e-06,
      "loss": 0.2767,
      "step": 27183
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4578091538453519,
      "learning_rate": 1.4355925179869523e-06,
      "loss": 0.2167,
      "step": 27184
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.5788107363819224,
      "learning_rate": 1.4350805048816495e-06,
      "loss": 0.7223,
      "step": 27185
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39829556638936076,
      "learning_rate": 1.4345685760408445e-06,
      "loss": 0.1167,
      "step": 27186
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.435409790761252,
      "learning_rate": 1.434056731469573e-06,
      "loss": 0.2139,
      "step": 27187
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9632704165039562,
      "learning_rate": 1.433544971172869e-06,
      "loss": 0.3746,
      "step": 27188
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.30237917561677347,
      "learning_rate": 1.433033295155768e-06,
      "loss": 0.2208,
      "step": 27189
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4358914815856033,
      "learning_rate": 1.4325217034233063e-06,
      "loss": 0.2066,
      "step": 27190
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.28399490473576716,
      "learning_rate": 1.43201019598051e-06,
      "loss": 0.1686,
      "step": 27191
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.44950474107212224,
      "learning_rate": 1.4314987728324214e-06,
      "loss": 0.2279,
      "step": 27192
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.1474120841178626,
      "learning_rate": 1.4309874339840657e-06,
      "loss": 0.4173,
      "step": 27193
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5116960788606928,
      "learning_rate": 1.4304761794404753e-06,
      "loss": 0.3648,
      "step": 27194
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3436557039309541,
      "learning_rate": 1.4299650092066818e-06,
      "loss": 0.1698,
      "step": 27195
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.530825666976251,
      "learning_rate": 1.4294539232877103e-06,
      "loss": 0.3022,
      "step": 27196
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5866715393513989,
      "learning_rate": 1.428942921688591e-06,
      "loss": 0.2296,
      "step": 27197
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.0326716821285074,
      "learning_rate": 1.4284320044143518e-06,
      "loss": 0.3664,
      "step": 27198
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.28592845293841324,
      "learning_rate": 1.4279211714700215e-06,
      "loss": 0.0719,
      "step": 27199
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.39212481746721367,
      "learning_rate": 1.4274104228606179e-06,
      "loss": 0.1653,
      "step": 27200
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2707546042668226,
      "learning_rate": 1.4268997585911759e-06,
      "loss": 0.2221,
      "step": 27201
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42209032196960766,
      "learning_rate": 1.4263891786667127e-06,
      "loss": 0.2462,
      "step": 27202
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4208958702144956,
      "learning_rate": 1.4258786830922544e-06,
      "loss": 0.5598,
      "step": 27203
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8014561127565472,
      "learning_rate": 1.4253682718728257e-06,
      "loss": 0.188,
      "step": 27204
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3891716079231489,
      "learning_rate": 1.4248579450134404e-06,
      "loss": 0.2588,
      "step": 27205
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5229200799774607,
      "learning_rate": 1.4243477025191277e-06,
      "loss": 0.2398,
      "step": 27206
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.48323972746292027,
      "learning_rate": 1.4238375443949027e-06,
      "loss": 0.3158,
      "step": 27207
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6475253928828074,
      "learning_rate": 1.4233274706457857e-06,
      "loss": 0.0392,
      "step": 27208
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34762407975919835,
      "learning_rate": 1.422817481276798e-06,
      "loss": 0.2109,
      "step": 27209
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24541672602204878,
      "learning_rate": 1.4223075762929507e-06,
      "loss": 0.1517,
      "step": 27210
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3631942093273648,
      "learning_rate": 1.4217977556992646e-06,
      "loss": 0.6899,
      "step": 27211
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.45490580788517515,
      "learning_rate": 1.421288019500755e-06,
      "loss": 0.2312,
      "step": 27212
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41226606893793477,
      "learning_rate": 1.420778367702439e-06,
      "loss": 0.2909,
      "step": 27213
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3157360638632833,
      "learning_rate": 1.4202688003093257e-06,
      "loss": 0.1978,
      "step": 27214
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7683032967480018,
      "learning_rate": 1.4197593173264313e-06,
      "loss": 0.2488,
      "step": 27215
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2975703041415445,
      "learning_rate": 1.4192499187587693e-06,
      "loss": 0.5612,
      "step": 27216
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.41338032773859323,
      "learning_rate": 1.4187406046113473e-06,
      "loss": 0.0798,
      "step": 27217
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4203112351492493,
      "learning_rate": 1.4182313748891819e-06,
      "loss": 0.2454,
      "step": 27218
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36925916621450344,
      "learning_rate": 1.4177222295972781e-06,
      "loss": 0.2118,
      "step": 27219
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.32580504812976324,
      "learning_rate": 1.4172131687406465e-06,
      "loss": 0.2387,
      "step": 27220
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.879305489349426,
      "learning_rate": 1.4167041923242996e-06,
      "loss": 0.3591,
      "step": 27221
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.681462661553089,
      "learning_rate": 1.4161953003532381e-06,
      "loss": 0.3262,
      "step": 27222
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3660705169430965,
      "learning_rate": 1.415686492832471e-06,
      "loss": 0.163,
      "step": 27223
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4750785818159638,
      "learning_rate": 1.4151777697670054e-06,
      "loss": 0.2827,
      "step": 27224
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3062918600453273,
      "learning_rate": 1.4146691311618476e-06,
      "loss": 0.2119,
      "step": 27225
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.7015673240748255,
      "learning_rate": 1.4141605770219978e-06,
      "loss": 0.8093,
      "step": 27226
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.18046495961696987,
      "learning_rate": 1.4136521073524611e-06,
      "loss": 0.067,
      "step": 27227
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7473954517265831,
      "learning_rate": 1.413143722158241e-06,
      "loss": 0.1563,
      "step": 27228
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3082169618686774,
      "learning_rate": 1.4126354214443383e-06,
      "loss": 0.7469,
      "step": 27229
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3274481455525057,
      "learning_rate": 1.4121272052157553e-06,
      "loss": 0.2348,
      "step": 27230
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.9933318834301377,
      "learning_rate": 1.411619073477487e-06,
      "loss": 0.4996,
      "step": 27231
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2945497569778919,
      "learning_rate": 1.4111110262345406e-06,
      "loss": 0.1778,
      "step": 27232
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2032872766279679,
      "learning_rate": 1.4106030634919098e-06,
      "loss": 0.4424,
      "step": 27233
    },
    {
      "epoch": 0.83,
      "grad_norm": 3.4424886146980294,
      "learning_rate": 1.4100951852545875e-06,
      "loss": 0.0995,
      "step": 27234
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4874298149140286,
      "learning_rate": 1.4095873915275803e-06,
      "loss": 0.5022,
      "step": 27235
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3364596909585087,
      "learning_rate": 1.409079682315876e-06,
      "loss": 0.173,
      "step": 27236
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2897299897021109,
      "learning_rate": 1.4085720576244733e-06,
      "loss": 0.2449,
      "step": 27237
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.34500878140450303,
      "learning_rate": 1.4080645174583652e-06,
      "loss": 0.1642,
      "step": 27238
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.073013719610034,
      "learning_rate": 1.4075570618225487e-06,
      "loss": 0.4841,
      "step": 27239
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8931165135154014,
      "learning_rate": 1.40704969072201e-06,
      "loss": 0.2953,
      "step": 27240
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3073951157455106,
      "learning_rate": 1.4065424041617438e-06,
      "loss": 0.1868,
      "step": 27241
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.4445775951916129,
      "learning_rate": 1.4060352021467438e-06,
      "loss": 0.23,
      "step": 27242
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.30059039164747775,
      "learning_rate": 1.405528084681992e-06,
      "loss": 0.1934,
      "step": 27243
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.3447748744664871,
      "learning_rate": 1.4050210517724882e-06,
      "loss": 0.5186,
      "step": 27244
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.36948136514440133,
      "learning_rate": 1.4045141034232125e-06,
      "loss": 0.1607,
      "step": 27245
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3399178616149894,
      "learning_rate": 1.4040072396391557e-06,
      "loss": 0.2261,
      "step": 27246
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2025512071431737,
      "learning_rate": 1.4035004604253066e-06,
      "loss": 0.3173,
      "step": 27247
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3314514217785238,
      "learning_rate": 1.4029937657866466e-06,
      "loss": 0.2946,
      "step": 27248
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.6980384425945625,
      "learning_rate": 1.4024871557281616e-06,
      "loss": 0.2328,
      "step": 27249
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5223891680033068,
      "learning_rate": 1.4019806302548378e-06,
      "loss": 0.3024,
      "step": 27250
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.24669440665572362,
      "learning_rate": 1.40147418937166e-06,
      "loss": 0.1434,
      "step": 27251
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.4622440028571322,
      "learning_rate": 1.4009678330836052e-06,
      "loss": 0.4486,
      "step": 27252
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.42243348631071853,
      "learning_rate": 1.4004615613956597e-06,
      "loss": 0.1894,
      "step": 27253
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.37762628010741967,
      "learning_rate": 1.3999553743128013e-06,
      "loss": 0.1917,
      "step": 27254
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.3610903744236838,
      "learning_rate": 1.3994492718400132e-06,
      "loss": 0.2629,
      "step": 27255
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7403040220498261,
      "learning_rate": 1.3989432539822744e-06,
      "loss": 0.2862,
      "step": 27256
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7468922954750301,
      "learning_rate": 1.39843732074456e-06,
      "loss": 0.4184,
      "step": 27257
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.220337497057835,
      "learning_rate": 1.3979314721318504e-06,
      "loss": 0.4604,
      "step": 27258
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.2987156801315399,
      "learning_rate": 1.3974257081491227e-06,
      "loss": 0.1335,
      "step": 27259
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.23494589728866777,
      "learning_rate": 1.3969200288013473e-06,
      "loss": 0.1811,
      "step": 27260
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.5554173791655161,
      "learning_rate": 1.3964144340935081e-06,
      "loss": 0.3056,
      "step": 27261
    },
    {
      "epoch": 0.83,
      "grad_norm": 1.2906385818103807,
      "learning_rate": 1.3959089240305724e-06,
      "loss": 0.0768,
      "step": 27262
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.8271002623824738,
      "learning_rate": 1.3954034986175148e-06,
      "loss": 0.3639,
      "step": 27263
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.32233091748049986,
      "learning_rate": 1.3948981578593123e-06,
      "loss": 0.194,
      "step": 27264
    },
    {
      "epoch": 0.83,
      "grad_norm": 0.7607402696882711,
      "learning_rate": 1.3943929017609303e-06,
      "loss": 0.3533,
      "step": 27265
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4553619825421782,
      "learning_rate": 1.3938877303273436e-06,
      "loss": 0.2233,
      "step": 27266
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39462243692068055,
      "learning_rate": 1.3933826435635201e-06,
      "loss": 0.2409,
      "step": 27267
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3874186577901896,
      "learning_rate": 1.3928776414744328e-06,
      "loss": 0.2408,
      "step": 27268
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.280366746233653,
      "learning_rate": 1.392372724065042e-06,
      "loss": 0.1295,
      "step": 27269
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41608673192999324,
      "learning_rate": 1.391867891340326e-06,
      "loss": 0.192,
      "step": 27270
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42762205570642,
      "learning_rate": 1.3913631433052444e-06,
      "loss": 0.2011,
      "step": 27271
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44045441139003577,
      "learning_rate": 1.390858479964764e-06,
      "loss": 0.2914,
      "step": 27272
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3318090046922236,
      "learning_rate": 1.3903539013238531e-06,
      "loss": 0.1865,
      "step": 27273
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8528493830580843,
      "learning_rate": 1.3898494073874714e-06,
      "loss": 0.384,
      "step": 27274
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8578420368186612,
      "learning_rate": 1.3893449981605834e-06,
      "loss": 0.3097,
      "step": 27275
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1548021654412237,
      "learning_rate": 1.388840673648154e-06,
      "loss": 0.5436,
      "step": 27276
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3090190417554733,
      "learning_rate": 1.3883364338551418e-06,
      "loss": 0.1448,
      "step": 27277
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3271785731859671,
      "learning_rate": 1.387832278786513e-06,
      "loss": 0.2059,
      "step": 27278
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24329240763558208,
      "learning_rate": 1.3873282084472216e-06,
      "loss": 0.1736,
      "step": 27279
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0250360707205979,
      "learning_rate": 1.386824222842229e-06,
      "loss": 0.2294,
      "step": 27280
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.981013717974452,
      "learning_rate": 1.3863203219764942e-06,
      "loss": 0.51,
      "step": 27281
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2693398116235684,
      "learning_rate": 1.3858165058549756e-06,
      "loss": 0.1656,
      "step": 27282
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8485425132187678,
      "learning_rate": 1.3853127744826278e-06,
      "loss": 0.404,
      "step": 27283
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3231536776925724,
      "learning_rate": 1.384809127864406e-06,
      "loss": 0.2435,
      "step": 27284
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.87149422523392,
      "learning_rate": 1.3843055660052708e-06,
      "loss": 0.7507,
      "step": 27285
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4600701030837765,
      "learning_rate": 1.3838020889101667e-06,
      "loss": 0.1189,
      "step": 27286
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38032731915786494,
      "learning_rate": 1.3832986965840566e-06,
      "loss": 0.2664,
      "step": 27287
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1962785722454234,
      "learning_rate": 1.3827953890318878e-06,
      "loss": 0.0692,
      "step": 27288
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.032179553986651,
      "learning_rate": 1.3822921662586142e-06,
      "loss": 0.3683,
      "step": 27289
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33029881189324506,
      "learning_rate": 1.3817890282691871e-06,
      "loss": 0.2184,
      "step": 27290
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35170172707719144,
      "learning_rate": 1.3812859750685536e-06,
      "loss": 0.2517,
      "step": 27291
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8503085929022537,
      "learning_rate": 1.3807830066616633e-06,
      "loss": 0.2302,
      "step": 27292
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.136953298687,
      "learning_rate": 1.3802801230534678e-06,
      "loss": 0.4372,
      "step": 27293
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4751061585505896,
      "learning_rate": 1.3797773242489131e-06,
      "loss": 0.6726,
      "step": 27294
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.390279043892866,
      "learning_rate": 1.3792746102529442e-06,
      "loss": 0.1421,
      "step": 27295
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.27743482597746394,
      "learning_rate": 1.378771981070509e-06,
      "loss": 0.2407,
      "step": 27296
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1487436960390805,
      "learning_rate": 1.378269436706553e-06,
      "loss": 0.0682,
      "step": 27297
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.080583039395072,
      "learning_rate": 1.3777669771660152e-06,
      "loss": 0.4213,
      "step": 27298
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8042635833175662,
      "learning_rate": 1.377264602453847e-06,
      "loss": 0.2535,
      "step": 27299
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35201700025727095,
      "learning_rate": 1.3767623125749852e-06,
      "loss": 0.2576,
      "step": 27300
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.648040360920985,
      "learning_rate": 1.3762601075343728e-06,
      "loss": 0.0709,
      "step": 27301
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32178021684219077,
      "learning_rate": 1.375757987336951e-06,
      "loss": 0.2492,
      "step": 27302
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.105460680214478,
      "learning_rate": 1.3752559519876607e-06,
      "loss": 0.6024,
      "step": 27303
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5984721411869637,
      "learning_rate": 1.3747540014914418e-06,
      "loss": 0.2499,
      "step": 27304
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36324189207932184,
      "learning_rate": 1.374252135853229e-06,
      "loss": 0.2079,
      "step": 27305
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2961274932921217,
      "learning_rate": 1.3737503550779619e-06,
      "loss": 0.1064,
      "step": 27306
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.48546616377273283,
      "learning_rate": 1.3732486591705762e-06,
      "loss": 0.2898,
      "step": 27307
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4805358858531355,
      "learning_rate": 1.3727470481360127e-06,
      "loss": 0.2136,
      "step": 27308
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37825063776323975,
      "learning_rate": 1.3722455219791998e-06,
      "loss": 0.2269,
      "step": 27309
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5101119029092426,
      "learning_rate": 1.371744080705073e-06,
      "loss": 0.1887,
      "step": 27310
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1238931769318092,
      "learning_rate": 1.371242724318571e-06,
      "loss": 0.5356,
      "step": 27311
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.083407337731395,
      "learning_rate": 1.3707414528246165e-06,
      "loss": 0.4367,
      "step": 27312
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42323251302912784,
      "learning_rate": 1.3702402662281521e-06,
      "loss": 0.2862,
      "step": 27313
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2568633941716216,
      "learning_rate": 1.3697391645341019e-06,
      "loss": 0.1961,
      "step": 27314
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9261708756784547,
      "learning_rate": 1.369238147747397e-06,
      "loss": 0.4031,
      "step": 27315
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7342398807923195,
      "learning_rate": 1.3687372158729694e-06,
      "loss": 0.2505,
      "step": 27316
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5649232836520475,
      "learning_rate": 1.3682363689157429e-06,
      "loss": 0.1786,
      "step": 27317
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3308878192382644,
      "learning_rate": 1.3677356068806469e-06,
      "loss": 0.161,
      "step": 27318
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33025714207149914,
      "learning_rate": 1.3672349297726084e-06,
      "loss": 0.1994,
      "step": 27319
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.48343049339989647,
      "learning_rate": 1.366734337596556e-06,
      "loss": 0.2932,
      "step": 27320
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9880358653680885,
      "learning_rate": 1.3662338303574086e-06,
      "loss": 0.5962,
      "step": 27321
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6735245071652552,
      "learning_rate": 1.3657334080600948e-06,
      "loss": 0.2999,
      "step": 27322
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3633449781916321,
      "learning_rate": 1.3652330707095385e-06,
      "loss": 0.2122,
      "step": 27323
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9860924339735969,
      "learning_rate": 1.3647328183106556e-06,
      "loss": 0.4665,
      "step": 27324
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5201669520300367,
      "learning_rate": 1.3642326508683778e-06,
      "loss": 0.2577,
      "step": 27325
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22850286403109704,
      "learning_rate": 1.363732568387619e-06,
      "loss": 0.1745,
      "step": 27326
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2849885323465367,
      "learning_rate": 1.3632325708732996e-06,
      "loss": 0.1772,
      "step": 27327
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.366441745180277,
      "learning_rate": 1.3627326583303424e-06,
      "loss": 0.4432,
      "step": 27328
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9403711525500694,
      "learning_rate": 1.36223283076366e-06,
      "loss": 0.4138,
      "step": 27329
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.396354409087848,
      "learning_rate": 1.3617330881781765e-06,
      "loss": 0.623,
      "step": 27330
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2650798540997443,
      "learning_rate": 1.3612334305788034e-06,
      "loss": 0.192,
      "step": 27331
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3189786428984944,
      "learning_rate": 1.3607338579704588e-06,
      "loss": 0.1842,
      "step": 27332
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8422846285669542,
      "learning_rate": 1.360234370358059e-06,
      "loss": 0.3636,
      "step": 27333
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8488854196517054,
      "learning_rate": 1.3597349677465145e-06,
      "loss": 0.2629,
      "step": 27334
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2572257396655062,
      "learning_rate": 1.35923565014074e-06,
      "loss": 0.1458,
      "step": 27335
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6233821625348285,
      "learning_rate": 1.3587364175456485e-06,
      "loss": 0.0815,
      "step": 27336
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37143507831165534,
      "learning_rate": 1.358237269966154e-06,
      "loss": 0.2661,
      "step": 27337
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32721715609494445,
      "learning_rate": 1.35773820740716e-06,
      "loss": 0.2407,
      "step": 27338
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.46857722467431,
      "learning_rate": 1.3572392298735849e-06,
      "loss": 0.7214,
      "step": 27339
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7033910161213219,
      "learning_rate": 1.3567403373703325e-06,
      "loss": 0.1937,
      "step": 27340
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3413883972976017,
      "learning_rate": 1.3562415299023124e-06,
      "loss": 0.244,
      "step": 27341
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7479601879290473,
      "learning_rate": 1.355742807474435e-06,
      "loss": 0.2607,
      "step": 27342
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.468318254128343,
      "learning_rate": 1.3552441700916018e-06,
      "loss": 0.679,
      "step": 27343
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2290104613159556,
      "learning_rate": 1.3547456177587214e-06,
      "loss": 0.1722,
      "step": 27344
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.22307005455967285,
      "learning_rate": 1.3542471504806976e-06,
      "loss": 0.1171,
      "step": 27345
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.593831455980174,
      "learning_rate": 1.3537487682624372e-06,
      "loss": 0.3199,
      "step": 27346
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0822123640018093,
      "learning_rate": 1.3532504711088402e-06,
      "loss": 0.4375,
      "step": 27347
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3494683309420146,
      "learning_rate": 1.35275225902481e-06,
      "loss": 0.6094,
      "step": 27348
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3218410379868765,
      "learning_rate": 1.3522541320152504e-06,
      "loss": 0.2098,
      "step": 27349
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3313118800786284,
      "learning_rate": 1.3517560900850557e-06,
      "loss": 0.2574,
      "step": 27350
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9166522733309652,
      "learning_rate": 1.3512581332391339e-06,
      "loss": 0.2588,
      "step": 27351
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0476677796006555,
      "learning_rate": 1.350760261482379e-06,
      "loss": 0.4583,
      "step": 27352
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.15798596722292968,
      "learning_rate": 1.3502624748196901e-06,
      "loss": 0.0682,
      "step": 27353
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4128977915691679,
      "learning_rate": 1.3497647732559683e-06,
      "loss": 0.2675,
      "step": 27354
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5463856179824382,
      "learning_rate": 1.3492671567961013e-06,
      "loss": 0.1915,
      "step": 27355
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3205413193440464,
      "learning_rate": 1.3487696254449955e-06,
      "loss": 0.2631,
      "step": 27356
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9036740526855427,
      "learning_rate": 1.3482721792075392e-06,
      "loss": 0.3474,
      "step": 27357
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8493466140325348,
      "learning_rate": 1.3477748180886273e-06,
      "loss": 0.2444,
      "step": 27358
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44985376768605106,
      "learning_rate": 1.3472775420931561e-06,
      "loss": 0.2442,
      "step": 27359
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4897283220066832,
      "learning_rate": 1.3467803512260125e-06,
      "loss": 0.2183,
      "step": 27360
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34068441415912537,
      "learning_rate": 1.3462832454920917e-06,
      "loss": 0.2701,
      "step": 27361
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9077933876084663,
      "learning_rate": 1.3457862248962838e-06,
      "loss": 0.0347,
      "step": 27362
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38571086126115606,
      "learning_rate": 1.3452892894434811e-06,
      "loss": 0.2113,
      "step": 27363
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.20742649859564544,
      "learning_rate": 1.344792439138568e-06,
      "loss": 0.1448,
      "step": 27364
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3563296026145095,
      "learning_rate": 1.3442956739864343e-06,
      "loss": 0.7671,
      "step": 27365
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8564282008718972,
      "learning_rate": 1.3437989939919705e-06,
      "loss": 0.2712,
      "step": 27366
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34380251297979253,
      "learning_rate": 1.3433023991600558e-06,
      "loss": 0.2825,
      "step": 27367
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3589126694995363,
      "learning_rate": 1.3428058894955854e-06,
      "loss": 0.1812,
      "step": 27368
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5728432872529431,
      "learning_rate": 1.3423094650034364e-06,
      "loss": 0.342,
      "step": 27369
    },
    {
      "epoch": 0.84,
      "grad_norm": 2.2870642129463516,
      "learning_rate": 1.3418131256884958e-06,
      "loss": 0.2129,
      "step": 27370
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3478654972260496,
      "learning_rate": 1.341316871555647e-06,
      "loss": 0.0808,
      "step": 27371
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3941109226532974,
      "learning_rate": 1.3408207026097742e-06,
      "loss": 0.1951,
      "step": 27372
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24369862766956568,
      "learning_rate": 1.3403246188557538e-06,
      "loss": 0.2155,
      "step": 27373
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.6465354117701008,
      "learning_rate": 1.339828620298469e-06,
      "loss": 0.7473,
      "step": 27374
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6903753886891101,
      "learning_rate": 1.339332706942803e-06,
      "loss": 0.3164,
      "step": 27375
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7596449703427488,
      "learning_rate": 1.3388368787936258e-06,
      "loss": 0.3954,
      "step": 27376
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2913929367052522,
      "learning_rate": 1.3383411358558275e-06,
      "loss": 0.1771,
      "step": 27377
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7237978886441765,
      "learning_rate": 1.3378454781342754e-06,
      "loss": 0.747,
      "step": 27378
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2917980153794332,
      "learning_rate": 1.3373499056338502e-06,
      "loss": 0.201,
      "step": 27379
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9698419600372514,
      "learning_rate": 1.336854418359429e-06,
      "loss": 0.3465,
      "step": 27380
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.215894429365791,
      "learning_rate": 1.3363590163158802e-06,
      "loss": 0.071,
      "step": 27381
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6688097666737391,
      "learning_rate": 1.3358636995080875e-06,
      "loss": 0.3178,
      "step": 27382
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5454833874232249,
      "learning_rate": 1.3353684679409151e-06,
      "loss": 0.2447,
      "step": 27383
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44935436117487976,
      "learning_rate": 1.33487332161924e-06,
      "loss": 0.2458,
      "step": 27384
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41604935240302726,
      "learning_rate": 1.3343782605479338e-06,
      "loss": 0.3024,
      "step": 27385
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3719281170989113,
      "learning_rate": 1.3338832847318627e-06,
      "loss": 0.1677,
      "step": 27386
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3321009664732878,
      "learning_rate": 1.3333883941759006e-06,
      "loss": 0.2252,
      "step": 27387
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5907218622457714,
      "learning_rate": 1.3328935888849148e-06,
      "loss": 0.0206,
      "step": 27388
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.49388391138982607,
      "learning_rate": 1.3323988688637757e-06,
      "loss": 0.2169,
      "step": 27389
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5010870930042635,
      "learning_rate": 1.3319042341173471e-06,
      "loss": 0.1381,
      "step": 27390
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36936687365075216,
      "learning_rate": 1.3314096846504953e-06,
      "loss": 0.2549,
      "step": 27391
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3224469633439255,
      "learning_rate": 1.330915220468091e-06,
      "loss": 0.2417,
      "step": 27392
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8756187736622053,
      "learning_rate": 1.3304208415749909e-06,
      "loss": 0.3747,
      "step": 27393
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.1409168307757795,
      "learning_rate": 1.329926547976067e-06,
      "loss": 0.5354,
      "step": 27394
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31861228990347296,
      "learning_rate": 1.3294323396761765e-06,
      "loss": 0.1721,
      "step": 27395
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3715811010811804,
      "learning_rate": 1.3289382166801834e-06,
      "loss": 0.1945,
      "step": 27396
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.30181916421023863,
      "learning_rate": 1.3284441789929513e-06,
      "loss": 0.2062,
      "step": 27397
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2253527684168863,
      "learning_rate": 1.3279502266193366e-06,
      "loss": 0.477,
      "step": 27398
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.78623287412672,
      "learning_rate": 1.327456359564202e-06,
      "loss": 0.161,
      "step": 27399
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.382251820614481,
      "learning_rate": 1.3269625778324047e-06,
      "loss": 0.2457,
      "step": 27400
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7101955017921394,
      "learning_rate": 1.3264688814288051e-06,
      "loss": 0.2575,
      "step": 27401
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.946723299792887,
      "learning_rate": 1.325975270358255e-06,
      "loss": 0.4199,
      "step": 27402
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.29774175340757414,
      "learning_rate": 1.3254817446256184e-06,
      "loss": 0.2256,
      "step": 27403
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38196165267253485,
      "learning_rate": 1.3249883042357447e-06,
      "loss": 0.2361,
      "step": 27404
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3119038373350044,
      "learning_rate": 1.32449494919349e-06,
      "loss": 0.1354,
      "step": 27405
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44041780002611075,
      "learning_rate": 1.3240016795037114e-06,
      "loss": 0.2282,
      "step": 27406
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5198616061825314,
      "learning_rate": 1.323508495171254e-06,
      "loss": 0.0939,
      "step": 27407
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4144473212069082,
      "learning_rate": 1.3230153962009796e-06,
      "loss": 0.2706,
      "step": 27408
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.37145088789126457,
      "learning_rate": 1.3225223825977329e-06,
      "loss": 0.1655,
      "step": 27409
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3573832717668462,
      "learning_rate": 1.322029454366367e-06,
      "loss": 0.2623,
      "step": 27410
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8078072060206105,
      "learning_rate": 1.3215366115117313e-06,
      "loss": 0.2986,
      "step": 27411
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.25646485113871564,
      "learning_rate": 1.3210438540386728e-06,
      "loss": 0.1238,
      "step": 27412
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9392336530581522,
      "learning_rate": 1.32055118195204e-06,
      "loss": 0.4626,
      "step": 27413
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24774763634425037,
      "learning_rate": 1.3200585952566814e-06,
      "loss": 0.1591,
      "step": 27414
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35029510227650357,
      "learning_rate": 1.3195660939574439e-06,
      "loss": 0.2616,
      "step": 27415
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9244764793881978,
      "learning_rate": 1.3190736780591695e-06,
      "loss": 0.2198,
      "step": 27416
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9762309538207994,
      "learning_rate": 1.3185813475667052e-06,
      "loss": 0.4527,
      "step": 27417
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.28754370136567947,
      "learning_rate": 1.3180891024848963e-06,
      "loss": 0.1718,
      "step": 27418
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7749165664596647,
      "learning_rate": 1.3175969428185786e-06,
      "loss": 0.4055,
      "step": 27419
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4756648443834988,
      "learning_rate": 1.3171048685726029e-06,
      "loss": 0.2543,
      "step": 27420
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.494283624823729,
      "learning_rate": 1.316612879751805e-06,
      "loss": 0.2911,
      "step": 27421
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.352431942798355,
      "learning_rate": 1.3161209763610273e-06,
      "loss": 0.1113,
      "step": 27422
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3002755461601941,
      "learning_rate": 1.3156291584051107e-06,
      "loss": 0.2302,
      "step": 27423
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2159586798243627,
      "learning_rate": 1.3151374258888893e-06,
      "loss": 0.0641,
      "step": 27424
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9602239071783241,
      "learning_rate": 1.314645778817204e-06,
      "loss": 0.2322,
      "step": 27425
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3568525539516458,
      "learning_rate": 1.3141542171948918e-06,
      "loss": 0.2662,
      "step": 27426
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.296865998260702,
      "learning_rate": 1.31366274102679e-06,
      "loss": 0.1677,
      "step": 27427
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8732386183829043,
      "learning_rate": 1.3131713503177301e-06,
      "loss": 0.3696,
      "step": 27428
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.240610379874976,
      "learning_rate": 1.3126800450725485e-06,
      "loss": 0.2709,
      "step": 27429
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2424688134926143,
      "learning_rate": 1.31218882529608e-06,
      "loss": 0.5953,
      "step": 27430
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3663561834780784,
      "learning_rate": 1.3116976909931556e-06,
      "loss": 0.1494,
      "step": 27431
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5092724195997528,
      "learning_rate": 1.311206642168611e-06,
      "loss": 0.2893,
      "step": 27432
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19302149600782884,
      "learning_rate": 1.3107156788272724e-06,
      "loss": 0.1474,
      "step": 27433
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9869166755057065,
      "learning_rate": 1.3102248009739716e-06,
      "loss": 0.4775,
      "step": 27434
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6117693985207011,
      "learning_rate": 1.3097340086135414e-06,
      "loss": 0.2374,
      "step": 27435
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0245960006230248,
      "learning_rate": 1.3092433017508021e-06,
      "loss": 0.4927,
      "step": 27436
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3552067905459756,
      "learning_rate": 1.3087526803905914e-06,
      "loss": 0.1945,
      "step": 27437
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31727067376442936,
      "learning_rate": 1.3082621445377308e-06,
      "loss": 0.2334,
      "step": 27438
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3891438859670469,
      "learning_rate": 1.3077716941970453e-06,
      "loss": 0.7228,
      "step": 27439
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4446121604702518,
      "learning_rate": 1.3072813293733633e-06,
      "loss": 0.0704,
      "step": 27440
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3708831414014654,
      "learning_rate": 1.3067910500715098e-06,
      "loss": 0.2363,
      "step": 27441
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2492441703645218,
      "learning_rate": 1.3063008562963042e-06,
      "loss": 0.1068,
      "step": 27442
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6788260545116213,
      "learning_rate": 1.3058107480525705e-06,
      "loss": 0.3397,
      "step": 27443
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32118238097067275,
      "learning_rate": 1.305320725345135e-06,
      "loss": 0.228,
      "step": 27444
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6706222524570981,
      "learning_rate": 1.3048307881788091e-06,
      "loss": 0.263,
      "step": 27445
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35096087974894385,
      "learning_rate": 1.3043409365584247e-06,
      "loss": 0.193,
      "step": 27446
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.401940442627436,
      "learning_rate": 1.3038511704887935e-06,
      "loss": 0.7663,
      "step": 27447
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8940432991625961,
      "learning_rate": 1.3033614899747349e-06,
      "loss": 0.2638,
      "step": 27448
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8711675649789795,
      "learning_rate": 1.3028718950210694e-06,
      "loss": 0.3341,
      "step": 27449
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.21251905111278344,
      "learning_rate": 1.3023823856326112e-06,
      "loss": 0.1433,
      "step": 27450
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.242217065908208,
      "learning_rate": 1.3018929618141763e-06,
      "loss": 0.1806,
      "step": 27451
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7356855607054484,
      "learning_rate": 1.3014036235705797e-06,
      "loss": 0.3853,
      "step": 27452
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7179124862450138,
      "learning_rate": 1.3009143709066397e-06,
      "loss": 0.2476,
      "step": 27453
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4166899449646625,
      "learning_rate": 1.3004252038271636e-06,
      "loss": 0.2696,
      "step": 27454
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4930088356022185,
      "learning_rate": 1.2999361223369677e-06,
      "loss": 0.2127,
      "step": 27455
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5010362596215295,
      "learning_rate": 1.2994471264408625e-06,
      "loss": 0.335,
      "step": 27456
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4105596073560209,
      "learning_rate": 1.2989582161436598e-06,
      "loss": 0.2289,
      "step": 27457
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8847431495691216,
      "learning_rate": 1.29846939145017e-06,
      "loss": 0.3571,
      "step": 27458
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41037401725820105,
      "learning_rate": 1.2979806523652006e-06,
      "loss": 0.1657,
      "step": 27459
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.555130349906309,
      "learning_rate": 1.2974919988935607e-06,
      "loss": 0.3162,
      "step": 27460
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4170116907353713,
      "learning_rate": 1.2970034310400592e-06,
      "loss": 0.1831,
      "step": 27461
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2635853871222664,
      "learning_rate": 1.2965149488094986e-06,
      "loss": 0.2171,
      "step": 27462
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3164876355583422,
      "learning_rate": 1.2960265522066916e-06,
      "loss": 0.07,
      "step": 27463
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.38859022748658595,
      "learning_rate": 1.295538241236437e-06,
      "loss": 0.181,
      "step": 27464
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4435773142951513,
      "learning_rate": 1.2950500159035416e-06,
      "loss": 0.6899,
      "step": 27465
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.091981889088549,
      "learning_rate": 1.294561876212811e-06,
      "loss": 0.1622,
      "step": 27466
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8570115390347595,
      "learning_rate": 1.2940738221690419e-06,
      "loss": 0.3573,
      "step": 27467
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2754968212599845,
      "learning_rate": 1.2935858537770397e-06,
      "loss": 0.1675,
      "step": 27468
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35592983553661856,
      "learning_rate": 1.2930979710416047e-06,
      "loss": 0.2652,
      "step": 27469
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42914534887732747,
      "learning_rate": 1.292610173967539e-06,
      "loss": 0.167,
      "step": 27470
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.489933384265979,
      "learning_rate": 1.2921224625596351e-06,
      "loss": 0.2572,
      "step": 27471
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2708263366809712,
      "learning_rate": 1.2916348368227004e-06,
      "loss": 0.068,
      "step": 27472
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3814547703361879,
      "learning_rate": 1.2911472967615246e-06,
      "loss": 0.2848,
      "step": 27473
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2973949235038601,
      "learning_rate": 1.290659842380908e-06,
      "loss": 0.2235,
      "step": 27474
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.9099683419417066,
      "learning_rate": 1.2901724736856481e-06,
      "loss": 0.8436,
      "step": 27475
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9984483208197034,
      "learning_rate": 1.2896851906805342e-06,
      "loss": 0.2406,
      "step": 27476
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.28277275595792484,
      "learning_rate": 1.2891979933703636e-06,
      "loss": 0.1616,
      "step": 27477
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7344247255154306,
      "learning_rate": 1.2887108817599302e-06,
      "loss": 0.3277,
      "step": 27478
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4444521746015915,
      "learning_rate": 1.2882238558540271e-06,
      "loss": 0.0824,
      "step": 27479
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2929482002805032,
      "learning_rate": 1.2877369156574415e-06,
      "loss": 0.24,
      "step": 27480
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.19904151118438507,
      "learning_rate": 1.2872500611749684e-06,
      "loss": 0.1142,
      "step": 27481
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5558552106947715,
      "learning_rate": 1.286763292411395e-06,
      "loss": 0.3377,
      "step": 27482
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0551776548975316,
      "learning_rate": 1.2862766093715119e-06,
      "loss": 0.5197,
      "step": 27483
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4135742692089743,
      "learning_rate": 1.2857900120601074e-06,
      "loss": 0.5319,
      "step": 27484
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.44839282492147936,
      "learning_rate": 1.285303500481968e-06,
      "loss": 0.2224,
      "step": 27485
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.605735252556589,
      "learning_rate": 1.2848170746418787e-06,
      "loss": 0.259,
      "step": 27486
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35810876330991676,
      "learning_rate": 1.28433073454463e-06,
      "loss": 0.2291,
      "step": 27487
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.070220800418751,
      "learning_rate": 1.2838444801949979e-06,
      "loss": 0.442,
      "step": 27488
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1452639917336042,
      "learning_rate": 1.2833583115977765e-06,
      "loss": 0.0673,
      "step": 27489
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4468550612813459,
      "learning_rate": 1.2828722287577433e-06,
      "loss": 0.1445,
      "step": 27490
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36356774966484157,
      "learning_rate": 1.2823862316796797e-06,
      "loss": 0.2557,
      "step": 27491
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31519944969058616,
      "learning_rate": 1.2819003203683721e-06,
      "loss": 0.2281,
      "step": 27492
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9703345499323093,
      "learning_rate": 1.2814144948285956e-06,
      "loss": 0.4293,
      "step": 27493
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9863707705611404,
      "learning_rate": 1.2809287550651317e-06,
      "loss": 0.2847,
      "step": 27494
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41945225247928525,
      "learning_rate": 1.280443101082759e-06,
      "loss": 0.2627,
      "step": 27495
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.47428413674930825,
      "learning_rate": 1.279957532886259e-06,
      "loss": 0.2148,
      "step": 27496
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5371155158620329,
      "learning_rate": 1.2794720504804037e-06,
      "loss": 0.2465,
      "step": 27497
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4985130520569408,
      "learning_rate": 1.2789866538699724e-06,
      "loss": 0.2043,
      "step": 27498
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2513355831205306,
      "learning_rate": 1.2785013430597415e-06,
      "loss": 0.1399,
      "step": 27499
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33192962618408905,
      "learning_rate": 1.2780161180544804e-06,
      "loss": 0.2046,
      "step": 27500
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3192583058030887,
      "learning_rate": 1.2775309788589708e-06,
      "loss": 0.8276,
      "step": 27501
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9982121247098515,
      "learning_rate": 1.2770459254779788e-06,
      "loss": 0.2092,
      "step": 27502
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31137928750747024,
      "learning_rate": 1.2765609579162797e-06,
      "loss": 0.2336,
      "step": 27503
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41997324858869606,
      "learning_rate": 1.2760760761786462e-06,
      "loss": 0.2709,
      "step": 27504
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4654744615996962,
      "learning_rate": 1.2755912802698457e-06,
      "loss": 0.2042,
      "step": 27505
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.878925851234329,
      "learning_rate": 1.2751065701946475e-06,
      "loss": 0.7246,
      "step": 27506
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34478883402712956,
      "learning_rate": 1.2746219459578225e-06,
      "loss": 0.0811,
      "step": 27507
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9221450494411357,
      "learning_rate": 1.2741374075641378e-06,
      "loss": 0.4507,
      "step": 27508
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.31111771048642123,
      "learning_rate": 1.2736529550183595e-06,
      "loss": 0.1801,
      "step": 27509
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4802628565871495,
      "learning_rate": 1.2731685883252576e-06,
      "loss": 0.3252,
      "step": 27510
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4970044237862048,
      "learning_rate": 1.2726843074895933e-06,
      "loss": 0.2592,
      "step": 27511
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4774011203784167,
      "learning_rate": 1.272200112516131e-06,
      "loss": 0.2533,
      "step": 27512
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.30979759850516625,
      "learning_rate": 1.2717160034096386e-06,
      "loss": 0.149,
      "step": 27513
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5848110762712452,
      "learning_rate": 1.2712319801748729e-06,
      "loss": 0.3191,
      "step": 27514
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.32354244234768,
      "learning_rate": 1.270748042816602e-06,
      "loss": 0.1027,
      "step": 27515
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24114320100208336,
      "learning_rate": 1.2702641913395818e-06,
      "loss": 0.1738,
      "step": 27516
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9259115867184804,
      "learning_rate": 1.2697804257485757e-06,
      "loss": 0.4782,
      "step": 27517
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.29591818327860975,
      "learning_rate": 1.2692967460483452e-06,
      "loss": 0.1804,
      "step": 27518
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.854541693479534,
      "learning_rate": 1.2688131522436432e-06,
      "loss": 0.4404,
      "step": 27519
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6184088273222167,
      "learning_rate": 1.26832964433923e-06,
      "loss": 0.2858,
      "step": 27520
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5325236701438506,
      "learning_rate": 1.2678462223398636e-06,
      "loss": 0.3864,
      "step": 27521
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.23364645684416627,
      "learning_rate": 1.2673628862503008e-06,
      "loss": 0.1423,
      "step": 27522
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4153295093707042,
      "learning_rate": 1.2668796360752933e-06,
      "loss": 0.2508,
      "step": 27523
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.869228635847098,
      "learning_rate": 1.2663964718195976e-06,
      "loss": 0.1306,
      "step": 27524
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.312153001322637,
      "learning_rate": 1.2659133934879698e-06,
      "loss": 0.5263,
      "step": 27525
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5893233123369213,
      "learning_rate": 1.2654304010851548e-06,
      "loss": 0.1638,
      "step": 27526
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.335986068040919,
      "learning_rate": 1.2649474946159146e-06,
      "loss": 0.2652,
      "step": 27527
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2934328165491775,
      "learning_rate": 1.264464674084992e-06,
      "loss": 0.22,
      "step": 27528
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8564143673381576,
      "learning_rate": 1.263981939497142e-06,
      "loss": 0.2557,
      "step": 27529
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4401107713132352,
      "learning_rate": 1.263499290857113e-06,
      "loss": 0.2183,
      "step": 27530
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2217934193435571,
      "learning_rate": 1.2630167281696481e-06,
      "loss": 0.1279,
      "step": 27531
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.548413102441723,
      "learning_rate": 1.2625342514395045e-06,
      "loss": 0.327,
      "step": 27532
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.3503882137893861,
      "learning_rate": 1.2620518606714217e-06,
      "loss": 0.0886,
      "step": 27533
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35682234902297977,
      "learning_rate": 1.261569555870148e-06,
      "loss": 0.2803,
      "step": 27534
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5059154341292674,
      "learning_rate": 1.2610873370404308e-06,
      "loss": 0.1993,
      "step": 27535
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33101321146932877,
      "learning_rate": 1.2606052041870087e-06,
      "loss": 0.2217,
      "step": 27536
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8021592233970196,
      "learning_rate": 1.260123157314629e-06,
      "loss": 0.2705,
      "step": 27537
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7983616225336392,
      "learning_rate": 1.259641196428033e-06,
      "loss": 0.3613,
      "step": 27538
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24601261679278966,
      "learning_rate": 1.2591593215319654e-06,
      "loss": 0.1752,
      "step": 27539
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33557936724843196,
      "learning_rate": 1.2586775326311606e-06,
      "loss": 0.191,
      "step": 27540
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.33744599929847674,
      "learning_rate": 1.2581958297303653e-06,
      "loss": 0.1932,
      "step": 27541
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.508148610248517,
      "learning_rate": 1.2577142128343155e-06,
      "loss": 0.0728,
      "step": 27542
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.5344313244235903,
      "learning_rate": 1.2572326819477486e-06,
      "loss": 0.5063,
      "step": 27543
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.6868701760387996,
      "learning_rate": 1.2567512370754065e-06,
      "loss": 0.2329,
      "step": 27544
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.47887693248808816,
      "learning_rate": 1.2562698782220196e-06,
      "loss": 0.2963,
      "step": 27545
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2723602265395484,
      "learning_rate": 1.2557886053923274e-06,
      "loss": 0.2147,
      "step": 27546
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8976179723297318,
      "learning_rate": 1.255307418591063e-06,
      "loss": 0.4905,
      "step": 27547
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.8337475480835541,
      "learning_rate": 1.2548263178229646e-06,
      "loss": 0.212,
      "step": 27548
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.39833238058080767,
      "learning_rate": 1.254345303092761e-06,
      "loss": 0.1515,
      "step": 27549
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.34517811040169094,
      "learning_rate": 1.253864374405185e-06,
      "loss": 0.1734,
      "step": 27550
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5019981529236796,
      "learning_rate": 1.2533835317649711e-06,
      "loss": 0.3038,
      "step": 27551
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3970098099717557,
      "learning_rate": 1.252902775176844e-06,
      "loss": 0.1934,
      "step": 27552
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2668848441802567,
      "learning_rate": 1.252422104645542e-06,
      "loss": 0.4294,
      "step": 27553
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3419146763358695,
      "learning_rate": 1.2519415201757878e-06,
      "loss": 0.1722,
      "step": 27554
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5114416760868643,
      "learning_rate": 1.251461021772311e-06,
      "loss": 0.2403,
      "step": 27555
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.777928160598094,
      "learning_rate": 1.2509806094398414e-06,
      "loss": 0.742,
      "step": 27556
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.30586268276676254,
      "learning_rate": 1.2505002831830993e-06,
      "loss": 0.23,
      "step": 27557
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5579914165728602,
      "learning_rate": 1.2500200430068187e-06,
      "loss": 0.1832,
      "step": 27558
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.1977792191102809,
      "learning_rate": 1.2495398889157183e-06,
      "loss": 0.1337,
      "step": 27559
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.7848656305320705,
      "learning_rate": 1.2490598209145232e-06,
      "loss": 0.5915,
      "step": 27560
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.233599144052367,
      "learning_rate": 1.2485798390079596e-06,
      "loss": 0.2169,
      "step": 27561
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4400388186719467,
      "learning_rate": 1.2480999432007445e-06,
      "loss": 0.2827,
      "step": 27562
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.30188867665217084,
      "learning_rate": 1.2476201334976012e-06,
      "loss": 0.1724,
      "step": 27563
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.46342576048839024,
      "learning_rate": 1.2471404099032525e-06,
      "loss": 0.2945,
      "step": 27564
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9050472259364727,
      "learning_rate": 1.2466607724224177e-06,
      "loss": 0.3102,
      "step": 27565
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4071366430432093,
      "learning_rate": 1.2461812210598124e-06,
      "loss": 0.7028,
      "step": 27566
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.15536335063323045,
      "learning_rate": 1.2457017558201567e-06,
      "loss": 0.0693,
      "step": 27567
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.35561887138155335,
      "learning_rate": 1.2452223767081706e-06,
      "loss": 0.1996,
      "step": 27568
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5221416176698412,
      "learning_rate": 1.2447430837285623e-06,
      "loss": 0.3202,
      "step": 27569
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5025275321768913,
      "learning_rate": 1.2442638768860572e-06,
      "loss": 0.247,
      "step": 27570
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8533864398655957,
      "learning_rate": 1.2437847561853622e-06,
      "loss": 0.4106,
      "step": 27571
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.3495200724521826,
      "learning_rate": 1.243305721631195e-06,
      "loss": 0.1499,
      "step": 27572
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5575288434212289,
      "learning_rate": 1.242826773228266e-06,
      "loss": 0.2783,
      "step": 27573
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.9976535805728214,
      "learning_rate": 1.242347910981292e-06,
      "loss": 0.557,
      "step": 27574
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.32748194970220346,
      "learning_rate": 1.2418691348949786e-06,
      "loss": 0.2527,
      "step": 27575
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.7559611367561377,
      "learning_rate": 1.2413904449740376e-06,
      "loss": 0.0823,
      "step": 27576
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.345557727524457,
      "learning_rate": 1.2409118412231825e-06,
      "loss": 0.2575,
      "step": 27577
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.24053090155952334,
      "learning_rate": 1.2404333236471144e-06,
      "loss": 0.0771,
      "step": 27578
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.0678638095172006,
      "learning_rate": 1.23995489225055e-06,
      "loss": 0.3643,
      "step": 27579
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.42725472636145334,
      "learning_rate": 1.2394765470381886e-06,
      "loss": 0.2267,
      "step": 27580
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.2744773439277559,
      "learning_rate": 1.2389982880147412e-06,
      "loss": 0.1822,
      "step": 27581
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.5035081846927857,
      "learning_rate": 1.2385201151849125e-06,
      "loss": 0.2644,
      "step": 27582
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.2002406977849733,
      "learning_rate": 1.2380420285534012e-06,
      "loss": 0.4511,
      "step": 27583
    },
    {
      "epoch": 0.84,
      "grad_norm": 1.4034838450434604,
      "learning_rate": 1.2375640281249202e-06,
      "loss": 0.5869,
      "step": 27584
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.12588729553475667,
      "learning_rate": 1.2370861139041656e-06,
      "loss": 0.0691,
      "step": 27585
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.36235288522776504,
      "learning_rate": 1.2366082858958405e-06,
      "loss": 0.2705,
      "step": 27586
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.41627540707179755,
      "learning_rate": 1.236130544104649e-06,
      "loss": 0.2192,
      "step": 27587
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4591455723997289,
      "learning_rate": 1.235652888535287e-06,
      "loss": 0.3138,
      "step": 27588
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.8101784257845134,
      "learning_rate": 1.2351753191924565e-06,
      "loss": 0.2534,
      "step": 27589
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.4379272078532396,
      "learning_rate": 1.2346978360808537e-06,
      "loss": 0.2708,
      "step": 27590
    },
    {
      "epoch": 0.84,
      "grad_norm": 0.480057048449488,
      "learning_rate": 1.2342204392051816e-06,
      "loss": 0.1744,
      "step": 27591
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3032642009225108,
      "learning_rate": 1.2337431285701295e-06,
      "loss": 0.6711,
      "step": 27592
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3167399138428205,
      "learning_rate": 1.2332659041803974e-06,
      "loss": 0.2243,
      "step": 27593
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.046480670931439,
      "learning_rate": 1.2327887660406823e-06,
      "loss": 0.4744,
      "step": 27594
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.39453137648410264,
      "learning_rate": 1.2323117141556717e-06,
      "loss": 0.1711,
      "step": 27595
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3159585443562187,
      "learning_rate": 1.2318347485300676e-06,
      "loss": 0.1692,
      "step": 27596
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5269567138046933,
      "learning_rate": 1.2313578691685567e-06,
      "loss": 0.2632,
      "step": 27597
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4481423957000554,
      "learning_rate": 1.2308810760758316e-06,
      "loss": 0.1833,
      "step": 27598
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.46654466644945425,
      "learning_rate": 1.230404369256587e-06,
      "loss": 0.2867,
      "step": 27599
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3024351872849505,
      "learning_rate": 1.229927748715507e-06,
      "loss": 0.1747,
      "step": 27600
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3091349177863003,
      "learning_rate": 1.2294512144572833e-06,
      "loss": 0.6953,
      "step": 27601
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2068068893414705,
      "learning_rate": 1.2289747664866047e-06,
      "loss": 0.2989,
      "step": 27602
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5448244419498862,
      "learning_rate": 1.2284984048081606e-06,
      "loss": 0.6067,
      "step": 27603
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2596693815055394,
      "learning_rate": 1.2280221294266336e-06,
      "loss": 0.1643,
      "step": 27604
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3472072629306681,
      "learning_rate": 1.2275459403467106e-06,
      "loss": 0.3039,
      "step": 27605
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8164249852906199,
      "learning_rate": 1.2270698375730794e-06,
      "loss": 0.2677,
      "step": 27606
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2961024901366213,
      "learning_rate": 1.226593821110418e-06,
      "loss": 0.1473,
      "step": 27607
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5987438940294114,
      "learning_rate": 1.2261178909634175e-06,
      "loss": 0.0972,
      "step": 27608
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37129005630933376,
      "learning_rate": 1.2256420471367515e-06,
      "loss": 0.2231,
      "step": 27609
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6789083165175729,
      "learning_rate": 1.2251662896351114e-06,
      "loss": 0.7958,
      "step": 27610
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.32126483651283955,
      "learning_rate": 1.22469061846317e-06,
      "loss": 0.2145,
      "step": 27611
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6033721113369646,
      "learning_rate": 1.2242150336256097e-06,
      "loss": 0.4411,
      "step": 27612
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2892601800098719,
      "learning_rate": 1.2237395351271108e-06,
      "loss": 0.1651,
      "step": 27613
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.042254219744726,
      "learning_rate": 1.2232641229723497e-06,
      "loss": 0.3972,
      "step": 27614
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4400019648328002,
      "learning_rate": 1.2227887971660025e-06,
      "loss": 0.1606,
      "step": 27615
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3554486911280414,
      "learning_rate": 1.2223135577127487e-06,
      "loss": 0.24,
      "step": 27616
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33217689838544967,
      "learning_rate": 1.2218384046172637e-06,
      "loss": 0.1584,
      "step": 27617
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.586617153009709,
      "learning_rate": 1.2213633378842183e-06,
      "loss": 0.3444,
      "step": 27618
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49144390790720743,
      "learning_rate": 1.2208883575182896e-06,
      "loss": 0.26,
      "step": 27619
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4682885168967466,
      "learning_rate": 1.2204134635241516e-06,
      "loss": 0.7674,
      "step": 27620
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5490523848455549,
      "learning_rate": 1.2199386559064708e-06,
      "loss": 0.24,
      "step": 27621
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40479687394406644,
      "learning_rate": 1.2194639346699266e-06,
      "loss": 0.1755,
      "step": 27622
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2870800312792844,
      "learning_rate": 1.2189892998191833e-06,
      "loss": 0.2469,
      "step": 27623
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8950099331039001,
      "learning_rate": 1.2185147513589136e-06,
      "loss": 0.2183,
      "step": 27624
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.30502534208662474,
      "learning_rate": 1.218040289293786e-06,
      "loss": 0.1456,
      "step": 27625
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4042790365421398,
      "learning_rate": 1.2175659136284657e-06,
      "loss": 0.1399,
      "step": 27626
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37500370421813134,
      "learning_rate": 1.2170916243676223e-06,
      "loss": 0.2706,
      "step": 27627
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3903745929811135,
      "learning_rate": 1.216617421515921e-06,
      "loss": 0.2503,
      "step": 27628
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5217816822778406,
      "learning_rate": 1.2161433050780303e-06,
      "loss": 0.3027,
      "step": 27629
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7765629464615856,
      "learning_rate": 1.2156692750586097e-06,
      "loss": 0.2619,
      "step": 27630
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5301308777184524,
      "learning_rate": 1.2151953314623255e-06,
      "loss": 0.2797,
      "step": 27631
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3841727876515897,
      "learning_rate": 1.2147214742938418e-06,
      "loss": 0.2175,
      "step": 27632
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4435852818059648,
      "learning_rate": 1.214247703557815e-06,
      "loss": 0.2017,
      "step": 27633
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2375920048744018,
      "learning_rate": 1.2137740192589143e-06,
      "loss": 0.1442,
      "step": 27634
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.31337604293292853,
      "learning_rate": 1.2133004214017941e-06,
      "loss": 0.1641,
      "step": 27635
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3745346162925058,
      "learning_rate": 1.2128269099911149e-06,
      "loss": 0.2659,
      "step": 27636
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3215823361527752,
      "learning_rate": 1.2123534850315388e-06,
      "loss": 0.4765,
      "step": 27637
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8790632472756443,
      "learning_rate": 1.2118801465277153e-06,
      "loss": 0.4407,
      "step": 27638
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6429955765305054,
      "learning_rate": 1.2114068944843116e-06,
      "loss": 0.2747,
      "step": 27639
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.39150247396765014,
      "learning_rate": 1.2109337289059764e-06,
      "loss": 0.2372,
      "step": 27640
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33466424600915345,
      "learning_rate": 1.210460649797367e-06,
      "loss": 0.2083,
      "step": 27641
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49545101328979463,
      "learning_rate": 1.2099876571631386e-06,
      "loss": 0.3014,
      "step": 27642
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.1828617137105642,
      "learning_rate": 1.2095147510079452e-06,
      "loss": 0.068,
      "step": 27643
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8496781644237643,
      "learning_rate": 1.2090419313364366e-06,
      "loss": 0.4219,
      "step": 27644
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3725551482320107,
      "learning_rate": 1.2085691981532656e-06,
      "loss": 0.1584,
      "step": 27645
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5694075660757344,
      "learning_rate": 1.208096551463086e-06,
      "loss": 0.3794,
      "step": 27646
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33446957891092943,
      "learning_rate": 1.2076239912705412e-06,
      "loss": 0.2352,
      "step": 27647
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7828301358447031,
      "learning_rate": 1.2071515175802894e-06,
      "loss": 0.2564,
      "step": 27648
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9458635419455739,
      "learning_rate": 1.2066791303969716e-06,
      "loss": 0.4026,
      "step": 27649
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3550697915107334,
      "learning_rate": 1.2062068297252371e-06,
      "loss": 0.1897,
      "step": 27650
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.604007624675187,
      "learning_rate": 1.2057346155697369e-06,
      "loss": 0.5573,
      "step": 27651
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.27916628834344825,
      "learning_rate": 1.2052624879351105e-06,
      "loss": 0.181,
      "step": 27652
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6710073340594382,
      "learning_rate": 1.2047904468260053e-06,
      "loss": 0.8238,
      "step": 27653
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2879928200986796,
      "learning_rate": 1.2043184922470664e-06,
      "loss": 0.1675,
      "step": 27654
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3413554460495878,
      "learning_rate": 1.2038466242029378e-06,
      "loss": 0.7506,
      "step": 27655
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6096146068013107,
      "learning_rate": 1.2033748426982584e-06,
      "loss": 0.2885,
      "step": 27656
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4857055913075181,
      "learning_rate": 1.2029031477376717e-06,
      "loss": 0.2231,
      "step": 27657
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3574013302472369,
      "learning_rate": 1.2024315393258202e-06,
      "loss": 0.1725,
      "step": 27658
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29517717815522077,
      "learning_rate": 1.2019600174673386e-06,
      "loss": 0.2478,
      "step": 27659
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3862484183654666,
      "learning_rate": 1.201488582166873e-06,
      "loss": 0.0942,
      "step": 27660
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49079036948294086,
      "learning_rate": 1.201017233429056e-06,
      "loss": 0.095,
      "step": 27661
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.306638476314503,
      "learning_rate": 1.2005459712585265e-06,
      "loss": 0.7183,
      "step": 27662
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28202550570984175,
      "learning_rate": 1.2000747956599235e-06,
      "loss": 0.1598,
      "step": 27663
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4768347274365427,
      "learning_rate": 1.1996037066378764e-06,
      "loss": 0.3191,
      "step": 27664
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5555074290373001,
      "learning_rate": 1.1991327041970268e-06,
      "loss": 0.2422,
      "step": 27665
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.47746800553465746,
      "learning_rate": 1.198661788342005e-06,
      "loss": 0.245,
      "step": 27666
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.322416808492959,
      "learning_rate": 1.1981909590774431e-06,
      "loss": 0.1423,
      "step": 27667
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6074804976451351,
      "learning_rate": 1.197720216407977e-06,
      "loss": 0.323,
      "step": 27668
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7973767436142605,
      "learning_rate": 1.197249560338234e-06,
      "loss": 0.1191,
      "step": 27669
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3540746867192776,
      "learning_rate": 1.1967789908728466e-06,
      "loss": 0.2831,
      "step": 27670
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40949661306933266,
      "learning_rate": 1.196308508016444e-06,
      "loss": 0.1813,
      "step": 27671
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9656493527593359,
      "learning_rate": 1.1958381117736574e-06,
      "loss": 0.4856,
      "step": 27672
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4102120658569001,
      "learning_rate": 1.1953678021491067e-06,
      "loss": 0.2321,
      "step": 27673
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8644782137696084,
      "learning_rate": 1.1948975791474305e-06,
      "loss": 0.3078,
      "step": 27674
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2535035800688353,
      "learning_rate": 1.1944274427732483e-06,
      "loss": 0.1943,
      "step": 27675
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.365899645218191,
      "learning_rate": 1.1939573930311831e-06,
      "loss": 0.1936,
      "step": 27676
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3376917757850987,
      "learning_rate": 1.1934874299258658e-06,
      "loss": 0.241,
      "step": 27677
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9599663403224892,
      "learning_rate": 1.193017553461916e-06,
      "loss": 0.0915,
      "step": 27678
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.484534261006595,
      "learning_rate": 1.1925477636439563e-06,
      "loss": 0.62,
      "step": 27679
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9704613437020214,
      "learning_rate": 1.1920780604766102e-06,
      "loss": 0.2776,
      "step": 27680
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5738155517733958,
      "learning_rate": 1.1916084439645003e-06,
      "loss": 0.3173,
      "step": 27681
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28644095018727267,
      "learning_rate": 1.191138914112242e-06,
      "loss": 0.1999,
      "step": 27682
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4532202328142832,
      "learning_rate": 1.1906694709244582e-06,
      "loss": 0.284,
      "step": 27683
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23191265119413204,
      "learning_rate": 1.1902001144057674e-06,
      "loss": 0.0958,
      "step": 27684
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41065899526788396,
      "learning_rate": 1.1897308445607859e-06,
      "loss": 0.2203,
      "step": 27685
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3388353800918661,
      "learning_rate": 1.1892616613941343e-06,
      "loss": 0.1916,
      "step": 27686
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.872117659990225,
      "learning_rate": 1.1887925649104226e-06,
      "loss": 0.1108,
      "step": 27687
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3300536719060751,
      "learning_rate": 1.1883235551142691e-06,
      "loss": 0.2613,
      "step": 27688
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7917877032031585,
      "learning_rate": 1.1878546320102902e-06,
      "loss": 0.2565,
      "step": 27689
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4007887775792019,
      "learning_rate": 1.1873857956030933e-06,
      "loss": 0.2571,
      "step": 27690
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4846470447912771,
      "learning_rate": 1.186917045897298e-06,
      "loss": 0.2392,
      "step": 27691
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3934956460773877,
      "learning_rate": 1.1864483828975103e-06,
      "loss": 0.6959,
      "step": 27692
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2889628980471956,
      "learning_rate": 1.185979806608344e-06,
      "loss": 0.1636,
      "step": 27693
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38926170362741125,
      "learning_rate": 1.1855113170344113e-06,
      "loss": 0.21,
      "step": 27694
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.31656743248440683,
      "learning_rate": 1.1850429141803154e-06,
      "loss": 0.1832,
      "step": 27695
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.65138345891557,
      "learning_rate": 1.1845745980506685e-06,
      "loss": 0.6487,
      "step": 27696
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0795065886839728,
      "learning_rate": 1.1841063686500775e-06,
      "loss": 0.2553,
      "step": 27697
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6697296851634484,
      "learning_rate": 1.1836382259831503e-06,
      "loss": 0.4025,
      "step": 27698
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28590459560612114,
      "learning_rate": 1.1831701700544896e-06,
      "loss": 0.0694,
      "step": 27699
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26682496921947313,
      "learning_rate": 1.1827022008687016e-06,
      "loss": 0.2133,
      "step": 27700
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49156270585684,
      "learning_rate": 1.182234318430393e-06,
      "loss": 0.3508,
      "step": 27701
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21047029817289248,
      "learning_rate": 1.18176652274416e-06,
      "loss": 0.0865,
      "step": 27702
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.46201464748394366,
      "learning_rate": 1.1812988138146143e-06,
      "loss": 0.2134,
      "step": 27703
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2872835944172717,
      "learning_rate": 1.18083119164635e-06,
      "loss": 0.1712,
      "step": 27704
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.189750297525782,
      "learning_rate": 1.18036365624397e-06,
      "loss": 0.3509,
      "step": 27705
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3348182048953785,
      "learning_rate": 1.179896207612078e-06,
      "loss": 0.2264,
      "step": 27706
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9952520174588204,
      "learning_rate": 1.179428845755266e-06,
      "loss": 0.5049,
      "step": 27707
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5965782093103518,
      "learning_rate": 1.1789615706781354e-06,
      "loss": 0.203,
      "step": 27708
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3975433649039297,
      "learning_rate": 1.1784943823852835e-06,
      "loss": 0.2832,
      "step": 27709
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1248830478662681,
      "learning_rate": 1.1780272808813065e-06,
      "loss": 0.3258,
      "step": 27710
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4795217308465507,
      "learning_rate": 1.1775602661708008e-06,
      "loss": 0.5569,
      "step": 27711
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20027664005276005,
      "learning_rate": 1.1770933382583604e-06,
      "loss": 0.1549,
      "step": 27712
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22483705846167593,
      "learning_rate": 1.1766264971485787e-06,
      "loss": 0.1473,
      "step": 27713
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.6371711116260628,
      "learning_rate": 1.176159742846048e-06,
      "loss": 0.5617,
      "step": 27714
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7617320548440842,
      "learning_rate": 1.175693075355363e-06,
      "loss": 0.2621,
      "step": 27715
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7517260332521746,
      "learning_rate": 1.1752264946811087e-06,
      "loss": 0.3958,
      "step": 27716
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3283064246259937,
      "learning_rate": 1.1747600008278848e-06,
      "loss": 0.1719,
      "step": 27717
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28516588299342693,
      "learning_rate": 1.174293593800273e-06,
      "loss": 0.2599,
      "step": 27718
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3340406906501823,
      "learning_rate": 1.1738272736028656e-06,
      "loss": 0.4597,
      "step": 27719
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5047828080789276,
      "learning_rate": 1.1733610402402518e-06,
      "loss": 0.7581,
      "step": 27720
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.17193053851261625,
      "learning_rate": 1.1728948937170136e-06,
      "loss": 0.0699,
      "step": 27721
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3589377883815003,
      "learning_rate": 1.172428834037741e-06,
      "loss": 0.2386,
      "step": 27722
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8716433902404526,
      "learning_rate": 1.1719628612070178e-06,
      "loss": 0.2318,
      "step": 27723
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3589881077887354,
      "learning_rate": 1.1714969752294304e-06,
      "loss": 0.2706,
      "step": 27724
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.107356247686122,
      "learning_rate": 1.1710311761095583e-06,
      "loss": 0.2086,
      "step": 27725
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49553091735934,
      "learning_rate": 1.1705654638519881e-06,
      "loss": 0.148,
      "step": 27726
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3685032386859997,
      "learning_rate": 1.1700998384613006e-06,
      "loss": 0.2511,
      "step": 27727
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0052212510112704,
      "learning_rate": 1.1696342999420728e-06,
      "loss": 0.3252,
      "step": 27728
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40512099882568786,
      "learning_rate": 1.1691688482988928e-06,
      "loss": 0.2948,
      "step": 27729
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23083454009034232,
      "learning_rate": 1.1687034835363332e-06,
      "loss": 0.0706,
      "step": 27730
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4010729335600695,
      "learning_rate": 1.1682382056589748e-06,
      "loss": 0.2597,
      "step": 27731
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3109839786120384,
      "learning_rate": 1.1677730146713962e-06,
      "loss": 0.1752,
      "step": 27732
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7090456924193096,
      "learning_rate": 1.1673079105781715e-06,
      "loss": 0.3697,
      "step": 27733
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7048587953779082,
      "learning_rate": 1.1668428933838781e-06,
      "loss": 0.1564,
      "step": 27734
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4282604904786298,
      "learning_rate": 1.1663779630930906e-06,
      "loss": 0.3001,
      "step": 27735
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.257789151285159,
      "learning_rate": 1.1659131197103824e-06,
      "loss": 0.1984,
      "step": 27736
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3597608249796245,
      "learning_rate": 1.1654483632403313e-06,
      "loss": 0.759,
      "step": 27737
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9846895257627837,
      "learning_rate": 1.1649836936875024e-06,
      "loss": 0.2025,
      "step": 27738
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43878094840958526,
      "learning_rate": 1.1645191110564723e-06,
      "loss": 0.1306,
      "step": 27739
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.45465952072954946,
      "learning_rate": 1.1640546153518107e-06,
      "loss": 0.2593,
      "step": 27740
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35763476357560586,
      "learning_rate": 1.163590206578088e-06,
      "loss": 0.2274,
      "step": 27741
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4984007159504635,
      "learning_rate": 1.1631258847398697e-06,
      "loss": 0.3142,
      "step": 27742
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3961415396407337,
      "learning_rate": 1.16266164984173e-06,
      "loss": 0.1402,
      "step": 27743
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7029492228004806,
      "learning_rate": 1.1621975018882303e-06,
      "loss": 0.4068,
      "step": 27744
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34843397805484355,
      "learning_rate": 1.161733440883941e-06,
      "loss": 0.227,
      "step": 27745
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.625964880578087,
      "learning_rate": 1.1612694668334268e-06,
      "loss": 0.6772,
      "step": 27746
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33388417705413753,
      "learning_rate": 1.160805579741251e-06,
      "loss": 0.2161,
      "step": 27747
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9168015586020882,
      "learning_rate": 1.1603417796119777e-06,
      "loss": 0.4485,
      "step": 27748
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3249697109924064,
      "learning_rate": 1.1598780664501709e-06,
      "loss": 0.1606,
      "step": 27749
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4883779592527696,
      "learning_rate": 1.1594144402603946e-06,
      "loss": 0.3145,
      "step": 27750
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4658082920282791,
      "learning_rate": 1.1589509010472055e-06,
      "loss": 0.1864,
      "step": 27751
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23833038252987793,
      "learning_rate": 1.1584874488151665e-06,
      "loss": 0.1486,
      "step": 27752
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4400159293912947,
      "learning_rate": 1.1580240835688406e-06,
      "loss": 0.2867,
      "step": 27753
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33049730061333993,
      "learning_rate": 1.1575608053127775e-06,
      "loss": 0.1834,
      "step": 27754
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3650054227525832,
      "learning_rate": 1.1570976140515455e-06,
      "loss": 0.7722,
      "step": 27755
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2021664747392953,
      "learning_rate": 1.1566345097896947e-06,
      "loss": 0.1184,
      "step": 27756
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3735774417053912,
      "learning_rate": 1.1561714925317836e-06,
      "loss": 0.4025,
      "step": 27757
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5348129377657758,
      "learning_rate": 1.1557085622823694e-06,
      "loss": 0.1644,
      "step": 27758
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3997726310873571,
      "learning_rate": 1.155245719046002e-06,
      "loss": 0.2844,
      "step": 27759
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23774085045020868,
      "learning_rate": 1.1547829628272378e-06,
      "loss": 0.1683,
      "step": 27760
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3952348421017556,
      "learning_rate": 1.1543202936306297e-06,
      "loss": 0.1804,
      "step": 27761
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6497044262103959,
      "learning_rate": 1.1538577114607286e-06,
      "loss": 0.1707,
      "step": 27762
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35156153286127134,
      "learning_rate": 1.1533952163220886e-06,
      "loss": 0.2533,
      "step": 27763
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2400634082768844,
      "learning_rate": 1.1529328082192547e-06,
      "loss": 0.26,
      "step": 27764
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.30911290721559503,
      "learning_rate": 1.1524704871567794e-06,
      "loss": 0.2183,
      "step": 27765
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8051186837105084,
      "learning_rate": 1.1520082531392106e-06,
      "loss": 0.3309,
      "step": 27766
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36062522664995494,
      "learning_rate": 1.1515461061710976e-06,
      "loss": 0.1661,
      "step": 27767
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5583072437817552,
      "learning_rate": 1.1510840462569839e-06,
      "loss": 0.3325,
      "step": 27768
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8347712498881491,
      "learning_rate": 1.1506220734014173e-06,
      "loss": 0.0406,
      "step": 27769
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29635761961788737,
      "learning_rate": 1.1501601876089442e-06,
      "loss": 0.2001,
      "step": 27770
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29526474498540994,
      "learning_rate": 1.149698388884103e-06,
      "loss": 0.2179,
      "step": 27771
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.372704255745021,
      "learning_rate": 1.1492366772314455e-06,
      "loss": 0.257,
      "step": 27772
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0182104489157644,
      "learning_rate": 1.1487750526555087e-06,
      "loss": 0.4252,
      "step": 27773
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9116495696764984,
      "learning_rate": 1.1483135151608339e-06,
      "loss": 0.3669,
      "step": 27774
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8158668448289915,
      "learning_rate": 1.1478520647519664e-06,
      "loss": 0.2373,
      "step": 27775
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4179465746705734,
      "learning_rate": 1.1473907014334395e-06,
      "loss": 0.2649,
      "step": 27776
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.583174081809219,
      "learning_rate": 1.1469294252097972e-06,
      "loss": 0.2035,
      "step": 27777
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33758652579197185,
      "learning_rate": 1.146468236085576e-06,
      "loss": 0.265,
      "step": 27778
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.18628116202233727,
      "learning_rate": 1.1460071340653155e-06,
      "loss": 0.0816,
      "step": 27779
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.47593940226153125,
      "learning_rate": 1.1455461191535455e-06,
      "loss": 0.2109,
      "step": 27780
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.48828491588139883,
      "learning_rate": 1.1450851913548111e-06,
      "loss": 0.3346,
      "step": 27781
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.48187945208648747,
      "learning_rate": 1.1446243506736398e-06,
      "loss": 0.2078,
      "step": 27782
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3255769595892368,
      "learning_rate": 1.1441635971145682e-06,
      "loss": 0.2656,
      "step": 27783
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8885871529642713,
      "learning_rate": 1.1437029306821313e-06,
      "loss": 0.2574,
      "step": 27784
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9333813633253172,
      "learning_rate": 1.1432423513808566e-06,
      "loss": 0.3672,
      "step": 27785
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3322517636527608,
      "learning_rate": 1.1427818592152773e-06,
      "loss": 0.1889,
      "step": 27786
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.283536694561807,
      "learning_rate": 1.1423214541899253e-06,
      "loss": 0.3921,
      "step": 27787
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2606896761262485,
      "learning_rate": 1.1418611363093302e-06,
      "loss": 0.0998,
      "step": 27788
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34754027397906045,
      "learning_rate": 1.1414009055780206e-06,
      "loss": 0.2899,
      "step": 27789
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3288513915008616,
      "learning_rate": 1.140940762000522e-06,
      "loss": 0.1645,
      "step": 27790
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33764617123214286,
      "learning_rate": 1.1404807055813637e-06,
      "loss": 0.2287,
      "step": 27791
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8141293570239911,
      "learning_rate": 1.1400207363250715e-06,
      "loss": 0.2438,
      "step": 27792
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0898591801542554,
      "learning_rate": 1.1395608542361714e-06,
      "loss": 0.2698,
      "step": 27793
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6249552655409163,
      "learning_rate": 1.1391010593191865e-06,
      "loss": 0.2543,
      "step": 27794
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.24111777750023292,
      "learning_rate": 1.13864135157864e-06,
      "loss": 0.2128,
      "step": 27795
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4939755635539553,
      "learning_rate": 1.138181731019058e-06,
      "loss": 0.5618,
      "step": 27796
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.32329226607015676,
      "learning_rate": 1.137722197644956e-06,
      "loss": 0.0932,
      "step": 27797
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4495754540080459,
      "learning_rate": 1.1372627514608636e-06,
      "loss": 0.7244,
      "step": 27798
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2940902820742104,
      "learning_rate": 1.136803392471294e-06,
      "loss": 0.1756,
      "step": 27799
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.0498359779358422,
      "learning_rate": 1.136344120680769e-06,
      "loss": 0.4384,
      "step": 27800
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3569980831050955,
      "learning_rate": 1.1358849360938095e-06,
      "loss": 0.2297,
      "step": 27801
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5330931329702917,
      "learning_rate": 1.1354258387149275e-06,
      "loss": 0.2555,
      "step": 27802
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2547530499966026,
      "learning_rate": 1.1349668285486447e-06,
      "loss": 0.0656,
      "step": 27803
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5615332076528433,
      "learning_rate": 1.1345079055994735e-06,
      "loss": 0.3003,
      "step": 27804
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5515875145303968,
      "learning_rate": 1.1340490698719341e-06,
      "loss": 0.1974,
      "step": 27805
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.30638308601737496,
      "learning_rate": 1.1335903213705345e-06,
      "loss": 0.2217,
      "step": 27806
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8822322322068429,
      "learning_rate": 1.133131660099791e-06,
      "loss": 0.5189,
      "step": 27807
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4094143537023366,
      "learning_rate": 1.1326730860642178e-06,
      "loss": 0.1744,
      "step": 27808
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5550560153403077,
      "learning_rate": 1.1322145992683198e-06,
      "loss": 0.3152,
      "step": 27809
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.46820056837087504,
      "learning_rate": 1.1317561997166171e-06,
      "loss": 0.1858,
      "step": 27810
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5462448507253866,
      "learning_rate": 1.1312978874136106e-06,
      "loss": 0.1964,
      "step": 27811
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.393788437009173,
      "learning_rate": 1.1308396623638186e-06,
      "loss": 0.2057,
      "step": 27812
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3600567323322439,
      "learning_rate": 1.1303815245717408e-06,
      "loss": 0.2679,
      "step": 27813
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4050688674661403,
      "learning_rate": 1.1299234740418896e-06,
      "loss": 0.1774,
      "step": 27814
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7651751542567264,
      "learning_rate": 1.1294655107787712e-06,
      "loss": 0.7653,
      "step": 27815
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8961943373255805,
      "learning_rate": 1.1290076347868873e-06,
      "loss": 0.3378,
      "step": 27816
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6079626683940218,
      "learning_rate": 1.1285498460707468e-06,
      "loss": 0.2578,
      "step": 27817
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.33981780902369396,
      "learning_rate": 1.1280921446348503e-06,
      "loss": 0.2241,
      "step": 27818
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29131199733278595,
      "learning_rate": 1.1276345304837055e-06,
      "loss": 0.2103,
      "step": 27819
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.429217842016299,
      "learning_rate": 1.1271770036218089e-06,
      "loss": 0.2285,
      "step": 27820
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3422726195378288,
      "learning_rate": 1.1267195640536654e-06,
      "loss": 0.0863,
      "step": 27821
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36455537417753414,
      "learning_rate": 1.1262622117837763e-06,
      "loss": 0.2398,
      "step": 27822
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.7353647692020266,
      "learning_rate": 1.1258049468166342e-06,
      "loss": 0.0701,
      "step": 27823
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.856123786730776,
      "learning_rate": 1.1253477691567483e-06,
      "loss": 0.6875,
      "step": 27824
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35229602788953757,
      "learning_rate": 1.1248906788086089e-06,
      "loss": 0.2325,
      "step": 27825
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.41075377902140414,
      "learning_rate": 1.124433675776716e-06,
      "loss": 0.2511,
      "step": 27826
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5384116730931381,
      "learning_rate": 1.123976760065566e-06,
      "loss": 0.2142,
      "step": 27827
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.896401810483216,
      "learning_rate": 1.123519931679652e-06,
      "loss": 0.4285,
      "step": 27828
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.23341488629611262,
      "learning_rate": 1.123063190623469e-06,
      "loss": 0.0849,
      "step": 27829
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.30961355312404937,
      "learning_rate": 1.1226065369015116e-06,
      "loss": 0.2444,
      "step": 27830
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3748725763950862,
      "learning_rate": 1.1221499705182736e-06,
      "loss": 0.1694,
      "step": 27831
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.49837422413800264,
      "learning_rate": 1.1216934914782441e-06,
      "loss": 0.1841,
      "step": 27832
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.095868832908679,
      "learning_rate": 1.1212370997859146e-06,
      "loss": 0.4628,
      "step": 27833
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7480087565652125,
      "learning_rate": 1.1207807954457784e-06,
      "loss": 0.2755,
      "step": 27834
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8840341317149816,
      "learning_rate": 1.1203245784623185e-06,
      "loss": 0.3518,
      "step": 27835
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3252080474528789,
      "learning_rate": 1.1198684488400313e-06,
      "loss": 0.1952,
      "step": 27836
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.36717928388004945,
      "learning_rate": 1.119412406583399e-06,
      "loss": 0.284,
      "step": 27837
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22140335356507657,
      "learning_rate": 1.1189564516969077e-06,
      "loss": 0.0912,
      "step": 27838
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.42802334051224533,
      "learning_rate": 1.1185005841850483e-06,
      "loss": 0.2157,
      "step": 27839
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2972642386045187,
      "learning_rate": 1.1180448040522984e-06,
      "loss": 0.1709,
      "step": 27840
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.2414193302971135,
      "learning_rate": 1.11758911130315e-06,
      "loss": 0.4652,
      "step": 27841
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4621985141875338,
      "learning_rate": 1.1171335059420808e-06,
      "loss": 0.1992,
      "step": 27842
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4634291199632974,
      "learning_rate": 1.1166779879735756e-06,
      "loss": 0.3031,
      "step": 27843
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5473824679332243,
      "learning_rate": 1.1162225574021147e-06,
      "loss": 0.1947,
      "step": 27844
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3369230976281598,
      "learning_rate": 1.1157672142321807e-06,
      "loss": 0.2196,
      "step": 27845
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.4930416580173966,
      "learning_rate": 1.1153119584682514e-06,
      "loss": 0.7914,
      "step": 27846
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.21754889474734426,
      "learning_rate": 1.1148567901148054e-06,
      "loss": 0.0742,
      "step": 27847
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35572658507068916,
      "learning_rate": 1.1144017091763248e-06,
      "loss": 0.2831,
      "step": 27848
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.30366468948810144,
      "learning_rate": 1.1139467156572804e-06,
      "loss": 0.1713,
      "step": 27849
    },
    {
      "epoch": 0.85,
      "grad_norm": 2.0014556335493285,
      "learning_rate": 1.113491809562155e-06,
      "loss": 0.6122,
      "step": 27850
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9185331648204503,
      "learning_rate": 1.1130369908954198e-06,
      "loss": 0.2707,
      "step": 27851
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7389295281683946,
      "learning_rate": 1.11258225966155e-06,
      "loss": 0.3475,
      "step": 27852
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34222367597935094,
      "learning_rate": 1.1121276158650229e-06,
      "loss": 0.1764,
      "step": 27853
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5700725592617845,
      "learning_rate": 1.1116730595103076e-06,
      "loss": 0.3229,
      "step": 27854
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3008609046650234,
      "learning_rate": 1.1112185906018768e-06,
      "loss": 0.2241,
      "step": 27855
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5567711608949544,
      "learning_rate": 1.1107642091442017e-06,
      "loss": 0.6795,
      "step": 27856
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20444832182030528,
      "learning_rate": 1.1103099151417563e-06,
      "loss": 0.0759,
      "step": 27857
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4005349719412491,
      "learning_rate": 1.1098557085990035e-06,
      "loss": 0.1695,
      "step": 27858
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5866504132848095,
      "learning_rate": 1.109401589520417e-06,
      "loss": 0.3238,
      "step": 27859
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3303396968297677,
      "learning_rate": 1.1089475579104646e-06,
      "loss": 0.2423,
      "step": 27860
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8639205781253543,
      "learning_rate": 1.108493613773607e-06,
      "loss": 0.4419,
      "step": 27861
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4309798897829643,
      "learning_rate": 1.1080397571143197e-06,
      "loss": 0.1375,
      "step": 27862
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3366663980080193,
      "learning_rate": 1.107585987937061e-06,
      "loss": 0.2299,
      "step": 27863
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.093191533257418,
      "learning_rate": 1.1071323062462969e-06,
      "loss": 0.3356,
      "step": 27864
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.47408634815024686,
      "learning_rate": 1.1066787120464938e-06,
      "loss": 0.2213,
      "step": 27865
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3061392517223459,
      "learning_rate": 1.1062252053421074e-06,
      "loss": 0.2072,
      "step": 27866
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.31627820610088087,
      "learning_rate": 1.1057717861376083e-06,
      "loss": 0.1885,
      "step": 27867
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.38142872589303206,
      "learning_rate": 1.1053184544374506e-06,
      "loss": 0.2085,
      "step": 27868
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.7365542387017922,
      "learning_rate": 1.1048652102460978e-06,
      "loss": 0.3469,
      "step": 27869
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3711509169643592,
      "learning_rate": 1.104412053568008e-06,
      "loss": 0.1289,
      "step": 27870
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3366604882067017,
      "learning_rate": 1.1039589844076393e-06,
      "loss": 0.1534,
      "step": 27871
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.31655222725650467,
      "learning_rate": 1.1035060027694478e-06,
      "loss": 0.2514,
      "step": 27872
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4369313707554535,
      "learning_rate": 1.1030531086578922e-06,
      "loss": 0.2597,
      "step": 27873
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.805330781605561,
      "learning_rate": 1.1026003020774301e-06,
      "loss": 0.6797,
      "step": 27874
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.2550534189615268,
      "learning_rate": 1.1021475830325112e-06,
      "loss": 0.1057,
      "step": 27875
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.387756722114242,
      "learning_rate": 1.101694951527592e-06,
      "loss": 0.2438,
      "step": 27876
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5293273732250968,
      "learning_rate": 1.1012424075671291e-06,
      "loss": 0.2238,
      "step": 27877
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.43418125737551333,
      "learning_rate": 1.1007899511555665e-06,
      "loss": 0.3089,
      "step": 27878
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4686147614339202,
      "learning_rate": 1.1003375822973639e-06,
      "loss": 0.1893,
      "step": 27879
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9058343300061169,
      "learning_rate": 1.0998853009969678e-06,
      "loss": 0.438,
      "step": 27880
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.40434417709452025,
      "learning_rate": 1.0994331072588282e-06,
      "loss": 0.1793,
      "step": 27881
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5286964173444986,
      "learning_rate": 1.0989810010873937e-06,
      "loss": 0.3392,
      "step": 27882
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8595202197708135,
      "learning_rate": 1.098528982487116e-06,
      "loss": 0.0493,
      "step": 27883
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.22405172545434432,
      "learning_rate": 1.0980770514624361e-06,
      "loss": 0.1637,
      "step": 27884
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9373053261655254,
      "learning_rate": 1.097625208017804e-06,
      "loss": 0.3796,
      "step": 27885
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3814795661115586,
      "learning_rate": 1.097173452157666e-06,
      "loss": 0.2339,
      "step": 27886
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4836348238346194,
      "learning_rate": 1.0967217838864619e-06,
      "loss": 0.2657,
      "step": 27887
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.5419321847937744,
      "learning_rate": 1.0962702032086414e-06,
      "loss": 0.2792,
      "step": 27888
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4876446370206538,
      "learning_rate": 1.0958187101286421e-06,
      "loss": 0.2906,
      "step": 27889
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.28896220747910917,
      "learning_rate": 1.0953673046509094e-06,
      "loss": 0.1682,
      "step": 27890
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5056946072694074,
      "learning_rate": 1.0949159867798843e-06,
      "loss": 0.3426,
      "step": 27891
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.1017595110219678,
      "learning_rate": 1.0944647565200028e-06,
      "loss": 0.1365,
      "step": 27892
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.9681082732733098,
      "learning_rate": 1.0940136138757107e-06,
      "loss": 0.4386,
      "step": 27893
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5409091261215927,
      "learning_rate": 1.0935625588514408e-06,
      "loss": 0.2051,
      "step": 27894
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37997489181717026,
      "learning_rate": 1.0931115914516333e-06,
      "loss": 0.2788,
      "step": 27895
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.20334170289511108,
      "learning_rate": 1.0926607116807263e-06,
      "loss": 0.1618,
      "step": 27896
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5975528376338611,
      "learning_rate": 1.0922099195431535e-06,
      "loss": 0.0256,
      "step": 27897
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.3736604106877819,
      "learning_rate": 1.0917592150433486e-06,
      "loss": 0.7235,
      "step": 27898
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.26205628924137847,
      "learning_rate": 1.0913085981857496e-06,
      "loss": 0.1637,
      "step": 27899
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.618139042380533,
      "learning_rate": 1.0908580689747894e-06,
      "loss": 0.6642,
      "step": 27900
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8680061812843977,
      "learning_rate": 1.0904076274148966e-06,
      "loss": 0.214,
      "step": 27901
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.357963617946223,
      "learning_rate": 1.0899572735105047e-06,
      "loss": 0.295,
      "step": 27902
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.34993084785893824,
      "learning_rate": 1.0895070072660473e-06,
      "loss": 0.1745,
      "step": 27903
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5307672586426854,
      "learning_rate": 1.0890568286859493e-06,
      "loss": 0.2851,
      "step": 27904
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.35420406465889165,
      "learning_rate": 1.0886067377746445e-06,
      "loss": 0.0667,
      "step": 27905
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4987206788137643,
      "learning_rate": 1.0881567345365574e-06,
      "loss": 0.2884,
      "step": 27906
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.37785371602733775,
      "learning_rate": 1.0877068189761175e-06,
      "loss": 0.2329,
      "step": 27907
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4557093537789931,
      "learning_rate": 1.0872569910977515e-06,
      "loss": 0.2556,
      "step": 27908
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3387870249073962,
      "learning_rate": 1.0868072509058824e-06,
      "loss": 0.2142,
      "step": 27909
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.8883722422724705,
      "learning_rate": 1.0863575984049358e-06,
      "loss": 0.2355,
      "step": 27910
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.6532811986246978,
      "learning_rate": 1.0859080335993367e-06,
      "loss": 0.355,
      "step": 27911
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.736438720574692,
      "learning_rate": 1.0854585564935094e-06,
      "loss": 0.0666,
      "step": 27912
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3532213445036079,
      "learning_rate": 1.0850091670918706e-06,
      "loss": 0.2379,
      "step": 27913
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.3138948230935659,
      "learning_rate": 1.0845598653988477e-06,
      "loss": 0.2017,
      "step": 27914
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.29716580991701,
      "learning_rate": 1.084110651418857e-06,
      "loss": 0.1652,
      "step": 27915
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.136287543531528,
      "learning_rate": 1.08366152515632e-06,
      "loss": 0.4392,
      "step": 27916
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.4280383483168482,
      "learning_rate": 1.0832124866156568e-06,
      "loss": 0.2757,
      "step": 27917
    },
    {
      "epoch": 0.85,
      "grad_norm": 0.5177518476491727,
      "learning_rate": 1.082763535801279e-06,
      "loss": 0.2064,
      "step": 27918
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.434993286635051,
      "learning_rate": 1.0823146727176115e-06,
      "loss": 0.294,
      "step": 27919
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4858285725916551,
      "learning_rate": 1.0818658973690644e-06,
      "loss": 0.2372,
      "step": 27920
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8064485151194228,
      "learning_rate": 1.081417209760055e-06,
      "loss": 0.3072,
      "step": 27921
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31088538669126153,
      "learning_rate": 1.080968609895e-06,
      "loss": 0.1774,
      "step": 27922
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2776705262872827,
      "learning_rate": 1.0805200977783081e-06,
      "loss": 0.0796,
      "step": 27923
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6887661018594915,
      "learning_rate": 1.0800716734143947e-06,
      "loss": 0.7365,
      "step": 27924
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31631087048552126,
      "learning_rate": 1.0796233368076715e-06,
      "loss": 0.2365,
      "step": 27925
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3999530219218617,
      "learning_rate": 1.0791750879625506e-06,
      "loss": 0.2455,
      "step": 27926
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4764747873582053,
      "learning_rate": 1.0787269268834388e-06,
      "loss": 0.2244,
      "step": 27927
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.813356383030613,
      "learning_rate": 1.0782788535747468e-06,
      "loss": 0.3772,
      "step": 27928
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8505682439416521,
      "learning_rate": 1.0778308680408855e-06,
      "loss": 0.2313,
      "step": 27929
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8081506999655206,
      "learning_rate": 1.0773829702862549e-06,
      "loss": 0.2969,
      "step": 27930
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31099795146414433,
      "learning_rate": 1.076935160315271e-06,
      "loss": 0.1844,
      "step": 27931
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35837936046693003,
      "learning_rate": 1.0764874381323342e-06,
      "loss": 0.2928,
      "step": 27932
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22533226297462847,
      "learning_rate": 1.0760398037418495e-06,
      "loss": 0.0955,
      "step": 27933
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.351884224002997,
      "learning_rate": 1.0755922571482237e-06,
      "loss": 0.7806,
      "step": 27934
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3238037335168881,
      "learning_rate": 1.075144798355855e-06,
      "loss": 0.1382,
      "step": 27935
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36754801585718644,
      "learning_rate": 1.074697427369149e-06,
      "loss": 0.1601,
      "step": 27936
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6016511206780385,
      "learning_rate": 1.0742501441925069e-06,
      "loss": 0.2667,
      "step": 27937
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5050555584972715,
      "learning_rate": 1.073802948830328e-06,
      "loss": 0.21,
      "step": 27938
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3984718067479487,
      "learning_rate": 1.073355841287015e-06,
      "loss": 0.5075,
      "step": 27939
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.287642975992204,
      "learning_rate": 1.0729088215669626e-06,
      "loss": 0.1631,
      "step": 27940
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.324875408604782,
      "learning_rate": 1.072461889674571e-06,
      "loss": 0.5111,
      "step": 27941
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2583991831953483,
      "learning_rate": 1.0720150456142364e-06,
      "loss": 0.1518,
      "step": 27942
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5271374228150466,
      "learning_rate": 1.0715682893903578e-06,
      "loss": 0.3527,
      "step": 27943
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34079400763343487,
      "learning_rate": 1.0711216210073238e-06,
      "loss": 0.188,
      "step": 27944
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3970716686340705,
      "learning_rate": 1.0706750404695375e-06,
      "loss": 0.2614,
      "step": 27945
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5318898949022675,
      "learning_rate": 1.0702285477813867e-06,
      "loss": 0.1794,
      "step": 27946
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6607607432459455,
      "learning_rate": 1.0697821429472655e-06,
      "loss": 0.7442,
      "step": 27947
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37921533400732216,
      "learning_rate": 1.069335825971568e-06,
      "loss": 0.0599,
      "step": 27948
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.33390876859325086,
      "learning_rate": 1.068889596858681e-06,
      "loss": 0.1737,
      "step": 27949
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.349136978601067,
      "learning_rate": 1.0684434556129963e-06,
      "loss": 0.2585,
      "step": 27950
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.62112210233165,
      "learning_rate": 1.067997402238905e-06,
      "loss": 0.191,
      "step": 27951
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9060862307535694,
      "learning_rate": 1.0675514367407957e-06,
      "loss": 0.4967,
      "step": 27952
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37801260244721674,
      "learning_rate": 1.0671055591230528e-06,
      "loss": 0.1785,
      "step": 27953
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.529279632102125,
      "learning_rate": 1.0666597693900637e-06,
      "loss": 0.3048,
      "step": 27954
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5194152481303445,
      "learning_rate": 1.0662140675462185e-06,
      "loss": 0.2513,
      "step": 27955
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22586981105916173,
      "learning_rate": 1.0657684535958945e-06,
      "loss": 0.1771,
      "step": 27956
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.658185679961105,
      "learning_rate": 1.065322927543484e-06,
      "loss": 0.0688,
      "step": 27957
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3542971535947317,
      "learning_rate": 1.0648774893933644e-06,
      "loss": 0.2502,
      "step": 27958
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7178338913294444,
      "learning_rate": 1.0644321391499201e-06,
      "loss": 0.1431,
      "step": 27959
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.263278522856177,
      "learning_rate": 1.0639868768175354e-06,
      "loss": 0.573,
      "step": 27960
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3525376452302443,
      "learning_rate": 1.0635417024005855e-06,
      "loss": 0.2475,
      "step": 27961
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.647836599805089,
      "learning_rate": 1.063096615903453e-06,
      "loss": 0.3141,
      "step": 27962
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3610715331589873,
      "learning_rate": 1.0626516173305158e-06,
      "loss": 0.2002,
      "step": 27963
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.873377449550361,
      "learning_rate": 1.062206706686154e-06,
      "loss": 0.2894,
      "step": 27964
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.26593699906043083,
      "learning_rate": 1.0617618839747446e-06,
      "loss": 0.139,
      "step": 27965
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8822487387603397,
      "learning_rate": 1.0613171492006612e-06,
      "loss": 0.037,
      "step": 27966
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4166869573429881,
      "learning_rate": 1.0608725023682809e-06,
      "loss": 0.2467,
      "step": 27967
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2731387110260131,
      "learning_rate": 1.0604279434819786e-06,
      "loss": 0.2098,
      "step": 27968
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3090421239791699,
      "learning_rate": 1.059983472546129e-06,
      "loss": 0.4976,
      "step": 27969
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8176650297843431,
      "learning_rate": 1.0595390895651025e-06,
      "loss": 0.245,
      "step": 27970
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7746097274169951,
      "learning_rate": 1.0590947945432716e-06,
      "loss": 0.3277,
      "step": 27971
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2966099967497828,
      "learning_rate": 1.0586505874850106e-06,
      "loss": 0.193,
      "step": 27972
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3403093004290276,
      "learning_rate": 1.0582064683946825e-06,
      "loss": 0.2605,
      "step": 27973
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.270457784330974,
      "learning_rate": 1.057762437276666e-06,
      "loss": 0.0614,
      "step": 27974
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47299467689634794,
      "learning_rate": 1.057318494135322e-06,
      "loss": 0.2068,
      "step": 27975
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35388611535129116,
      "learning_rate": 1.056874638975023e-06,
      "loss": 0.1586,
      "step": 27976
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4623113019920094,
      "learning_rate": 1.0564308718001349e-06,
      "loss": 0.2113,
      "step": 27977
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.048420169989116,
      "learning_rate": 1.05598719261502e-06,
      "loss": 0.4908,
      "step": 27978
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3341092680986848,
      "learning_rate": 1.0555436014240471e-06,
      "loss": 0.2442,
      "step": 27979
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6113974327519642,
      "learning_rate": 1.0551000982315784e-06,
      "loss": 0.2775,
      "step": 27980
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3333971850955641,
      "learning_rate": 1.0546566830419814e-06,
      "loss": 0.2137,
      "step": 27981
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8157320425921495,
      "learning_rate": 1.0542133558596112e-06,
      "loss": 0.8645,
      "step": 27982
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.16271593848678167,
      "learning_rate": 1.053770116688837e-06,
      "loss": 0.0713,
      "step": 27983
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4928475932337069,
      "learning_rate": 1.0533269655340139e-06,
      "loss": 0.2918,
      "step": 27984
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2951428972020236,
      "learning_rate": 1.052883902399504e-06,
      "loss": 0.1725,
      "step": 27985
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5282058059878254,
      "learning_rate": 1.0524409272896696e-06,
      "loss": 0.3191,
      "step": 27986
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1294909304602763,
      "learning_rate": 1.0519980402088626e-06,
      "loss": 0.2257,
      "step": 27987
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6936533095099567,
      "learning_rate": 1.0515552411614428e-06,
      "loss": 0.3583,
      "step": 27988
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5656375574767921,
      "learning_rate": 1.0511125301517666e-06,
      "loss": 0.2198,
      "step": 27989
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.33308717449858843,
      "learning_rate": 1.0506699071841909e-06,
      "loss": 0.1868,
      "step": 27990
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4660454438829983,
      "learning_rate": 1.0502273722630707e-06,
      "loss": 0.278,
      "step": 27991
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20890443077405974,
      "learning_rate": 1.0497849253927573e-06,
      "loss": 0.1302,
      "step": 27992
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5923093251257499,
      "learning_rate": 1.0493425665776035e-06,
      "loss": 0.5108,
      "step": 27993
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3550530553477348,
      "learning_rate": 1.048900295821964e-06,
      "loss": 0.1622,
      "step": 27994
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5983036792845272,
      "learning_rate": 1.0484581131301906e-06,
      "loss": 0.3246,
      "step": 27995
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4433111138199011,
      "learning_rate": 1.048016018506629e-06,
      "loss": 0.2361,
      "step": 27996
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5798785880052338,
      "learning_rate": 1.047574011955632e-06,
      "loss": 0.3297,
      "step": 27997
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4965133724160684,
      "learning_rate": 1.0471320934815488e-06,
      "loss": 0.2021,
      "step": 27998
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34292292068901076,
      "learning_rate": 1.046690263088722e-06,
      "loss": 0.2474,
      "step": 27999
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6901847078852665,
      "learning_rate": 1.0462485207815076e-06,
      "loss": 0.3026,
      "step": 28000
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.45794617301869406,
      "learning_rate": 1.045806866564244e-06,
      "loss": 0.1946,
      "step": 28001
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2683057291104409,
      "learning_rate": 1.0453653004412778e-06,
      "loss": 0.1507,
      "step": 28002
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.26301197886236827,
      "learning_rate": 1.0449238224169566e-06,
      "loss": 0.1711,
      "step": 28003
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.533866263460278,
      "learning_rate": 1.0444824324956205e-06,
      "loss": 0.3275,
      "step": 28004
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7127346777413512,
      "learning_rate": 1.0440411306816123e-06,
      "loss": 0.27,
      "step": 28005
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1208331895669943,
      "learning_rate": 1.0435999169792742e-06,
      "loss": 0.4418,
      "step": 28006
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.48563690035896795,
      "learning_rate": 1.0431587913929487e-06,
      "loss": 0.2556,
      "step": 28007
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36051131367916917,
      "learning_rate": 1.0427177539269729e-06,
      "loss": 0.2901,
      "step": 28008
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.48734190476742295,
      "learning_rate": 1.0422768045856867e-06,
      "loss": 0.1913,
      "step": 28009
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4946052014331551,
      "learning_rate": 1.0418359433734316e-06,
      "loss": 0.349,
      "step": 28010
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.152996580354177,
      "learning_rate": 1.0413951702945379e-06,
      "loss": 0.0732,
      "step": 28011
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4106266110581985,
      "learning_rate": 1.0409544853533494e-06,
      "loss": 0.2451,
      "step": 28012
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5079988729740588,
      "learning_rate": 1.0405138885541976e-06,
      "loss": 0.2162,
      "step": 28013
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6215715851604003,
      "learning_rate": 1.0400733799014172e-06,
      "loss": 0.3398,
      "step": 28014
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30635657250543324,
      "learning_rate": 1.0396329593993436e-06,
      "loss": 0.219,
      "step": 28015
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.46958073110385806,
      "learning_rate": 1.03919262705231e-06,
      "loss": 0.1888,
      "step": 28016
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47356602633250733,
      "learning_rate": 1.0387523828646495e-06,
      "loss": 0.2929,
      "step": 28017
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47333528183899903,
      "learning_rate": 1.0383122268406886e-06,
      "loss": 0.1988,
      "step": 28018
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5027833771921535,
      "learning_rate": 1.0378721589847618e-06,
      "loss": 0.2284,
      "step": 28019
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2354571312889223,
      "learning_rate": 1.0374321793011977e-06,
      "loss": 0.1728,
      "step": 28020
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7765901747392325,
      "learning_rate": 1.0369922877943272e-06,
      "loss": 0.3562,
      "step": 28021
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40919862449536565,
      "learning_rate": 1.0365524844684727e-06,
      "loss": 0.1892,
      "step": 28022
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7640129167026796,
      "learning_rate": 1.036112769327965e-06,
      "loss": 0.3974,
      "step": 28023
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6434728223083899,
      "learning_rate": 1.0356731423771315e-06,
      "loss": 0.305,
      "step": 28024
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.417234000688941,
      "learning_rate": 1.0352336036202904e-06,
      "loss": 0.7385,
      "step": 28025
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2889831642758261,
      "learning_rate": 1.0347941530617755e-06,
      "loss": 0.1776,
      "step": 28026
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.399208454660195,
      "learning_rate": 1.0343547907059038e-06,
      "loss": 0.2886,
      "step": 28027
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5981009630798566,
      "learning_rate": 1.0339155165569993e-06,
      "loss": 0.0137,
      "step": 28028
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9193499344850112,
      "learning_rate": 1.0334763306193863e-06,
      "loss": 0.2744,
      "step": 28029
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5282561979707343,
      "learning_rate": 1.0330372328973815e-06,
      "loss": 0.2545,
      "step": 28030
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31182372101540606,
      "learning_rate": 1.032598223395308e-06,
      "loss": 0.1801,
      "step": 28031
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.42524104300047755,
      "learning_rate": 1.0321593021174824e-06,
      "loss": 0.2096,
      "step": 28032
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2854423498279787,
      "learning_rate": 1.031720469068227e-06,
      "loss": 0.2138,
      "step": 28033
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4287596747915652,
      "learning_rate": 1.0312817242518546e-06,
      "loss": 0.795,
      "step": 28034
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.28507151047680757,
      "learning_rate": 1.0308430676726833e-06,
      "loss": 0.1654,
      "step": 28035
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6732369375316176,
      "learning_rate": 1.0304044993350314e-06,
      "loss": 0.6304,
      "step": 28036
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4468863598774857,
      "learning_rate": 1.0299660192432081e-06,
      "loss": 0.1093,
      "step": 28037
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3581897548096933,
      "learning_rate": 1.0295276274015332e-06,
      "loss": 0.2633,
      "step": 28038
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.48264528465365786,
      "learning_rate": 1.0290893238143162e-06,
      "loss": 0.1838,
      "step": 28039
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6335338301048148,
      "learning_rate": 1.028651108485871e-06,
      "loss": 0.3533,
      "step": 28040
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2294186528457756,
      "learning_rate": 1.0282129814205088e-06,
      "loss": 0.1321,
      "step": 28041
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1952512366237984,
      "learning_rate": 1.0277749426225358e-06,
      "loss": 0.3349,
      "step": 28042
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9200726910692207,
      "learning_rate": 1.0273369920962683e-06,
      "loss": 0.3748,
      "step": 28043
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37289889566584705,
      "learning_rate": 1.0268991298460108e-06,
      "loss": 0.1567,
      "step": 28044
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.29526225867388317,
      "learning_rate": 1.0264613558760728e-06,
      "loss": 0.2428,
      "step": 28045
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0620814579271387,
      "learning_rate": 1.02602367019076e-06,
      "loss": 0.2338,
      "step": 28046
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6537241356900572,
      "learning_rate": 1.0255860727943812e-06,
      "loss": 0.3485,
      "step": 28047
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9326655135755297,
      "learning_rate": 1.0251485636912373e-06,
      "loss": 0.1322,
      "step": 28048
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37479427334454657,
      "learning_rate": 1.0247111428856349e-06,
      "loss": 0.2601,
      "step": 28049
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20457274285436056,
      "learning_rate": 1.0242738103818805e-06,
      "loss": 0.1265,
      "step": 28050
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.49691992757073805,
      "learning_rate": 1.0238365661842685e-06,
      "loss": 0.3254,
      "step": 28051
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.251905806799213,
      "learning_rate": 1.0233994102971101e-06,
      "loss": 0.4698,
      "step": 28052
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6170034233918136,
      "learning_rate": 1.0229623427247004e-06,
      "loss": 0.2756,
      "step": 28053
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37003493020970507,
      "learning_rate": 1.0225253634713416e-06,
      "loss": 0.1981,
      "step": 28054
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.869486801451955,
      "learning_rate": 1.0220884725413337e-06,
      "loss": 0.2564,
      "step": 28055
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3487871863351942,
      "learning_rate": 1.0216516699389712e-06,
      "loss": 0.2863,
      "step": 28056
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5436402318679633,
      "learning_rate": 1.0212149556685546e-06,
      "loss": 0.1498,
      "step": 28057
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39089118044255544,
      "learning_rate": 1.0207783297343787e-06,
      "loss": 0.2671,
      "step": 28058
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2275184028437813,
      "learning_rate": 1.0203417921407432e-06,
      "loss": 0.0608,
      "step": 28059
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.48109321826094853,
      "learning_rate": 1.019905342891937e-06,
      "loss": 0.2836,
      "step": 28060
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3027753470982158,
      "learning_rate": 1.0194689819922565e-06,
      "loss": 0.1895,
      "step": 28061
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.42449585138364,
      "learning_rate": 1.0190327094459985e-06,
      "loss": 0.2461,
      "step": 28062
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3381159666185879,
      "learning_rate": 1.018596525257447e-06,
      "loss": 0.2149,
      "step": 28063
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.135089690691161,
      "learning_rate": 1.0181604294309022e-06,
      "loss": 0.3699,
      "step": 28064
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.856532867235484,
      "learning_rate": 1.0177244219706483e-06,
      "loss": 0.2724,
      "step": 28065
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8859849241930006,
      "learning_rate": 1.0172885028809776e-06,
      "loss": 0.396,
      "step": 28066
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3678348946669953,
      "learning_rate": 1.0168526721661798e-06,
      "loss": 0.1594,
      "step": 28067
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24976839532739345,
      "learning_rate": 1.0164169298305383e-06,
      "loss": 0.17,
      "step": 28068
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31315566930642763,
      "learning_rate": 1.015981275878346e-06,
      "loss": 0.2318,
      "step": 28069
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9580946636448947,
      "learning_rate": 1.0155457103138844e-06,
      "loss": 0.3238,
      "step": 28070
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9045219350448593,
      "learning_rate": 1.015110233141441e-06,
      "loss": 0.3818,
      "step": 28071
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3324962789996853,
      "learning_rate": 1.0146748443653e-06,
      "loss": 0.1902,
      "step": 28072
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0125490510668396,
      "learning_rate": 1.0142395439897435e-06,
      "loss": 0.3402,
      "step": 28073
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.350926491725098,
      "learning_rate": 1.0138043320190548e-06,
      "loss": 0.2151,
      "step": 28074
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7134001464197117,
      "learning_rate": 1.0133692084575164e-06,
      "loss": 0.7336,
      "step": 28075
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.27699765878704263,
      "learning_rate": 1.01293417330941e-06,
      "loss": 0.1702,
      "step": 28076
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4705835324824448,
      "learning_rate": 1.0124992265790124e-06,
      "loss": 0.6041,
      "step": 28077
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23958163472539837,
      "learning_rate": 1.0120643682706056e-06,
      "loss": 0.1135,
      "step": 28078
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.352921501997727,
      "learning_rate": 1.0116295983884684e-06,
      "loss": 0.7043,
      "step": 28079
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2658951950888659,
      "learning_rate": 1.0111949169368728e-06,
      "loss": 0.1896,
      "step": 28080
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4561719737386793,
      "learning_rate": 1.0107603239201025e-06,
      "loss": 0.2311,
      "step": 28081
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7812568268833567,
      "learning_rate": 1.0103258193424292e-06,
      "loss": 0.3791,
      "step": 28082
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19756852484714013,
      "learning_rate": 1.0098914032081275e-06,
      "loss": 0.0625,
      "step": 28083
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3324256884746084,
      "learning_rate": 1.0094570755214727e-06,
      "loss": 0.5211,
      "step": 28084
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2854647324319521,
      "learning_rate": 1.0090228362867404e-06,
      "loss": 0.1758,
      "step": 28085
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5627476667538213,
      "learning_rate": 1.0085886855081962e-06,
      "loss": 0.3384,
      "step": 28086
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2732460947483796,
      "learning_rate": 1.0081546231901163e-06,
      "loss": 0.1621,
      "step": 28087
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4772676640247722,
      "learning_rate": 1.007720649336772e-06,
      "loss": 0.8029,
      "step": 28088
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3894654124076137,
      "learning_rate": 1.0072867639524263e-06,
      "loss": 0.1858,
      "step": 28089
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6200481494373223,
      "learning_rate": 1.0068529670413562e-06,
      "loss": 0.3159,
      "step": 28090
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47036130394842024,
      "learning_rate": 1.0064192586078236e-06,
      "loss": 0.1697,
      "step": 28091
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35526841966619394,
      "learning_rate": 1.0059856386560984e-06,
      "loss": 0.2606,
      "step": 28092
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7579074256176704,
      "learning_rate": 1.0055521071904473e-06,
      "loss": 0.0763,
      "step": 28093
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3507843768287771,
      "learning_rate": 1.0051186642151301e-06,
      "loss": 0.1514,
      "step": 28094
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6153675294288371,
      "learning_rate": 1.00468530973442e-06,
      "loss": 0.3898,
      "step": 28095
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0250819567254272,
      "learning_rate": 1.0042520437525727e-06,
      "loss": 0.5094,
      "step": 28096
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3994957181654508,
      "learning_rate": 1.0038188662738547e-06,
      "loss": 0.2691,
      "step": 28097
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6722969911749243,
      "learning_rate": 1.003385777302528e-06,
      "loss": 0.1995,
      "step": 28098
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3399889689044886,
      "learning_rate": 1.0029527768428505e-06,
      "loss": 0.244,
      "step": 28099
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4470500033707478,
      "learning_rate": 1.0025198648990842e-06,
      "loss": 0.1768,
      "step": 28100
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4948850502586298,
      "learning_rate": 1.0020870414754879e-06,
      "loss": 0.2246,
      "step": 28101
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.8265089221790234,
      "learning_rate": 1.0016543065763218e-06,
      "loss": 0.1745,
      "step": 28102
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6141417824002988,
      "learning_rate": 1.00122166020584e-06,
      "loss": 0.2687,
      "step": 28103
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.22534662275433057,
      "learning_rate": 1.0007891023683004e-06,
      "loss": 0.2129,
      "step": 28104
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2678214471774736,
      "learning_rate": 1.0003566330679603e-06,
      "loss": 0.5167,
      "step": 28105
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8065693926787383,
      "learning_rate": 9.99924252309069e-07,
      "loss": 0.2584,
      "step": 28106
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.528270909811325,
      "learning_rate": 9.994919600958886e-07,
      "loss": 0.2038,
      "step": 28107
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3623431017384782,
      "learning_rate": 9.990597564326666e-07,
      "loss": 0.2655,
      "step": 28108
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3845502274381516,
      "learning_rate": 9.986276413236552e-07,
      "loss": 0.0963,
      "step": 28109
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.34105685107841033,
      "learning_rate": 9.981956147731086e-07,
      "loss": 0.2866,
      "step": 28110
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.21459166810530436,
      "learning_rate": 9.977636767852749e-07,
      "loss": 0.0661,
      "step": 28111
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.42589320452889323,
      "learning_rate": 9.973318273644028e-07,
      "loss": 0.2546,
      "step": 28112
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4650427350611318,
      "learning_rate": 9.969000665147432e-07,
      "loss": 0.1955,
      "step": 28113
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5150676526967781,
      "learning_rate": 9.964683942405462e-07,
      "loss": 0.7309,
      "step": 28114
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4505525575457401,
      "learning_rate": 9.960368105460506e-07,
      "loss": 0.2423,
      "step": 28115
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5450670142924968,
      "learning_rate": 9.956053154355117e-07,
      "loss": 0.2856,
      "step": 28116
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37440414061324095,
      "learning_rate": 9.951739089131685e-07,
      "loss": 0.1633,
      "step": 28117
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5215519216780521,
      "learning_rate": 9.947425909832664e-07,
      "loss": 0.3518,
      "step": 28118
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23590871873055413,
      "learning_rate": 9.943113616500533e-07,
      "loss": 0.0833,
      "step": 28119
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8147840908081252,
      "learning_rate": 9.938802209177633e-07,
      "loss": 0.0818,
      "step": 28120
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8836332722662767,
      "learning_rate": 9.934491687906478e-07,
      "loss": 0.3891,
      "step": 28121
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2264484424203632,
      "learning_rate": 9.930182052729408e-07,
      "loss": 0.2045,
      "step": 28122
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.856910893702017,
      "learning_rate": 9.925873303688837e-07,
      "loss": 0.3597,
      "step": 28123
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7267368718561109,
      "learning_rate": 9.921565440827197e-07,
      "loss": 0.2682,
      "step": 28124
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6070483910373756,
      "learning_rate": 9.917258464186807e-07,
      "loss": 0.7464,
      "step": 28125
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.28177718652871053,
      "learning_rate": 9.91295237381007e-07,
      "loss": 0.1686,
      "step": 28126
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3732765135879779,
      "learning_rate": 9.90864716973935e-07,
      "loss": 0.478,
      "step": 28127
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.20623129363588769,
      "learning_rate": 9.904342852017023e-07,
      "loss": 0.1611,
      "step": 28128
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.1711618049240027,
      "learning_rate": 9.900039420685393e-07,
      "loss": 0.8143,
      "step": 28129
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5537697488197753,
      "learning_rate": 9.89573687578682e-07,
      "loss": 0.0906,
      "step": 28130
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3965424171854165,
      "learning_rate": 9.891435217363666e-07,
      "loss": 0.2759,
      "step": 28131
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6668184163906892,
      "learning_rate": 9.887134445458179e-07,
      "loss": 0.2793,
      "step": 28132
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4057419995113274,
      "learning_rate": 9.882834560112753e-07,
      "loss": 0.2344,
      "step": 28133
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4838292921230723,
      "learning_rate": 9.878535561369629e-07,
      "loss": 0.3254,
      "step": 28134
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2853686890960853,
      "learning_rate": 9.87423744927113e-07,
      "loss": 0.1668,
      "step": 28135
    },
    {
      "epoch": 0.86,
      "grad_norm": 2.2405942194878095,
      "learning_rate": 9.869940223859554e-07,
      "loss": 0.7219,
      "step": 28136
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.19623489566356034,
      "learning_rate": 9.865643885177146e-07,
      "loss": 0.0704,
      "step": 28137
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2682647999544385,
      "learning_rate": 9.861348433266193e-07,
      "loss": 0.4526,
      "step": 28138
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2907554158710518,
      "learning_rate": 9.857053868168953e-07,
      "loss": 0.1962,
      "step": 28139
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.38654688527005543,
      "learning_rate": 9.852760189927702e-07,
      "loss": 0.2846,
      "step": 28140
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7553015979334793,
      "learning_rate": 9.848467398584627e-07,
      "loss": 0.2618,
      "step": 28141
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.777532725869268,
      "learning_rate": 9.844175494182006e-07,
      "loss": 0.4176,
      "step": 28142
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44340115209097913,
      "learning_rate": 9.839884476762052e-07,
      "loss": 0.2521,
      "step": 28143
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.558650539077011,
      "learning_rate": 9.835594346366973e-07,
      "loss": 0.2936,
      "step": 28144
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4749135818007819,
      "learning_rate": 9.831305103039003e-07,
      "loss": 0.1909,
      "step": 28145
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24668608383841,
      "learning_rate": 9.827016746820295e-07,
      "loss": 0.1883,
      "step": 28146
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4171162431656278,
      "learning_rate": 9.822729277753096e-07,
      "loss": 0.1875,
      "step": 28147
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.65258056733258,
      "learning_rate": 9.818442695879559e-07,
      "loss": 0.1852,
      "step": 28148
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.37151456798704263,
      "learning_rate": 9.81415700124182e-07,
      "loss": 0.2364,
      "step": 28149
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5749917538503132,
      "learning_rate": 9.80987219388212e-07,
      "loss": 0.2536,
      "step": 28150
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35829262506128395,
      "learning_rate": 9.805588273842548e-07,
      "loss": 0.2721,
      "step": 28151
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1408741084449192,
      "learning_rate": 9.801305241165281e-07,
      "loss": 0.4298,
      "step": 28152
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.41136820725514783,
      "learning_rate": 9.797023095892456e-07,
      "loss": 0.2423,
      "step": 28153
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4931349464031299,
      "learning_rate": 9.792741838066212e-07,
      "loss": 0.199,
      "step": 28154
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4636190324881385,
      "learning_rate": 9.788461467728637e-07,
      "loss": 0.2133,
      "step": 28155
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.36154142831902436,
      "learning_rate": 9.78418198492187e-07,
      "loss": 0.1396,
      "step": 28156
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.40477802017233666,
      "learning_rate": 9.779903389688017e-07,
      "loss": 0.2705,
      "step": 28157
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.25886634383440826,
      "learning_rate": 9.775625682069123e-07,
      "loss": 0.1877,
      "step": 28158
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7981260104284073,
      "learning_rate": 9.771348862107355e-07,
      "loss": 0.3773,
      "step": 28159
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.117783137954517,
      "learning_rate": 9.767072929844711e-07,
      "loss": 0.1526,
      "step": 28160
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9170459711535374,
      "learning_rate": 9.762797885323316e-07,
      "loss": 0.4838,
      "step": 28161
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39188544376374845,
      "learning_rate": 9.758523728585211e-07,
      "loss": 0.2247,
      "step": 28162
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.294856664315389,
      "learning_rate": 9.754250459672431e-07,
      "loss": 0.2177,
      "step": 28163
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6153298607604603,
      "learning_rate": 9.74997807862702e-07,
      "loss": 0.4133,
      "step": 28164
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31855782452126863,
      "learning_rate": 9.745706585491033e-07,
      "loss": 0.1101,
      "step": 28165
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5992151166985175,
      "learning_rate": 9.741435980306502e-07,
      "loss": 0.3063,
      "step": 28166
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35967866211114485,
      "learning_rate": 9.737166263115394e-07,
      "loss": 0.202,
      "step": 28167
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.015133484342494,
      "learning_rate": 9.732897433959754e-07,
      "loss": 0.4175,
      "step": 28168
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31053415450035654,
      "learning_rate": 9.728629492881557e-07,
      "loss": 0.2112,
      "step": 28169
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3351598633671053,
      "learning_rate": 9.724362439922818e-07,
      "loss": 0.7866,
      "step": 28170
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.31956289758409023,
      "learning_rate": 9.720096275125524e-07,
      "loss": 0.1535,
      "step": 28171
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5800580063739273,
      "learning_rate": 9.715830998531605e-07,
      "loss": 0.3151,
      "step": 28172
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0212194874908391,
      "learning_rate": 9.711566610183054e-07,
      "loss": 0.1126,
      "step": 28173
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.47566461947362915,
      "learning_rate": 9.707303110121835e-07,
      "loss": 0.2476,
      "step": 28174
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4605928774453049,
      "learning_rate": 9.703040498389838e-07,
      "loss": 0.3017,
      "step": 28175
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3292302378226058,
      "learning_rate": 9.698778775029084e-07,
      "loss": 0.1902,
      "step": 28176
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.30752665063191903,
      "learning_rate": 9.694517940081438e-07,
      "loss": 0.1727,
      "step": 28177
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0410832779386072,
      "learning_rate": 9.690257993588837e-07,
      "loss": 0.4987,
      "step": 28178
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.773447098095149,
      "learning_rate": 9.685998935593211e-07,
      "loss": 0.8243,
      "step": 28179
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3260135860932268,
      "learning_rate": 9.681740766136426e-07,
      "loss": 0.0667,
      "step": 28180
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3092334049791709,
      "learning_rate": 9.677483485260398e-07,
      "loss": 0.2537,
      "step": 28181
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4533107656648146,
      "learning_rate": 9.673227093006998e-07,
      "loss": 0.2195,
      "step": 28182
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0604224284508947,
      "learning_rate": 9.66897158941814e-07,
      "loss": 0.5086,
      "step": 28183
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7086694287566576,
      "learning_rate": 9.664716974535615e-07,
      "loss": 0.2114,
      "step": 28184
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.35775648862178216,
      "learning_rate": 9.660463248401375e-07,
      "loss": 0.2537,
      "step": 28185
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.15543050919371337,
      "learning_rate": 9.656210411057198e-07,
      "loss": 0.081,
      "step": 28186
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.32110786202588476,
      "learning_rate": 9.651958462544953e-07,
      "loss": 0.2185,
      "step": 28187
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.3701464763503872,
      "learning_rate": 9.647707402906492e-07,
      "loss": 0.8041,
      "step": 28188
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.32372702542021364,
      "learning_rate": 9.643457232183596e-07,
      "loss": 0.1599,
      "step": 28189
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6259702588021745,
      "learning_rate": 9.639207950418095e-07,
      "loss": 0.2989,
      "step": 28190
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0684288078765465,
      "learning_rate": 9.634959557651801e-07,
      "loss": 0.2298,
      "step": 28191
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.7418027287936716,
      "learning_rate": 9.630712053926506e-07,
      "loss": 0.3624,
      "step": 28192
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2974779728711146,
      "learning_rate": 9.626465439284028e-07,
      "loss": 0.2137,
      "step": 28193
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3408279370044144,
      "learning_rate": 9.62221971376609e-07,
      "loss": 0.2288,
      "step": 28194
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.23892308324897868,
      "learning_rate": 9.617974877414494e-07,
      "loss": 0.0676,
      "step": 28195
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4616818304432976,
      "learning_rate": 9.613730930271003e-07,
      "loss": 0.2907,
      "step": 28196
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8718436369701922,
      "learning_rate": 9.609487872377387e-07,
      "loss": 0.2501,
      "step": 28197
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4239665196277939,
      "learning_rate": 9.605245703775356e-07,
      "loss": 0.2448,
      "step": 28198
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5727901345617816,
      "learning_rate": 9.601004424506667e-07,
      "loss": 0.209,
      "step": 28199
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.44084433718267124,
      "learning_rate": 9.596764034613049e-07,
      "loss": 0.2341,
      "step": 28200
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.2327791023692267,
      "learning_rate": 9.592524534136183e-07,
      "loss": 0.3989,
      "step": 28201
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9440045300389742,
      "learning_rate": 9.588285923117835e-07,
      "loss": 0.1143,
      "step": 28202
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5379562452520591,
      "learning_rate": 9.584048201599672e-07,
      "loss": 0.2821,
      "step": 28203
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.26080410935803355,
      "learning_rate": 9.579811369623393e-07,
      "loss": 0.1459,
      "step": 28204
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3182024085168223,
      "learning_rate": 9.5755754272307e-07,
      "loss": 0.2603,
      "step": 28205
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.1910823105103614,
      "learning_rate": 9.571340374463222e-07,
      "loss": 0.4391,
      "step": 28206
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3366673383865153,
      "learning_rate": 9.567106211362664e-07,
      "loss": 0.175,
      "step": 28207
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3461554282640658,
      "learning_rate": 9.56287293797067e-07,
      "loss": 0.2042,
      "step": 28208
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8323628815119417,
      "learning_rate": 9.558640554328901e-07,
      "loss": 0.3792,
      "step": 28209
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.166626383738089,
      "learning_rate": 9.554409060478965e-07,
      "loss": 0.076,
      "step": 28210
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3488801818376141,
      "learning_rate": 9.550178456462523e-07,
      "loss": 0.2541,
      "step": 28211
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3228813013134233,
      "learning_rate": 9.545948742321198e-07,
      "loss": 0.1529,
      "step": 28212
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.28909365672316834,
      "learning_rate": 9.54171991809656e-07,
      "loss": 0.1382,
      "step": 28213
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.5556630362806603,
      "learning_rate": 9.537491983830283e-07,
      "loss": 0.746,
      "step": 28214
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.350071703863931,
      "learning_rate": 9.533264939563914e-07,
      "loss": 0.1165,
      "step": 28215
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6520758499173686,
      "learning_rate": 9.52903878533904e-07,
      "loss": 0.3227,
      "step": 28216
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.25080099054976485,
      "learning_rate": 9.524813521197251e-07,
      "loss": 0.2073,
      "step": 28217
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.7957600786502572,
      "learning_rate": 9.520589147180126e-07,
      "loss": 0.4477,
      "step": 28218
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.9038288559120637,
      "learning_rate": 9.51636566332923e-07,
      "loss": 0.2511,
      "step": 28219
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6098134275747762,
      "learning_rate": 9.512143069686087e-07,
      "loss": 0.53,
      "step": 28220
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.2737032609875659,
      "learning_rate": 9.50792136629225e-07,
      "loss": 0.1709,
      "step": 28221
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.6171640310589288,
      "learning_rate": 9.503700553189266e-07,
      "loss": 0.5491,
      "step": 28222
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.32936462164395597,
      "learning_rate": 9.499480630418667e-07,
      "loss": 0.2298,
      "step": 28223
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.4552771451333336,
      "learning_rate": 9.495261598021932e-07,
      "loss": 0.2648,
      "step": 28224
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.6412896579377761,
      "learning_rate": 9.491043456040594e-07,
      "loss": 0.1946,
      "step": 28225
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3592627683349069,
      "learning_rate": 9.486826204516175e-07,
      "loss": 0.2021,
      "step": 28226
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5155607556569463,
      "learning_rate": 9.482609843490109e-07,
      "loss": 0.2701,
      "step": 28227
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3043997047349197,
      "learning_rate": 9.478394373003941e-07,
      "loss": 0.2211,
      "step": 28228
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8956799888330572,
      "learning_rate": 9.474179793099103e-07,
      "loss": 0.3418,
      "step": 28229
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.39762360233884114,
      "learning_rate": 9.469966103817074e-07,
      "loss": 0.1659,
      "step": 28230
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5887118597774098,
      "learning_rate": 9.465753305199321e-07,
      "loss": 0.3165,
      "step": 28231
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.0035092793631093,
      "learning_rate": 9.461541397287255e-07,
      "loss": 0.5533,
      "step": 28232
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.8105126708341501,
      "learning_rate": 9.457330380122332e-07,
      "loss": 0.4076,
      "step": 28233
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3951749608285688,
      "learning_rate": 9.453120253745996e-07,
      "loss": 0.2054,
      "step": 28234
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3493256599595572,
      "learning_rate": 9.448911018199669e-07,
      "loss": 0.2461,
      "step": 28235
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.24737871209043633,
      "learning_rate": 9.44470267352473e-07,
      "loss": 0.1476,
      "step": 28236
    },
    {
      "epoch": 0.86,
      "grad_norm": 1.4919800397699132,
      "learning_rate": 9.440495219762603e-07,
      "loss": 0.4529,
      "step": 28237
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5893820646605595,
      "learning_rate": 9.436288656954706e-07,
      "loss": 0.0155,
      "step": 28238
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.350522582750178,
      "learning_rate": 9.432082985142365e-07,
      "loss": 0.1514,
      "step": 28239
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5691068085254815,
      "learning_rate": 9.427878204367025e-07,
      "loss": 0.3456,
      "step": 28240
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3142592002155146,
      "learning_rate": 9.423674314669995e-07,
      "loss": 0.2202,
      "step": 28241
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.919598156508292,
      "learning_rate": 9.419471316092676e-07,
      "loss": 0.5157,
      "step": 28242
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.5964558777298732,
      "learning_rate": 9.415269208676414e-07,
      "loss": 0.2356,
      "step": 28243
    },
    {
      "epoch": 0.86,
      "grad_norm": 0.3882380118521257,
      "learning_rate": 9.411067992462497e-07,
      "loss": 0.2497,
      "step": 28244
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.24581354812819362,
      "learning_rate": 9.406867667492347e-07,
      "loss": 0.0654,
      "step": 28245
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3000164442534244,
      "learning_rate": 9.402668233807222e-07,
      "loss": 0.2173,
      "step": 28246
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.42398643606703357,
      "learning_rate": 9.398469691448453e-07,
      "loss": 0.1782,
      "step": 28247
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4291295310512128,
      "learning_rate": 9.394272040457375e-07,
      "loss": 0.2552,
      "step": 28248
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4860578085913464,
      "learning_rate": 9.390075280875244e-07,
      "loss": 0.176,
      "step": 28249
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4383734864524698,
      "learning_rate": 9.38587941274336e-07,
      "loss": 0.7422,
      "step": 28250
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.62857054720167,
      "learning_rate": 9.381684436103012e-07,
      "loss": 0.2719,
      "step": 28251
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28332251604584296,
      "learning_rate": 9.377490350995499e-07,
      "loss": 0.2007,
      "step": 28252
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5711138935018021,
      "learning_rate": 9.373297157462003e-07,
      "loss": 0.3071,
      "step": 28253
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4319269757394869,
      "learning_rate": 9.369104855543876e-07,
      "loss": 0.1959,
      "step": 28254
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4298318339198069,
      "learning_rate": 9.364913445282298e-07,
      "loss": 0.1574,
      "step": 28255
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0655964951258745,
      "learning_rate": 9.360722926718513e-07,
      "loss": 0.0759,
      "step": 28256
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0260244604013988,
      "learning_rate": 9.356533299893788e-07,
      "loss": 0.4898,
      "step": 28257
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3377159012582051,
      "learning_rate": 9.352344564849303e-07,
      "loss": 0.2017,
      "step": 28258
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3446030243764936,
      "learning_rate": 9.348156721626266e-07,
      "loss": 0.2759,
      "step": 28259
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6846908589926414,
      "learning_rate": 9.343969770265893e-07,
      "loss": 0.2709,
      "step": 28260
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5636153212468837,
      "learning_rate": 9.339783710809402e-07,
      "loss": 0.7648,
      "step": 28261
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.31027622946839206,
      "learning_rate": 9.33559854329793e-07,
      "loss": 0.1872,
      "step": 28262
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5330247470670582,
      "learning_rate": 9.331414267772665e-07,
      "loss": 0.2339,
      "step": 28263
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23817402188719294,
      "learning_rate": 9.327230884274807e-07,
      "loss": 0.1716,
      "step": 28264
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.559294771958378,
      "learning_rate": 9.323048392845447e-07,
      "loss": 0.0679,
      "step": 28265
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6302480245890105,
      "learning_rate": 9.318866793525805e-07,
      "loss": 0.2986,
      "step": 28266
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3755045142096255,
      "learning_rate": 9.314686086356972e-07,
      "loss": 0.237,
      "step": 28267
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1240201347809249,
      "learning_rate": 9.310506271380104e-07,
      "loss": 0.3951,
      "step": 28268
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6980966661634795,
      "learning_rate": 9.306327348636335e-07,
      "loss": 0.2724,
      "step": 28269
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3839899178753199,
      "learning_rate": 9.302149318166709e-07,
      "loss": 0.2853,
      "step": 28270
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.30965185056274525,
      "learning_rate": 9.297972180012426e-07,
      "loss": 0.174,
      "step": 28271
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8975032934955898,
      "learning_rate": 9.293795934214522e-07,
      "loss": 0.8737,
      "step": 28272
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2929815863737625,
      "learning_rate": 9.289620580814107e-07,
      "loss": 0.085,
      "step": 28273
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.741304272469138,
      "learning_rate": 9.28544611985226e-07,
      "loss": 0.5764,
      "step": 28274
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4675521162445943,
      "learning_rate": 9.281272551370024e-07,
      "loss": 0.1711,
      "step": 28275
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5694377957485127,
      "learning_rate": 9.277099875408479e-07,
      "loss": 0.2897,
      "step": 28276
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.27395824173880756,
      "learning_rate": 9.272928092008682e-07,
      "loss": 0.2173,
      "step": 28277
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9036706926833838,
      "learning_rate": 9.268757201211697e-07,
      "loss": 0.2905,
      "step": 28278
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9670723360529676,
      "learning_rate": 9.264587203058506e-07,
      "loss": 0.5124,
      "step": 28279
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3756860464220241,
      "learning_rate": 9.260418097590162e-07,
      "loss": 0.1696,
      "step": 28280
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6682403731924581,
      "learning_rate": 9.256249884847701e-07,
      "loss": 0.2738,
      "step": 28281
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19821282037096966,
      "learning_rate": 9.252082564872067e-07,
      "loss": 0.1579,
      "step": 28282
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2001802310884708,
      "learning_rate": 9.247916137704349e-07,
      "loss": 0.3572,
      "step": 28283
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6667122054830087,
      "learning_rate": 9.243750603385482e-07,
      "loss": 0.1882,
      "step": 28284
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3500929085609812,
      "learning_rate": 9.239585961956455e-07,
      "loss": 0.2524,
      "step": 28285
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7691210837375339,
      "learning_rate": 9.235422213458245e-07,
      "loss": 0.28,
      "step": 28286
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5039474573755587,
      "learning_rate": 9.231259357931832e-07,
      "loss": 0.3534,
      "step": 28287
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.46332304838928223,
      "learning_rate": 9.227097395418139e-07,
      "loss": 0.2456,
      "step": 28288
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6485039964898057,
      "learning_rate": 9.222936325958132e-07,
      "loss": 0.2686,
      "step": 28289
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34495781370992934,
      "learning_rate": 9.218776149592756e-07,
      "loss": 0.1913,
      "step": 28290
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.14817580702249203,
      "learning_rate": 9.214616866362902e-07,
      "loss": 0.0675,
      "step": 28291
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.578384744978483,
      "learning_rate": 9.210458476309558e-07,
      "loss": 0.8149,
      "step": 28292
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5550620715046355,
      "learning_rate": 9.206300979473581e-07,
      "loss": 0.1962,
      "step": 28293
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2985160398421109,
      "learning_rate": 9.202144375895882e-07,
      "loss": 0.2553,
      "step": 28294
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4301988574720083,
      "learning_rate": 9.197988665617386e-07,
      "loss": 0.2212,
      "step": 28295
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9823970128002536,
      "learning_rate": 9.193833848678923e-07,
      "loss": 0.4501,
      "step": 28296
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3045029188252717,
      "learning_rate": 9.189679925121431e-07,
      "loss": 0.2368,
      "step": 28297
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3865194751194348,
      "learning_rate": 9.185526894985742e-07,
      "loss": 0.2351,
      "step": 28298
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.49828256319572517,
      "learning_rate": 9.181374758312711e-07,
      "loss": 0.1817,
      "step": 28299
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5068654513310954,
      "learning_rate": 9.177223515143219e-07,
      "loss": 0.2825,
      "step": 28300
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23960493025376173,
      "learning_rate": 9.173073165518064e-07,
      "loss": 0.1523,
      "step": 28301
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6872850097177919,
      "learning_rate": 9.168923709478106e-07,
      "loss": 0.331,
      "step": 28302
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3842216673795802,
      "learning_rate": 9.164775147064165e-07,
      "loss": 0.1709,
      "step": 28303
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4754786491611773,
      "learning_rate": 9.160627478317064e-07,
      "loss": 0.2095,
      "step": 28304
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.47353362191623055,
      "learning_rate": 9.156480703277593e-07,
      "loss": 0.3415,
      "step": 28305
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4192035817246284,
      "learning_rate": 9.152334821986552e-07,
      "loss": 0.2092,
      "step": 28306
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0091338305911652,
      "learning_rate": 9.148189834484755e-07,
      "loss": 0.4443,
      "step": 28307
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3663370751339476,
      "learning_rate": 9.144045740812913e-07,
      "loss": 0.1915,
      "step": 28308
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4496422581571439,
      "learning_rate": 9.139902541011892e-07,
      "loss": 0.2397,
      "step": 28309
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6283087580827985,
      "learning_rate": 9.135760235122392e-07,
      "loss": 0.2489,
      "step": 28310
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5078943884634886,
      "learning_rate": 9.131618823185173e-07,
      "loss": 0.3242,
      "step": 28311
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2865741062139096,
      "learning_rate": 9.127478305241e-07,
      "loss": 0.1748,
      "step": 28312
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3130656781289595,
      "learning_rate": 9.123338681330595e-07,
      "loss": 0.2316,
      "step": 28313
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9711542079067494,
      "learning_rate": 9.119199951494672e-07,
      "loss": 0.5646,
      "step": 28314
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.782490439203468,
      "learning_rate": 9.115062115773976e-07,
      "loss": 0.7255,
      "step": 28315
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4777864826858195,
      "learning_rate": 9.110925174209206e-07,
      "loss": 0.0749,
      "step": 28316
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3840149148991374,
      "learning_rate": 9.106789126841042e-07,
      "loss": 0.1959,
      "step": 28317
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.35799450680775724,
      "learning_rate": 9.102653973710219e-07,
      "loss": 0.2834,
      "step": 28318
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7376780074316166,
      "learning_rate": 9.098519714857379e-07,
      "loss": 0.2675,
      "step": 28319
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44058036427769837,
      "learning_rate": 9.094386350323214e-07,
      "loss": 0.2291,
      "step": 28320
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.292983428642135,
      "learning_rate": 9.090253880148403e-07,
      "loss": 0.1731,
      "step": 28321
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.44391597874509403,
      "learning_rate": 9.086122304373557e-07,
      "loss": 0.1671,
      "step": 28322
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9993795204215119,
      "learning_rate": 9.081991623039388e-07,
      "loss": 0.4975,
      "step": 28323
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3219793488102532,
      "learning_rate": 9.077861836186486e-07,
      "loss": 0.2645,
      "step": 28324
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.31358428060727406,
      "learning_rate": 9.073732943855496e-07,
      "loss": 0.1514,
      "step": 28325
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6390890795619856,
      "learning_rate": 9.069604946087052e-07,
      "loss": 0.3395,
      "step": 28326
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0094169188972235,
      "learning_rate": 9.065477842921744e-07,
      "loss": 0.2486,
      "step": 28327
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6713522832823834,
      "learning_rate": 9.061351634400184e-07,
      "loss": 0.3594,
      "step": 28328
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3031055094211404,
      "learning_rate": 9.057226320562961e-07,
      "loss": 0.1958,
      "step": 28329
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39655686362264086,
      "learning_rate": 9.0531019014507e-07,
      "loss": 0.1626,
      "step": 28330
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.277340121065452,
      "learning_rate": 9.048978377103923e-07,
      "loss": 0.1823,
      "step": 28331
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1693695410955922,
      "learning_rate": 9.044855747563231e-07,
      "loss": 0.3367,
      "step": 28332
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1268544210502722,
      "learning_rate": 9.040734012869179e-07,
      "loss": 0.4131,
      "step": 28333
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.29654475775286004,
      "learning_rate": 9.036613173062292e-07,
      "loss": 0.0602,
      "step": 28334
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.35969868422065865,
      "learning_rate": 9.03249322818317e-07,
      "loss": 0.2409,
      "step": 28335
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37183081539822677,
      "learning_rate": 9.028374178272293e-07,
      "loss": 0.2294,
      "step": 28336
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8181855983984325,
      "learning_rate": 9.024256023370204e-07,
      "loss": 0.3872,
      "step": 28337
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.9019973731414006,
      "learning_rate": 9.020138763517438e-07,
      "loss": 0.0921,
      "step": 28338
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39467727250129286,
      "learning_rate": 9.016022398754465e-07,
      "loss": 0.2235,
      "step": 28339
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.18959711387972514,
      "learning_rate": 9.011906929121794e-07,
      "loss": 0.1205,
      "step": 28340
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5040138050991615,
      "learning_rate": 9.007792354659928e-07,
      "loss": 0.3292,
      "step": 28341
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4220230613667356,
      "learning_rate": 9.003678675409366e-07,
      "loss": 0.2291,
      "step": 28342
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9576029298660262,
      "learning_rate": 8.999565891410534e-07,
      "loss": 0.3931,
      "step": 28343
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34153461862505824,
      "learning_rate": 8.995454002703907e-07,
      "loss": 0.1891,
      "step": 28344
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.702657506503416,
      "learning_rate": 8.991343009329956e-07,
      "loss": 0.2568,
      "step": 28345
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8937670531831443,
      "learning_rate": 8.987232911329114e-07,
      "loss": 0.4119,
      "step": 28346
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3308522771043094,
      "learning_rate": 8.983123708741837e-07,
      "loss": 0.2123,
      "step": 28347
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5852258281170811,
      "learning_rate": 8.979015401608526e-07,
      "loss": 0.2175,
      "step": 28348
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19050462541980703,
      "learning_rate": 8.974907989969617e-07,
      "loss": 0.1116,
      "step": 28349
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4009071982311587,
      "learning_rate": 8.97080147386552e-07,
      "loss": 0.7026,
      "step": 28350
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1648989041272457,
      "learning_rate": 8.966695853336593e-07,
      "loss": 0.2154,
      "step": 28351
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5551939561357332,
      "learning_rate": 8.962591128423315e-07,
      "loss": 0.3117,
      "step": 28352
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.29983194756461473,
      "learning_rate": 8.958487299165997e-07,
      "loss": 0.1804,
      "step": 28353
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3807058508011955,
      "learning_rate": 8.95438436560504e-07,
      "loss": 0.2691,
      "step": 28354
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7357114647453683,
      "learning_rate": 8.950282327780802e-07,
      "loss": 0.2253,
      "step": 28355
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4085242589981994,
      "learning_rate": 8.94618118573366e-07,
      "loss": 0.5982,
      "step": 28356
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3020135099810314,
      "learning_rate": 8.942080939503939e-07,
      "loss": 0.0681,
      "step": 28357
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.25414832684968963,
      "learning_rate": 8.937981589131983e-07,
      "loss": 0.17,
      "step": 28358
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4667047048057012,
      "learning_rate": 8.93388313465815e-07,
      "loss": 0.3312,
      "step": 28359
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28971459859634585,
      "learning_rate": 8.929785576122696e-07,
      "loss": 0.1782,
      "step": 28360
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5443589761722525,
      "learning_rate": 8.925688913566022e-07,
      "loss": 0.4467,
      "step": 28361
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28106811813232147,
      "learning_rate": 8.921593147028362e-07,
      "loss": 0.1757,
      "step": 28362
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.798611502521836,
      "learning_rate": 8.917498276550052e-07,
      "loss": 0.3415,
      "step": 28363
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2406932503712305,
      "learning_rate": 8.913404302171369e-07,
      "loss": 0.1098,
      "step": 28364
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5098527902190729,
      "learning_rate": 8.909311223932571e-07,
      "loss": 0.3465,
      "step": 28365
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.29373374088029647,
      "learning_rate": 8.905219041873936e-07,
      "loss": 0.1562,
      "step": 28366
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4148941765162413,
      "learning_rate": 8.901127756035743e-07,
      "loss": 0.2509,
      "step": 28367
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9604911857163323,
      "learning_rate": 8.897037366458239e-07,
      "loss": 0.5687,
      "step": 28368
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4003794343710396,
      "learning_rate": 8.892947873181646e-07,
      "loss": 0.1963,
      "step": 28369
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.576237461392184,
      "learning_rate": 8.888859276246197e-07,
      "loss": 0.2466,
      "step": 28370
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.25348171392598784,
      "learning_rate": 8.884771575692142e-07,
      "loss": 0.1931,
      "step": 28371
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3506896401316596,
      "learning_rate": 8.880684771559689e-07,
      "loss": 0.2637,
      "step": 28372
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9660697451852756,
      "learning_rate": 8.876598863889052e-07,
      "loss": 0.0247,
      "step": 28373
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7988346861229636,
      "learning_rate": 8.872513852720399e-07,
      "loss": 0.7621,
      "step": 28374
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3668545682696071,
      "learning_rate": 8.868429738093942e-07,
      "loss": 0.1601,
      "step": 28375
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8848444126195161,
      "learning_rate": 8.864346520049882e-07,
      "loss": 0.3514,
      "step": 28376
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.45886479321187373,
      "learning_rate": 8.860264198628333e-07,
      "loss": 0.2186,
      "step": 28377
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5034307768919266,
      "learning_rate": 8.856182773869526e-07,
      "loss": 0.3233,
      "step": 28378
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5813541714324364,
      "learning_rate": 8.852102245813554e-07,
      "loss": 0.2375,
      "step": 28379
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3748381686085292,
      "learning_rate": 8.848022614500606e-07,
      "loss": 0.2601,
      "step": 28380
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.15913343745845376,
      "learning_rate": 8.843943879970818e-07,
      "loss": 0.0699,
      "step": 28381
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2996705447202042,
      "learning_rate": 8.839866042264289e-07,
      "loss": 0.391,
      "step": 28382
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3230781504429835,
      "learning_rate": 8.835789101421155e-07,
      "loss": 0.2219,
      "step": 28383
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5608450684996527,
      "learning_rate": 8.831713057481517e-07,
      "loss": 0.0618,
      "step": 28384
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39991581426760114,
      "learning_rate": 8.82763791048551e-07,
      "loss": 0.2258,
      "step": 28385
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8533420249222294,
      "learning_rate": 8.823563660473167e-07,
      "loss": 0.3489,
      "step": 28386
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6942568435208476,
      "learning_rate": 8.819490307484635e-07,
      "loss": 0.3517,
      "step": 28387
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39943869640220125,
      "learning_rate": 8.815417851559937e-07,
      "loss": 0.2198,
      "step": 28388
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4421819578963638,
      "learning_rate": 8.811346292739176e-07,
      "loss": 0.2658,
      "step": 28389
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23443337747737628,
      "learning_rate": 8.807275631062396e-07,
      "loss": 0.1614,
      "step": 28390
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4682033285625088,
      "learning_rate": 8.803205866569641e-07,
      "loss": 0.2225,
      "step": 28391
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.835299819891796,
      "learning_rate": 8.79913699930095e-07,
      "loss": 0.149,
      "step": 28392
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8992316729720293,
      "learning_rate": 8.795069029296354e-07,
      "loss": 0.4346,
      "step": 28393
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3309774839007576,
      "learning_rate": 8.791001956595901e-07,
      "loss": 0.2035,
      "step": 28394
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3396216191726033,
      "learning_rate": 8.78693578123957e-07,
      "loss": 0.2859,
      "step": 28395
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7406165227205602,
      "learning_rate": 8.782870503267371e-07,
      "loss": 0.259,
      "step": 28396
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.777736744906364,
      "learning_rate": 8.778806122719308e-07,
      "loss": 0.2786,
      "step": 28397
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3997362051970457,
      "learning_rate": 8.774742639635369e-07,
      "loss": 0.2417,
      "step": 28398
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19413481396803342,
      "learning_rate": 8.770680054055536e-07,
      "loss": 0.1142,
      "step": 28399
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3908939397085658,
      "learning_rate": 8.766618366019763e-07,
      "loss": 0.4389,
      "step": 28400
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3305694752601113,
      "learning_rate": 8.762557575568021e-07,
      "loss": 0.2017,
      "step": 28401
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6118389726213915,
      "learning_rate": 8.758497682740263e-07,
      "loss": 0.2508,
      "step": 28402
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37010586959363007,
      "learning_rate": 8.754438687576405e-07,
      "loss": 0.2214,
      "step": 28403
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0672367076732843,
      "learning_rate": 8.750380590116436e-07,
      "loss": 0.4958,
      "step": 28404
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7478326273494874,
      "learning_rate": 8.746323390400225e-07,
      "loss": 0.285,
      "step": 28405
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5064874406888342,
      "learning_rate": 8.742267088467716e-07,
      "loss": 0.3298,
      "step": 28406
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.31723776978571455,
      "learning_rate": 8.738211684358821e-07,
      "loss": 0.155,
      "step": 28407
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4253183789662708,
      "learning_rate": 8.734157178113423e-07,
      "loss": 0.2836,
      "step": 28408
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17110971666963457,
      "learning_rate": 8.730103569771408e-07,
      "loss": 0.0664,
      "step": 28409
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1334746775516582,
      "learning_rate": 8.72605085937267e-07,
      "loss": 0.1666,
      "step": 28410
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7257190447974617,
      "learning_rate": 8.721999046957097e-07,
      "loss": 0.3734,
      "step": 28411
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3262591840642797,
      "learning_rate": 8.717948132564502e-07,
      "loss": 0.1867,
      "step": 28412
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.35163268971813366,
      "learning_rate": 8.713898116234776e-07,
      "loss": 0.2834,
      "step": 28413
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7805471029260582,
      "learning_rate": 8.709848998007786e-07,
      "loss": 0.3057,
      "step": 28414
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.563003735492601,
      "learning_rate": 8.70580077792329e-07,
      "loss": 0.7166,
      "step": 28415
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4917118988160253,
      "learning_rate": 8.7017534560212e-07,
      "loss": 0.0708,
      "step": 28416
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3769913034238619,
      "learning_rate": 8.697707032341296e-07,
      "loss": 0.268,
      "step": 28417
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23454403723223127,
      "learning_rate": 8.693661506923379e-07,
      "loss": 0.1324,
      "step": 28418
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5490447029606736,
      "learning_rate": 8.689616879807295e-07,
      "loss": 0.3246,
      "step": 28419
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8911521102777387,
      "learning_rate": 8.685573151032778e-07,
      "loss": 0.1434,
      "step": 28420
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36426411073982146,
      "learning_rate": 8.681530320639642e-07,
      "loss": 0.2379,
      "step": 28421
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.628135626317629,
      "learning_rate": 8.677488388667654e-07,
      "loss": 0.2643,
      "step": 28422
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1632564163406909,
      "learning_rate": 8.673447355156595e-07,
      "loss": 0.4328,
      "step": 28423
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.49045716929962124,
      "learning_rate": 8.669407220146197e-07,
      "loss": 0.3402,
      "step": 28424
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.29854932215216395,
      "learning_rate": 8.665367983676254e-07,
      "loss": 0.1871,
      "step": 28425
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.555100761976005,
      "learning_rate": 8.661329645786443e-07,
      "loss": 0.3407,
      "step": 28426
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.17150164859805458,
      "learning_rate": 8.657292206516532e-07,
      "loss": 0.0713,
      "step": 28427
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.789058909766445,
      "learning_rate": 8.653255665906257e-07,
      "loss": 0.6946,
      "step": 28428
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6590419184924504,
      "learning_rate": 8.649220023995275e-07,
      "loss": 0.2305,
      "step": 28429
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34754713511432334,
      "learning_rate": 8.645185280823354e-07,
      "loss": 0.237,
      "step": 28430
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.29423448143070297,
      "learning_rate": 8.641151436430139e-07,
      "loss": 0.2001,
      "step": 28431
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0914360848057574,
      "learning_rate": 8.637118490855345e-07,
      "loss": 0.5235,
      "step": 28432
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3628405765888656,
      "learning_rate": 8.633086444138661e-07,
      "loss": 0.2953,
      "step": 28433
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4218348656468177,
      "learning_rate": 8.62905529631971e-07,
      "loss": 0.2601,
      "step": 28434
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6382416479182885,
      "learning_rate": 8.625025047438185e-07,
      "loss": 0.1796,
      "step": 28435
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.21879637357755885,
      "learning_rate": 8.620995697533729e-07,
      "loss": 0.1336,
      "step": 28436
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4881359679637633,
      "learning_rate": 8.616967246646002e-07,
      "loss": 0.332,
      "step": 28437
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7330477320269291,
      "learning_rate": 8.612939694814604e-07,
      "loss": 0.2116,
      "step": 28438
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.41551430179937454,
      "learning_rate": 8.60891304207917e-07,
      "loss": 0.247,
      "step": 28439
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5189911558745526,
      "learning_rate": 8.604887288479347e-07,
      "loss": 0.2332,
      "step": 28440
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.507691897969773,
      "learning_rate": 8.600862434054668e-07,
      "loss": 0.7396,
      "step": 28441
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.31259685553831895,
      "learning_rate": 8.596838478844827e-07,
      "loss": 0.2144,
      "step": 28442
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8896105206994608,
      "learning_rate": 8.592815422889344e-07,
      "loss": 0.3694,
      "step": 28443
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.30324892855010016,
      "learning_rate": 8.588793266227813e-07,
      "loss": 0.1829,
      "step": 28444
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3187346738251666,
      "learning_rate": 8.584772008899833e-07,
      "loss": 0.1513,
      "step": 28445
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1379285001715276,
      "learning_rate": 8.58075165094493e-07,
      "loss": 0.2269,
      "step": 28446
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6940676454764362,
      "learning_rate": 8.57673219240267e-07,
      "loss": 0.3436,
      "step": 28447
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.27101942715683,
      "learning_rate": 8.572713633312601e-07,
      "loss": 0.1668,
      "step": 28448
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.30408229908077905,
      "learning_rate": 8.56869597371427e-07,
      "loss": 0.2147,
      "step": 28449
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6309960331132223,
      "learning_rate": 8.564679213647198e-07,
      "loss": 0.7713,
      "step": 28450
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.288188737167173,
      "learning_rate": 8.560663353150889e-07,
      "loss": 0.2296,
      "step": 28451
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6171142383178215,
      "learning_rate": 8.556648392264854e-07,
      "loss": 0.255,
      "step": 28452
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.33599935446771495,
      "learning_rate": 8.552634331028598e-07,
      "loss": 0.1961,
      "step": 28453
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.430959767492536,
      "learning_rate": 8.548621169481641e-07,
      "loss": 0.2074,
      "step": 28454
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38358010974518025,
      "learning_rate": 8.544608907663399e-07,
      "loss": 0.2307,
      "step": 28455
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5692177781078042,
      "learning_rate": 8.540597545613427e-07,
      "loss": 0.2497,
      "step": 28456
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.28939595527695405,
      "learning_rate": 8.536587083371139e-07,
      "loss": 0.1693,
      "step": 28457
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.8951808401033308,
      "learning_rate": 8.532577520975993e-07,
      "loss": 0.835,
      "step": 28458
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0139650461829637,
      "learning_rate": 8.528568858467468e-07,
      "loss": 0.5575,
      "step": 28459
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.327779447293671,
      "learning_rate": 8.524561095884965e-07,
      "loss": 0.278,
      "step": 28460
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6409614706966332,
      "learning_rate": 8.520554233267919e-07,
      "loss": 0.1014,
      "step": 28461
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3660816158106221,
      "learning_rate": 8.516548270655766e-07,
      "loss": 0.1991,
      "step": 28462
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9443621346083072,
      "learning_rate": 8.512543208087942e-07,
      "loss": 0.3573,
      "step": 28463
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7694994797419797,
      "learning_rate": 8.508539045603792e-07,
      "loss": 0.2505,
      "step": 28464
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2562384575517803,
      "learning_rate": 8.50453578324274e-07,
      "loss": 0.1417,
      "step": 28465
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34811590741401827,
      "learning_rate": 8.500533421044188e-07,
      "loss": 0.1555,
      "step": 28466
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.304484087700567,
      "learning_rate": 8.496531959047472e-07,
      "loss": 0.2684,
      "step": 28467
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0733662679540714,
      "learning_rate": 8.492531397292003e-07,
      "loss": 0.5537,
      "step": 28468
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3853171013934584,
      "learning_rate": 8.488531735817107e-07,
      "loss": 0.3693,
      "step": 28469
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.46866658881959516,
      "learning_rate": 8.484532974662152e-07,
      "loss": 0.0578,
      "step": 28470
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3702157603796492,
      "learning_rate": 8.480535113866495e-07,
      "loss": 0.2689,
      "step": 28471
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3410377256078395,
      "learning_rate": 8.476538153469405e-07,
      "loss": 0.2184,
      "step": 28472
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9056346616016456,
      "learning_rate": 8.472542093510283e-07,
      "loss": 0.3736,
      "step": 28473
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4442113501204977,
      "learning_rate": 8.468546934028388e-07,
      "loss": 0.1266,
      "step": 28474
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.20109297975108614,
      "learning_rate": 8.464552675063054e-07,
      "loss": 0.1155,
      "step": 28475
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.609647901213655,
      "learning_rate": 8.460559316653583e-07,
      "loss": 0.3869,
      "step": 28476
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1021429809417318,
      "learning_rate": 8.456566858839233e-07,
      "loss": 0.4231,
      "step": 28477
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36750651174423965,
      "learning_rate": 8.452575301659294e-07,
      "loss": 0.2732,
      "step": 28478
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2693581352506611,
      "learning_rate": 8.448584645153046e-07,
      "loss": 0.0679,
      "step": 28479
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3737463994052222,
      "learning_rate": 8.444594889359769e-07,
      "loss": 0.2745,
      "step": 28480
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8876777044503171,
      "learning_rate": 8.440606034318677e-07,
      "loss": 0.2462,
      "step": 28481
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8274137077043878,
      "learning_rate": 8.436618080069026e-07,
      "loss": 0.3837,
      "step": 28482
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3509563530752616,
      "learning_rate": 8.432631026650073e-07,
      "loss": 0.1624,
      "step": 28483
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3808637710091031,
      "learning_rate": 8.428644874100989e-07,
      "loss": 0.2218,
      "step": 28484
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.33409718421357354,
      "learning_rate": 8.424659622461062e-07,
      "loss": 0.2087,
      "step": 28485
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.364660925062894,
      "learning_rate": 8.420675271769452e-07,
      "loss": 0.6947,
      "step": 28486
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.32918110314765836,
      "learning_rate": 8.416691822065371e-07,
      "loss": 0.0933,
      "step": 28487
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5381960943274136,
      "learning_rate": 8.412709273388009e-07,
      "loss": 0.1858,
      "step": 28488
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.527988152023042,
      "learning_rate": 8.408727625776569e-07,
      "loss": 0.2876,
      "step": 28489
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37015590492527806,
      "learning_rate": 8.404746879270175e-07,
      "loss": 0.2344,
      "step": 28490
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4835068378588639,
      "learning_rate": 8.400767033908031e-07,
      "loss": 0.3123,
      "step": 28491
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34393758058274687,
      "learning_rate": 8.396788089729302e-07,
      "loss": 0.0986,
      "step": 28492
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.1626925087231996,
      "learning_rate": 8.39281004677307e-07,
      "loss": 0.4154,
      "step": 28493
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3234123835514813,
      "learning_rate": 8.388832905078558e-07,
      "loss": 0.191,
      "step": 28494
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5369307929413539,
      "learning_rate": 8.384856664684826e-07,
      "loss": 0.2742,
      "step": 28495
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3225480809123536,
      "learning_rate": 8.38088132563103e-07,
      "loss": 0.2048,
      "step": 28496
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0301034666544235,
      "learning_rate": 8.376906887956283e-07,
      "loss": 0.4653,
      "step": 28497
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2597873077095862,
      "learning_rate": 8.372933351699641e-07,
      "loss": 0.1624,
      "step": 28498
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7146548606269256,
      "learning_rate": 8.368960716900254e-07,
      "loss": 0.3851,
      "step": 28499
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.598251643388104,
      "learning_rate": 8.364988983597177e-07,
      "loss": 0.0177,
      "step": 28500
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4232470376305897,
      "learning_rate": 8.36101815182948e-07,
      "loss": 0.2011,
      "step": 28501
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3763282302140995,
      "learning_rate": 8.357048221636266e-07,
      "loss": 0.2592,
      "step": 28502
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3554223678137926,
      "learning_rate": 8.353079193056546e-07,
      "loss": 0.2203,
      "step": 28503
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4010751339299472,
      "learning_rate": 8.34911106612939e-07,
      "loss": 0.7791,
      "step": 28504
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.283033844199393,
      "learning_rate": 8.345143840893833e-07,
      "loss": 0.1128,
      "step": 28505
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6428634354589771,
      "learning_rate": 8.341177517388932e-07,
      "loss": 0.3356,
      "step": 28506
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.289480217744044,
      "learning_rate": 8.337212095653669e-07,
      "loss": 0.1769,
      "step": 28507
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5173184396968485,
      "learning_rate": 8.333247575727066e-07,
      "loss": 0.314,
      "step": 28508
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4619997752991904,
      "learning_rate": 8.32928395764816e-07,
      "loss": 0.2232,
      "step": 28509
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.285938288398158,
      "learning_rate": 8.325321241455886e-07,
      "loss": 0.3225,
      "step": 28510
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.35992957641774986,
      "learning_rate": 8.321359427189302e-07,
      "loss": 0.1708,
      "step": 28511
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6140365950896529,
      "learning_rate": 8.317398514887332e-07,
      "loss": 0.3689,
      "step": 28512
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9850774612320617,
      "learning_rate": 8.313438504588956e-07,
      "loss": 0.4726,
      "step": 28513
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3013462153394509,
      "learning_rate": 8.309479396333165e-07,
      "loss": 0.2249,
      "step": 28514
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6374770850692432,
      "learning_rate": 8.305521190158872e-07,
      "loss": 0.3098,
      "step": 28515
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37002551231893466,
      "learning_rate": 8.301563886105024e-07,
      "loss": 0.1601,
      "step": 28516
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2655189739285792,
      "learning_rate": 8.297607484210568e-07,
      "loss": 0.1921,
      "step": 28517
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3831308402461393,
      "learning_rate": 8.293651984514451e-07,
      "loss": 0.1557,
      "step": 28518
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.36574004367232044,
      "learning_rate": 8.289697387055529e-07,
      "loss": 0.2855,
      "step": 28519
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4529015401734205,
      "learning_rate": 8.285743691872749e-07,
      "loss": 0.1386,
      "step": 28520
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3552830583758763,
      "learning_rate": 8.281790899005015e-07,
      "loss": 0.2592,
      "step": 28521
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.416464559545101,
      "learning_rate": 8.277839008491162e-07,
      "loss": 0.5364,
      "step": 28522
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6825036269250822,
      "learning_rate": 8.273888020370146e-07,
      "loss": 0.3421,
      "step": 28523
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.40682194797545845,
      "learning_rate": 8.269937934680772e-07,
      "loss": 0.1333,
      "step": 28524
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3689546955749279,
      "learning_rate": 8.265988751461962e-07,
      "loss": 0.2126,
      "step": 28525
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.24296062368276358,
      "learning_rate": 8.26204047075253e-07,
      "loss": 0.218,
      "step": 28526
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3045145950703554,
      "learning_rate": 8.258093092591324e-07,
      "loss": 0.4952,
      "step": 28527
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5895810804968291,
      "learning_rate": 8.2541466170172e-07,
      "loss": 0.088,
      "step": 28528
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34091283269060124,
      "learning_rate": 8.250201044068951e-07,
      "loss": 0.1492,
      "step": 28529
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39502792415503707,
      "learning_rate": 8.246256373785421e-07,
      "loss": 0.28,
      "step": 28530
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.8294103164163965,
      "learning_rate": 8.242312606205404e-07,
      "loss": 0.2845,
      "step": 28531
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3294405557547434,
      "learning_rate": 8.238369741367735e-07,
      "loss": 0.2629,
      "step": 28532
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7926017038326245,
      "learning_rate": 8.234427779311161e-07,
      "loss": 0.2532,
      "step": 28533
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5625556163803553,
      "learning_rate": 8.230486720074471e-07,
      "loss": 0.2656,
      "step": 28534
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2868970305917398,
      "learning_rate": 8.226546563696469e-07,
      "loss": 0.1418,
      "step": 28535
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5018331959754585,
      "learning_rate": 8.222607310215869e-07,
      "loss": 0.2527,
      "step": 28536
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.325285856114357,
      "learning_rate": 8.218668959671483e-07,
      "loss": 0.2171,
      "step": 28537
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.887504925762142,
      "learning_rate": 8.214731512102026e-07,
      "loss": 0.4494,
      "step": 28538
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.37955741429350337,
      "learning_rate": 8.210794967546232e-07,
      "loss": 0.1582,
      "step": 28539
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5260779767099248,
      "learning_rate": 8.206859326042859e-07,
      "loss": 0.3334,
      "step": 28540
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.6493261527641441,
      "learning_rate": 8.202924587630601e-07,
      "loss": 0.2677,
      "step": 28541
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.23668071681348174,
      "learning_rate": 8.198990752348168e-07,
      "loss": 0.0689,
      "step": 28542
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0063181814025464,
      "learning_rate": 8.195057820234265e-07,
      "loss": 0.4176,
      "step": 28543
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.22895921748471715,
      "learning_rate": 8.191125791327614e-07,
      "loss": 0.2035,
      "step": 28544
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5426976038549827,
      "learning_rate": 8.187194665666864e-07,
      "loss": 0.1827,
      "step": 28545
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.3554677151554666,
      "learning_rate": 8.183264443290695e-07,
      "loss": 0.3127,
      "step": 28546
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.6314797756340855,
      "learning_rate": 8.179335124237808e-07,
      "loss": 0.7188,
      "step": 28547
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2784977506951622,
      "learning_rate": 8.175406708546796e-07,
      "loss": 0.1699,
      "step": 28548
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7136530765523355,
      "learning_rate": 8.171479196256382e-07,
      "loss": 0.3707,
      "step": 28549
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.316362550602012,
      "learning_rate": 8.167552587405147e-07,
      "loss": 0.2346,
      "step": 28550
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.638836935685643,
      "learning_rate": 8.16362688203175e-07,
      "loss": 0.7033,
      "step": 28551
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.27334386643473796,
      "learning_rate": 8.159702080174836e-07,
      "loss": 0.0659,
      "step": 28552
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2680011430249506,
      "learning_rate": 8.155778181872953e-07,
      "loss": 0.1948,
      "step": 28553
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.38412129169506015,
      "learning_rate": 8.15185518716477e-07,
      "loss": 0.0768,
      "step": 28554
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.34131190984895415,
      "learning_rate": 8.147933096088856e-07,
      "loss": 0.2171,
      "step": 28555
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.9061417047178937,
      "learning_rate": 8.144011908683791e-07,
      "loss": 0.4168,
      "step": 28556
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2784850384251492,
      "learning_rate": 8.140091624988155e-07,
      "loss": 0.1682,
      "step": 28557
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7246726476448054,
      "learning_rate": 8.136172245040553e-07,
      "loss": 0.3704,
      "step": 28558
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.0611544237748054,
      "learning_rate": 8.132253768879494e-07,
      "loss": 0.4391,
      "step": 28559
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.4157149094596377,
      "learning_rate": 8.128336196543551e-07,
      "loss": 0.4648,
      "step": 28560
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.2917332729259865,
      "learning_rate": 8.124419528071281e-07,
      "loss": 0.1907,
      "step": 28561
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.39827557759617727,
      "learning_rate": 8.120503763501175e-07,
      "loss": 0.2915,
      "step": 28562
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.19821123660713008,
      "learning_rate": 8.116588902871814e-07,
      "loss": 0.0718,
      "step": 28563
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.2679700315725524,
      "learning_rate": 8.112674946221666e-07,
      "loss": 0.3546,
      "step": 28564
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7771392557071212,
      "learning_rate": 8.108761893589256e-07,
      "loss": 0.2291,
      "step": 28565
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.4502271064263745,
      "learning_rate": 8.10484974501311e-07,
      "loss": 0.2415,
      "step": 28566
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.5865803037309747,
      "learning_rate": 8.100938500531664e-07,
      "loss": 0.2334,
      "step": 28567
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3182932945651035,
      "learning_rate": 8.097028160183429e-07,
      "loss": 0.2402,
      "step": 28568
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.5035165060463083,
      "learning_rate": 8.093118724006865e-07,
      "loss": 0.5652,
      "step": 28569
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.3550011106056973,
      "learning_rate": 8.089210192040464e-07,
      "loss": 0.0587,
      "step": 28570
    },
    {
      "epoch": 0.87,
      "grad_norm": 0.7973207690694287,
      "learning_rate": 8.085302564322628e-07,
      "loss": 0.2495,
      "step": 28571
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.15272638260398008,
      "learning_rate": 8.081395840891837e-07,
      "loss": 0.0691,
      "step": 28572
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34388693226290545,
      "learning_rate": 8.077490021786539e-07,
      "loss": 0.2841,
      "step": 28573
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7412512422556886,
      "learning_rate": 8.073585107045101e-07,
      "loss": 0.2342,
      "step": 28574
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4067943381416462,
      "learning_rate": 8.069681096706017e-07,
      "loss": 0.2252,
      "step": 28575
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45626217178605444,
      "learning_rate": 8.065777990807644e-07,
      "loss": 0.2351,
      "step": 28576
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4375216420362444,
      "learning_rate": 8.061875789388396e-07,
      "loss": 0.6868,
      "step": 28577
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2512643412656794,
      "learning_rate": 8.057974492486675e-07,
      "loss": 0.0786,
      "step": 28578
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3193481676477278,
      "learning_rate": 8.054074100140829e-07,
      "loss": 0.2598,
      "step": 28579
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3343473736085999,
      "learning_rate": 8.050174612389294e-07,
      "loss": 0.1849,
      "step": 28580
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.15633816827177646,
      "learning_rate": 8.046276029270373e-07,
      "loss": 0.0664,
      "step": 28581
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0449822637026722,
      "learning_rate": 8.042378350822445e-07,
      "loss": 0.4099,
      "step": 28582
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7086580766877841,
      "learning_rate": 8.038481577083879e-07,
      "loss": 0.2643,
      "step": 28583
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4134146957629615,
      "learning_rate": 8.034585708092968e-07,
      "loss": 0.2702,
      "step": 28584
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3262928484974597,
      "learning_rate": 8.030690743888058e-07,
      "loss": 0.213,
      "step": 28585
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5659892931919327,
      "learning_rate": 8.026796684507487e-07,
      "loss": 0.3217,
      "step": 28586
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4472660725650464,
      "learning_rate": 8.022903529989567e-07,
      "loss": 0.3139,
      "step": 28587
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.441813193138568,
      "learning_rate": 8.019011280372546e-07,
      "loss": 0.2844,
      "step": 28588
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3577459351734804,
      "learning_rate": 8.015119935694793e-07,
      "loss": 0.2025,
      "step": 28589
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.29890869172864687,
      "learning_rate": 8.011229495994544e-07,
      "loss": 0.179,
      "step": 28590
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35183767258741566,
      "learning_rate": 8.00733996131009e-07,
      "loss": 0.2349,
      "step": 28591
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8892185309303708,
      "learning_rate": 8.003451331679701e-07,
      "loss": 0.3695,
      "step": 28592
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.31499243916743563,
      "learning_rate": 7.999563607141625e-07,
      "loss": 0.1481,
      "step": 28593
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47300108624745013,
      "learning_rate": 7.995676787734107e-07,
      "loss": 0.204,
      "step": 28594
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2630886236720777,
      "learning_rate": 7.991790873495397e-07,
      "loss": 0.7477,
      "step": 28595
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3431556270916874,
      "learning_rate": 7.987905864463741e-07,
      "loss": 0.2178,
      "step": 28596
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.930895731630085,
      "learning_rate": 7.984021760677318e-07,
      "loss": 0.3615,
      "step": 28597
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2853705981267769,
      "learning_rate": 7.980138562174378e-07,
      "loss": 0.1787,
      "step": 28598
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5096792790596417,
      "learning_rate": 7.97625626899311e-07,
      "loss": 0.197,
      "step": 28599
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6184800360590502,
      "learning_rate": 7.972374881171707e-07,
      "loss": 0.2653,
      "step": 28600
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45806914405297017,
      "learning_rate": 7.968494398748383e-07,
      "loss": 0.2138,
      "step": 28601
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39347951900983713,
      "learning_rate": 7.964614821761274e-07,
      "loss": 0.1115,
      "step": 28602
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2983409514236434,
      "learning_rate": 7.960736150248571e-07,
      "loss": 0.2488,
      "step": 28603
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39402911850100697,
      "learning_rate": 7.956858384248445e-07,
      "loss": 0.2423,
      "step": 28604
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.668905685470416,
      "learning_rate": 7.952981523798997e-07,
      "loss": 0.815,
      "step": 28605
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3566345270998186,
      "learning_rate": 7.949105568938442e-07,
      "loss": 0.0619,
      "step": 28606
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3201687612637985,
      "learning_rate": 7.94523051970485e-07,
      "loss": 0.1863,
      "step": 28607
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7396569294597196,
      "learning_rate": 7.941356376136377e-07,
      "loss": 0.3596,
      "step": 28608
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33265794186610653,
      "learning_rate": 7.937483138271151e-07,
      "loss": 0.2177,
      "step": 28609
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.41415962860526423,
      "learning_rate": 7.933610806147241e-07,
      "loss": 0.1833,
      "step": 28610
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.20852019956399487,
      "learning_rate": 7.929739379802759e-07,
      "loss": 0.113,
      "step": 28611
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5598685518469897,
      "learning_rate": 7.925868859275787e-07,
      "loss": 0.332,
      "step": 28612
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1184342698175134,
      "learning_rate": 7.921999244604439e-07,
      "loss": 0.4118,
      "step": 28613
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.519814259414604,
      "learning_rate": 7.918130535826751e-07,
      "loss": 0.2973,
      "step": 28614
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37329491485973354,
      "learning_rate": 7.914262732980782e-07,
      "loss": 0.1672,
      "step": 28615
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.341008862986304,
      "learning_rate": 7.910395836104623e-07,
      "loss": 0.2366,
      "step": 28616
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7966999488646522,
      "learning_rate": 7.906529845236244e-07,
      "loss": 0.2458,
      "step": 28617
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7709397391084513,
      "learning_rate": 7.90266476041377e-07,
      "loss": 0.3637,
      "step": 28618
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16828257418377307,
      "learning_rate": 7.89880058167517e-07,
      "loss": 0.0835,
      "step": 28619
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36585828524838093,
      "learning_rate": 7.89493730905847e-07,
      "loss": 0.1529,
      "step": 28620
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3455168865243204,
      "learning_rate": 7.891074942601707e-07,
      "loss": 0.2401,
      "step": 28621
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3987898816911998,
      "learning_rate": 7.887213482342825e-07,
      "loss": 0.2464,
      "step": 28622
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8839491105321118,
      "learning_rate": 7.883352928319865e-07,
      "loss": 0.7192,
      "step": 28623
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0342792539661958,
      "learning_rate": 7.879493280570771e-07,
      "loss": 0.2,
      "step": 28624
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4074932601258918,
      "learning_rate": 7.875634539133548e-07,
      "loss": 0.2134,
      "step": 28625
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5739544056794708,
      "learning_rate": 7.871776704046141e-07,
      "loss": 0.2436,
      "step": 28626
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36572890092176497,
      "learning_rate": 7.867919775346522e-07,
      "loss": 0.2694,
      "step": 28627
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.890777514184877,
      "learning_rate": 7.864063753072604e-07,
      "loss": 0.029,
      "step": 28628
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37439340296204504,
      "learning_rate": 7.860208637262356e-07,
      "loss": 0.1844,
      "step": 28629
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30344464449378644,
      "learning_rate": 7.856354427953705e-07,
      "loss": 0.1857,
      "step": 28630
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43500685238020786,
      "learning_rate": 7.852501125184519e-07,
      "loss": 0.2586,
      "step": 28631
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48791733795012343,
      "learning_rate": 7.848648728992781e-07,
      "loss": 0.0141,
      "step": 28632
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34393463024720183,
      "learning_rate": 7.844797239416347e-07,
      "loss": 0.235,
      "step": 28633
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4347249308913892,
      "learning_rate": 7.84094665649312e-07,
      "loss": 0.2309,
      "step": 28634
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5722638394519245,
      "learning_rate": 7.837096980261005e-07,
      "loss": 0.2295,
      "step": 28635
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.8798535342757812,
      "learning_rate": 7.833248210757827e-07,
      "loss": 0.7667,
      "step": 28636
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.24818021053171813,
      "learning_rate": 7.829400348021488e-07,
      "loss": 0.0942,
      "step": 28637
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3780785493216878,
      "learning_rate": 7.825553392089835e-07,
      "loss": 0.2388,
      "step": 28638
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26123612944926594,
      "learning_rate": 7.82170734300074e-07,
      "loss": 0.2142,
      "step": 28639
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6443168246084352,
      "learning_rate": 7.817862200791992e-07,
      "loss": 0.6964,
      "step": 28640
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4363867526858571,
      "learning_rate": 7.814017965501453e-07,
      "loss": 0.1357,
      "step": 28641
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7030275136697699,
      "learning_rate": 7.810174637166956e-07,
      "loss": 0.3502,
      "step": 28642
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2776736188695978,
      "learning_rate": 7.806332215826262e-07,
      "loss": 0.171,
      "step": 28643
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.442147998587065,
      "learning_rate": 7.80249070151724e-07,
      "loss": 0.5458,
      "step": 28644
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3076870029929464,
      "learning_rate": 7.798650094277627e-07,
      "loss": 0.2147,
      "step": 28645
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1210849560804372,
      "learning_rate": 7.794810394145246e-07,
      "loss": 0.1304,
      "step": 28646
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6810068892496094,
      "learning_rate": 7.79097160115787e-07,
      "loss": 0.2988,
      "step": 28647
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33951537784629476,
      "learning_rate": 7.787133715353234e-07,
      "loss": 0.1981,
      "step": 28648
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5081499804909213,
      "learning_rate": 7.78329673676913e-07,
      "loss": 0.3388,
      "step": 28649
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26592117539046983,
      "learning_rate": 7.779460665443284e-07,
      "loss": 0.1908,
      "step": 28650
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8429173719382985,
      "learning_rate": 7.775625501413453e-07,
      "loss": 0.3381,
      "step": 28651
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.32423281461147974,
      "learning_rate": 7.771791244717386e-07,
      "loss": 0.1492,
      "step": 28652
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5313133435678098,
      "learning_rate": 7.767957895392763e-07,
      "loss": 0.2917,
      "step": 28653
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6176754471297121,
      "learning_rate": 7.764125453477323e-07,
      "loss": 0.0404,
      "step": 28654
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7409556289477752,
      "learning_rate": 7.760293919008766e-07,
      "loss": 0.7221,
      "step": 28655
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2958037811398663,
      "learning_rate": 7.756463292024796e-07,
      "loss": 0.2053,
      "step": 28656
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3684792316968412,
      "learning_rate": 7.752633572563062e-07,
      "loss": 0.2729,
      "step": 28657
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0666579388177364,
      "learning_rate": 7.74880476066131e-07,
      "loss": 0.4843,
      "step": 28658
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.810648174166691,
      "learning_rate": 7.744976856357156e-07,
      "loss": 0.2854,
      "step": 28659
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45369339553776256,
      "learning_rate": 7.741149859688268e-07,
      "loss": 0.2298,
      "step": 28660
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2444705915968891,
      "learning_rate": 7.737323770692329e-07,
      "loss": 0.1219,
      "step": 28661
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3675994148215929,
      "learning_rate": 7.73349858940694e-07,
      "loss": 0.2498,
      "step": 28662
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4864442452750057,
      "learning_rate": 7.729674315869751e-07,
      "loss": 0.2065,
      "step": 28663
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1077837328046805,
      "learning_rate": 7.725850950118385e-07,
      "loss": 0.4112,
      "step": 28664
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5019374929638564,
      "learning_rate": 7.72202849219047e-07,
      "loss": 0.2433,
      "step": 28665
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3778507348191856,
      "learning_rate": 7.718206942123596e-07,
      "loss": 0.2613,
      "step": 28666
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9695192403847648,
      "learning_rate": 7.714386299955357e-07,
      "loss": 0.2775,
      "step": 28667
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4633306537661925,
      "learning_rate": 7.710566565723376e-07,
      "loss": 0.2877,
      "step": 28668
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30962058885978666,
      "learning_rate": 7.70674773946517e-07,
      "loss": 0.1736,
      "step": 28669
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4184665214513133,
      "learning_rate": 7.702929821218386e-07,
      "loss": 0.2581,
      "step": 28670
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.29645121478136044,
      "learning_rate": 7.699112811020537e-07,
      "loss": 0.1348,
      "step": 28671
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4625563905850862,
      "learning_rate": 7.695296708909173e-07,
      "loss": 0.1701,
      "step": 28672
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.441632869967681,
      "learning_rate": 7.691481514921883e-07,
      "loss": 0.691,
      "step": 28673
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2902052576091925,
      "learning_rate": 7.687667229096141e-07,
      "loss": 0.2053,
      "step": 28674
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45073327690383885,
      "learning_rate": 7.683853851469514e-07,
      "loss": 0.2709,
      "step": 28675
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4682017775127465,
      "learning_rate": 7.680041382079506e-07,
      "loss": 0.239,
      "step": 28676
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6739103291776913,
      "learning_rate": 7.676229820963632e-07,
      "loss": 0.351,
      "step": 28677
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2897573115896198,
      "learning_rate": 7.672419168159395e-07,
      "loss": 0.0675,
      "step": 28678
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3782486092224632,
      "learning_rate": 7.668609423704266e-07,
      "loss": 0.2079,
      "step": 28679
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.25598527728975945,
      "learning_rate": 7.664800587635735e-07,
      "loss": 0.2019,
      "step": 28680
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48596608962729465,
      "learning_rate": 7.660992659991285e-07,
      "loss": 0.2918,
      "step": 28681
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4110908663060986,
      "learning_rate": 7.657185640808384e-07,
      "loss": 0.1605,
      "step": 28682
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4309650514589676,
      "learning_rate": 7.65337953012446e-07,
      "loss": 0.7101,
      "step": 28683
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30803719731263396,
      "learning_rate": 7.649574327976983e-07,
      "loss": 0.1774,
      "step": 28684
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5752911387798075,
      "learning_rate": 7.645770034403388e-07,
      "loss": 0.2716,
      "step": 28685
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5401900801737349,
      "learning_rate": 7.641966649441057e-07,
      "loss": 0.3551,
      "step": 28686
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.43133416210868425,
      "learning_rate": 7.638164173127494e-07,
      "loss": 0.2324,
      "step": 28687
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4812297016839614,
      "learning_rate": 7.634362605500045e-07,
      "loss": 0.2055,
      "step": 28688
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.24537494746646596,
      "learning_rate": 7.630561946596115e-07,
      "loss": 0.1559,
      "step": 28689
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.965948482332703,
      "learning_rate": 7.62676219645313e-07,
      "loss": 0.7117,
      "step": 28690
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4845046083878257,
      "learning_rate": 7.622963355108459e-07,
      "loss": 0.3133,
      "step": 28691
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3695174855321055,
      "learning_rate": 7.61916542259945e-07,
      "loss": 0.2945,
      "step": 28692
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.27816797389167114,
      "learning_rate": 7.615368398963496e-07,
      "loss": 0.1663,
      "step": 28693
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6070655563601188,
      "learning_rate": 7.611572284237955e-07,
      "loss": 0.4499,
      "step": 28694
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7679250340135452,
      "learning_rate": 7.607777078460144e-07,
      "loss": 0.2944,
      "step": 28695
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1297333861466203,
      "learning_rate": 7.603982781667452e-07,
      "loss": 0.3343,
      "step": 28696
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2336528181722569,
      "learning_rate": 7.600189393897162e-07,
      "loss": 0.1364,
      "step": 28697
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3360607798499166,
      "learning_rate": 7.59639691518661e-07,
      "loss": 0.1989,
      "step": 28698
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34027621385984075,
      "learning_rate": 7.592605345573123e-07,
      "loss": 0.2142,
      "step": 28699
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2904393346933565,
      "learning_rate": 7.588814685093982e-07,
      "loss": 0.2473,
      "step": 28700
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6703331725200505,
      "learning_rate": 7.585024933786489e-07,
      "loss": 0.3191,
      "step": 28701
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4002654245167609,
      "learning_rate": 7.581236091687927e-07,
      "loss": 0.1661,
      "step": 28702
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44670838493489573,
      "learning_rate": 7.577448158835576e-07,
      "loss": 0.3117,
      "step": 28703
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33822965385402615,
      "learning_rate": 7.573661135266719e-07,
      "loss": 0.2311,
      "step": 28704
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5089832450668423,
      "learning_rate": 7.569875021018569e-07,
      "loss": 0.6437,
      "step": 28705
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5605919333859789,
      "learning_rate": 7.566089816128408e-07,
      "loss": 0.0636,
      "step": 28706
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4656655196600679,
      "learning_rate": 7.562305520633473e-07,
      "loss": 0.2511,
      "step": 28707
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26187934395278045,
      "learning_rate": 7.558522134571011e-07,
      "loss": 0.0738,
      "step": 28708
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5466832812522244,
      "learning_rate": 7.554739657978206e-07,
      "loss": 0.2989,
      "step": 28709
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4210778004282679,
      "learning_rate": 7.550958090892291e-07,
      "loss": 0.2361,
      "step": 28710
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35362372662610747,
      "learning_rate": 7.547177433350484e-07,
      "loss": 0.1783,
      "step": 28711
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5178260443670282,
      "learning_rate": 7.543397685389941e-07,
      "loss": 0.3283,
      "step": 28712
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2613897776995717,
      "learning_rate": 7.539618847047891e-07,
      "loss": 0.4453,
      "step": 28713
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5597281101510037,
      "learning_rate": 7.535840918361492e-07,
      "loss": 0.5635,
      "step": 28714
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1687691643861288,
      "learning_rate": 7.532063899367914e-07,
      "loss": 0.0702,
      "step": 28715
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.29635161186981573,
      "learning_rate": 7.528287790104327e-07,
      "loss": 0.2449,
      "step": 28716
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2792342283570653,
      "learning_rate": 7.524512590607857e-07,
      "loss": 0.1622,
      "step": 28717
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0066767499303944,
      "learning_rate": 7.520738300915664e-07,
      "loss": 0.3906,
      "step": 28718
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6838769498528737,
      "learning_rate": 7.516964921064873e-07,
      "loss": 0.2692,
      "step": 28719
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3901690591840432,
      "learning_rate": 7.513192451092633e-07,
      "loss": 0.2515,
      "step": 28720
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.51427769737856,
      "learning_rate": 7.509420891036024e-07,
      "loss": 0.2286,
      "step": 28721
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34847683364256543,
      "learning_rate": 7.505650240932161e-07,
      "loss": 0.2858,
      "step": 28722
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5288769337378758,
      "learning_rate": 7.50188050081816e-07,
      "loss": 0.2533,
      "step": 28723
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9908420505710545,
      "learning_rate": 7.498111670731067e-07,
      "loss": 0.3361,
      "step": 28724
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4034741663032102,
      "learning_rate": 7.494343750708022e-07,
      "loss": 0.1724,
      "step": 28725
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2148540576534147,
      "learning_rate": 7.490576740786015e-07,
      "loss": 0.1167,
      "step": 28726
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4665512572720965,
      "learning_rate": 7.486810641002207e-07,
      "loss": 0.2873,
      "step": 28727
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4284332470021412,
      "learning_rate": 7.483045451393567e-07,
      "loss": 0.2205,
      "step": 28728
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9157285484288973,
      "learning_rate": 7.479281171997166e-07,
      "loss": 0.4394,
      "step": 28729
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3578347247696468,
      "learning_rate": 7.475517802850063e-07,
      "loss": 0.2176,
      "step": 28730
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.316895243427349,
      "learning_rate": 7.47175534398924e-07,
      "loss": 0.5627,
      "step": 28731
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3382953661598385,
      "learning_rate": 7.467993795451744e-07,
      "loss": 0.0751,
      "step": 28732
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4717531531061024,
      "learning_rate": 7.464233157274559e-07,
      "loss": 0.3172,
      "step": 28733
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.28872707612946125,
      "learning_rate": 7.460473429494719e-07,
      "loss": 0.1626,
      "step": 28734
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.22287675118667805,
      "learning_rate": 7.456714612149174e-07,
      "loss": 0.1793,
      "step": 28735
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6047596219827435,
      "learning_rate": 7.452956705274928e-07,
      "loss": 0.2693,
      "step": 28736
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.781753519894413,
      "learning_rate": 7.449199708908949e-07,
      "loss": 0.6312,
      "step": 28737
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.32253502413536755,
      "learning_rate": 7.445443623088167e-07,
      "loss": 0.1503,
      "step": 28738
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2750579484037936,
      "learning_rate": 7.441688447849604e-07,
      "loss": 0.2193,
      "step": 28739
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7358537733388059,
      "learning_rate": 7.437934183230145e-07,
      "loss": 0.3076,
      "step": 28740
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7719720856234207,
      "learning_rate": 7.434180829266746e-07,
      "loss": 0.02,
      "step": 28741
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3439221337620268,
      "learning_rate": 7.430428385996358e-07,
      "loss": 0.4014,
      "step": 28742
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2675792939178309,
      "learning_rate": 7.42667685345585e-07,
      "loss": 0.1692,
      "step": 28743
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8126151046853098,
      "learning_rate": 7.42292623168216e-07,
      "loss": 0.3739,
      "step": 28744
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4559155550179637,
      "learning_rate": 7.41917652071219e-07,
      "loss": 0.237,
      "step": 28745
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.25994748342848895,
      "learning_rate": 7.415427720582824e-07,
      "loss": 0.1832,
      "step": 28746
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.32982342160773753,
      "learning_rate": 7.411679831330943e-07,
      "loss": 0.1626,
      "step": 28747
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5848199773348357,
      "learning_rate": 7.407932852993416e-07,
      "loss": 0.3416,
      "step": 28748
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0569764348802706,
      "learning_rate": 7.404186785607125e-07,
      "loss": 0.2522,
      "step": 28749
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2691170107801777,
      "learning_rate": 7.400441629208887e-07,
      "loss": 0.5728,
      "step": 28750
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.296912913018194,
      "learning_rate": 7.396697383835605e-07,
      "loss": 0.2074,
      "step": 28751
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4013966450302051,
      "learning_rate": 7.39295404952406e-07,
      "loss": 0.1679,
      "step": 28752
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.512189479963427,
      "learning_rate": 7.38921162631111e-07,
      "loss": 0.2906,
      "step": 28753
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6793034364996414,
      "learning_rate": 7.385470114233595e-07,
      "loss": 0.2505,
      "step": 28754
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4351307964610447,
      "learning_rate": 7.381729513328251e-07,
      "loss": 0.2273,
      "step": 28755
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44996314240241186,
      "learning_rate": 7.37798982363197e-07,
      "loss": 0.1355,
      "step": 28756
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.357963779884799,
      "learning_rate": 7.37425104518148e-07,
      "loss": 0.2685,
      "step": 28757
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.23757979316018019,
      "learning_rate": 7.370513178013583e-07,
      "loss": 0.191,
      "step": 28758
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3486573413194358,
      "learning_rate": 7.366776222165062e-07,
      "loss": 0.5445,
      "step": 28759
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9972361754095285,
      "learning_rate": 7.363040177672698e-07,
      "loss": 0.2386,
      "step": 28760
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4006419833089272,
      "learning_rate": 7.359305044573206e-07,
      "loss": 0.2398,
      "step": 28761
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4702529244084941,
      "learning_rate": 7.355570822903357e-07,
      "loss": 0.2406,
      "step": 28762
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34699131010144946,
      "learning_rate": 7.351837512699889e-07,
      "loss": 0.2764,
      "step": 28763
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.358149954038913,
      "learning_rate": 7.348105113999515e-07,
      "loss": 0.0721,
      "step": 28764
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.33724214158999255,
      "learning_rate": 7.344373626838985e-07,
      "loss": 0.1497,
      "step": 28765
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38584886510757527,
      "learning_rate": 7.340643051254992e-07,
      "loss": 0.2813,
      "step": 28766
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3365629444754276,
      "learning_rate": 7.336913387284228e-07,
      "loss": 0.0931,
      "step": 28767
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.321664201810867,
      "learning_rate": 7.333184634963431e-07,
      "loss": 0.4299,
      "step": 28768
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.32079788819354427,
      "learning_rate": 7.329456794329226e-07,
      "loss": 0.2316,
      "step": 28769
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4103822418719072,
      "learning_rate": 7.325729865418329e-07,
      "loss": 0.2592,
      "step": 28770
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.48918592589778154,
      "learning_rate": 7.32200384826739e-07,
      "loss": 0.2294,
      "step": 28771
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2846862705503934,
      "learning_rate": 7.318278742913087e-07,
      "loss": 0.6358,
      "step": 28772
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.9950315660850205,
      "learning_rate": 7.314554549392039e-07,
      "loss": 0.4059,
      "step": 28773
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8677842880366317,
      "learning_rate": 7.310831267740904e-07,
      "loss": 0.4403,
      "step": 28774
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39073193123212313,
      "learning_rate": 7.307108897996329e-07,
      "loss": 0.162,
      "step": 28775
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2836695624666023,
      "learning_rate": 7.303387440194875e-07,
      "loss": 0.2449,
      "step": 28776
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.27078582934272716,
      "learning_rate": 7.299666894373236e-07,
      "loss": 0.0809,
      "step": 28777
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.684817003816922,
      "learning_rate": 7.295947260567948e-07,
      "loss": 0.2297,
      "step": 28778
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8908852688845529,
      "learning_rate": 7.292228538815649e-07,
      "loss": 0.371,
      "step": 28779
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3545341421247606,
      "learning_rate": 7.28851072915292e-07,
      "loss": 0.2142,
      "step": 28780
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.366307021536817,
      "learning_rate": 7.284793831616299e-07,
      "loss": 0.2728,
      "step": 28781
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.316111819293608,
      "learning_rate": 7.281077846242413e-07,
      "loss": 0.276,
      "step": 28782
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6060156544922293,
      "learning_rate": 7.277362773067776e-07,
      "loss": 0.6974,
      "step": 28783
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2810034821727676,
      "learning_rate": 7.273648612128958e-07,
      "loss": 0.1754,
      "step": 28784
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.27357808188796406,
      "learning_rate": 7.26993536346251e-07,
      "loss": 0.1873,
      "step": 28785
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4341024044982896,
      "learning_rate": 7.266223027104935e-07,
      "loss": 0.2276,
      "step": 28786
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.510187424253961,
      "learning_rate": 7.26251160309277e-07,
      "loss": 0.333,
      "step": 28787
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5230569341942416,
      "learning_rate": 7.25880109146252e-07,
      "loss": 0.1908,
      "step": 28788
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4003610143998965,
      "learning_rate": 7.255091492250732e-07,
      "loss": 0.2621,
      "step": 28789
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5258980108711475,
      "learning_rate": 7.251382805493833e-07,
      "loss": 0.0939,
      "step": 28790
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.210705031647771,
      "learning_rate": 7.247675031228385e-07,
      "loss": 0.4569,
      "step": 28791
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8679853766532196,
      "learning_rate": 7.243968169490823e-07,
      "loss": 0.462,
      "step": 28792
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.24953005240582773,
      "learning_rate": 7.240262220317595e-07,
      "loss": 0.1897,
      "step": 28793
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3494670867214093,
      "learning_rate": 7.236557183745219e-07,
      "loss": 0.2183,
      "step": 28794
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7949779563386953,
      "learning_rate": 7.232853059810096e-07,
      "loss": 0.2811,
      "step": 28795
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45802703543372986,
      "learning_rate": 7.2291498485487e-07,
      "loss": 0.2842,
      "step": 28796
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.34868315507087844,
      "learning_rate": 7.225447549997444e-07,
      "loss": 0.1486,
      "step": 28797
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.557214755185995,
      "learning_rate": 7.221746164192778e-07,
      "loss": 0.2987,
      "step": 28798
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4342038697136008,
      "learning_rate": 7.218045691171094e-07,
      "loss": 0.2028,
      "step": 28799
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4847388522511122,
      "learning_rate": 7.214346130968797e-07,
      "loss": 0.2935,
      "step": 28800
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.49882167979378295,
      "learning_rate": 7.210647483622313e-07,
      "loss": 0.2582,
      "step": 28801
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47681505784409334,
      "learning_rate": 7.20694974916798e-07,
      "loss": 0.2704,
      "step": 28802
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4978611899040442,
      "learning_rate": 7.203252927642235e-07,
      "loss": 0.1756,
      "step": 28803
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42562899961692746,
      "learning_rate": 7.199557019081415e-07,
      "loss": 0.2374,
      "step": 28804
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3283287200154243,
      "learning_rate": 7.195862023521882e-07,
      "loss": 0.2248,
      "step": 28805
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.1796433762728559,
      "learning_rate": 7.192167941000016e-07,
      "loss": 0.0667,
      "step": 28806
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3472518835863204,
      "learning_rate": 7.188474771552101e-07,
      "loss": 0.2528,
      "step": 28807
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1679639418776473,
      "learning_rate": 7.18478251521455e-07,
      "loss": 0.0419,
      "step": 28808
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.3862237611180006,
      "learning_rate": 7.181091172023624e-07,
      "loss": 0.8549,
      "step": 28809
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4076680524492538,
      "learning_rate": 7.177400742015672e-07,
      "loss": 0.2172,
      "step": 28810
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.407652787797251,
      "learning_rate": 7.173711225226998e-07,
      "loss": 0.2455,
      "step": 28811
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3287207071204681,
      "learning_rate": 7.170022621693895e-07,
      "loss": 0.209,
      "step": 28812
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6486572611178595,
      "learning_rate": 7.166334931452646e-07,
      "loss": 0.3659,
      "step": 28813
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.25996921600741674,
      "learning_rate": 7.162648154539542e-07,
      "loss": 0.0715,
      "step": 28814
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.42535824782986187,
      "learning_rate": 7.158962290990878e-07,
      "loss": 0.1645,
      "step": 28815
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35677169913177526,
      "learning_rate": 7.155277340842859e-07,
      "loss": 0.1752,
      "step": 28816
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3243521002272337,
      "learning_rate": 7.151593304131788e-07,
      "loss": 0.2167,
      "step": 28817
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.6773339264424654,
      "learning_rate": 7.147910180893902e-07,
      "loss": 0.7654,
      "step": 28818
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8888535452184241,
      "learning_rate": 7.144227971165397e-07,
      "loss": 0.3484,
      "step": 28819
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.46892060014996056,
      "learning_rate": 7.140546674982563e-07,
      "loss": 0.2636,
      "step": 28820
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.47957762782851954,
      "learning_rate": 7.136866292381562e-07,
      "loss": 0.2408,
      "step": 28821
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.1110616973595726,
      "learning_rate": 7.13318682339863e-07,
      "loss": 0.7454,
      "step": 28822
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.38847147214367084,
      "learning_rate": 7.129508268069985e-07,
      "loss": 0.2188,
      "step": 28823
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3393378434579318,
      "learning_rate": 7.125830626431774e-07,
      "loss": 0.1632,
      "step": 28824
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2626411961557736,
      "learning_rate": 7.122153898520212e-07,
      "loss": 0.1617,
      "step": 28825
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2267725712181943,
      "learning_rate": 7.118478084371449e-07,
      "loss": 0.4591,
      "step": 28826
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.5878685711095866,
      "learning_rate": 7.114803184021679e-07,
      "loss": 0.3959,
      "step": 28827
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35260109124863115,
      "learning_rate": 7.111129197507005e-07,
      "loss": 0.2634,
      "step": 28828
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3117701895821374,
      "learning_rate": 7.107456124863643e-07,
      "loss": 0.1426,
      "step": 28829
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.45379461240728064,
      "learning_rate": 7.103783966127676e-07,
      "loss": 0.2379,
      "step": 28830
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8134246795531028,
      "learning_rate": 7.10011272133525e-07,
      "loss": 0.3919,
      "step": 28831
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.16702611547579732,
      "learning_rate": 7.096442390522496e-07,
      "loss": 0.0693,
      "step": 28832
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0133338618066818,
      "learning_rate": 7.092772973725481e-07,
      "loss": 0.4231,
      "step": 28833
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.29896493285854525,
      "learning_rate": 7.089104470980368e-07,
      "loss": 0.1739,
      "step": 28834
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3564668583838795,
      "learning_rate": 7.085436882323204e-07,
      "loss": 0.2807,
      "step": 28835
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2205887701931724,
      "learning_rate": 7.081770207790084e-07,
      "loss": 0.2618,
      "step": 28836
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7421643549542326,
      "learning_rate": 7.0781044474171e-07,
      "loss": 0.3639,
      "step": 28837
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5201518997944085,
      "learning_rate": 7.07443960124028e-07,
      "loss": 0.1572,
      "step": 28838
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.39438424698187874,
      "learning_rate": 7.070775669295693e-07,
      "loss": 0.2811,
      "step": 28839
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4305341857079328,
      "learning_rate": 7.067112651619401e-07,
      "loss": 0.2256,
      "step": 28840
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4991866986282342,
      "learning_rate": 7.063450548247453e-07,
      "loss": 0.3067,
      "step": 28841
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.26533393817957623,
      "learning_rate": 7.05978935921584e-07,
      "loss": 0.0708,
      "step": 28842
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.21576725828396923,
      "learning_rate": 7.056129084560593e-07,
      "loss": 0.1464,
      "step": 28843
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.7344393504556077,
      "learning_rate": 7.052469724317756e-07,
      "loss": 0.6443,
      "step": 28844
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8700896440696195,
      "learning_rate": 7.048811278523271e-07,
      "loss": 0.2802,
      "step": 28845
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3379639655605721,
      "learning_rate": 7.045153747213196e-07,
      "loss": 0.2926,
      "step": 28846
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37154611629301965,
      "learning_rate": 7.04149713042347e-07,
      "loss": 0.1735,
      "step": 28847
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5206092772240726,
      "learning_rate": 7.037841428190084e-07,
      "loss": 0.3251,
      "step": 28848
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1064830386250242,
      "learning_rate": 7.034186640549012e-07,
      "loss": 0.428,
      "step": 28849
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4680523716745446,
      "learning_rate": 7.030532767536191e-07,
      "loss": 0.4079,
      "step": 28850
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.13493299793513724,
      "learning_rate": 7.026879809187581e-07,
      "loss": 0.0684,
      "step": 28851
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.36594584713172995,
      "learning_rate": 7.02322776553912e-07,
      "loss": 0.2432,
      "step": 28852
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30171960015900845,
      "learning_rate": 7.019576636626734e-07,
      "loss": 0.1956,
      "step": 28853
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0541031357154562,
      "learning_rate": 7.015926422486363e-07,
      "loss": 0.3901,
      "step": 28854
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6380945335692243,
      "learning_rate": 7.012277123153888e-07,
      "loss": 0.2611,
      "step": 28855
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.334432172031476,
      "learning_rate": 7.008628738665235e-07,
      "loss": 0.155,
      "step": 28856
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5923668320364487,
      "learning_rate": 7.004981269056277e-07,
      "loss": 0.3447,
      "step": 28857
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4681388234961065,
      "learning_rate": 7.001334714362939e-07,
      "loss": 0.2774,
      "step": 28858
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5896942175160009,
      "learning_rate": 6.997689074621039e-07,
      "loss": 0.2704,
      "step": 28859
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.31117255796228493,
      "learning_rate": 6.994044349866503e-07,
      "loss": 0.0856,
      "step": 28860
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4871568663986731,
      "learning_rate": 6.990400540135144e-07,
      "loss": 0.2665,
      "step": 28861
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2872867043953246,
      "learning_rate": 6.986757645462838e-07,
      "loss": 0.1313,
      "step": 28862
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6255571304864449,
      "learning_rate": 6.983115665885431e-07,
      "loss": 0.3435,
      "step": 28863
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2956463300478484,
      "learning_rate": 6.979474601438719e-07,
      "loss": 0.2058,
      "step": 28864
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.7719416447928823,
      "learning_rate": 6.975834452158536e-07,
      "loss": 0.4218,
      "step": 28865
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3163788831915715,
      "learning_rate": 6.972195218080713e-07,
      "loss": 0.1877,
      "step": 28866
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.4499426950935312,
      "learning_rate": 6.968556899241053e-07,
      "loss": 0.7548,
      "step": 28867
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5833145291492583,
      "learning_rate": 6.964919495675327e-07,
      "loss": 0.0352,
      "step": 28868
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2710789450200399,
      "learning_rate": 6.961283007419339e-07,
      "loss": 0.0686,
      "step": 28869
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3530447819958319,
      "learning_rate": 6.957647434508874e-07,
      "loss": 0.2624,
      "step": 28870
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.2682511446401957,
      "learning_rate": 6.954012776979658e-07,
      "loss": 0.1964,
      "step": 28871
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.020418713846527,
      "learning_rate": 6.950379034867516e-07,
      "loss": 0.4137,
      "step": 28872
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8092230270893767,
      "learning_rate": 6.946746208208144e-07,
      "loss": 0.2664,
      "step": 28873
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6202697720860335,
      "learning_rate": 6.943114297037301e-07,
      "loss": 0.3409,
      "step": 28874
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3521956091400251,
      "learning_rate": 6.939483301390737e-07,
      "loss": 0.2192,
      "step": 28875
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.2408791721863233,
      "learning_rate": 6.935853221304145e-07,
      "loss": 0.5107,
      "step": 28876
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.37021872865232336,
      "learning_rate": 6.932224056813241e-07,
      "loss": 0.2091,
      "step": 28877
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4727565553124487,
      "learning_rate": 6.928595807953742e-07,
      "loss": 0.2403,
      "step": 28878
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.27429522592383904,
      "learning_rate": 6.924968474761352e-07,
      "loss": 0.165,
      "step": 28879
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44382273637416964,
      "learning_rate": 6.921342057271751e-07,
      "loss": 0.2291,
      "step": 28880
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6157608711130563,
      "learning_rate": 6.917716555520604e-07,
      "loss": 0.2765,
      "step": 28881
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.30667619375866556,
      "learning_rate": 6.914091969543591e-07,
      "loss": 0.2163,
      "step": 28882
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.6532801798320773,
      "learning_rate": 6.910468299376372e-07,
      "loss": 0.306,
      "step": 28883
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.35975639529974623,
      "learning_rate": 6.906845545054608e-07,
      "loss": 0.2033,
      "step": 28884
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.1749115239700756,
      "learning_rate": 6.903223706613915e-07,
      "loss": 0.6166,
      "step": 28885
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.8833792222548696,
      "learning_rate": 6.899602784089943e-07,
      "loss": 0.0393,
      "step": 28886
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.44764886634175366,
      "learning_rate": 6.895982777518318e-07,
      "loss": 0.2934,
      "step": 28887
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4074867350496975,
      "learning_rate": 6.892363686934622e-07,
      "loss": 0.1686,
      "step": 28888
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3869984379653894,
      "learning_rate": 6.888745512374529e-07,
      "loss": 0.2803,
      "step": 28889
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.40357598342854917,
      "learning_rate": 6.885128253873574e-07,
      "loss": 0.1859,
      "step": 28890
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.4699018347793941,
      "learning_rate": 6.881511911467365e-07,
      "loss": 0.2617,
      "step": 28891
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.5913003222244101,
      "learning_rate": 6.877896485191505e-07,
      "loss": 0.2123,
      "step": 28892
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3522589602345445,
      "learning_rate": 6.874281975081509e-07,
      "loss": 0.2527,
      "step": 28893
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3299931814515816,
      "learning_rate": 6.870668381172984e-07,
      "loss": 0.237,
      "step": 28894
    },
    {
      "epoch": 0.88,
      "grad_norm": 2.170162599814209,
      "learning_rate": 6.867055703501457e-07,
      "loss": 0.1057,
      "step": 28895
    },
    {
      "epoch": 0.88,
      "grad_norm": 1.0334535607431643,
      "learning_rate": 6.86344394210251e-07,
      "loss": 0.4185,
      "step": 28896
    },
    {
      "epoch": 0.88,
      "grad_norm": 0.3799588221829388,
      "learning_rate": 6.859833097011603e-07,
      "loss": 0.1773,
      "step": 28897
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.481124490276345,
      "learning_rate": 6.856223168264342e-07,
      "loss": 0.2957,
      "step": 28898
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22435250563181755,
      "learning_rate": 6.852614155896187e-07,
      "loss": 0.0691,
      "step": 28899
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.24322743709514508,
      "learning_rate": 6.849006059942664e-07,
      "loss": 0.2026,
      "step": 28900
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3397222009093001,
      "learning_rate": 6.845398880439292e-07,
      "loss": 0.1497,
      "step": 28901
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4119458080615939,
      "learning_rate": 6.841792617421516e-07,
      "loss": 0.2899,
      "step": 28902
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2278802926757837,
      "learning_rate": 6.838187270924834e-07,
      "loss": 0.3119,
      "step": 28903
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4164809657723665,
      "learning_rate": 6.834582840984727e-07,
      "loss": 0.5717,
      "step": 28904
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3507346755759751,
      "learning_rate": 6.830979327636655e-07,
      "loss": 0.2166,
      "step": 28905
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.42030667398520294,
      "learning_rate": 6.827376730916068e-07,
      "loss": 0.2205,
      "step": 28906
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5113506666231686,
      "learning_rate": 6.823775050858394e-07,
      "loss": 0.221,
      "step": 28907
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7920834613761404,
      "learning_rate": 6.820174287499081e-07,
      "loss": 0.3509,
      "step": 28908
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.27437868855173364,
      "learning_rate": 6.816574440873557e-07,
      "loss": 0.1096,
      "step": 28909
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3314035205290087,
      "learning_rate": 6.81297551101725e-07,
      "loss": 0.1468,
      "step": 28910
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5820105575178965,
      "learning_rate": 6.80937749796553e-07,
      "loss": 0.3419,
      "step": 28911
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21227907578142502,
      "learning_rate": 6.805780401753826e-07,
      "loss": 0.1814,
      "step": 28912
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4211867993990694,
      "learning_rate": 6.802184222417529e-07,
      "loss": 0.4904,
      "step": 28913
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6192173529236916,
      "learning_rate": 6.798588959991981e-07,
      "loss": 0.2503,
      "step": 28914
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7328098041079778,
      "learning_rate": 6.794994614512617e-07,
      "loss": 0.3584,
      "step": 28915
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34210952770262215,
      "learning_rate": 6.791401186014745e-07,
      "loss": 0.1902,
      "step": 28916
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7445189453779177,
      "learning_rate": 6.787808674533747e-07,
      "loss": 0.8146,
      "step": 28917
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3174115521329316,
      "learning_rate": 6.78421708010496e-07,
      "loss": 0.2283,
      "step": 28918
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.428247917508287,
      "learning_rate": 6.780626402763713e-07,
      "loss": 0.8036,
      "step": 28919
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2971832757338609,
      "learning_rate": 6.777036642545321e-07,
      "loss": 0.1743,
      "step": 28920
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3564826525451837,
      "learning_rate": 6.773447799485133e-07,
      "loss": 0.1945,
      "step": 28921
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8887044889690959,
      "learning_rate": 6.769859873618445e-07,
      "loss": 0.2309,
      "step": 28922
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4463931134000759,
      "learning_rate": 6.766272864980549e-07,
      "loss": 0.2237,
      "step": 28923
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3821775724689144,
      "learning_rate": 6.762686773606731e-07,
      "loss": 0.2283,
      "step": 28924
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37680426999863703,
      "learning_rate": 6.759101599532291e-07,
      "loss": 0.1887,
      "step": 28925
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4701446595720016,
      "learning_rate": 6.755517342792461e-07,
      "loss": 0.5621,
      "step": 28926
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0747293736080543,
      "learning_rate": 6.751934003422567e-07,
      "loss": 0.4238,
      "step": 28927
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.47052236777898254,
      "learning_rate": 6.748351581457813e-07,
      "loss": 0.2783,
      "step": 28928
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2941971807712203,
      "learning_rate": 6.74477007693346e-07,
      "loss": 0.1815,
      "step": 28929
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5628777510806905,
      "learning_rate": 6.741189489884747e-07,
      "loss": 0.3064,
      "step": 28930
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.31323680600092585,
      "learning_rate": 6.737609820346891e-07,
      "loss": 0.1742,
      "step": 28931
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7949748576983895,
      "learning_rate": 6.734031068355141e-07,
      "loss": 0.344,
      "step": 28932
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.36713502509986845,
      "learning_rate": 6.730453233944667e-07,
      "loss": 0.1459,
      "step": 28933
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5705644520392028,
      "learning_rate": 6.726876317150677e-07,
      "loss": 0.3469,
      "step": 28934
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.038174217442608,
      "learning_rate": 6.723300318008363e-07,
      "loss": 0.0473,
      "step": 28935
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3246063757510962,
      "learning_rate": 6.719725236552943e-07,
      "loss": 0.2525,
      "step": 28936
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9129680019814308,
      "learning_rate": 6.716151072819532e-07,
      "loss": 0.528,
      "step": 28937
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3878553809015126,
      "learning_rate": 6.712577826843325e-07,
      "loss": 0.1644,
      "step": 28938
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3385859895148802,
      "learning_rate": 6.709005498659493e-07,
      "loss": 0.2292,
      "step": 28939
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6943553729194835,
      "learning_rate": 6.70543408830312e-07,
      "loss": 0.2756,
      "step": 28940
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2548823102869607,
      "learning_rate": 6.701863595809433e-07,
      "loss": 0.2215,
      "step": 28941
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.32691787252735965,
      "learning_rate": 6.69829402121348e-07,
      "loss": 0.0585,
      "step": 28942
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3525362926192627,
      "learning_rate": 6.694725364550414e-07,
      "loss": 0.2217,
      "step": 28943
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2151768566127341,
      "learning_rate": 6.69115762585536e-07,
      "loss": 0.0676,
      "step": 28944
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3756934193634902,
      "learning_rate": 6.68759080516338e-07,
      "loss": 0.7699,
      "step": 28945
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0521083101611175,
      "learning_rate": 6.68402490250959e-07,
      "loss": 0.2979,
      "step": 28946
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4040307749860505,
      "learning_rate": 6.680459917929061e-07,
      "loss": 0.248,
      "step": 28947
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2742140081231132,
      "learning_rate": 6.6768958514569e-07,
      "loss": 0.2102,
      "step": 28948
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6507506336922712,
      "learning_rate": 6.673332703128122e-07,
      "loss": 0.2799,
      "step": 28949
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.25914124674779904,
      "learning_rate": 6.66977047297781e-07,
      "loss": 0.1618,
      "step": 28950
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3004191872865495,
      "learning_rate": 6.666209161041015e-07,
      "loss": 0.0668,
      "step": 28951
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39194978294637006,
      "learning_rate": 6.662648767352742e-07,
      "loss": 0.2657,
      "step": 28952
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5948850822500367,
      "learning_rate": 6.659089291948062e-07,
      "loss": 0.1735,
      "step": 28953
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3614358780052171,
      "learning_rate": 6.65553073486197e-07,
      "loss": 0.3024,
      "step": 28954
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0280149917867265,
      "learning_rate": 6.651973096129483e-07,
      "loss": 0.3076,
      "step": 28955
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4064544938597317,
      "learning_rate": 6.648416375785605e-07,
      "loss": 0.2497,
      "step": 28956
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.49961801627511127,
      "learning_rate": 6.644860573865297e-07,
      "loss": 0.2323,
      "step": 28957
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6825579416590538,
      "learning_rate": 6.641305690403599e-07,
      "loss": 0.3803,
      "step": 28958
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2707953046364102,
      "learning_rate": 6.637751725435437e-07,
      "loss": 0.1867,
      "step": 28959
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4387283337694912,
      "learning_rate": 6.634198678995807e-07,
      "loss": 0.1967,
      "step": 28960
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38725424209134646,
      "learning_rate": 6.630646551119635e-07,
      "loss": 0.1656,
      "step": 28961
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5157061478412389,
      "learning_rate": 6.627095341841905e-07,
      "loss": 0.1779,
      "step": 28962
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.414629028176952,
      "learning_rate": 6.623545051197522e-07,
      "loss": 0.7407,
      "step": 28963
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5352516863981887,
      "learning_rate": 6.619995679221425e-07,
      "loss": 0.2191,
      "step": 28964
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0100468789871413,
      "learning_rate": 6.616447225948564e-07,
      "loss": 0.3767,
      "step": 28965
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23131063871373983,
      "learning_rate": 6.612899691413777e-07,
      "loss": 0.2149,
      "step": 28966
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8680555320822865,
      "learning_rate": 6.609353075652047e-07,
      "loss": 0.3849,
      "step": 28967
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1477814433889206,
      "learning_rate": 6.605807378698215e-07,
      "loss": 0.0692,
      "step": 28968
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.8025040897505429,
      "learning_rate": 6.602262600587184e-07,
      "loss": 0.7671,
      "step": 28969
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2865417309726815,
      "learning_rate": 6.598718741353838e-07,
      "loss": 0.1746,
      "step": 28970
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5841179785279313,
      "learning_rate": 6.595175801033016e-07,
      "loss": 0.2657,
      "step": 28971
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.46881696928278827,
      "learning_rate": 6.591633779659579e-07,
      "loss": 0.2737,
      "step": 28972
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0703911915385416,
      "learning_rate": 6.588092677268398e-07,
      "loss": 0.4999,
      "step": 28973
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.545127804989939,
      "learning_rate": 6.584552493894303e-07,
      "loss": 0.1884,
      "step": 28974
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3746037996478891,
      "learning_rate": 6.581013229572109e-07,
      "loss": 0.223,
      "step": 28975
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7609134748445754,
      "learning_rate": 6.577474884336643e-07,
      "loss": 0.7111,
      "step": 28976
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3032437606255563,
      "learning_rate": 6.573937458222745e-07,
      "loss": 0.2041,
      "step": 28977
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3506948115493621,
      "learning_rate": 6.570400951265143e-07,
      "loss": 0.1698,
      "step": 28978
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2796099068311362,
      "learning_rate": 6.56686536349872e-07,
      "loss": 0.1639,
      "step": 28979
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3047691021306282,
      "learning_rate": 6.563330694958203e-07,
      "loss": 0.6664,
      "step": 28980
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0592406727470007,
      "learning_rate": 6.559796945678387e-07,
      "loss": 0.519,
      "step": 28981
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7196085825729873,
      "learning_rate": 6.556264115694045e-07,
      "loss": 0.3333,
      "step": 28982
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.28335382520945385,
      "learning_rate": 6.552732205039891e-07,
      "loss": 0.1806,
      "step": 28983
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.373987441166401,
      "learning_rate": 6.549201213750745e-07,
      "loss": 0.2815,
      "step": 28984
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0886754490286876,
      "learning_rate": 6.545671141861287e-07,
      "loss": 0.2604,
      "step": 28985
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2981405766821645,
      "learning_rate": 6.542141989406259e-07,
      "loss": 0.4568,
      "step": 28986
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.15929709595241973,
      "learning_rate": 6.538613756420409e-07,
      "loss": 0.0715,
      "step": 28987
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.42886639691859546,
      "learning_rate": 6.535086442938421e-07,
      "loss": 0.1722,
      "step": 28988
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.40114819578230193,
      "learning_rate": 6.531560048995001e-07,
      "loss": 0.2693,
      "step": 28989
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4603846056635678,
      "learning_rate": 6.528034574624854e-07,
      "loss": 0.2513,
      "step": 28990
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7456396516823903,
      "learning_rate": 6.524510019862673e-07,
      "loss": 0.3274,
      "step": 28991
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4309223232576421,
      "learning_rate": 6.5209863847431e-07,
      "loss": 0.142,
      "step": 28992
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3612240648621643,
      "learning_rate": 6.517463669300817e-07,
      "loss": 0.2543,
      "step": 28993
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4443489127519946,
      "learning_rate": 6.513941873570517e-07,
      "loss": 0.1641,
      "step": 28994
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.352193299391787,
      "learning_rate": 6.510420997586775e-07,
      "loss": 0.2737,
      "step": 28995
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.18203390522473456,
      "learning_rate": 6.506901041384306e-07,
      "loss": 0.0696,
      "step": 28996
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4317814129798316,
      "learning_rate": 6.503382004997694e-07,
      "loss": 0.2748,
      "step": 28997
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4774679053527536,
      "learning_rate": 6.499863888461577e-07,
      "loss": 0.2067,
      "step": 28998
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7632339682221847,
      "learning_rate": 6.496346691810562e-07,
      "loss": 0.3381,
      "step": 28999
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9014306964664318,
      "learning_rate": 6.492830415079266e-07,
      "loss": 0.2733,
      "step": 29000
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2613844290224157,
      "learning_rate": 6.489315058302259e-07,
      "loss": 0.1891,
      "step": 29001
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4014011637873447,
      "learning_rate": 6.485800621514149e-07,
      "loss": 0.2743,
      "step": 29002
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2490620826598762,
      "learning_rate": 6.482287104749507e-07,
      "loss": 0.3277,
      "step": 29003
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5366404953863118,
      "learning_rate": 6.478774508042873e-07,
      "loss": 0.4688,
      "step": 29004
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.24672237303034522,
      "learning_rate": 6.475262831428852e-07,
      "loss": 0.0731,
      "step": 29005
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4212464049848613,
      "learning_rate": 6.47175207494195e-07,
      "loss": 0.2706,
      "step": 29006
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33207431908116536,
      "learning_rate": 6.46824223861674e-07,
      "loss": 0.2238,
      "step": 29007
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.490253066161318,
      "learning_rate": 6.464733322487749e-07,
      "loss": 0.3009,
      "step": 29008
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8412722528564224,
      "learning_rate": 6.461225326589449e-07,
      "loss": 0.2686,
      "step": 29009
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6097473055091049,
      "learning_rate": 6.457718250956435e-07,
      "loss": 0.3008,
      "step": 29010
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33820214036432034,
      "learning_rate": 6.454212095623146e-07,
      "loss": 0.2142,
      "step": 29011
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3875394303864153,
      "learning_rate": 6.450706860624101e-07,
      "loss": 0.5654,
      "step": 29012
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.29959583017257907,
      "learning_rate": 6.447202545993792e-07,
      "loss": 0.2041,
      "step": 29013
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23035568345751567,
      "learning_rate": 6.44369915176668e-07,
      "loss": 0.0725,
      "step": 29014
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.509019663058434,
      "learning_rate": 6.44019667797724e-07,
      "loss": 0.3004,
      "step": 29015
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3573608189158832,
      "learning_rate": 6.436695124659931e-07,
      "loss": 0.1737,
      "step": 29016
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7595653158638446,
      "learning_rate": 6.433194491849215e-07,
      "loss": 0.3795,
      "step": 29017
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.29123655406651894,
      "learning_rate": 6.429694779579498e-07,
      "loss": 0.2191,
      "step": 29018
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.707660833254806,
      "learning_rate": 6.426195987885242e-07,
      "loss": 0.7863,
      "step": 29019
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2907151664397884,
      "learning_rate": 6.422698116800874e-07,
      "loss": 0.1687,
      "step": 29020
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3073369581999537,
      "learning_rate": 6.419201166360755e-07,
      "loss": 0.5188,
      "step": 29021
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.924878580747918,
      "learning_rate": 6.41570513659937e-07,
      "loss": 0.0924,
      "step": 29022
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.49082090529015937,
      "learning_rate": 6.412210027551047e-07,
      "loss": 0.2602,
      "step": 29023
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5033614197908247,
      "learning_rate": 6.408715839250191e-07,
      "loss": 0.1877,
      "step": 29024
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.30582383316071926,
      "learning_rate": 6.405222571731206e-07,
      "loss": 0.263,
      "step": 29025
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2997574114334025,
      "learning_rate": 6.401730225028424e-07,
      "loss": 0.1867,
      "step": 29026
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0744477183975136,
      "learning_rate": 6.398238799176215e-07,
      "loss": 0.4114,
      "step": 29027
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9340410854116554,
      "learning_rate": 6.394748294208941e-07,
      "loss": 0.5081,
      "step": 29028
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3167164944999531,
      "learning_rate": 6.391258710160941e-07,
      "loss": 0.1687,
      "step": 29029
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4818846668782201,
      "learning_rate": 6.3877700470665e-07,
      "loss": 0.2589,
      "step": 29030
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4373032756523324,
      "learning_rate": 6.384282304960021e-07,
      "loss": 0.2099,
      "step": 29031
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6083456318594265,
      "learning_rate": 6.380795483875757e-07,
      "loss": 0.6983,
      "step": 29032
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3477577740667381,
      "learning_rate": 6.377309583848035e-07,
      "loss": 0.1678,
      "step": 29033
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.46943371947782847,
      "learning_rate": 6.373824604911161e-07,
      "loss": 0.2978,
      "step": 29034
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.38968327313590967,
      "learning_rate": 6.370340547099385e-07,
      "loss": 0.1238,
      "step": 29035
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.549675343791949,
      "learning_rate": 6.366857410447025e-07,
      "loss": 0.3519,
      "step": 29036
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3713406859330288,
      "learning_rate": 6.363375194988319e-07,
      "loss": 0.2175,
      "step": 29037
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3720499431209208,
      "learning_rate": 6.359893900757542e-07,
      "loss": 0.2687,
      "step": 29038
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41643776442594244,
      "learning_rate": 6.356413527788952e-07,
      "loss": 0.0841,
      "step": 29039
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2003424296164817,
      "learning_rate": 6.352934076116768e-07,
      "loss": 0.1252,
      "step": 29040
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.638036135883793,
      "learning_rate": 6.34945554577523e-07,
      "loss": 0.37,
      "step": 29041
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2973509651476002,
      "learning_rate": 6.345977936798565e-07,
      "loss": 0.0687,
      "step": 29042
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3013289043235345,
      "learning_rate": 6.342501249221e-07,
      "loss": 0.2637,
      "step": 29043
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.31981672161165403,
      "learning_rate": 6.339025483076699e-07,
      "loss": 0.1697,
      "step": 29044
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3265728578799882,
      "learning_rate": 6.335550638399901e-07,
      "loss": 0.737,
      "step": 29045
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2346415304971582,
      "learning_rate": 6.332076715224778e-07,
      "loss": 0.0882,
      "step": 29046
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4221092775191488,
      "learning_rate": 6.328603713585479e-07,
      "loss": 0.225,
      "step": 29047
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.48562522322338675,
      "learning_rate": 6.325131633516235e-07,
      "loss": 0.1893,
      "step": 29048
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33815617240827933,
      "learning_rate": 6.32166047505115e-07,
      "loss": 0.2687,
      "step": 29049
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7961492628032808,
      "learning_rate": 6.318190238224386e-07,
      "loss": 0.2422,
      "step": 29050
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6103789990922778,
      "learning_rate": 6.314720923070116e-07,
      "loss": 0.3372,
      "step": 29051
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33956267941138824,
      "learning_rate": 6.311252529622436e-07,
      "loss": 0.1712,
      "step": 29052
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3729090625860227,
      "learning_rate": 6.307785057915472e-07,
      "loss": 0.3006,
      "step": 29053
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.48650129450447926,
      "learning_rate": 6.304318507983354e-07,
      "loss": 0.2988,
      "step": 29054
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.265866236451922,
      "learning_rate": 6.300852879860187e-07,
      "loss": 0.1666,
      "step": 29055
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5356058812659372,
      "learning_rate": 6.297388173580054e-07,
      "loss": 0.2512,
      "step": 29056
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.27878697618521076,
      "learning_rate": 6.293924389177041e-07,
      "loss": 0.1256,
      "step": 29057
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.76495094129883,
      "learning_rate": 6.290461526685243e-07,
      "loss": 0.7611,
      "step": 29058
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6319805898521692,
      "learning_rate": 6.286999586138709e-07,
      "loss": 0.2795,
      "step": 29059
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3783699971153285,
      "learning_rate": 6.283538567571524e-07,
      "loss": 0.2524,
      "step": 29060
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.28410271988185243,
      "learning_rate": 6.280078471017703e-07,
      "loss": 0.2041,
      "step": 29061
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7103312498369094,
      "learning_rate": 6.276619296511333e-07,
      "loss": 0.6285,
      "step": 29062
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0060960111248904,
      "learning_rate": 6.273161044086418e-07,
      "loss": 0.3115,
      "step": 29063
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4611383160643572,
      "learning_rate": 6.269703713776954e-07,
      "loss": 0.2497,
      "step": 29064
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.267505358837944,
      "learning_rate": 6.266247305617024e-07,
      "loss": 0.1681,
      "step": 29065
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.24928329220709222,
      "learning_rate": 6.262791819640579e-07,
      "loss": 0.0681,
      "step": 29066
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.350827437161122,
      "learning_rate": 6.259337255881626e-07,
      "loss": 0.2726,
      "step": 29067
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.634383704392517,
      "learning_rate": 6.255883614374159e-07,
      "loss": 0.2519,
      "step": 29068
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8439119026984316,
      "learning_rate": 6.252430895152173e-07,
      "loss": 0.336,
      "step": 29069
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3195194252671836,
      "learning_rate": 6.248979098249608e-07,
      "loss": 0.1844,
      "step": 29070
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5496058476345413,
      "learning_rate": 6.245528223700425e-07,
      "loss": 0.6509,
      "step": 29071
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2909664316977152,
      "learning_rate": 6.24207827153861e-07,
      "loss": 0.2196,
      "step": 29072
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.6317909949134584,
      "learning_rate": 6.238629241798044e-07,
      "loss": 0.7438,
      "step": 29073
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3190405195002561,
      "learning_rate": 6.235181134512725e-07,
      "loss": 0.147,
      "step": 29074
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.26428569960172127,
      "learning_rate": 6.231733949716523e-07,
      "loss": 0.1809,
      "step": 29075
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6819779996095703,
      "learning_rate": 6.22828768744339e-07,
      "loss": 0.2625,
      "step": 29076
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3875969417192338,
      "learning_rate": 6.224842347727223e-07,
      "loss": 0.5025,
      "step": 29077
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2754875926226827,
      "learning_rate": 6.221397930601902e-07,
      "loss": 0.2173,
      "step": 29078
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35790938939287337,
      "learning_rate": 6.217954436101326e-07,
      "loss": 0.1864,
      "step": 29079
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.69925024181505,
      "learning_rate": 6.214511864259376e-07,
      "loss": 0.7689,
      "step": 29080
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9953118141446127,
      "learning_rate": 6.211070215109927e-07,
      "loss": 0.4262,
      "step": 29081
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.597497905163283,
      "learning_rate": 6.207629488686817e-07,
      "loss": 0.8645,
      "step": 29082
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.20962863318327005,
      "learning_rate": 6.204189685023909e-07,
      "loss": 0.1204,
      "step": 29083
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3780360163413972,
      "learning_rate": 6.200750804155054e-07,
      "loss": 0.2917,
      "step": 29084
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.45379929659060325,
      "learning_rate": 6.197312846114068e-07,
      "loss": 0.2188,
      "step": 29085
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4775850254610951,
      "learning_rate": 6.193875810934791e-07,
      "loss": 0.2579,
      "step": 29086
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8021684030820957,
      "learning_rate": 6.19043969865103e-07,
      "loss": 0.0907,
      "step": 29087
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.33539782685959907,
      "learning_rate": 6.187004509296579e-07,
      "loss": 0.2307,
      "step": 29088
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1265285016700697,
      "learning_rate": 6.183570242905268e-07,
      "loss": 0.0409,
      "step": 29089
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5100287519921416,
      "learning_rate": 6.180136899510836e-07,
      "loss": 0.3459,
      "step": 29090
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44349886844971365,
      "learning_rate": 6.176704479147111e-07,
      "loss": 0.2423,
      "step": 29091
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39473000366643796,
      "learning_rate": 6.173272981847822e-07,
      "loss": 0.2256,
      "step": 29092
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.48905333907882487,
      "learning_rate": 6.169842407646753e-07,
      "loss": 0.2208,
      "step": 29093
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.45423718894212617,
      "learning_rate": 6.166412756577667e-07,
      "loss": 0.1802,
      "step": 29094
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3250749844798035,
      "learning_rate": 6.162984028674268e-07,
      "loss": 0.2328,
      "step": 29095
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.41091421905120806,
      "learning_rate": 6.159556223970309e-07,
      "loss": 0.1878,
      "step": 29096
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4425374592898566,
      "learning_rate": 6.156129342499506e-07,
      "loss": 0.2748,
      "step": 29097
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4848620864365501,
      "learning_rate": 6.1527033842956e-07,
      "loss": 0.191,
      "step": 29098
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3332997160016953,
      "learning_rate": 6.149278349392252e-07,
      "loss": 0.7469,
      "step": 29099
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.802802290979408,
      "learning_rate": 6.145854237823212e-07,
      "loss": 0.2835,
      "step": 29100
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7298379381885682,
      "learning_rate": 6.142431049622133e-07,
      "loss": 0.3137,
      "step": 29101
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2500034656991599,
      "learning_rate": 6.139008784822687e-07,
      "loss": 0.1932,
      "step": 29102
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4637864552066347,
      "learning_rate": 6.135587443458579e-07,
      "loss": 0.2887,
      "step": 29103
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2346055553835851,
      "learning_rate": 6.13216702556344e-07,
      "loss": 0.0721,
      "step": 29104
    },
    {
      "epoch": 0.89,
      "grad_norm": 2.005629106014136,
      "learning_rate": 6.12874753117092e-07,
      "loss": 0.7202,
      "step": 29105
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2800145882303086,
      "learning_rate": 6.125328960314669e-07,
      "loss": 0.1637,
      "step": 29106
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.311263397416216,
      "learning_rate": 6.121911313028317e-07,
      "loss": 0.3073,
      "step": 29107
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5274583347976994,
      "learning_rate": 6.118494589345514e-07,
      "loss": 0.3377,
      "step": 29108
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4243794630428869,
      "learning_rate": 6.115078789299833e-07,
      "loss": 0.2544,
      "step": 29109
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6274814374379145,
      "learning_rate": 6.111663912924903e-07,
      "loss": 0.2898,
      "step": 29110
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3562096500229018,
      "learning_rate": 6.108249960254309e-07,
      "loss": 0.2119,
      "step": 29111
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.9250120238251478,
      "learning_rate": 6.104836931321667e-07,
      "loss": 0.5147,
      "step": 29112
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3053454727915741,
      "learning_rate": 6.101424826160518e-07,
      "loss": 0.1502,
      "step": 29113
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.35034648193193413,
      "learning_rate": 6.098013644804445e-07,
      "loss": 0.2361,
      "step": 29114
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2754630081657088,
      "learning_rate": 6.094603387287035e-07,
      "loss": 0.1646,
      "step": 29115
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4408046539905393,
      "learning_rate": 6.09119405364178e-07,
      "loss": 0.6322,
      "step": 29116
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.049181513471768,
      "learning_rate": 6.087785643902289e-07,
      "loss": 0.4984,
      "step": 29117
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7067352838887018,
      "learning_rate": 6.084378158102056e-07,
      "loss": 0.3513,
      "step": 29118
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3693957173316032,
      "learning_rate": 6.08097159627461e-07,
      "loss": 0.1971,
      "step": 29119
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37705919735756843,
      "learning_rate": 6.07756595845348e-07,
      "loss": 0.209,
      "step": 29120
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4912273283000568,
      "learning_rate": 6.074161244672139e-07,
      "loss": 0.3201,
      "step": 29121
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.25333691759354027,
      "learning_rate": 6.070757454964116e-07,
      "loss": 0.0787,
      "step": 29122
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9649568225011812,
      "learning_rate": 6.067354589362883e-07,
      "loss": 0.3839,
      "step": 29123
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4254149261364012,
      "learning_rate": 6.063952647901939e-07,
      "loss": 0.1657,
      "step": 29124
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5325627770722503,
      "learning_rate": 6.060551630614731e-07,
      "loss": 0.3257,
      "step": 29125
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3453497665342344,
      "learning_rate": 6.057151537534712e-07,
      "loss": 0.2384,
      "step": 29126
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6960271623838818,
      "learning_rate": 6.053752368695376e-07,
      "loss": 0.3522,
      "step": 29127
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4752071596121989,
      "learning_rate": 6.050354124130098e-07,
      "loss": 0.1454,
      "step": 29128
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3686518053111217,
      "learning_rate": 6.046956803872384e-07,
      "loss": 0.2577,
      "step": 29129
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1474846254766555,
      "learning_rate": 6.043560407955606e-07,
      "loss": 0.2714,
      "step": 29130
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5827041404536053,
      "learning_rate": 6.040164936413195e-07,
      "loss": 0.222,
      "step": 29131
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23262223731440956,
      "learning_rate": 6.036770389278556e-07,
      "loss": 0.1708,
      "step": 29132
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3556507910883254,
      "learning_rate": 6.033376766585098e-07,
      "loss": 0.1395,
      "step": 29133
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6532567832645644,
      "learning_rate": 6.029984068366212e-07,
      "loss": 0.3182,
      "step": 29134
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3751147754054855,
      "learning_rate": 6.026592294655242e-07,
      "loss": 0.2732,
      "step": 29135
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6844895900334724,
      "learning_rate": 6.023201445485571e-07,
      "loss": 0.3384,
      "step": 29136
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.32548199553297524,
      "learning_rate": 6.019811520890584e-07,
      "loss": 0.183,
      "step": 29137
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5581108201915684,
      "learning_rate": 6.01642252090362e-07,
      "loss": 0.254,
      "step": 29138
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4252260090410898,
      "learning_rate": 6.013034445557997e-07,
      "loss": 0.1796,
      "step": 29139
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.254007583258874,
      "learning_rate": 6.009647294887077e-07,
      "loss": 0.4585,
      "step": 29140
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.21721616833075186,
      "learning_rate": 6.00626106892418e-07,
      "loss": 0.0721,
      "step": 29141
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3787661143154337,
      "learning_rate": 6.002875767702588e-07,
      "loss": 0.2079,
      "step": 29142
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4735798317334827,
      "learning_rate": 5.999491391255674e-07,
      "loss": 0.222,
      "step": 29143
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3386375232604796,
      "learning_rate": 5.996107939616669e-07,
      "loss": 0.2701,
      "step": 29144
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7047219935955864,
      "learning_rate": 5.992725412818879e-07,
      "loss": 0.2652,
      "step": 29145
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.34219164578317346,
      "learning_rate": 5.989343810895609e-07,
      "loss": 0.1522,
      "step": 29146
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6184094628587885,
      "learning_rate": 5.985963133880102e-07,
      "loss": 0.3148,
      "step": 29147
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6241428910243845,
      "learning_rate": 5.982583381805618e-07,
      "loss": 0.1828,
      "step": 29148
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4963224429182276,
      "learning_rate": 5.979204554705409e-07,
      "loss": 0.2686,
      "step": 29149
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22463566092619136,
      "learning_rate": 5.975826652612737e-07,
      "loss": 0.1386,
      "step": 29150
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9494072631967748,
      "learning_rate": 5.97244967556081e-07,
      "loss": 0.4255,
      "step": 29151
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3232118894653676,
      "learning_rate": 5.969073623582867e-07,
      "loss": 0.1909,
      "step": 29152
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.7324756900800811,
      "learning_rate": 5.965698496712124e-07,
      "loss": 0.3327,
      "step": 29153
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.47835663348065416,
      "learning_rate": 5.962324294981759e-07,
      "loss": 0.2908,
      "step": 29154
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.49634524316163336,
      "learning_rate": 5.958951018425018e-07,
      "loss": 0.3456,
      "step": 29155
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2878590374910748,
      "learning_rate": 5.955578667075046e-07,
      "loss": 0.1663,
      "step": 29156
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5999304716299192,
      "learning_rate": 5.952207240965036e-07,
      "loss": 0.4852,
      "step": 29157
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3593629614960643,
      "learning_rate": 5.948836740128161e-07,
      "loss": 0.1235,
      "step": 29158
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.16816947420532738,
      "learning_rate": 5.945467164597563e-07,
      "loss": 0.0694,
      "step": 29159
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6889630022876699,
      "learning_rate": 5.942098514406424e-07,
      "loss": 0.3044,
      "step": 29160
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.29212939763000767,
      "learning_rate": 5.938730789587866e-07,
      "loss": 0.2138,
      "step": 29161
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.49055905428810387,
      "learning_rate": 5.935363990175014e-07,
      "loss": 0.2985,
      "step": 29162
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8933847312707294,
      "learning_rate": 5.931998116201021e-07,
      "loss": 0.2827,
      "step": 29163
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.566519395859832,
      "learning_rate": 5.928633167698971e-07,
      "loss": 0.7707,
      "step": 29164
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2913713805678658,
      "learning_rate": 5.925269144701973e-07,
      "loss": 0.1622,
      "step": 29165
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.7758614957946262,
      "learning_rate": 5.921906047243142e-07,
      "loss": 0.8443,
      "step": 29166
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.437371605918563,
      "learning_rate": 5.918543875355565e-07,
      "loss": 0.1794,
      "step": 29167
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.29901237324118174,
      "learning_rate": 5.91518262907228e-07,
      "loss": 0.1985,
      "step": 29168
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44476534628106373,
      "learning_rate": 5.911822308426418e-07,
      "loss": 0.1794,
      "step": 29169
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.605514319099169,
      "learning_rate": 5.908462913450997e-07,
      "loss": 0.3105,
      "step": 29170
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4572311600627502,
      "learning_rate": 5.905104444179067e-07,
      "loss": 0.1867,
      "step": 29171
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.0829155300055002,
      "learning_rate": 5.901746900643701e-07,
      "loss": 0.555,
      "step": 29172
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3336272475156206,
      "learning_rate": 5.898390282877897e-07,
      "loss": 0.2539,
      "step": 29173
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.39937838564458855,
      "learning_rate": 5.895034590914695e-07,
      "loss": 0.1704,
      "step": 29174
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.530644886446884,
      "learning_rate": 5.891679824787111e-07,
      "loss": 0.258,
      "step": 29175
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.372281374695753,
      "learning_rate": 5.888325984528154e-07,
      "loss": 0.0499,
      "step": 29176
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6772920533558124,
      "learning_rate": 5.884973070170796e-07,
      "loss": 0.3347,
      "step": 29177
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6383282134883597,
      "learning_rate": 5.881621081748045e-07,
      "loss": 0.1653,
      "step": 29178
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3621009828306357,
      "learning_rate": 5.878270019292898e-07,
      "loss": 0.2453,
      "step": 29179
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.24138318038234463,
      "learning_rate": 5.87491988283826e-07,
      "loss": 0.1737,
      "step": 29180
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3403529483999124,
      "learning_rate": 5.87157067241716e-07,
      "loss": 0.7396,
      "step": 29181
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22219296490530555,
      "learning_rate": 5.868222388062506e-07,
      "loss": 0.0873,
      "step": 29182
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37212883598948227,
      "learning_rate": 5.86487502980726e-07,
      "loss": 0.2041,
      "step": 29183
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5361403955320401,
      "learning_rate": 5.861528597684363e-07,
      "loss": 0.1783,
      "step": 29184
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.331782122992706,
      "learning_rate": 5.858183091726688e-07,
      "loss": 0.2106,
      "step": 29185
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9890661193180122,
      "learning_rate": 5.854838511967209e-07,
      "loss": 0.3981,
      "step": 29186
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5289883780267979,
      "learning_rate": 5.8514948584388e-07,
      "loss": 0.1817,
      "step": 29187
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4017166361895788,
      "learning_rate": 5.848152131174367e-07,
      "loss": 0.2619,
      "step": 29188
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.23007713107443556,
      "learning_rate": 5.844810330206807e-07,
      "loss": 0.0951,
      "step": 29189
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5765834355172093,
      "learning_rate": 5.84146945556896e-07,
      "loss": 0.7441,
      "step": 29190
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.22063751020947509,
      "learning_rate": 5.838129507293721e-07,
      "loss": 0.1692,
      "step": 29191
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44200105089305824,
      "learning_rate": 5.834790485413954e-07,
      "loss": 0.2643,
      "step": 29192
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5637190022015006,
      "learning_rate": 5.831452389962522e-07,
      "loss": 0.1727,
      "step": 29193
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.4050182512045326,
      "learning_rate": 5.828115220972219e-07,
      "loss": 0.5059,
      "step": 29194
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6402976083359846,
      "learning_rate": 5.824778978475909e-07,
      "loss": 0.2604,
      "step": 29195
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.43991284387936364,
      "learning_rate": 5.821443662506432e-07,
      "loss": 0.272,
      "step": 29196
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.44220375222110886,
      "learning_rate": 5.81810927309654e-07,
      "loss": 0.1808,
      "step": 29197
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1933491120089683,
      "learning_rate": 5.814775810279116e-07,
      "loss": 0.1985,
      "step": 29198
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3704473697071577,
      "learning_rate": 5.811443274086903e-07,
      "loss": 0.7157,
      "step": 29199
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.1785841319411916,
      "learning_rate": 5.808111664552706e-07,
      "loss": 0.0787,
      "step": 29200
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8942777410114968,
      "learning_rate": 5.804780981709301e-07,
      "loss": 0.4244,
      "step": 29201
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3617127263310758,
      "learning_rate": 5.801451225589461e-07,
      "loss": 0.1922,
      "step": 29202
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.37005003607648895,
      "learning_rate": 5.798122396225935e-07,
      "loss": 0.2767,
      "step": 29203
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6521888481104725,
      "learning_rate": 5.794794493651468e-07,
      "loss": 0.2567,
      "step": 29204
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.5257559010185715,
      "learning_rate": 5.79146751789883e-07,
      "loss": 0.5234,
      "step": 29205
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2915074450044713,
      "learning_rate": 5.788141469000708e-07,
      "loss": 0.1661,
      "step": 29206
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.3408204683256744,
      "learning_rate": 5.784816346989863e-07,
      "loss": 0.5799,
      "step": 29207
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4422878680879871,
      "learning_rate": 5.781492151898993e-07,
      "loss": 0.2815,
      "step": 29208
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5307741676013773,
      "learning_rate": 5.778168883760804e-07,
      "loss": 0.3308,
      "step": 29209
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3540277795242509,
      "learning_rate": 5.774846542608014e-07,
      "loss": 0.0919,
      "step": 29210
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.2572213698984168,
      "learning_rate": 5.771525128473254e-07,
      "loss": 0.1644,
      "step": 29211
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.6925987424411105,
      "learning_rate": 5.768204641389264e-07,
      "loss": 0.3523,
      "step": 29212
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.1373791721269815,
      "learning_rate": 5.76488508138866e-07,
      "loss": 0.2601,
      "step": 29213
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9432800734971144,
      "learning_rate": 5.76156644850413e-07,
      "loss": 0.3988,
      "step": 29214
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.271511669413666,
      "learning_rate": 5.758248742768335e-07,
      "loss": 0.201,
      "step": 29215
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5049420186439194,
      "learning_rate": 5.754931964213883e-07,
      "loss": 0.3165,
      "step": 29216
    },
    {
      "epoch": 0.89,
      "grad_norm": 1.2979982406115333,
      "learning_rate": 5.751616112873426e-07,
      "loss": 0.4505,
      "step": 29217
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.47193387755102956,
      "learning_rate": 5.74830118877957e-07,
      "loss": 0.2233,
      "step": 29218
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.26854145262909773,
      "learning_rate": 5.744987191964968e-07,
      "loss": 0.1231,
      "step": 29219
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.5724228314378165,
      "learning_rate": 5.741674122462171e-07,
      "loss": 0.3228,
      "step": 29220
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.3539627967464617,
      "learning_rate": 5.738361980303797e-07,
      "loss": 0.2251,
      "step": 29221
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.8332162062160943,
      "learning_rate": 5.735050765522455e-07,
      "loss": 0.3264,
      "step": 29222
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.9080975824547045,
      "learning_rate": 5.731740478150671e-07,
      "loss": 0.0771,
      "step": 29223
    },
    {
      "epoch": 0.89,
      "grad_norm": 0.4100447401330723,
      "learning_rate": 5.728431118221067e-07,
      "loss": 0.1765,
      "step": 29224
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.59435949000613,
      "learning_rate": 5.725122685766171e-07,
      "loss": 0.3389,
      "step": 29225
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4206089061724056,
      "learning_rate": 5.721815180818535e-07,
      "loss": 0.246,
      "step": 29226
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.519887081604077,
      "learning_rate": 5.718508603410711e-07,
      "loss": 0.3209,
      "step": 29227
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7115422839416286,
      "learning_rate": 5.715202953575216e-07,
      "loss": 0.1751,
      "step": 29228
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3357952226123632,
      "learning_rate": 5.71189823134457e-07,
      "loss": 0.2352,
      "step": 29229
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.47662939580992975,
      "learning_rate": 5.708594436751302e-07,
      "loss": 0.187,
      "step": 29230
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4563625191179621,
      "learning_rate": 5.70529156982792e-07,
      "loss": 0.2008,
      "step": 29231
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.385405755147455,
      "learning_rate": 5.701989630606875e-07,
      "loss": 0.1806,
      "step": 29232
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41622569270513615,
      "learning_rate": 5.698688619120718e-07,
      "loss": 0.2316,
      "step": 29233
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.30125680718997006,
      "learning_rate": 5.695388535401891e-07,
      "loss": 0.2065,
      "step": 29234
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6814239490297203,
      "learning_rate": 5.692089379482835e-07,
      "loss": 0.7892,
      "step": 29235
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7362054355090578,
      "learning_rate": 5.688791151396056e-07,
      "loss": 0.0296,
      "step": 29236
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5834309214282205,
      "learning_rate": 5.685493851173951e-07,
      "loss": 0.1861,
      "step": 29237
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3992851211743438,
      "learning_rate": 5.682197478849027e-07,
      "loss": 0.2719,
      "step": 29238
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23847810874913117,
      "learning_rate": 5.678902034453671e-07,
      "loss": 0.1817,
      "step": 29239
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40174490978674265,
      "learning_rate": 5.675607518020299e-07,
      "loss": 0.1882,
      "step": 29240
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1754692255641075,
      "learning_rate": 5.672313929581364e-07,
      "loss": 0.1013,
      "step": 29241
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3985805297878164,
      "learning_rate": 5.669021269169218e-07,
      "loss": 0.2423,
      "step": 29242
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.44980391696264393,
      "learning_rate": 5.665729536816278e-07,
      "loss": 0.2455,
      "step": 29243
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7881931878246053,
      "learning_rate": 5.662438732554942e-07,
      "loss": 0.8148,
      "step": 29244
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3258310940111015,
      "learning_rate": 5.659148856417585e-07,
      "loss": 0.2202,
      "step": 29245
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5540110749309511,
      "learning_rate": 5.655859908436545e-07,
      "loss": 0.2866,
      "step": 29246
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40336277190061387,
      "learning_rate": 5.652571888644198e-07,
      "loss": 0.2127,
      "step": 29247
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0087379667997727,
      "learning_rate": 5.649284797072918e-07,
      "loss": 0.4861,
      "step": 29248
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.15296254411180385,
      "learning_rate": 5.645998633754979e-07,
      "loss": 0.0675,
      "step": 29249
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.30557447535597104,
      "learning_rate": 5.642713398722788e-07,
      "loss": 0.2065,
      "step": 29250
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4633940225506395,
      "learning_rate": 5.639429092008609e-07,
      "loss": 0.2509,
      "step": 29251
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.476209661387192,
      "learning_rate": 5.636145713644781e-07,
      "loss": 0.1994,
      "step": 29252
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3364643911280794,
      "learning_rate": 5.632863263663613e-07,
      "loss": 0.6923,
      "step": 29253
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5974124511937072,
      "learning_rate": 5.629581742097379e-07,
      "loss": 0.2596,
      "step": 29254
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8485515973504377,
      "learning_rate": 5.626301148978364e-07,
      "loss": 0.2949,
      "step": 29255
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.30107090744012316,
      "learning_rate": 5.623021484338853e-07,
      "loss": 0.1771,
      "step": 29256
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3709798280002447,
      "learning_rate": 5.619742748211133e-07,
      "loss": 0.2497,
      "step": 29257
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2770286769327483,
      "learning_rate": 5.61646494062742e-07,
      "loss": 0.0685,
      "step": 29258
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8512738260124046,
      "learning_rate": 5.613188061619979e-07,
      "loss": 0.8071,
      "step": 29259
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3356051439163449,
      "learning_rate": 5.60991211122105e-07,
      "loss": 0.1099,
      "step": 29260
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.420004195199273,
      "learning_rate": 5.606637089462863e-07,
      "loss": 0.2808,
      "step": 29261
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.29189522171612964,
      "learning_rate": 5.60336299637766e-07,
      "loss": 0.2259,
      "step": 29262
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8174004721022773,
      "learning_rate": 5.600089831997613e-07,
      "loss": 0.2439,
      "step": 29263
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6296162182208036,
      "learning_rate": 5.596817596354942e-07,
      "loss": 0.3625,
      "step": 29264
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2894440329989321,
      "learning_rate": 5.593546289481866e-07,
      "loss": 0.1707,
      "step": 29265
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3368544904080752,
      "learning_rate": 5.590275911410504e-07,
      "loss": 0.5075,
      "step": 29266
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.15778170561045915,
      "learning_rate": 5.587006462173117e-07,
      "loss": 0.0664,
      "step": 29267
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34881667803640437,
      "learning_rate": 5.583737941801804e-07,
      "loss": 0.2652,
      "step": 29268
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38146996822274,
      "learning_rate": 5.580470350328737e-07,
      "loss": 0.1579,
      "step": 29269
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5593297029317762,
      "learning_rate": 5.577203687786081e-07,
      "loss": 0.3867,
      "step": 29270
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1021024158099682,
      "learning_rate": 5.573937954205977e-07,
      "loss": 0.2891,
      "step": 29271
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6382848824266435,
      "learning_rate": 5.57067314962052e-07,
      "loss": 0.3532,
      "step": 29272
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4074769582178985,
      "learning_rate": 5.567409274061852e-07,
      "loss": 0.1349,
      "step": 29273
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3540541989104841,
      "learning_rate": 5.564146327562114e-07,
      "loss": 0.2566,
      "step": 29274
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.31560615779799556,
      "learning_rate": 5.560884310153337e-07,
      "loss": 0.2214,
      "step": 29275
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.647037994802941,
      "learning_rate": 5.557623221867681e-07,
      "loss": 0.3711,
      "step": 29276
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.14774845066230072,
      "learning_rate": 5.554363062737201e-07,
      "loss": 0.0905,
      "step": 29277
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33262651021413864,
      "learning_rate": 5.551103832793958e-07,
      "loss": 0.1475,
      "step": 29278
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5864063044052247,
      "learning_rate": 5.547845532070062e-07,
      "loss": 0.3576,
      "step": 29279
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3487755190567289,
      "learning_rate": 5.54458816059752e-07,
      "loss": 0.229,
      "step": 29280
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8135589238113008,
      "learning_rate": 5.541331718408394e-07,
      "loss": 0.4036,
      "step": 29281
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9146627790632389,
      "learning_rate": 5.538076205534737e-07,
      "loss": 0.3026,
      "step": 29282
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.41101163833556414,
      "learning_rate": 5.534821622008579e-07,
      "loss": 0.2598,
      "step": 29283
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4934042588762592,
      "learning_rate": 5.531567967861918e-07,
      "loss": 0.1888,
      "step": 29284
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2863804865741819,
      "learning_rate": 5.52831524312677e-07,
      "loss": 0.1512,
      "step": 29285
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33072460520886676,
      "learning_rate": 5.525063447835144e-07,
      "loss": 0.2241,
      "step": 29286
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8775456231736422,
      "learning_rate": 5.521812582019026e-07,
      "loss": 0.3355,
      "step": 29287
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34432592556998814,
      "learning_rate": 5.518562645710423e-07,
      "loss": 0.198,
      "step": 29288
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0153815942575326,
      "learning_rate": 5.515313638941278e-07,
      "loss": 0.3535,
      "step": 29289
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8831429227284264,
      "learning_rate": 5.512065561743563e-07,
      "loss": 0.2702,
      "step": 29290
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.47616440594905024,
      "learning_rate": 5.508818414149253e-07,
      "loss": 0.2132,
      "step": 29291
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4560120809012299,
      "learning_rate": 5.505572196190245e-07,
      "loss": 0.2837,
      "step": 29292
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.31761183057765174,
      "learning_rate": 5.502326907898548e-07,
      "loss": 0.2006,
      "step": 29293
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8980683751645617,
      "learning_rate": 5.499082549306034e-07,
      "loss": 0.6998,
      "step": 29294
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.20271517530714697,
      "learning_rate": 5.495839120444635e-07,
      "loss": 0.0848,
      "step": 29295
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8946321444280545,
      "learning_rate": 5.492596621346291e-07,
      "loss": 0.2588,
      "step": 29296
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3387798522800397,
      "learning_rate": 5.489355052042855e-07,
      "loss": 0.2063,
      "step": 29297
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0685720449608995,
      "learning_rate": 5.486114412566245e-07,
      "loss": 0.4069,
      "step": 29298
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3251597362837109,
      "learning_rate": 5.482874702948349e-07,
      "loss": 0.2447,
      "step": 29299
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4676446414106805,
      "learning_rate": 5.479635923221038e-07,
      "loss": 0.529,
      "step": 29300
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.29188552115420724,
      "learning_rate": 5.476398073416145e-07,
      "loss": 0.1722,
      "step": 29301
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0552640964006943,
      "learning_rate": 5.473161153565576e-07,
      "loss": 0.7532,
      "step": 29302
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1273122640321245,
      "learning_rate": 5.46992516370114e-07,
      "loss": 0.0709,
      "step": 29303
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2632646609965442,
      "learning_rate": 5.466690103854677e-07,
      "loss": 0.197,
      "step": 29304
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6764956524007476,
      "learning_rate": 5.46345597405804e-07,
      "loss": 0.265,
      "step": 29305
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34912893896108776,
      "learning_rate": 5.460222774343016e-07,
      "loss": 0.1976,
      "step": 29306
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8209757362981058,
      "learning_rate": 5.456990504741433e-07,
      "loss": 0.3842,
      "step": 29307
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2638215970467182,
      "learning_rate": 5.453759165285088e-07,
      "loss": 0.1494,
      "step": 29308
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5065978205243166,
      "learning_rate": 5.450528756005779e-07,
      "loss": 0.2947,
      "step": 29309
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.32850828111683,
      "learning_rate": 5.447299276935259e-07,
      "loss": 0.1468,
      "step": 29310
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.36276763458008215,
      "learning_rate": 5.444070728105333e-07,
      "loss": 0.2585,
      "step": 29311
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.1739610528508857,
      "learning_rate": 5.440843109547744e-07,
      "loss": 0.1681,
      "step": 29312
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4833104530878796,
      "learning_rate": 5.437616421294256e-07,
      "loss": 0.2191,
      "step": 29313
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6230482714334524,
      "learning_rate": 5.434390663376621e-07,
      "loss": 0.2054,
      "step": 29314
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3609914046113867,
      "learning_rate": 5.431165835826557e-07,
      "loss": 0.2496,
      "step": 29315
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3468552092915789,
      "learning_rate": 5.427941938675807e-07,
      "loss": 0.2337,
      "step": 29316
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2240364566747912,
      "learning_rate": 5.424718971956089e-07,
      "loss": 0.4221,
      "step": 29317
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.46683464644413175,
      "learning_rate": 5.421496935699089e-07,
      "loss": 0.2552,
      "step": 29318
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3353029245701547,
      "learning_rate": 5.418275829936537e-07,
      "loss": 0.1459,
      "step": 29319
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6493419781168815,
      "learning_rate": 5.415055654700097e-07,
      "loss": 0.3583,
      "step": 29320
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3574083270841846,
      "learning_rate": 5.411836410021464e-07,
      "loss": 0.1229,
      "step": 29321
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3566398037304142,
      "learning_rate": 5.408618095932328e-07,
      "loss": 0.2739,
      "step": 29322
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5857716122311086,
      "learning_rate": 5.405400712464315e-07,
      "loss": 0.1864,
      "step": 29323
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4194812853496826,
      "learning_rate": 5.402184259649091e-07,
      "loss": 0.2718,
      "step": 29324
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4761430305577155,
      "learning_rate": 5.398968737518307e-07,
      "loss": 0.1714,
      "step": 29325
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2935803032621063,
      "learning_rate": 5.395754146103615e-07,
      "loss": 0.722,
      "step": 29326
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21584283571003132,
      "learning_rate": 5.392540485436604e-07,
      "loss": 0.1628,
      "step": 29327
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39617865450135864,
      "learning_rate": 5.389327755548901e-07,
      "loss": 0.2479,
      "step": 29328
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4727552525220005,
      "learning_rate": 5.386115956472148e-07,
      "loss": 0.199,
      "step": 29329
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.040479257623516,
      "learning_rate": 5.382905088237889e-07,
      "loss": 0.2528,
      "step": 29330
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7481295298355748,
      "learning_rate": 5.379695150877773e-07,
      "loss": 0.355,
      "step": 29331
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3685628270904163,
      "learning_rate": 5.376486144423332e-07,
      "loss": 0.0582,
      "step": 29332
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4865710405571946,
      "learning_rate": 5.373278068906151e-07,
      "loss": 0.2902,
      "step": 29333
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2666956557606304,
      "learning_rate": 5.370070924357807e-07,
      "loss": 0.2169,
      "step": 29334
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4249463160342626,
      "learning_rate": 5.36686471080985e-07,
      "loss": 0.7024,
      "step": 29335
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.25990024937879574,
      "learning_rate": 5.363659428293811e-07,
      "loss": 0.0715,
      "step": 29336
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9764959054401212,
      "learning_rate": 5.36045507684122e-07,
      "loss": 0.3573,
      "step": 29337
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.32142342278051456,
      "learning_rate": 5.357251656483619e-07,
      "loss": 0.184,
      "step": 29338
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8539996059308089,
      "learning_rate": 5.354049167252506e-07,
      "loss": 0.3337,
      "step": 29339
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33145254929749646,
      "learning_rate": 5.350847609179432e-07,
      "loss": 0.2266,
      "step": 29340
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2727970776882793,
      "learning_rate": 5.347646982295839e-07,
      "loss": 0.502,
      "step": 29341
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.28438241543310094,
      "learning_rate": 5.344447286633236e-07,
      "loss": 0.1719,
      "step": 29342
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1441166128495253,
      "learning_rate": 5.341248522223119e-07,
      "loss": 0.4214,
      "step": 29343
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5781217709672688,
      "learning_rate": 5.33805068909693e-07,
      "loss": 0.7513,
      "step": 29344
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2531047893490258,
      "learning_rate": 5.334853787286165e-07,
      "loss": 0.1746,
      "step": 29345
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3474401570426795,
      "learning_rate": 5.331657816822245e-07,
      "loss": 0.1931,
      "step": 29346
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3700335674944233,
      "learning_rate": 5.328462777736631e-07,
      "loss": 0.1913,
      "step": 29347
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0038226749955914,
      "learning_rate": 5.325268670060757e-07,
      "loss": 0.4403,
      "step": 29348
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6778677094025851,
      "learning_rate": 5.322075493826029e-07,
      "loss": 0.2387,
      "step": 29349
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2557759585652617,
      "learning_rate": 5.318883249063878e-07,
      "loss": 0.4207,
      "step": 29350
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26054446134357095,
      "learning_rate": 5.315691935805712e-07,
      "loss": 0.162,
      "step": 29351
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5106343920202018,
      "learning_rate": 5.312501554082938e-07,
      "loss": 0.3334,
      "step": 29352
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4601420425023292,
      "learning_rate": 5.309312103926911e-07,
      "loss": 0.2746,
      "step": 29353
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5382889124222824,
      "learning_rate": 5.306123585369027e-07,
      "loss": 0.223,
      "step": 29354
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23839760293724616,
      "learning_rate": 5.302935998440673e-07,
      "loss": 0.1281,
      "step": 29355
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.591272463472248,
      "learning_rate": 5.299749343173166e-07,
      "loss": 0.2142,
      "step": 29356
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6603373974167687,
      "learning_rate": 5.296563619597916e-07,
      "loss": 0.3644,
      "step": 29357
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3212616948170202,
      "learning_rate": 5.293378827746221e-07,
      "loss": 0.2089,
      "step": 29358
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8992337169590958,
      "learning_rate": 5.290194967649431e-07,
      "loss": 0.419,
      "step": 29359
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3895657556724991,
      "learning_rate": 5.287012039338868e-07,
      "loss": 0.1682,
      "step": 29360
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6130621705294731,
      "learning_rate": 5.283830042845839e-07,
      "loss": 0.3496,
      "step": 29361
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8249171174492379,
      "learning_rate": 5.280648978201663e-07,
      "loss": 0.3137,
      "step": 29362
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2748528503371481,
      "learning_rate": 5.277468845437617e-07,
      "loss": 0.2324,
      "step": 29363
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1766871254409586,
      "learning_rate": 5.274289644585007e-07,
      "loss": 0.0667,
      "step": 29364
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5878145315387067,
      "learning_rate": 5.271111375675119e-07,
      "loss": 0.2559,
      "step": 29365
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5821165550016986,
      "learning_rate": 5.267934038739186e-07,
      "loss": 0.2558,
      "step": 29366
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.006307434319344,
      "learning_rate": 5.264757633808492e-07,
      "loss": 0.7072,
      "step": 29367
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5305043773632723,
      "learning_rate": 5.261582160914291e-07,
      "loss": 0.1247,
      "step": 29368
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2719819185948372,
      "learning_rate": 5.258407620087824e-07,
      "loss": 0.1823,
      "step": 29369
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.3133996871827,
      "learning_rate": 5.255234011360288e-07,
      "loss": 0.2686,
      "step": 29370
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1343988785039545,
      "learning_rate": 5.25206133476297e-07,
      "loss": 0.423,
      "step": 29371
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3550942075967398,
      "learning_rate": 5.248889590327033e-07,
      "loss": 0.4112,
      "step": 29372
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5108516693256765,
      "learning_rate": 5.245718778083697e-07,
      "loss": 0.1733,
      "step": 29373
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4039788275873487,
      "learning_rate": 5.24254889806417e-07,
      "loss": 0.2685,
      "step": 29374
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3677391731702256,
      "learning_rate": 5.239379950299617e-07,
      "loss": 0.1263,
      "step": 29375
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.27525604155284467,
      "learning_rate": 5.236211934821222e-07,
      "loss": 0.2223,
      "step": 29376
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5800092430732176,
      "learning_rate": 5.233044851660152e-07,
      "loss": 0.0356,
      "step": 29377
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4199207053683903,
      "learning_rate": 5.22987870084758e-07,
      "loss": 0.2734,
      "step": 29378
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4901379040104126,
      "learning_rate": 5.226713482414636e-07,
      "loss": 0.2132,
      "step": 29379
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.5770258356270819,
      "learning_rate": 5.223549196392474e-07,
      "loss": 0.7227,
      "step": 29380
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3725104066524981,
      "learning_rate": 5.220385842812226e-07,
      "loss": 0.22,
      "step": 29381
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.645573246116967,
      "learning_rate": 5.217223421704987e-07,
      "loss": 0.1799,
      "step": 29382
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.545278527925588,
      "learning_rate": 5.214061933101922e-07,
      "loss": 0.3168,
      "step": 29383
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3309507283657288,
      "learning_rate": 5.210901377034083e-07,
      "loss": 0.147,
      "step": 29384
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5275516602051213,
      "learning_rate": 5.207741753532591e-07,
      "loss": 0.3034,
      "step": 29385
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39077938172223337,
      "learning_rate": 5.204583062628532e-07,
      "loss": 0.1587,
      "step": 29386
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8832350713154007,
      "learning_rate": 5.201425304352959e-07,
      "loss": 0.3875,
      "step": 29387
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2694155874558147,
      "learning_rate": 5.198268478736978e-07,
      "loss": 0.2031,
      "step": 29388
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.558627604747759,
      "learning_rate": 5.195112585811612e-07,
      "loss": 0.771,
      "step": 29389
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6567306323915084,
      "learning_rate": 5.191957625607924e-07,
      "loss": 0.2508,
      "step": 29390
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0896503787757055,
      "learning_rate": 5.188803598156977e-07,
      "loss": 0.4423,
      "step": 29391
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2792507422689899,
      "learning_rate": 5.185650503489748e-07,
      "loss": 0.1702,
      "step": 29392
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5153835322337228,
      "learning_rate": 5.182498341637299e-07,
      "loss": 0.3074,
      "step": 29393
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.19667515373209685,
      "learning_rate": 5.179347112630617e-07,
      "loss": 0.1234,
      "step": 29394
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2293014808467106,
      "learning_rate": 5.176196816500744e-07,
      "loss": 0.0779,
      "step": 29395
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6592388488840587,
      "learning_rate": 5.173047453278634e-07,
      "loss": 0.298,
      "step": 29396
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34605229905726964,
      "learning_rate": 5.169899022995284e-07,
      "loss": 0.2181,
      "step": 29397
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9393891302595837,
      "learning_rate": 5.166751525681679e-07,
      "loss": 0.4079,
      "step": 29398
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33326754201058467,
      "learning_rate": 5.163604961368751e-07,
      "loss": 0.2255,
      "step": 29399
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7706848153480172,
      "learning_rate": 5.160459330087508e-07,
      "loss": 0.3662,
      "step": 29400
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2955982781882607,
      "learning_rate": 5.15731463186887e-07,
      "loss": 0.1739,
      "step": 29401
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6386568434495277,
      "learning_rate": 5.154170866743768e-07,
      "loss": 0.4919,
      "step": 29402
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.14708143370691265,
      "learning_rate": 5.151028034743144e-07,
      "loss": 0.0672,
      "step": 29403
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4652964021862656,
      "learning_rate": 5.147886135897929e-07,
      "loss": 0.2704,
      "step": 29404
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.28887262652694556,
      "learning_rate": 5.144745170239007e-07,
      "loss": 0.1805,
      "step": 29405
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5831597698902917,
      "learning_rate": 5.14160513779729e-07,
      "loss": 0.352,
      "step": 29406
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9434893038607587,
      "learning_rate": 5.138466038603695e-07,
      "loss": 0.3205,
      "step": 29407
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7107550798229061,
      "learning_rate": 5.135327872689055e-07,
      "loss": 0.2891,
      "step": 29408
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8981068546503741,
      "learning_rate": 5.1321906400843e-07,
      "loss": 0.4173,
      "step": 29409
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37580104945842563,
      "learning_rate": 5.12905434082025e-07,
      "loss": 0.1555,
      "step": 29410
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3749788597010827,
      "learning_rate": 5.12591897492779e-07,
      "loss": 0.2976,
      "step": 29411
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.21056335854566816,
      "learning_rate": 5.122784542437764e-07,
      "loss": 0.1178,
      "step": 29412
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.723321158026322,
      "learning_rate": 5.119651043380979e-07,
      "loss": 0.7159,
      "step": 29413
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43255105693787926,
      "learning_rate": 5.116518477788313e-07,
      "loss": 0.1365,
      "step": 29414
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3340017412904482,
      "learning_rate": 5.113386845690538e-07,
      "loss": 0.224,
      "step": 29415
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7229599945108354,
      "learning_rate": 5.110256147118498e-07,
      "loss": 0.2643,
      "step": 29416
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.36413488600573507,
      "learning_rate": 5.107126382102989e-07,
      "loss": 0.2729,
      "step": 29417
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.164458634516425,
      "learning_rate": 5.103997550674777e-07,
      "loss": 0.1575,
      "step": 29418
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4166952971662028,
      "learning_rate": 5.100869652864659e-07,
      "loss": 0.2438,
      "step": 29419
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6213212462586281,
      "learning_rate": 5.097742688703423e-07,
      "loss": 0.1887,
      "step": 29420
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2388378602763843,
      "learning_rate": 5.094616658221829e-07,
      "loss": 0.0725,
      "step": 29421
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5285562016422449,
      "learning_rate": 5.091491561450612e-07,
      "loss": 0.3385,
      "step": 29422
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2887119776583468,
      "learning_rate": 5.088367398420524e-07,
      "loss": 0.1759,
      "step": 29423
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.39826455732260907,
      "learning_rate": 5.085244169162329e-07,
      "loss": 0.284,
      "step": 29424
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7684943876543052,
      "learning_rate": 5.082121873706702e-07,
      "loss": 0.2469,
      "step": 29425
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.07598895674448,
      "learning_rate": 5.07900051208442e-07,
      "loss": 0.4658,
      "step": 29426
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0194779408624048,
      "learning_rate": 5.075880084326157e-07,
      "loss": 0.2854,
      "step": 29427
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3874328912065958,
      "learning_rate": 5.072760590462611e-07,
      "loss": 0.2239,
      "step": 29428
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33298445725510895,
      "learning_rate": 5.069642030524502e-07,
      "loss": 0.2256,
      "step": 29429
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2361805823000978,
      "learning_rate": 5.066524404542473e-07,
      "loss": 0.4105,
      "step": 29430
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.171162717386735,
      "learning_rate": 5.063407712547208e-07,
      "loss": 0.0773,
      "step": 29431
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7237679309528517,
      "learning_rate": 5.060291954569385e-07,
      "loss": 0.3385,
      "step": 29432
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3198916267569479,
      "learning_rate": 5.057177130639657e-07,
      "loss": 0.1881,
      "step": 29433
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9210040251227671,
      "learning_rate": 5.054063240788643e-07,
      "loss": 0.2705,
      "step": 29434
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33965463139985347,
      "learning_rate": 5.050950285047007e-07,
      "loss": 0.2737,
      "step": 29435
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4310332686869541,
      "learning_rate": 5.047838263445382e-07,
      "loss": 0.1303,
      "step": 29436
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9825063039130724,
      "learning_rate": 5.04472717601433e-07,
      "loss": 0.4163,
      "step": 29437
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3496973849485298,
      "learning_rate": 5.041617022784528e-07,
      "loss": 0.1949,
      "step": 29438
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.500620470481038,
      "learning_rate": 5.038507803786519e-07,
      "loss": 0.1958,
      "step": 29439
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26282069007390335,
      "learning_rate": 5.035399519050932e-07,
      "loss": 0.1847,
      "step": 29440
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9239876380045392,
      "learning_rate": 5.032292168608332e-07,
      "loss": 0.4712,
      "step": 29441
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2872511985862447,
      "learning_rate": 5.029185752489285e-07,
      "loss": 0.1657,
      "step": 29442
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7254872447656078,
      "learning_rate": 5.026080270724365e-07,
      "loss": 0.3803,
      "step": 29443
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2316229752317924,
      "learning_rate": 5.022975723344115e-07,
      "loss": 0.5106,
      "step": 29444
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7954014875394146,
      "learning_rate": 5.019872110379076e-07,
      "loss": 0.7152,
      "step": 29445
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3678158468273021,
      "learning_rate": 5.01676943185978e-07,
      "loss": 0.1714,
      "step": 29446
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2642467037839501,
      "learning_rate": 5.013667687816781e-07,
      "loss": 0.1907,
      "step": 29447
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.46922897342038694,
      "learning_rate": 5.010566878280565e-07,
      "loss": 0.2195,
      "step": 29448
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3540374096952572,
      "learning_rate": 5.007467003281641e-07,
      "loss": 0.1201,
      "step": 29449
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6901019039266367,
      "learning_rate": 5.004368062850529e-07,
      "loss": 0.3099,
      "step": 29450
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3639716166462678,
      "learning_rate": 5.001270057017682e-07,
      "loss": 0.191,
      "step": 29451
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.48990294897347475,
      "learning_rate": 4.99817298581362e-07,
      "loss": 0.3247,
      "step": 29452
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4434217536646743,
      "learning_rate": 4.995076849268776e-07,
      "loss": 0.246,
      "step": 29453
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2776874227164738,
      "learning_rate": 4.991981647413624e-07,
      "loss": 0.4522,
      "step": 29454
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3537018203672373,
      "learning_rate": 4.98888738027864e-07,
      "loss": 0.1588,
      "step": 29455
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5505148733443633,
      "learning_rate": 4.985794047894221e-07,
      "loss": 0.308,
      "step": 29456
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0198781760388285,
      "learning_rate": 4.982701650290833e-07,
      "loss": 0.0329,
      "step": 29457
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5026317146804844,
      "learning_rate": 4.979610187498884e-07,
      "loss": 0.3038,
      "step": 29458
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34646916953775425,
      "learning_rate": 4.976519659548818e-07,
      "loss": 0.2027,
      "step": 29459
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4231321277659387,
      "learning_rate": 4.973430066470996e-07,
      "loss": 0.1722,
      "step": 29460
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.32477078143220894,
      "learning_rate": 4.970341408295842e-07,
      "loss": 0.2137,
      "step": 29461
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.26691421460118114,
      "learning_rate": 4.967253685053752e-07,
      "loss": 0.1431,
      "step": 29462
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5074347472872826,
      "learning_rate": 4.964166896775069e-07,
      "loss": 0.2801,
      "step": 29463
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.28370326353804753,
      "learning_rate": 4.961081043490201e-07,
      "loss": 0.0664,
      "step": 29464
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2976497373217971,
      "learning_rate": 4.95799612522948e-07,
      "loss": 0.2458,
      "step": 29465
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.503902136974457,
      "learning_rate": 4.954912142023261e-07,
      "loss": 0.0671,
      "step": 29466
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.71199525679801,
      "learning_rate": 4.951829093901906e-07,
      "loss": 0.3163,
      "step": 29467
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1309083262816277,
      "learning_rate": 4.948746980895703e-07,
      "loss": 0.2277,
      "step": 29468
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4359841669987317,
      "learning_rate": 4.945665803035038e-07,
      "loss": 0.2982,
      "step": 29469
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.31629406524014697,
      "learning_rate": 4.942585560350166e-07,
      "loss": 0.2286,
      "step": 29470
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4715162340628153,
      "learning_rate": 4.939506252871417e-07,
      "loss": 0.3321,
      "step": 29471
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.16305247592035602,
      "learning_rate": 4.936427880629092e-07,
      "loss": 0.0703,
      "step": 29472
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.8608580905200498,
      "learning_rate": 4.933350443653484e-07,
      "loss": 0.3546,
      "step": 29473
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3218912904934713,
      "learning_rate": 4.930273941974828e-07,
      "loss": 0.168,
      "step": 29474
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6731383899248307,
      "learning_rate": 4.927198375623432e-07,
      "loss": 0.2629,
      "step": 29475
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5210665575924194,
      "learning_rate": 4.92412374462955e-07,
      "loss": 0.3246,
      "step": 29476
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4898092459012832,
      "learning_rate": 4.921050049023401e-07,
      "loss": 0.2143,
      "step": 29477
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4546563654206692,
      "learning_rate": 4.917977288835263e-07,
      "loss": 0.2407,
      "step": 29478
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.432227063270077,
      "learning_rate": 4.914905464095343e-07,
      "loss": 0.2416,
      "step": 29479
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.6314924262469115,
      "learning_rate": 4.911834574833874e-07,
      "loss": 0.7317,
      "step": 29480
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24033173835889862,
      "learning_rate": 4.908764621081074e-07,
      "loss": 0.0738,
      "step": 29481
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3774644275173079,
      "learning_rate": 4.905695602867122e-07,
      "loss": 0.2502,
      "step": 29482
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2939113524836162,
      "learning_rate": 4.902627520222225e-07,
      "loss": 0.2137,
      "step": 29483
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0319654858553504,
      "learning_rate": 4.899560373176582e-07,
      "loss": 0.4477,
      "step": 29484
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7014076852700839,
      "learning_rate": 4.896494161760357e-07,
      "loss": 0.2598,
      "step": 29485
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.144318796229655,
      "learning_rate": 4.893428886003703e-07,
      "loss": 0.5159,
      "step": 29486
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3250549861048624,
      "learning_rate": 4.890364545936799e-07,
      "loss": 0.1503,
      "step": 29487
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33211811003496255,
      "learning_rate": 4.887301141589784e-07,
      "loss": 0.2141,
      "step": 29488
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5009853534248914,
      "learning_rate": 4.88423867299278e-07,
      "loss": 0.3489,
      "step": 29489
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.22339029332951013,
      "learning_rate": 4.881177140175963e-07,
      "loss": 0.0696,
      "step": 29490
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9023522106696558,
      "learning_rate": 4.878116543169398e-07,
      "loss": 0.3996,
      "step": 29491
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.29140874426470087,
      "learning_rate": 4.875056882003226e-07,
      "loss": 0.1701,
      "step": 29492
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.7391400134242038,
      "learning_rate": 4.871998156707558e-07,
      "loss": 0.3776,
      "step": 29493
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.35147174285163524,
      "learning_rate": 4.868940367312447e-07,
      "loss": 0.1968,
      "step": 29494
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.7645253936569651,
      "learning_rate": 4.865883513848035e-07,
      "loss": 0.6904,
      "step": 29495
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3491641313255697,
      "learning_rate": 4.862827596344344e-07,
      "loss": 0.1564,
      "step": 29496
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5690801154772934,
      "learning_rate": 4.85977261483147e-07,
      "loss": 0.3511,
      "step": 29497
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.293377580647156,
      "learning_rate": 4.856718569339458e-07,
      "loss": 0.1302,
      "step": 29498
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4777331335879464,
      "learning_rate": 4.853665459898349e-07,
      "loss": 0.2262,
      "step": 29499
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.33264863872808453,
      "learning_rate": 4.850613286538186e-07,
      "loss": 0.2156,
      "step": 29500
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.35572545470919137,
      "learning_rate": 4.847562049289001e-07,
      "loss": 0.2015,
      "step": 29501
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6636957301110467,
      "learning_rate": 4.844511748180814e-07,
      "loss": 0.3415,
      "step": 29502
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.2421686097167384,
      "learning_rate": 4.841462383243612e-07,
      "loss": 0.1041,
      "step": 29503
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3847294943778024,
      "learning_rate": 4.838413954507437e-07,
      "loss": 0.4348,
      "step": 29504
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34042992691814516,
      "learning_rate": 4.835366462002245e-07,
      "loss": 0.1682,
      "step": 29505
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.29581929650923405,
      "learning_rate": 4.832319905758031e-07,
      "loss": 0.2526,
      "step": 29506
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3201320239285643,
      "learning_rate": 4.829274285804786e-07,
      "loss": 0.4542,
      "step": 29507
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40742712876969983,
      "learning_rate": 4.826229602172438e-07,
      "loss": 0.213,
      "step": 29508
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5668317568199096,
      "learning_rate": 4.823185854890955e-07,
      "loss": 0.2327,
      "step": 29509
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.34931332850045727,
      "learning_rate": 4.820143043990289e-07,
      "loss": 0.2604,
      "step": 29510
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37733414282911204,
      "learning_rate": 4.817101169500393e-07,
      "loss": 0.1289,
      "step": 29511
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3468227479677816,
      "learning_rate": 4.814060231451157e-07,
      "loss": 0.2609,
      "step": 29512
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3013511068154424,
      "learning_rate": 4.81102022987252e-07,
      "loss": 0.1781,
      "step": 29513
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.37297917433824085,
      "learning_rate": 4.807981164794384e-07,
      "loss": 0.148,
      "step": 29514
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5688716158719089,
      "learning_rate": 4.804943036246656e-07,
      "loss": 0.3223,
      "step": 29515
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.0939680737155812,
      "learning_rate": 4.801905844259235e-07,
      "loss": 0.1939,
      "step": 29516
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.49376487551997944,
      "learning_rate": 4.798869588861976e-07,
      "loss": 0.3645,
      "step": 29517
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4461638085987638,
      "learning_rate": 4.795834270084754e-07,
      "loss": 0.2069,
      "step": 29518
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4762075645118586,
      "learning_rate": 4.792799887957466e-07,
      "loss": 0.2525,
      "step": 29519
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3876590012787412,
      "learning_rate": 4.789766442509902e-07,
      "loss": 0.1786,
      "step": 29520
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.4543483525006469,
      "learning_rate": 4.786733933771981e-07,
      "loss": 0.2207,
      "step": 29521
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.4040285852003667,
      "learning_rate": 4.783702361773479e-07,
      "loss": 0.3029,
      "step": 29522
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.006622482089628,
      "learning_rate": 4.78067172654425e-07,
      "loss": 0.479,
      "step": 29523
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.24839402412970976,
      "learning_rate": 4.777642028114116e-07,
      "loss": 0.1937,
      "step": 29524
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5310257375624288,
      "learning_rate": 4.774613266512851e-07,
      "loss": 0.341,
      "step": 29525
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.120093297475772,
      "learning_rate": 4.771585441770288e-07,
      "loss": 0.2199,
      "step": 29526
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5773470721513405,
      "learning_rate": 4.768558553916191e-07,
      "loss": 0.2258,
      "step": 29527
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3754235948555445,
      "learning_rate": 4.7655326029803715e-07,
      "loss": 0.2472,
      "step": 29528
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.2002867040933168,
      "learning_rate": 4.7625075889925596e-07,
      "loss": 0.1195,
      "step": 29529
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5233198913567378,
      "learning_rate": 4.759483511982532e-07,
      "loss": 0.3129,
      "step": 29530
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1591049033528174,
      "learning_rate": 4.756460371980065e-07,
      "loss": 0.1701,
      "step": 29531
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6690600364658287,
      "learning_rate": 4.7534381690148456e-07,
      "loss": 0.3463,
      "step": 29532
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.38203347473016114,
      "learning_rate": 4.7504169031166724e-07,
      "loss": 0.2356,
      "step": 29533
    },
    {
      "epoch": 0.9,
      "grad_norm": 2.0027681490294897,
      "learning_rate": 4.7473965743152215e-07,
      "loss": 0.6345,
      "step": 29534
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.43589718724166704,
      "learning_rate": 4.7443771826402365e-07,
      "loss": 0.2358,
      "step": 29535
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.48819361952739926,
      "learning_rate": 4.741358728121415e-07,
      "loss": 0.2706,
      "step": 29536
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.30184442643940235,
      "learning_rate": 4.738341210788444e-07,
      "loss": 0.1726,
      "step": 29537
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.3496118141518294,
      "learning_rate": 4.7353246306710123e-07,
      "loss": 0.5741,
      "step": 29538
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.1653850838409865,
      "learning_rate": 4.732308987798806e-07,
      "loss": 0.0684,
      "step": 29539
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1782140628089819,
      "learning_rate": 4.729294282201491e-07,
      "loss": 0.5146,
      "step": 29540
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.40166484575952976,
      "learning_rate": 4.726280513908721e-07,
      "loss": 0.2887,
      "step": 29541
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3574521540508903,
      "learning_rate": 4.723267682950172e-07,
      "loss": 0.1961,
      "step": 29542
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.5361071752788009,
      "learning_rate": 4.720255789355455e-07,
      "loss": 0.3327,
      "step": 29543
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6338093197113522,
      "learning_rate": 4.717244833154211e-07,
      "loss": 0.265,
      "step": 29544
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.9900794159189611,
      "learning_rate": 4.714234814376073e-07,
      "loss": 0.3756,
      "step": 29545
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.393263375033095,
      "learning_rate": 4.7112257330506174e-07,
      "loss": 0.165,
      "step": 29546
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.3557687281967307,
      "learning_rate": 4.7082175892075086e-07,
      "loss": 0.2274,
      "step": 29547
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.23887004202187237,
      "learning_rate": 4.7052103828763017e-07,
      "loss": 0.1711,
      "step": 29548
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.1765755720024955,
      "learning_rate": 4.702204114086584e-07,
      "loss": 0.4934,
      "step": 29549
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6044194484184016,
      "learning_rate": 4.699198782867953e-07,
      "loss": 0.2069,
      "step": 29550
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37143629379637216,
      "learning_rate": 4.696194389249953e-07,
      "loss": 0.2635,
      "step": 29551
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7278319017926799,
      "learning_rate": 4.6931909332621485e-07,
      "loss": 0.2654,
      "step": 29552
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4711927795841733,
      "learning_rate": 4.690188414934094e-07,
      "loss": 0.2321,
      "step": 29553
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5108428021984822,
      "learning_rate": 4.687186834295343e-07,
      "loss": 0.275,
      "step": 29554
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4060358564525599,
      "learning_rate": 4.684186191375395e-07,
      "loss": 0.1541,
      "step": 29555
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5431342543622569,
      "learning_rate": 4.6811864862037816e-07,
      "loss": 0.3177,
      "step": 29556
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1438209155184205,
      "learning_rate": 4.678187718810034e-07,
      "loss": 0.0686,
      "step": 29557
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.497666498617882,
      "learning_rate": 4.675189889223619e-07,
      "loss": 0.6747,
      "step": 29558
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.32682081586191636,
      "learning_rate": 4.6721929974740675e-07,
      "loss": 0.1857,
      "step": 29559
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.29569358565606213,
      "learning_rate": 4.669197043590845e-07,
      "loss": 0.2517,
      "step": 29560
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.742514311046629,
      "learning_rate": 4.666202027603428e-07,
      "loss": 0.2591,
      "step": 29561
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8780516987437839,
      "learning_rate": 4.663207949541304e-07,
      "loss": 0.382,
      "step": 29562
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4638488683897894,
      "learning_rate": 4.660214809433883e-07,
      "loss": 0.0799,
      "step": 29563
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.395940078565413,
      "learning_rate": 4.657222607310652e-07,
      "loss": 0.2415,
      "step": 29564
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4554753030177237,
      "learning_rate": 4.654231343201032e-07,
      "loss": 0.1678,
      "step": 29565
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26655266862840976,
      "learning_rate": 4.6512410171344555e-07,
      "loss": 0.1895,
      "step": 29566
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4881859639689759,
      "learning_rate": 4.6482516291403546e-07,
      "loss": 0.3054,
      "step": 29567
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6680872083700766,
      "learning_rate": 4.645263179248127e-07,
      "loss": 0.1704,
      "step": 29568
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6019463940685255,
      "learning_rate": 4.6422756674871617e-07,
      "loss": 0.3348,
      "step": 29569
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5072330230007835,
      "learning_rate": 4.6392890938868787e-07,
      "loss": 0.246,
      "step": 29570
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3410080873848864,
      "learning_rate": 4.6363034584766543e-07,
      "loss": 0.2726,
      "step": 29571
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5768708152361162,
      "learning_rate": 4.633318761285821e-07,
      "loss": 0.0943,
      "step": 29572
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9903132515768783,
      "learning_rate": 4.630335002343811e-07,
      "loss": 0.4052,
      "step": 29573
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.332052183302796,
      "learning_rate": 4.6273521816799336e-07,
      "loss": 0.1908,
      "step": 29574
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.31939927178692623,
      "learning_rate": 4.624370299323544e-07,
      "loss": 0.1414,
      "step": 29575
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9158144309506108,
      "learning_rate": 4.621389355304007e-07,
      "loss": 0.3212,
      "step": 29576
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.32479426874886835,
      "learning_rate": 4.618409349650599e-07,
      "loss": 0.2469,
      "step": 29577
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2890318973953292,
      "learning_rate": 4.6154302823926745e-07,
      "loss": 0.1732,
      "step": 29578
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6296237333873637,
      "learning_rate": 4.6124521535595213e-07,
      "loss": 0.2657,
      "step": 29579
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.412353012468651,
      "learning_rate": 4.6094749631804714e-07,
      "loss": 0.5487,
      "step": 29580
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.9471207155378296,
      "learning_rate": 4.606498711284779e-07,
      "loss": 0.0812,
      "step": 29581
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6416888777253681,
      "learning_rate": 4.6035233979017436e-07,
      "loss": 0.2626,
      "step": 29582
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2957260312949148,
      "learning_rate": 4.6005490230606407e-07,
      "loss": 0.2062,
      "step": 29583
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4897042235076811,
      "learning_rate": 4.5975755867907036e-07,
      "loss": 0.3077,
      "step": 29584
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5057328041319781,
      "learning_rate": 4.594603089121241e-07,
      "loss": 0.1696,
      "step": 29585
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.661658523124853,
      "learning_rate": 4.5916315300814416e-07,
      "loss": 0.3328,
      "step": 29586
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2626669206602725,
      "learning_rate": 4.5886609097005817e-07,
      "loss": 0.1684,
      "step": 29587
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8930180095383846,
      "learning_rate": 4.585691228007871e-07,
      "loss": 0.855,
      "step": 29588
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3278091359215593,
      "learning_rate": 4.582722485032509e-07,
      "loss": 0.2503,
      "step": 29589
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.44594796525294356,
      "learning_rate": 4.5797546808037273e-07,
      "loss": 0.2248,
      "step": 29590
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3572965454071796,
      "learning_rate": 4.576787815350714e-07,
      "loss": 0.1606,
      "step": 29591
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.49077030725123166,
      "learning_rate": 4.573821888702668e-07,
      "loss": 0.1875,
      "step": 29592
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5479093340933308,
      "learning_rate": 4.570856900888765e-07,
      "loss": 0.2677,
      "step": 29593
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3195000830051484,
      "learning_rate": 4.5678928519381607e-07,
      "loss": 0.1875,
      "step": 29594
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.45000990624883225,
      "learning_rate": 4.564929741880031e-07,
      "loss": 0.3146,
      "step": 29595
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3967575673787257,
      "learning_rate": 4.56196757074352e-07,
      "loss": 0.167,
      "step": 29596
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5262095983557737,
      "learning_rate": 4.559006338557792e-07,
      "loss": 0.3113,
      "step": 29597
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4254698943005482,
      "learning_rate": 4.5560460453519475e-07,
      "loss": 0.3318,
      "step": 29598
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0749821480332407,
      "learning_rate": 4.5530866911551286e-07,
      "loss": 0.4312,
      "step": 29599
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3404222974462504,
      "learning_rate": 4.5501282759964684e-07,
      "loss": 0.0655,
      "step": 29600
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3349454075396594,
      "learning_rate": 4.547170799905021e-07,
      "loss": 0.2327,
      "step": 29601
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.32343348895630697,
      "learning_rate": 4.5442142629099404e-07,
      "loss": 0.2031,
      "step": 29602
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8024338531443649,
      "learning_rate": 4.541258665040282e-07,
      "loss": 0.2996,
      "step": 29603
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3981774281797402,
      "learning_rate": 4.538304006325123e-07,
      "loss": 0.1322,
      "step": 29604
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.29984409211286345,
      "learning_rate": 4.5353502867935493e-07,
      "loss": 0.1718,
      "step": 29605
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.389052464386264,
      "learning_rate": 4.532397506474617e-07,
      "loss": 0.7464,
      "step": 29606
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.31043190639085133,
      "learning_rate": 4.529445665397358e-07,
      "loss": 0.2258,
      "step": 29607
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4756351561965034,
      "learning_rate": 4.5264947635908387e-07,
      "loss": 0.2125,
      "step": 29608
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6080078418360452,
      "learning_rate": 4.523544801084079e-07,
      "loss": 0.0955,
      "step": 29609
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4019808881668437,
      "learning_rate": 4.5205957779060783e-07,
      "loss": 0.2807,
      "step": 29610
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8267774932983833,
      "learning_rate": 4.517647694085903e-07,
      "loss": 0.2511,
      "step": 29611
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7641434892108503,
      "learning_rate": 4.514700549652506e-07,
      "loss": 0.3516,
      "step": 29612
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.29587303582806107,
      "learning_rate": 4.511754344634911e-07,
      "loss": 0.1978,
      "step": 29613
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40196861860556554,
      "learning_rate": 4.508809079062104e-07,
      "loss": 0.2419,
      "step": 29614
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4662759090002463,
      "learning_rate": 4.505864752963052e-07,
      "loss": 0.2191,
      "step": 29615
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2621766428232457,
      "learning_rate": 4.5029213663667084e-07,
      "loss": 0.7573,
      "step": 29616
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18624583867710934,
      "learning_rate": 4.4999789193020614e-07,
      "loss": 0.0663,
      "step": 29617
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3508469996867065,
      "learning_rate": 4.497037411798033e-07,
      "loss": 0.1659,
      "step": 29618
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5348020616762433,
      "learning_rate": 4.4940968438835996e-07,
      "loss": 0.283,
      "step": 29619
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36499602172858353,
      "learning_rate": 4.4911572155876495e-07,
      "loss": 0.2103,
      "step": 29620
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6526551449509786,
      "learning_rate": 4.4882185269391254e-07,
      "loss": 0.3473,
      "step": 29621
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4248285508598167,
      "learning_rate": 4.4852807779669274e-07,
      "loss": 0.1312,
      "step": 29622
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9893996281931255,
      "learning_rate": 4.482343968699976e-07,
      "loss": 0.49,
      "step": 29623
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3338679658402174,
      "learning_rate": 4.4794080991671485e-07,
      "loss": 0.2114,
      "step": 29624
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3543090854032837,
      "learning_rate": 4.476473169397333e-07,
      "loss": 0.2796,
      "step": 29625
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.15912751005283807,
      "learning_rate": 4.4735391794194173e-07,
      "loss": 0.0673,
      "step": 29626
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8103716417118931,
      "learning_rate": 4.4706061292622336e-07,
      "loss": 0.7958,
      "step": 29627
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.27474334435340575,
      "learning_rate": 4.467674018954682e-07,
      "loss": 0.1699,
      "step": 29628
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6602457897620501,
      "learning_rate": 4.464742848525583e-07,
      "loss": 0.3359,
      "step": 29629
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3452430324386217,
      "learning_rate": 4.461812618003769e-07,
      "loss": 0.2163,
      "step": 29630
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7566702597408717,
      "learning_rate": 4.4588833274180955e-07,
      "loss": 0.0855,
      "step": 29631
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5950330885978732,
      "learning_rate": 4.45595497679735e-07,
      "loss": 0.2988,
      "step": 29632
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34844801395036756,
      "learning_rate": 4.453027566170365e-07,
      "loss": 0.2196,
      "step": 29633
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5141516531861455,
      "learning_rate": 4.450101095565928e-07,
      "loss": 0.2338,
      "step": 29634
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3321914018480508,
      "learning_rate": 4.447175565012851e-07,
      "loss": 0.0722,
      "step": 29635
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.42413633808277884,
      "learning_rate": 4.4442509745398877e-07,
      "loss": 0.2882,
      "step": 29636
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26326407322213846,
      "learning_rate": 4.441327324175815e-07,
      "loss": 0.1891,
      "step": 29637
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7442053286940858,
      "learning_rate": 4.438404613949432e-07,
      "loss": 0.3688,
      "step": 29638
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.013863289782503,
      "learning_rate": 4.435482843889438e-07,
      "loss": 0.2148,
      "step": 29639
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7414093560680945,
      "learning_rate": 4.4325620140246437e-07,
      "loss": 0.7803,
      "step": 29640
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3503069423579789,
      "learning_rate": 4.429642124383715e-07,
      "loss": 0.1648,
      "step": 29641
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5610426142692323,
      "learning_rate": 4.42672317499544e-07,
      "loss": 0.329,
      "step": 29642
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3425281882439307,
      "learning_rate": 4.423805165888506e-07,
      "loss": 0.2202,
      "step": 29643
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.1757269453713453,
      "learning_rate": 4.4208880970916137e-07,
      "loss": 0.0683,
      "step": 29644
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6770092309928435,
      "learning_rate": 4.4179719686334944e-07,
      "loss": 0.3401,
      "step": 29645
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4086087894962561,
      "learning_rate": 4.4150567805428034e-07,
      "loss": 0.1702,
      "step": 29646
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.56193179014068,
      "learning_rate": 4.412142532848229e-07,
      "loss": 0.3076,
      "step": 29647
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43450965958393145,
      "learning_rate": 4.4092292255784596e-07,
      "loss": 0.1793,
      "step": 29648
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.49630450004863025,
      "learning_rate": 4.4063168587621495e-07,
      "loss": 0.3485,
      "step": 29649
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4797199353913288,
      "learning_rate": 4.403405432427943e-07,
      "loss": 0.1309,
      "step": 29650
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35360342709975473,
      "learning_rate": 4.400494946604483e-07,
      "loss": 0.2367,
      "step": 29651
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6474666886232325,
      "learning_rate": 4.3975854013204366e-07,
      "loss": 0.3109,
      "step": 29652
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.47759719975629833,
      "learning_rate": 4.3946767966043577e-07,
      "loss": 0.274,
      "step": 29653
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.333244694325607,
      "learning_rate": 4.3917691324849466e-07,
      "loss": 0.2297,
      "step": 29654
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3924328416364141,
      "learning_rate": 4.388862408990757e-07,
      "loss": 0.2226,
      "step": 29655
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34232365242578955,
      "learning_rate": 4.3859566261503894e-07,
      "loss": 0.1646,
      "step": 29656
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3787728080598498,
      "learning_rate": 4.383051783992465e-07,
      "loss": 0.5382,
      "step": 29657
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.319170439558374,
      "learning_rate": 4.3801478825455155e-07,
      "loss": 0.1804,
      "step": 29658
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3429888356573345,
      "learning_rate": 4.3772449218381416e-07,
      "loss": 0.1442,
      "step": 29659
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4084226383942165,
      "learning_rate": 4.374342901898898e-07,
      "loss": 0.298,
      "step": 29660
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3718443118053552,
      "learning_rate": 4.3714418227563393e-07,
      "loss": 0.2243,
      "step": 29661
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1158162281505184,
      "learning_rate": 4.368541684438987e-07,
      "loss": 0.4589,
      "step": 29662
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7334991304800317,
      "learning_rate": 4.3656424869753967e-07,
      "loss": 0.2521,
      "step": 29663
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41863859579233687,
      "learning_rate": 4.3627442303941004e-07,
      "loss": 0.2508,
      "step": 29664
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2026312245998381,
      "learning_rate": 4.359846914723553e-07,
      "loss": 0.1132,
      "step": 29665
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2726977823853827,
      "learning_rate": 4.356950539992344e-07,
      "loss": 0.3581,
      "step": 29666
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.320641695345429,
      "learning_rate": 4.354055106228905e-07,
      "loss": 0.228,
      "step": 29667
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6829181230599202,
      "learning_rate": 4.351160613461747e-07,
      "loss": 0.3117,
      "step": 29668
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34460605865610927,
      "learning_rate": 4.348267061719358e-07,
      "loss": 0.2068,
      "step": 29669
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5332635226843387,
      "learning_rate": 4.34537445103016e-07,
      "loss": 0.7951,
      "step": 29670
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6807381914198982,
      "learning_rate": 4.3424827814226634e-07,
      "loss": 0.2691,
      "step": 29671
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33182065968118135,
      "learning_rate": 4.339592052925301e-07,
      "loss": 0.2202,
      "step": 29672
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5129335123095753,
      "learning_rate": 4.336702265566495e-07,
      "loss": 0.2763,
      "step": 29673
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.27333733872411353,
      "learning_rate": 4.3338134193746883e-07,
      "loss": 0.1335,
      "step": 29674
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4067239775957945,
      "learning_rate": 4.3309255143783256e-07,
      "loss": 0.1642,
      "step": 29675
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2713283329514191,
      "learning_rate": 4.3280385506057733e-07,
      "loss": 0.3205,
      "step": 29676
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5803668880811477,
      "learning_rate": 4.3251525280854636e-07,
      "loss": 0.8788,
      "step": 29677
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30021040499042045,
      "learning_rate": 4.322267446845807e-07,
      "loss": 0.1716,
      "step": 29678
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3394532922354313,
      "learning_rate": 4.319383306915126e-07,
      "loss": 0.27,
      "step": 29679
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5871545441854688,
      "learning_rate": 4.316500108321864e-07,
      "loss": 0.2748,
      "step": 29680
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5647503672643355,
      "learning_rate": 4.3136178510943426e-07,
      "loss": 0.5644,
      "step": 29681
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.552449235269578,
      "learning_rate": 4.31073653526094e-07,
      "loss": 0.0728,
      "step": 29682
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.27545169513805284,
      "learning_rate": 4.30785616085001e-07,
      "loss": 0.2115,
      "step": 29683
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2488144291495579,
      "learning_rate": 4.304976727889865e-07,
      "loss": 0.1821,
      "step": 29684
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9648029935337374,
      "learning_rate": 4.302098236408847e-07,
      "loss": 0.2808,
      "step": 29685
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9174645314714286,
      "learning_rate": 4.2992206864352683e-07,
      "loss": 0.4445,
      "step": 29686
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2979415591228301,
      "learning_rate": 4.296344077997461e-07,
      "loss": 0.1842,
      "step": 29687
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7710392413114121,
      "learning_rate": 4.2934684111237024e-07,
      "loss": 0.3565,
      "step": 29688
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6588071735746681,
      "learning_rate": 4.2905936858422926e-07,
      "loss": 0.2695,
      "step": 29689
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3706714418493761,
      "learning_rate": 4.287719902181531e-07,
      "loss": 0.2991,
      "step": 29690
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3419738245762755,
      "learning_rate": 4.284847060169639e-07,
      "loss": 0.153,
      "step": 29691
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3260848366926789,
      "learning_rate": 4.28197515983495e-07,
      "loss": 0.195,
      "step": 29692
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3646726323809021,
      "learning_rate": 4.279104201205675e-07,
      "loss": 0.0701,
      "step": 29693
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4626925052121338,
      "learning_rate": 4.2762341843100575e-07,
      "loss": 0.6987,
      "step": 29694
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36103164720659586,
      "learning_rate": 4.273365109176364e-07,
      "loss": 0.2055,
      "step": 29695
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37885285809374547,
      "learning_rate": 4.270496975832783e-07,
      "loss": 0.2529,
      "step": 29696
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.442391656184405,
      "learning_rate": 4.2676297843075696e-07,
      "loss": 0.2186,
      "step": 29697
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.796705991604951,
      "learning_rate": 4.2647635346289016e-07,
      "loss": 0.2658,
      "step": 29698
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.407502282967876,
      "learning_rate": 4.261898226825001e-07,
      "loss": 0.4617,
      "step": 29699
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2747411686375431,
      "learning_rate": 4.2590338609240444e-07,
      "loss": 0.0676,
      "step": 29700
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39179887636866784,
      "learning_rate": 4.2561704369542103e-07,
      "loss": 0.2636,
      "step": 29701
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19860798668418175,
      "learning_rate": 4.2533079549436865e-07,
      "loss": 0.1582,
      "step": 29702
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4705232487119686,
      "learning_rate": 4.250446414920606e-07,
      "loss": 0.6655,
      "step": 29703
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8397642525817987,
      "learning_rate": 4.247585816913158e-07,
      "loss": 0.2564,
      "step": 29704
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3932000993970131,
      "learning_rate": 4.244726160949453e-07,
      "loss": 0.2218,
      "step": 29705
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.436495196965922,
      "learning_rate": 4.2418674470576573e-07,
      "loss": 0.233,
      "step": 29706
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5677325114248788,
      "learning_rate": 4.2390096752658703e-07,
      "loss": 0.7097,
      "step": 29707
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30102898806177514,
      "learning_rate": 4.2361528456021925e-07,
      "loss": 0.2151,
      "step": 29708
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0572171271659343,
      "learning_rate": 4.23329695809479e-07,
      "loss": 0.3587,
      "step": 29709
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3426353168045697,
      "learning_rate": 4.2304420127717073e-07,
      "loss": 0.192,
      "step": 29710
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.25248852472323485,
      "learning_rate": 4.2275880096610434e-07,
      "loss": 0.0826,
      "step": 29711
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4755323454129192,
      "learning_rate": 4.224734948790876e-07,
      "loss": 0.7343,
      "step": 29712
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34872178402266846,
      "learning_rate": 4.2218828301893057e-07,
      "loss": 0.2308,
      "step": 29713
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4396054263919017,
      "learning_rate": 4.2190316538843537e-07,
      "loss": 0.252,
      "step": 29714
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5134662074480109,
      "learning_rate": 4.2161814199040753e-07,
      "loss": 0.2475,
      "step": 29715
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6531381824465776,
      "learning_rate": 4.213332128276537e-07,
      "loss": 0.7119,
      "step": 29716
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.156378307089244,
      "learning_rate": 4.2104837790297393e-07,
      "loss": 0.137,
      "step": 29717
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6748136775682267,
      "learning_rate": 4.2076363721917367e-07,
      "loss": 0.3259,
      "step": 29718
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2527313881574284,
      "learning_rate": 4.2047899077905184e-07,
      "loss": 0.1524,
      "step": 29719
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.47035415561182353,
      "learning_rate": 4.201944385854095e-07,
      "loss": 0.2781,
      "step": 29720
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2549809572512018,
      "learning_rate": 4.199099806410478e-07,
      "loss": 0.171,
      "step": 29721
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6963822588823061,
      "learning_rate": 4.196256169487611e-07,
      "loss": 0.3248,
      "step": 29722
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26704724018377846,
      "learning_rate": 4.193413475113528e-07,
      "loss": 0.1647,
      "step": 29723
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0130099532024264,
      "learning_rate": 4.1905717233161616e-07,
      "loss": 0.2853,
      "step": 29724
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.794481355039149,
      "learning_rate": 4.187730914123467e-07,
      "loss": 0.7852,
      "step": 29725
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.31589190841003856,
      "learning_rate": 4.184891047563422e-07,
      "loss": 0.1935,
      "step": 29726
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8364440612503284,
      "learning_rate": 4.1820521236639266e-07,
      "loss": 0.3602,
      "step": 29727
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2875668478749717,
      "learning_rate": 4.1792141424529367e-07,
      "loss": 0.1769,
      "step": 29728
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4574536957137247,
      "learning_rate": 4.176377103958362e-07,
      "loss": 0.1934,
      "step": 29729
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7147073702910662,
      "learning_rate": 4.1735410082081374e-07,
      "loss": 0.2682,
      "step": 29730
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.49121379134609455,
      "learning_rate": 4.1707058552301285e-07,
      "loss": 0.3404,
      "step": 29731
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30031470979888153,
      "learning_rate": 4.167871645052246e-07,
      "loss": 0.1441,
      "step": 29732
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37269903503400814,
      "learning_rate": 4.165038377702391e-07,
      "loss": 0.2875,
      "step": 29733
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.27526912797645897,
      "learning_rate": 4.1622060532084065e-07,
      "loss": 0.1158,
      "step": 29734
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.3319424011580823,
      "learning_rate": 4.159374671598182e-07,
      "loss": 0.5159,
      "step": 29735
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8416333582553256,
      "learning_rate": 4.156544232899562e-07,
      "loss": 0.08,
      "step": 29736
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.307446373732365,
      "learning_rate": 4.153714737140402e-07,
      "loss": 0.1825,
      "step": 29737
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2643739845265129,
      "learning_rate": 4.150886184348535e-07,
      "loss": 0.2101,
      "step": 29738
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7251695409523958,
      "learning_rate": 4.148058574551783e-07,
      "loss": 0.2364,
      "step": 29739
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8216588787843677,
      "learning_rate": 4.145231907777969e-07,
      "loss": 0.3988,
      "step": 29740
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34462550998828045,
      "learning_rate": 4.1424061840549034e-07,
      "loss": 0.1697,
      "step": 29741
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5405440735396022,
      "learning_rate": 4.139581403410398e-07,
      "loss": 0.3228,
      "step": 29742
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37692583662647255,
      "learning_rate": 4.136757565872207e-07,
      "loss": 0.081,
      "step": 29743
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35434546415343454,
      "learning_rate": 4.133934671468165e-07,
      "loss": 0.278,
      "step": 29744
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35554556922680963,
      "learning_rate": 4.1311127202260157e-07,
      "loss": 0.0679,
      "step": 29745
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4115115353700084,
      "learning_rate": 4.1282917121735155e-07,
      "loss": 0.2674,
      "step": 29746
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.5839352217233988,
      "learning_rate": 4.1254716473384304e-07,
      "loss": 0.0588,
      "step": 29747
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6619703203866962,
      "learning_rate": 4.122652525748494e-07,
      "loss": 0.3411,
      "step": 29748
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3270155726178368,
      "learning_rate": 4.119834347431462e-07,
      "loss": 0.2295,
      "step": 29749
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.27749858104366537,
      "learning_rate": 4.1170171124150447e-07,
      "loss": 0.1715,
      "step": 29750
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7280307716915912,
      "learning_rate": 4.114200820726955e-07,
      "loss": 0.745,
      "step": 29751
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.16612852935939015,
      "learning_rate": 4.111385472394924e-07,
      "loss": 0.0957,
      "step": 29752
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4272769685074884,
      "learning_rate": 4.108571067446621e-07,
      "loss": 0.5526,
      "step": 29753
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37958977786125114,
      "learning_rate": 4.1057576059097546e-07,
      "loss": 0.0597,
      "step": 29754
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5025537977634876,
      "learning_rate": 4.102945087811994e-07,
      "loss": 0.3082,
      "step": 29755
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.29669735962821653,
      "learning_rate": 4.100133513181026e-07,
      "loss": 0.2329,
      "step": 29756
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8209202506872445,
      "learning_rate": 4.0973228820444855e-07,
      "loss": 0.353,
      "step": 29757
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.170985851995042,
      "learning_rate": 4.0945131944300387e-07,
      "loss": 0.4229,
      "step": 29758
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9912998154254554,
      "learning_rate": 4.0917044503653415e-07,
      "loss": 0.4683,
      "step": 29759
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30808517713590383,
      "learning_rate": 4.0888966498779823e-07,
      "loss": 0.1883,
      "step": 29760
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.35857364497953215,
      "learning_rate": 4.086089792995651e-07,
      "loss": 0.2127,
      "step": 29761
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3025547791273483,
      "learning_rate": 4.083283879745914e-07,
      "loss": 0.1495,
      "step": 29762
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7866460596809047,
      "learning_rate": 4.0804789101563933e-07,
      "loss": 0.0986,
      "step": 29763
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33907074197244114,
      "learning_rate": 4.07767488425469e-07,
      "loss": 0.2323,
      "step": 29764
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5777062627616052,
      "learning_rate": 4.0748718020683697e-07,
      "loss": 0.2623,
      "step": 29765
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4010699457284235,
      "learning_rate": 4.072069663625033e-07,
      "loss": 0.5554,
      "step": 29766
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.285028436471582,
      "learning_rate": 4.0692684689522367e-07,
      "loss": 0.2062,
      "step": 29767
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7318150871308948,
      "learning_rate": 4.066468218077535e-07,
      "loss": 0.4127,
      "step": 29768
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3438868344935404,
      "learning_rate": 4.0636689110285066e-07,
      "loss": 0.217,
      "step": 29769
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7813648186750421,
      "learning_rate": 4.0608705478326514e-07,
      "loss": 0.7313,
      "step": 29770
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.18942594831802254,
      "learning_rate": 4.0580731285175256e-07,
      "loss": 0.0702,
      "step": 29771
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7741670274047808,
      "learning_rate": 4.05527665311064e-07,
      "loss": 0.3468,
      "step": 29772
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2893579325515428,
      "learning_rate": 4.052481121639529e-07,
      "loss": 0.182,
      "step": 29773
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5098417030173036,
      "learning_rate": 4.0496865341316583e-07,
      "loss": 0.3436,
      "step": 29774
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43406335442464566,
      "learning_rate": 4.0468928906145623e-07,
      "loss": 0.2454,
      "step": 29775
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.1910473948510192,
      "learning_rate": 4.044100191115696e-07,
      "loss": 0.5367,
      "step": 29776
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4788600446675519,
      "learning_rate": 4.0413084356625497e-07,
      "loss": 0.2824,
      "step": 29777
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.47295751020489785,
      "learning_rate": 4.0385176242825895e-07,
      "loss": 0.2143,
      "step": 29778
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5090743720072669,
      "learning_rate": 4.0357277570032606e-07,
      "loss": 0.2862,
      "step": 29779
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19369933779036483,
      "learning_rate": 4.032938833852018e-07,
      "loss": 0.1223,
      "step": 29780
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6352736908469583,
      "learning_rate": 4.0301508548563075e-07,
      "loss": 0.3567,
      "step": 29781
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3995285242156949,
      "learning_rate": 4.0273638200435504e-07,
      "loss": 0.1784,
      "step": 29782
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5476168502250762,
      "learning_rate": 4.0245777294411593e-07,
      "loss": 0.3084,
      "step": 29783
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8144278503120093,
      "learning_rate": 4.021792583076556e-07,
      "loss": 0.2281,
      "step": 29784
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3315621229387427,
      "learning_rate": 4.019008380977152e-07,
      "loss": 0.2692,
      "step": 29785
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.408212277569568,
      "learning_rate": 4.016225123170292e-07,
      "loss": 0.138,
      "step": 29786
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33943601575200627,
      "learning_rate": 4.0134428096834324e-07,
      "loss": 0.2395,
      "step": 29787
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.23391486981464368,
      "learning_rate": 4.010661440543884e-07,
      "loss": 0.0936,
      "step": 29788
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4415274305343561,
      "learning_rate": 4.007881015779025e-07,
      "loss": 0.1305,
      "step": 29789
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6846315429478278,
      "learning_rate": 4.005101535416245e-07,
      "loss": 0.3647,
      "step": 29790
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2656343984592969,
      "learning_rate": 4.0023229994828436e-07,
      "loss": 0.1748,
      "step": 29791
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3535335867988865,
      "learning_rate": 3.999545408006178e-07,
      "loss": 0.2494,
      "step": 29792
    },
    {
      "epoch": 0.91,
      "grad_norm": 2.2307488080924354,
      "learning_rate": 3.9967687610135695e-07,
      "loss": 0.1662,
      "step": 29793
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.544565713000799,
      "learning_rate": 3.993993058532342e-07,
      "loss": 0.7569,
      "step": 29794
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3252677813684557,
      "learning_rate": 3.9912183005898056e-07,
      "loss": 0.1439,
      "step": 29795
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4124108559283916,
      "learning_rate": 3.98844448721325e-07,
      "loss": 0.2826,
      "step": 29796
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.421800691496245,
      "learning_rate": 3.985671618429965e-07,
      "loss": 0.1912,
      "step": 29797
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.32588540605201805,
      "learning_rate": 3.982899694267239e-07,
      "loss": 0.2262,
      "step": 29798
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5656479757422174,
      "learning_rate": 3.9801287147523513e-07,
      "loss": 0.2611,
      "step": 29799
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.39121913526319785,
      "learning_rate": 3.9773586799125466e-07,
      "loss": 0.215,
      "step": 29800
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26556298685497454,
      "learning_rate": 3.97458958977508e-07,
      "loss": 0.1332,
      "step": 29801
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.197676225129645,
      "learning_rate": 3.9718214443671966e-07,
      "loss": 0.2943,
      "step": 29802
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3577831760391931,
      "learning_rate": 3.9690542437161193e-07,
      "loss": 0.2936,
      "step": 29803
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37270458410338164,
      "learning_rate": 3.9662879878491045e-07,
      "loss": 0.1495,
      "step": 29804
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.770562639902914,
      "learning_rate": 3.963522676793341e-07,
      "loss": 0.3303,
      "step": 29805
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6147252721846669,
      "learning_rate": 3.9607583105760407e-07,
      "loss": 0.1936,
      "step": 29806
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6831300826154851,
      "learning_rate": 3.9579948892244035e-07,
      "loss": 0.3563,
      "step": 29807
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3769882310286167,
      "learning_rate": 3.9552324127656085e-07,
      "loss": 0.1828,
      "step": 29808
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0601037227766912,
      "learning_rate": 3.9524708812268443e-07,
      "loss": 0.394,
      "step": 29809
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.22147082511872018,
      "learning_rate": 3.9497102946352784e-07,
      "loss": 0.1779,
      "step": 29810
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.46530466640416496,
      "learning_rate": 3.946950653018067e-07,
      "loss": 0.1745,
      "step": 29811
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0367088602456944,
      "learning_rate": 3.9441919564023435e-07,
      "loss": 0.5647,
      "step": 29812
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.437918176830193,
      "learning_rate": 3.941434204815287e-07,
      "loss": 0.4558,
      "step": 29813
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2824839066109168,
      "learning_rate": 3.938677398283997e-07,
      "loss": 0.1753,
      "step": 29814
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34969531445224855,
      "learning_rate": 3.9359215368356183e-07,
      "loss": 0.2141,
      "step": 29815
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.848092521435892,
      "learning_rate": 3.9331666204972527e-07,
      "loss": 0.3724,
      "step": 29816
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.164795537551242,
      "learning_rate": 3.930412649295989e-07,
      "loss": 0.2578,
      "step": 29817
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.320262632649375,
      "learning_rate": 3.9276596232589503e-07,
      "loss": 0.1729,
      "step": 29818
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.24294164124845039,
      "learning_rate": 3.9249075424131923e-07,
      "loss": 0.1512,
      "step": 29819
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.6341305275025095,
      "learning_rate": 3.922156406785815e-07,
      "loss": 0.7674,
      "step": 29820
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3237346518032906,
      "learning_rate": 3.919406216403887e-07,
      "loss": 0.2316,
      "step": 29821
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.4705944523217451,
      "learning_rate": 3.9166569712944524e-07,
      "loss": 0.6455,
      "step": 29822
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.29468609716977123,
      "learning_rate": 3.913908671484545e-07,
      "loss": 0.1719,
      "step": 29823
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9573005727446143,
      "learning_rate": 3.911161317001233e-07,
      "loss": 0.4315,
      "step": 29824
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7444449141990139,
      "learning_rate": 3.9084149078715495e-07,
      "loss": 0.2536,
      "step": 29825
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.46604284460403,
      "learning_rate": 3.905669444122473e-07,
      "loss": 0.2967,
      "step": 29826
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.26589364524494563,
      "learning_rate": 3.902924925781049e-07,
      "loss": 0.178,
      "step": 29827
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4491891490234709,
      "learning_rate": 3.900181352874277e-07,
      "loss": 0.1902,
      "step": 29828
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30148577733627546,
      "learning_rate": 3.8974387254291144e-07,
      "loss": 0.1797,
      "step": 29829
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.097385263240148,
      "learning_rate": 3.8946970434726063e-07,
      "loss": 0.5624,
      "step": 29830
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7512119282929637,
      "learning_rate": 3.8919563070316747e-07,
      "loss": 0.3249,
      "step": 29831
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3981423925881241,
      "learning_rate": 3.8892165161332986e-07,
      "loss": 0.1666,
      "step": 29832
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3928118739842193,
      "learning_rate": 3.886477670804445e-07,
      "loss": 0.2926,
      "step": 29833
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4186997873012552,
      "learning_rate": 3.8837397710720483e-07,
      "loss": 0.2219,
      "step": 29834
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.894925272168456,
      "learning_rate": 3.881002816963031e-07,
      "loss": 0.712,
      "step": 29835
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.19646882272071892,
      "learning_rate": 3.87826680850435e-07,
      "loss": 0.0681,
      "step": 29836
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.34693273076491243,
      "learning_rate": 3.875531745722916e-07,
      "loss": 0.2202,
      "step": 29837
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2689852558659061,
      "learning_rate": 3.872797628645608e-07,
      "loss": 0.1478,
      "step": 29838
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5053516461820806,
      "learning_rate": 3.870064457299361e-07,
      "loss": 0.3689,
      "step": 29839
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.0572015413030609,
      "learning_rate": 3.8673322317110517e-07,
      "loss": 0.2814,
      "step": 29840
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.440046521846945,
      "learning_rate": 3.864600951907538e-07,
      "loss": 0.2723,
      "step": 29841
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.43669586375332886,
      "learning_rate": 3.861870617915742e-07,
      "loss": 0.2292,
      "step": 29842
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9500689888531703,
      "learning_rate": 3.859141229762475e-07,
      "loss": 0.0345,
      "step": 29843
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5919310344323343,
      "learning_rate": 3.8564127874746283e-07,
      "loss": 0.3237,
      "step": 29844
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.36744293916842397,
      "learning_rate": 3.853685291079012e-07,
      "loss": 0.1779,
      "step": 29845
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.40426831210705116,
      "learning_rate": 3.850958740602484e-07,
      "loss": 0.2939,
      "step": 29846
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.17945492333185561,
      "learning_rate": 3.848233136071866e-07,
      "loss": 0.0879,
      "step": 29847
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.646859544743222,
      "learning_rate": 3.8455084775139593e-07,
      "loss": 0.7993,
      "step": 29848
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7031449778554476,
      "learning_rate": 3.842784764955587e-07,
      "loss": 0.2645,
      "step": 29849
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.4219574885559135,
      "learning_rate": 3.8400619984235276e-07,
      "loss": 0.2344,
      "step": 29850
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.2836553374603035,
      "learning_rate": 3.837340177944593e-07,
      "loss": 0.219,
      "step": 29851
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8737513297089925,
      "learning_rate": 3.8346193035455394e-07,
      "loss": 0.4041,
      "step": 29852
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7566197580948408,
      "learning_rate": 3.8318993752531455e-07,
      "loss": 0.0245,
      "step": 29853
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.8359485283145902,
      "learning_rate": 3.829180393094178e-07,
      "loss": 0.3413,
      "step": 29854
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.41931052698174903,
      "learning_rate": 3.826462357095351e-07,
      "loss": 0.1645,
      "step": 29855
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.20671685402398912,
      "learning_rate": 3.823745267283463e-07,
      "loss": 0.1228,
      "step": 29856
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3255249773256333,
      "learning_rate": 3.8210291236852047e-07,
      "loss": 0.2583,
      "step": 29857
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5764650195142956,
      "learning_rate": 3.8183139263272995e-07,
      "loss": 0.2442,
      "step": 29858
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.944801343610757,
      "learning_rate": 3.815599675236481e-07,
      "loss": 0.3672,
      "step": 29859
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.37356593882120853,
      "learning_rate": 3.8128863704394394e-07,
      "loss": 0.228,
      "step": 29860
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.478619194696662,
      "learning_rate": 3.810174011962864e-07,
      "loss": 0.543,
      "step": 29861
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3345715897958372,
      "learning_rate": 3.807462599833456e-07,
      "loss": 0.2144,
      "step": 29862
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.7398162694973691,
      "learning_rate": 3.8047521340778827e-07,
      "loss": 0.6203,
      "step": 29863
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.27384130779034255,
      "learning_rate": 3.802042614722812e-07,
      "loss": 0.1708,
      "step": 29864
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.30400760436210333,
      "learning_rate": 3.799334041794889e-07,
      "loss": 0.1547,
      "step": 29865
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6561636624245145,
      "learning_rate": 3.796626415320792e-07,
      "loss": 0.2626,
      "step": 29866
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.9443588248800954,
      "learning_rate": 3.7939197353271005e-07,
      "loss": 0.382,
      "step": 29867
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3159548448427711,
      "learning_rate": 3.7912140018405266e-07,
      "loss": 0.1856,
      "step": 29868
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3147505580037784,
      "learning_rate": 3.7885092148876145e-07,
      "loss": 0.227,
      "step": 29869
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.2818773218172288,
      "learning_rate": 3.785805374495022e-07,
      "loss": 0.51,
      "step": 29870
    },
    {
      "epoch": 0.91,
      "grad_norm": 1.8576765574235696,
      "learning_rate": 3.7831024806893377e-07,
      "loss": 0.1487,
      "step": 29871
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.7563361603952206,
      "learning_rate": 3.7804005334971306e-07,
      "loss": 0.267,
      "step": 29872
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.31782223381572233,
      "learning_rate": 3.7776995329450225e-07,
      "loss": 0.1811,
      "step": 29873
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.33226247739015846,
      "learning_rate": 3.77499947905956e-07,
      "loss": 0.2263,
      "step": 29874
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.5002426535147775,
      "learning_rate": 3.7723003718673214e-07,
      "loss": 0.2299,
      "step": 29875
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.6921382411047136,
      "learning_rate": 3.7696022113948517e-07,
      "loss": 0.3685,
      "step": 29876
    },
    {
      "epoch": 0.91,
      "grad_norm": 0.3649503559545466,
      "learning_rate": 3.766904997668708e-07,
      "loss": 0.1724,
      "step": 29877
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5894964202164373,
      "learning_rate": 3.7642087307154016e-07,
      "loss": 0.3435,
      "step": 29878
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3001738224863622,
      "learning_rate": 3.7615134105614795e-07,
      "loss": 0.0882,
      "step": 29879
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.33202098735284785,
      "learning_rate": 3.758819037233463e-07,
      "loss": 0.2679,
      "step": 29880
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5648621890365028,
      "learning_rate": 3.7561256107578323e-07,
      "loss": 0.0555,
      "step": 29881
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3783360664548203,
      "learning_rate": 3.7534331311611326e-07,
      "loss": 0.1663,
      "step": 29882
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3334199957552867,
      "learning_rate": 3.7507415984698205e-07,
      "loss": 0.19,
      "step": 29883
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5764946916203242,
      "learning_rate": 3.7480510127103744e-07,
      "loss": 0.2375,
      "step": 29884
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.49534250544993574,
      "learning_rate": 3.745361373909284e-07,
      "loss": 0.3526,
      "step": 29885
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3580867867039774,
      "learning_rate": 3.7426726820929847e-07,
      "loss": 0.0695,
      "step": 29886
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.29022630374693287,
      "learning_rate": 3.739984937287955e-07,
      "loss": 0.2482,
      "step": 29887
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3004657098232797,
      "learning_rate": 3.7372981395206176e-07,
      "loss": 0.1271,
      "step": 29888
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4899359444000382,
      "learning_rate": 3.7346122888174406e-07,
      "loss": 0.4714,
      "step": 29889
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0323802023776598,
      "learning_rate": 3.731927385204803e-07,
      "loss": 0.0567,
      "step": 29890
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42344116553379446,
      "learning_rate": 3.7292434287091394e-07,
      "loss": 0.2501,
      "step": 29891
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35730772915257564,
      "learning_rate": 3.7265604193568727e-07,
      "loss": 0.2281,
      "step": 29892
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.46929720954580817,
      "learning_rate": 3.723878357174349e-07,
      "loss": 0.319,
      "step": 29893
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9817795432777026,
      "learning_rate": 3.721197242188024e-07,
      "loss": 0.2799,
      "step": 29894
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3485596556077588,
      "learning_rate": 3.718517074424222e-07,
      "loss": 0.1546,
      "step": 29895
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3803922803933454,
      "learning_rate": 3.715837853909332e-07,
      "loss": 0.266,
      "step": 29896
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2219336236525375,
      "learning_rate": 3.7131595806697227e-07,
      "loss": 0.0618,
      "step": 29897
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5258363504385203,
      "learning_rate": 3.7104822547317065e-07,
      "loss": 0.353,
      "step": 29898
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38976838435560507,
      "learning_rate": 3.707805876121673e-07,
      "loss": 0.1999,
      "step": 29899
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.396783414640883,
      "learning_rate": 3.705130444865923e-07,
      "loss": 0.2542,
      "step": 29900
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4913386497460499,
      "learning_rate": 3.7024559609907806e-07,
      "loss": 0.236,
      "step": 29901
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1175323814059837,
      "learning_rate": 3.69978242452258e-07,
      "loss": 0.4525,
      "step": 29902
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4500007863954117,
      "learning_rate": 3.697109835487589e-07,
      "loss": 0.2528,
      "step": 29903
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37233155153604314,
      "learning_rate": 3.69443819391212e-07,
      "loss": 0.2703,
      "step": 29904
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3672090498554309,
      "learning_rate": 3.691767499822452e-07,
      "loss": 0.2152,
      "step": 29905
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4550575914726167,
      "learning_rate": 3.689097753244886e-07,
      "loss": 0.7915,
      "step": 29906
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21428903773930377,
      "learning_rate": 3.6864289542056564e-07,
      "loss": 0.0678,
      "step": 29907
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6517190752748254,
      "learning_rate": 3.6837611027310205e-07,
      "loss": 0.1851,
      "step": 29908
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4618790700772348,
      "learning_rate": 3.681094198847246e-07,
      "loss": 0.2717,
      "step": 29909
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3282045368256772,
      "learning_rate": 3.6784282425805453e-07,
      "loss": 0.2174,
      "step": 29910
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4632227812443227,
      "learning_rate": 3.6757632339571856e-07,
      "loss": 0.3156,
      "step": 29911
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1919148172247231,
      "learning_rate": 3.6730991730033474e-07,
      "loss": 0.5172,
      "step": 29912
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8867850921756095,
      "learning_rate": 3.6704360597452526e-07,
      "loss": 0.3591,
      "step": 29913
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3038327081404594,
      "learning_rate": 3.667773894209103e-07,
      "loss": 0.173,
      "step": 29914
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36013575121712116,
      "learning_rate": 3.665112676421101e-07,
      "loss": 0.1817,
      "step": 29915
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3312332054454795,
      "learning_rate": 3.662452406407413e-07,
      "loss": 0.2119,
      "step": 29916
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2011258393136608,
      "learning_rate": 3.6597930841942184e-07,
      "loss": 0.4548,
      "step": 29917
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41410083735582237,
      "learning_rate": 3.657134709807686e-07,
      "loss": 0.169,
      "step": 29918
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5297887106082207,
      "learning_rate": 3.6544772832739496e-07,
      "loss": 0.3241,
      "step": 29919
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5810591931662947,
      "learning_rate": 3.6518208046191774e-07,
      "loss": 0.2413,
      "step": 29920
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.41084976385311095,
      "learning_rate": 3.649165273869493e-07,
      "loss": 0.2239,
      "step": 29921
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.43767529636634117,
      "learning_rate": 3.6465106910510195e-07,
      "loss": 0.2834,
      "step": 29922
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.31482693677555207,
      "learning_rate": 3.643857056189892e-07,
      "loss": 0.1904,
      "step": 29923
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5485951952724261,
      "learning_rate": 3.64120436931219e-07,
      "loss": 0.3002,
      "step": 29924
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.30433697262594844,
      "learning_rate": 3.6385526304440364e-07,
      "loss": 0.0677,
      "step": 29925
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.662972220347793,
      "learning_rate": 3.6359018396114995e-07,
      "loss": 0.3786,
      "step": 29926
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37752159816080605,
      "learning_rate": 3.63325199684067e-07,
      "loss": 0.1723,
      "step": 29927
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.29278623269858134,
      "learning_rate": 3.630603102157626e-07,
      "loss": 0.2564,
      "step": 29928
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4011946997077285,
      "learning_rate": 3.627955155588414e-07,
      "loss": 0.0519,
      "step": 29929
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6018045923369097,
      "learning_rate": 3.6253081571590796e-07,
      "loss": 0.8798,
      "step": 29930
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6887769002528268,
      "learning_rate": 3.6226621068956803e-07,
      "loss": 0.1649,
      "step": 29931
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.342493283003947,
      "learning_rate": 3.62001700482425e-07,
      "loss": 0.249,
      "step": 29932
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2099556311870111,
      "learning_rate": 3.6173728509708015e-07,
      "loss": 0.0647,
      "step": 29933
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3208798424681166,
      "learning_rate": 3.6147296453613477e-07,
      "loss": 0.1899,
      "step": 29934
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6594943516282143,
      "learning_rate": 3.612087388021912e-07,
      "loss": 0.3304,
      "step": 29935
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36054985727043454,
      "learning_rate": 3.609446078978451e-07,
      "loss": 0.1554,
      "step": 29936
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5740209763983253,
      "learning_rate": 3.6068057182570114e-07,
      "loss": 0.3045,
      "step": 29937
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2830844077734698,
      "learning_rate": 3.604166305883505e-07,
      "loss": 0.1681,
      "step": 29938
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5457434414474546,
      "learning_rate": 3.601527841883945e-07,
      "loss": 0.3512,
      "step": 29939
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3279620579703506,
      "learning_rate": 3.598890326284277e-07,
      "loss": 0.1781,
      "step": 29940
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.48570909841293497,
      "learning_rate": 3.596253759110435e-07,
      "loss": 0.3,
      "step": 29941
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5285893776936647,
      "learning_rate": 3.5936181403883664e-07,
      "loss": 0.1648,
      "step": 29942
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6884561243874063,
      "learning_rate": 3.590983470144005e-07,
      "loss": 0.3435,
      "step": 29943
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4119158681183349,
      "learning_rate": 3.588349748403286e-07,
      "loss": 0.1377,
      "step": 29944
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.26935314031834423,
      "learning_rate": 3.5857169751920883e-07,
      "loss": 0.1889,
      "step": 29945
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24825633744415426,
      "learning_rate": 3.583085150536347e-07,
      "loss": 0.2027,
      "step": 29946
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6440036144965744,
      "learning_rate": 3.5804542744619196e-07,
      "loss": 0.2736,
      "step": 29947
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6590672429073658,
      "learning_rate": 3.577824346994718e-07,
      "loss": 0.7762,
      "step": 29948
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4423241948971824,
      "learning_rate": 3.575195368160622e-07,
      "loss": 0.1833,
      "step": 29949
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3481327005578447,
      "learning_rate": 3.572567337985455e-07,
      "loss": 0.2257,
      "step": 29950
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4414411705024834,
      "learning_rate": 3.5699402564951303e-07,
      "loss": 0.2129,
      "step": 29951
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4752864521502684,
      "learning_rate": 3.5673141237154484e-07,
      "loss": 0.304,
      "step": 29952
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4687364343939193,
      "learning_rate": 3.564688939672256e-07,
      "loss": 0.1812,
      "step": 29953
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37437990390692816,
      "learning_rate": 3.5620647043913993e-07,
      "loss": 0.1933,
      "step": 29954
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3681279805722674,
      "learning_rate": 3.559441417898668e-07,
      "loss": 0.1761,
      "step": 29955
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7564644088335093,
      "learning_rate": 3.5568190802198866e-07,
      "loss": 0.6756,
      "step": 29956
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.46462376339718986,
      "learning_rate": 3.5541976913808453e-07,
      "loss": 0.2697,
      "step": 29957
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5581193272517687,
      "learning_rate": 3.5515772514073677e-07,
      "loss": 0.3931,
      "step": 29958
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.28283810240114776,
      "learning_rate": 3.548957760325189e-07,
      "loss": 0.1665,
      "step": 29959
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1141367708697472,
      "learning_rate": 3.5463392181600997e-07,
      "loss": 0.255,
      "step": 29960
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6597493762302682,
      "learning_rate": 3.5437216249378794e-07,
      "loss": 0.3553,
      "step": 29961
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.27546318291662436,
      "learning_rate": 3.54110498068424e-07,
      "loss": 0.1366,
      "step": 29962
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3807068950658958,
      "learning_rate": 3.538489285424973e-07,
      "loss": 0.2512,
      "step": 29963
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3604362114018157,
      "learning_rate": 3.5358745391857685e-07,
      "loss": 0.2051,
      "step": 29964
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.45206833723319095,
      "learning_rate": 3.5332607419923835e-07,
      "loss": 0.24,
      "step": 29965
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2307607019658526,
      "learning_rate": 3.5306478938705204e-07,
      "loss": 0.5625,
      "step": 29966
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.075835662600225,
      "learning_rate": 3.52803599484588e-07,
      "loss": 0.4163,
      "step": 29967
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3319338540007538,
      "learning_rate": 3.525425044944164e-07,
      "loss": 0.1423,
      "step": 29968
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5091960218544552,
      "learning_rate": 3.522815044191064e-07,
      "loss": 0.3312,
      "step": 29969
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3241868212739384,
      "learning_rate": 3.5202059926122687e-07,
      "loss": 0.2204,
      "step": 29970
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.913486275548887,
      "learning_rate": 3.517597890233415e-07,
      "loss": 0.6531,
      "step": 29971
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3207819684654746,
      "learning_rate": 3.514990737080182e-07,
      "loss": 0.0681,
      "step": 29972
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24268495447855776,
      "learning_rate": 3.512384533178215e-07,
      "loss": 0.1499,
      "step": 29973
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3853862102957537,
      "learning_rate": 3.50977927855316e-07,
      "loss": 0.6213,
      "step": 29974
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4307852984544759,
      "learning_rate": 3.507174973230665e-07,
      "loss": 0.2384,
      "step": 29975
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5255219124192804,
      "learning_rate": 3.5045716172362967e-07,
      "loss": 0.299,
      "step": 29976
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4161285506885222,
      "learning_rate": 3.5019692105957347e-07,
      "loss": 0.1706,
      "step": 29977
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5304100334202805,
      "learning_rate": 3.499367753334548e-07,
      "loss": 0.3258,
      "step": 29978
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9806233752761548,
      "learning_rate": 3.4967672454783165e-07,
      "loss": 0.1867,
      "step": 29979
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2208343056876148,
      "learning_rate": 3.494167687052663e-07,
      "loss": 0.6438,
      "step": 29980
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21718586945618798,
      "learning_rate": 3.4915690780831347e-07,
      "loss": 0.1376,
      "step": 29981
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.31425339978286865,
      "learning_rate": 3.488971418595299e-07,
      "loss": 0.2667,
      "step": 29982
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1871568267502899,
      "learning_rate": 3.486374708614715e-07,
      "loss": 0.0648,
      "step": 29983
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.418891082441518,
      "learning_rate": 3.48377894816696e-07,
      "loss": 0.7588,
      "step": 29984
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6996578368066111,
      "learning_rate": 3.481184137277516e-07,
      "loss": 0.2757,
      "step": 29985
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.28592174485737915,
      "learning_rate": 3.4785902759719604e-07,
      "loss": 0.1788,
      "step": 29986
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9488374128970998,
      "learning_rate": 3.475997364275796e-07,
      "loss": 0.4369,
      "step": 29987
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.295332819086542,
      "learning_rate": 3.4734054022145024e-07,
      "loss": 0.2019,
      "step": 29988
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2539404296464711,
      "learning_rate": 3.4708143898136373e-07,
      "loss": 0.3819,
      "step": 29989
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.27007264701599576,
      "learning_rate": 3.468224327098646e-07,
      "loss": 0.0741,
      "step": 29990
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5757235099078295,
      "learning_rate": 3.465635214095031e-07,
      "loss": 0.3183,
      "step": 29991
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2544986501127547,
      "learning_rate": 3.463047050828272e-07,
      "loss": 0.1564,
      "step": 29992
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35895954720334455,
      "learning_rate": 3.460459837323804e-07,
      "loss": 0.2761,
      "step": 29993
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6465493149040182,
      "learning_rate": 3.457873573607107e-07,
      "loss": 0.2616,
      "step": 29994
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8369417092897957,
      "learning_rate": 3.4552882597036044e-07,
      "loss": 0.3456,
      "step": 29995
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2987253930598213,
      "learning_rate": 3.452703895638765e-07,
      "loss": 0.1841,
      "step": 29996
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5393429283922522,
      "learning_rate": 3.45012048143798e-07,
      "loss": 0.5481,
      "step": 29997
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4920592607337186,
      "learning_rate": 3.4475380171266725e-07,
      "loss": 0.1074,
      "step": 29998
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.290466901108866,
      "learning_rate": 3.4449565027302677e-07,
      "loss": 0.2074,
      "step": 29999
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.336751428576136,
      "learning_rate": 3.442375938274145e-07,
      "loss": 0.2432,
      "step": 30000
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1669986778846073,
      "learning_rate": 3.439796323783706e-07,
      "loss": 0.0969,
      "step": 30001
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0314049117373467,
      "learning_rate": 3.4372176592843197e-07,
      "loss": 0.5007,
      "step": 30002
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6614206047500704,
      "learning_rate": 3.4346399448013545e-07,
      "loss": 0.274,
      "step": 30003
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6959406160677714,
      "learning_rate": 3.43206318036019e-07,
      "loss": 0.3353,
      "step": 30004
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25007542703316943,
      "learning_rate": 3.4294873659861396e-07,
      "loss": 0.2089,
      "step": 30005
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6902373881316335,
      "learning_rate": 3.426912501704582e-07,
      "loss": 0.6905,
      "step": 30006
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1933018357781924,
      "learning_rate": 3.424338587540832e-07,
      "loss": 0.06,
      "step": 30007
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5633742862729735,
      "learning_rate": 3.421765623520223e-07,
      "loss": 0.1775,
      "step": 30008
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2840684543064462,
      "learning_rate": 3.41919360966807e-07,
      "loss": 0.1643,
      "step": 30009
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7796892259698829,
      "learning_rate": 3.4166225460096514e-07,
      "loss": 0.7219,
      "step": 30010
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3451110286685058,
      "learning_rate": 3.4140524325702917e-07,
      "loss": 0.23,
      "step": 30011
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6295691851185674,
      "learning_rate": 3.41148326937526e-07,
      "loss": 0.2751,
      "step": 30012
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.44624660304989816,
      "learning_rate": 3.4089150564498466e-07,
      "loss": 0.2377,
      "step": 30013
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5194852632133471,
      "learning_rate": 3.4063477938192866e-07,
      "loss": 0.2312,
      "step": 30014
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4769073809751473,
      "learning_rate": 3.403781481508883e-07,
      "loss": 0.2527,
      "step": 30015
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.49210516198147897,
      "learning_rate": 3.4012161195438487e-07,
      "loss": 0.2338,
      "step": 30016
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3847007071451682,
      "learning_rate": 3.3986517079494406e-07,
      "loss": 0.227,
      "step": 30017
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37120942988182454,
      "learning_rate": 3.3960882467508947e-07,
      "loss": 0.1693,
      "step": 30018
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5434179705889552,
      "learning_rate": 3.393525735973402e-07,
      "loss": 0.2993,
      "step": 30019
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7006332933902732,
      "learning_rate": 3.3909641756421975e-07,
      "loss": 0.2723,
      "step": 30020
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.289161803232096,
      "learning_rate": 3.388403565782472e-07,
      "loss": 0.7429,
      "step": 30021
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21518605817806916,
      "learning_rate": 3.385843906419439e-07,
      "loss": 0.0695,
      "step": 30022
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3458890896467662,
      "learning_rate": 3.3832851975782456e-07,
      "loss": 0.27,
      "step": 30023
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24470618916131093,
      "learning_rate": 3.3807274392840815e-07,
      "loss": 0.1746,
      "step": 30024
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2777656203436267,
      "learning_rate": 3.3781706315621165e-07,
      "loss": 0.1074,
      "step": 30025
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7466884323462313,
      "learning_rate": 3.375614774437508e-07,
      "loss": 0.5322,
      "step": 30026
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3402962260710566,
      "learning_rate": 3.373059867935402e-07,
      "loss": 0.1646,
      "step": 30027
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.561600702354059,
      "learning_rate": 3.3705059120809126e-07,
      "loss": 0.2878,
      "step": 30028
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.310054352256194,
      "learning_rate": 3.3679529068991857e-07,
      "loss": 0.2163,
      "step": 30029
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9880450891834606,
      "learning_rate": 3.3654008524153347e-07,
      "loss": 0.5042,
      "step": 30030
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3340512496632041,
      "learning_rate": 3.362849748654451e-07,
      "loss": 0.1533,
      "step": 30031
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.29073298080666965,
      "learning_rate": 3.360299595641681e-07,
      "loss": 0.2069,
      "step": 30032
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21899658559175936,
      "learning_rate": 3.357750393402059e-07,
      "loss": 0.0667,
      "step": 30033
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0307865571948343,
      "learning_rate": 3.355202141960701e-07,
      "loss": 0.8021,
      "step": 30034
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.30970168203687015,
      "learning_rate": 3.352654841342662e-07,
      "loss": 0.2063,
      "step": 30035
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38395695821416764,
      "learning_rate": 3.350108491573001e-07,
      "loss": 0.2463,
      "step": 30036
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5216985766025993,
      "learning_rate": 3.3475630926767757e-07,
      "loss": 0.2206,
      "step": 30037
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6894107757863654,
      "learning_rate": 3.3450186446790213e-07,
      "loss": 0.3326,
      "step": 30038
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1103159617115324,
      "learning_rate": 3.342475147604796e-07,
      "loss": 0.5589,
      "step": 30039
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35277819524401516,
      "learning_rate": 3.33993260147909e-07,
      "loss": 0.1589,
      "step": 30040
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6114948822586801,
      "learning_rate": 3.337391006326929e-07,
      "loss": 0.3527,
      "step": 30041
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.18930290136363181,
      "learning_rate": 3.334850362173336e-07,
      "loss": 0.1425,
      "step": 30042
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4025879716270606,
      "learning_rate": 3.33231066904327e-07,
      "loss": 0.5188,
      "step": 30043
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6469772797767774,
      "learning_rate": 3.329771926961767e-07,
      "loss": 0.1953,
      "step": 30044
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0476141824707381,
      "learning_rate": 3.327234135953761e-07,
      "loss": 0.3949,
      "step": 30045
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4076951627019662,
      "learning_rate": 3.3246972960442326e-07,
      "loss": 0.2225,
      "step": 30046
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.344253064260143,
      "learning_rate": 3.322161407258151e-07,
      "loss": 0.2821,
      "step": 30047
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1126639163036411,
      "learning_rate": 3.319626469620452e-07,
      "loss": 0.5493,
      "step": 30048
    },
    {
      "epoch": 0.92,
      "grad_norm": 2.0552473258402912,
      "learning_rate": 3.3170924831560815e-07,
      "loss": 0.7793,
      "step": 30049
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.28956846579085127,
      "learning_rate": 3.3145594478899646e-07,
      "loss": 0.1686,
      "step": 30050
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2983261555331947,
      "learning_rate": 3.312027363847026e-07,
      "loss": 0.1738,
      "step": 30051
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35683834977301127,
      "learning_rate": 3.3094962310521784e-07,
      "loss": 0.2215,
      "step": 30052
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3586165489901294,
      "learning_rate": 3.306966049530336e-07,
      "loss": 0.2773,
      "step": 30053
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.581990725684451,
      "learning_rate": 3.304436819306356e-07,
      "loss": 0.2593,
      "step": 30054
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3577104210984816,
      "learning_rate": 3.301908540405152e-07,
      "loss": 0.2195,
      "step": 30055
    },
    {
      "epoch": 0.92,
      "grad_norm": 3.8332478691540715,
      "learning_rate": 3.299381212851593e-07,
      "loss": 0.6436,
      "step": 30056
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2903303639926182,
      "learning_rate": 3.2968548366705264e-07,
      "loss": 0.5547,
      "step": 30057
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.48155069310668464,
      "learning_rate": 3.294329411886843e-07,
      "loss": 0.262,
      "step": 30058
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2623234975057334,
      "learning_rate": 3.2918049385253557e-07,
      "loss": 0.1975,
      "step": 30059
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4092737922621398,
      "learning_rate": 3.289281416610912e-07,
      "loss": 0.1606,
      "step": 30060
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2542698527000382,
      "learning_rate": 3.2867588461683475e-07,
      "loss": 0.0623,
      "step": 30061
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6803131429119009,
      "learning_rate": 3.2842372272224644e-07,
      "loss": 0.3095,
      "step": 30062
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3488030778028311,
      "learning_rate": 3.281716559798076e-07,
      "loss": 0.1566,
      "step": 30063
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5716930732573626,
      "learning_rate": 3.279196843919985e-07,
      "loss": 0.3076,
      "step": 30064
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.312025353838678,
      "learning_rate": 3.276678079612994e-07,
      "loss": 0.2,
      "step": 30065
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.987069879256774,
      "learning_rate": 3.2741602669018604e-07,
      "loss": 0.552,
      "step": 30066
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9332463918745753,
      "learning_rate": 3.2716434058113643e-07,
      "loss": 0.4261,
      "step": 30067
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.40300274977259826,
      "learning_rate": 3.2691274963662755e-07,
      "loss": 0.1656,
      "step": 30068
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37044635834372297,
      "learning_rate": 3.2666125385913295e-07,
      "loss": 0.258,
      "step": 30069
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.25453459696063657,
      "learning_rate": 3.264098532511284e-07,
      "loss": 0.1902,
      "step": 30070
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7686172689446665,
      "learning_rate": 3.261585478150864e-07,
      "loss": 0.327,
      "step": 30071
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7885114335818258,
      "learning_rate": 3.2590733755348046e-07,
      "loss": 0.0912,
      "step": 30072
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3502245076094345,
      "learning_rate": 3.2565622246878093e-07,
      "loss": 0.2322,
      "step": 30073
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3815919551451876,
      "learning_rate": 3.25405202563458e-07,
      "loss": 0.2668,
      "step": 30074
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3322891684514249,
      "learning_rate": 3.25154277839983e-07,
      "loss": 0.5527,
      "step": 30075
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.32612574492589397,
      "learning_rate": 3.2490344830082287e-07,
      "loss": 0.2353,
      "step": 30076
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42443992508129075,
      "learning_rate": 3.2465271394844675e-07,
      "loss": 0.2509,
      "step": 30077
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.29736721532313815,
      "learning_rate": 3.244020747853216e-07,
      "loss": 0.1413,
      "step": 30078
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.48877336083330725,
      "learning_rate": 3.2415153081391205e-07,
      "loss": 0.2019,
      "step": 30079
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.0089039289489388,
      "learning_rate": 3.239010820366828e-07,
      "loss": 0.4305,
      "step": 30080
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5205387929351878,
      "learning_rate": 3.2365072845609856e-07,
      "loss": 0.0683,
      "step": 30081
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38879402003903,
      "learning_rate": 3.234004700746229e-07,
      "loss": 0.2716,
      "step": 30082
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3216161550802723,
      "learning_rate": 3.2315030689471617e-07,
      "loss": 0.2233,
      "step": 30083
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5303123669676049,
      "learning_rate": 3.2290023891884295e-07,
      "loss": 0.7904,
      "step": 30084
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9085987543971611,
      "learning_rate": 3.2265026614946013e-07,
      "loss": 0.1532,
      "step": 30085
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4132318528896599,
      "learning_rate": 3.224003885890281e-07,
      "loss": 0.2273,
      "step": 30086
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5060463100621936,
      "learning_rate": 3.2215060624000703e-07,
      "loss": 0.2139,
      "step": 30087
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5043141328479487,
      "learning_rate": 3.2190091910485054e-07,
      "loss": 0.3156,
      "step": 30088
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.337424905666972,
      "learning_rate": 3.216513271860189e-07,
      "loss": 0.1929,
      "step": 30089
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3197158508350738,
      "learning_rate": 3.2140183048596453e-07,
      "loss": 0.1683,
      "step": 30090
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3413965791560243,
      "learning_rate": 3.211524290071455e-07,
      "loss": 0.1858,
      "step": 30091
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9949339122895983,
      "learning_rate": 3.2090312275201207e-07,
      "loss": 0.2309,
      "step": 30092
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.192862571101847,
      "learning_rate": 3.2065391172301896e-07,
      "loss": 0.5245,
      "step": 30093
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.29183257039974647,
      "learning_rate": 3.2040479592261863e-07,
      "loss": 0.2109,
      "step": 30094
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9596194615674268,
      "learning_rate": 3.2015577535325917e-07,
      "loss": 0.3691,
      "step": 30095
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38668139075837205,
      "learning_rate": 3.1990685001739405e-07,
      "loss": 0.2102,
      "step": 30096
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6565882507091836,
      "learning_rate": 3.1965801991746925e-07,
      "loss": 0.3692,
      "step": 30097
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1820092671152186,
      "learning_rate": 3.1940928505593496e-07,
      "loss": 0.0645,
      "step": 30098
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.7804135609976242,
      "learning_rate": 3.1916064543523694e-07,
      "loss": 0.5252,
      "step": 30099
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.28843905934020697,
      "learning_rate": 3.189121010578211e-07,
      "loss": 0.1711,
      "step": 30100
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3545102868205379,
      "learning_rate": 3.186636519261355e-07,
      "loss": 0.2921,
      "step": 30101
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.026486615556418,
      "learning_rate": 3.184152980426214e-07,
      "loss": 0.4636,
      "step": 30102
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3640469981313348,
      "learning_rate": 3.181670394097225e-07,
      "loss": 0.5708,
      "step": 30103
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.534565629178281,
      "learning_rate": 3.1791887602988346e-07,
      "loss": 0.1502,
      "step": 30104
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36426515699681417,
      "learning_rate": 3.176708079055435e-07,
      "loss": 0.2137,
      "step": 30105
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3354522592079877,
      "learning_rate": 3.1742283503914394e-07,
      "loss": 0.2669,
      "step": 30106
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6450393642959986,
      "learning_rate": 3.171749574331251e-07,
      "loss": 0.1233,
      "step": 30107
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2590269991950748,
      "learning_rate": 3.16927175089925e-07,
      "loss": 0.145,
      "step": 30108
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3269565071943735,
      "learning_rate": 3.1667948801198165e-07,
      "loss": 0.1824,
      "step": 30109
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5698473931271721,
      "learning_rate": 3.16431896201731e-07,
      "loss": 0.7064,
      "step": 30110
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.2401497770946006,
      "learning_rate": 3.1618439966161096e-07,
      "loss": 0.4555,
      "step": 30111
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34128968496756706,
      "learning_rate": 3.159369983940519e-07,
      "loss": 0.2699,
      "step": 30112
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38839406852594516,
      "learning_rate": 3.1568969240149514e-07,
      "loss": 0.1853,
      "step": 30113
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4761394221344183,
      "learning_rate": 3.154424816863677e-07,
      "loss": 0.3005,
      "step": 30114
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1414883260653677,
      "learning_rate": 3.151953662511031e-07,
      "loss": 0.2504,
      "step": 30115
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6747185512729104,
      "learning_rate": 3.1494834609813505e-07,
      "loss": 0.5222,
      "step": 30116
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.19204398402340847,
      "learning_rate": 3.1470142122989157e-07,
      "loss": 0.119,
      "step": 30117
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3147575744747023,
      "learning_rate": 3.1445459164880176e-07,
      "loss": 0.1782,
      "step": 30118
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5436553451529423,
      "learning_rate": 3.1420785735729487e-07,
      "loss": 0.3259,
      "step": 30119
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.045443132029343,
      "learning_rate": 3.139612183577989e-07,
      "loss": 0.5616,
      "step": 30120
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6385089790338648,
      "learning_rate": 3.1371467465273755e-07,
      "loss": 0.3267,
      "step": 30121
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3513667791255925,
      "learning_rate": 3.134682262445399e-07,
      "loss": 0.0667,
      "step": 30122
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38951910634365566,
      "learning_rate": 3.132218731356296e-07,
      "loss": 0.2628,
      "step": 30123
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3455560067226973,
      "learning_rate": 3.1297561532842804e-07,
      "loss": 0.2257,
      "step": 30124
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4158990953403918,
      "learning_rate": 3.1272945282536213e-07,
      "loss": 0.4705,
      "step": 30125
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.15597490757602436,
      "learning_rate": 3.1248338562884896e-07,
      "loss": 0.0697,
      "step": 30126
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4558536402028214,
      "learning_rate": 3.122374137413142e-07,
      "loss": 0.2461,
      "step": 30127
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4832431840800473,
      "learning_rate": 3.1199153716517385e-07,
      "loss": 0.2108,
      "step": 30128
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.47589458873068213,
      "learning_rate": 3.1174575590284917e-07,
      "loss": 0.3387,
      "step": 30129
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4721966303781943,
      "learning_rate": 3.115000699567583e-07,
      "loss": 0.2296,
      "step": 30130
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5236611830290675,
      "learning_rate": 3.1125447932931707e-07,
      "loss": 0.1969,
      "step": 30131
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.42206473770363573,
      "learning_rate": 3.1100898402294136e-07,
      "loss": 0.2521,
      "step": 30132
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4505759474403308,
      "learning_rate": 3.10763584040048e-07,
      "loss": 0.131,
      "step": 30133
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8895049813926386,
      "learning_rate": 3.105182793830519e-07,
      "loss": 0.7375,
      "step": 30134
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.24551045350767775,
      "learning_rate": 3.102730700543644e-07,
      "loss": 0.1784,
      "step": 30135
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3925253645993684,
      "learning_rate": 3.100279560563979e-07,
      "loss": 0.2276,
      "step": 30136
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2711441386668648,
      "learning_rate": 3.0978293739156615e-07,
      "loss": 0.1609,
      "step": 30137
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5708382275566857,
      "learning_rate": 3.0953801406227614e-07,
      "loss": 0.7144,
      "step": 30138
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6380576247386182,
      "learning_rate": 3.092931860709425e-07,
      "loss": 0.2608,
      "step": 30139
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6451734315888469,
      "learning_rate": 3.0904845341997003e-07,
      "loss": 0.3443,
      "step": 30140
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.37050512064845936,
      "learning_rate": 3.088038161117668e-07,
      "loss": 0.1951,
      "step": 30141
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5254576166538545,
      "learning_rate": 3.0855927414874197e-07,
      "loss": 0.292,
      "step": 30142
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.44725989650897097,
      "learning_rate": 3.083148275332992e-07,
      "loss": 0.2006,
      "step": 30143
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21534268111437257,
      "learning_rate": 3.0807047626784326e-07,
      "loss": 0.0755,
      "step": 30144
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4426578981301509,
      "learning_rate": 3.0782622035477993e-07,
      "loss": 0.2454,
      "step": 30145
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.4626751148885149,
      "learning_rate": 3.075820597965118e-07,
      "loss": 0.2317,
      "step": 30146
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6882242423688172,
      "learning_rate": 3.073379945954391e-07,
      "loss": 0.3472,
      "step": 30147
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.38819052152751893,
      "learning_rate": 3.0709402475396554e-07,
      "loss": 0.2428,
      "step": 30148
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9091613136108833,
      "learning_rate": 3.068501502744903e-07,
      "loss": 0.3737,
      "step": 30149
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3075449140611931,
      "learning_rate": 3.066063711594125e-07,
      "loss": 0.182,
      "step": 30150
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.686912635781829,
      "learning_rate": 3.063626874111325e-07,
      "loss": 0.8078,
      "step": 30151
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.372847794232582,
      "learning_rate": 3.0611909903204284e-07,
      "loss": 0.1223,
      "step": 30152
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3608905669581014,
      "learning_rate": 3.0587560602454604e-07,
      "loss": 0.2835,
      "step": 30153
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.31227162739204145,
      "learning_rate": 3.056322083910335e-07,
      "loss": 0.1483,
      "step": 30154
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34603527908631826,
      "learning_rate": 3.0538890613390105e-07,
      "loss": 0.224,
      "step": 30155
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6276997705258565,
      "learning_rate": 3.051456992555435e-07,
      "loss": 0.2734,
      "step": 30156
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.1006619506997504,
      "learning_rate": 3.0490258775835114e-07,
      "loss": 0.2861,
      "step": 30157
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3952184677843306,
      "learning_rate": 3.046595716447176e-07,
      "loss": 0.2158,
      "step": 30158
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.33308277212026466,
      "learning_rate": 3.044166509170332e-07,
      "loss": 0.1844,
      "step": 30159
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.35121985715485,
      "learning_rate": 3.041738255776894e-07,
      "loss": 0.2713,
      "step": 30160
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34232826771383035,
      "learning_rate": 3.039310956290731e-07,
      "loss": 0.0099,
      "step": 30161
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.4022360309939803,
      "learning_rate": 3.036884610735724e-07,
      "loss": 0.5152,
      "step": 30162
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3564536499664032,
      "learning_rate": 3.034459219135766e-07,
      "loss": 0.1662,
      "step": 30163
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5022444440651415,
      "learning_rate": 3.032034781514681e-07,
      "loss": 0.3035,
      "step": 30164
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8820638177409375,
      "learning_rate": 3.0296112978963733e-07,
      "loss": 0.2653,
      "step": 30165
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3476307244117542,
      "learning_rate": 3.027188768304645e-07,
      "loss": 0.2944,
      "step": 30166
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21534681684032958,
      "learning_rate": 3.024767192763334e-07,
      "loss": 0.1068,
      "step": 30167
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3467584339099606,
      "learning_rate": 3.0223465712962863e-07,
      "loss": 0.2691,
      "step": 30168
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.20641077069397074,
      "learning_rate": 3.019926903927295e-07,
      "loss": 0.0628,
      "step": 30169
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5117922525881573,
      "learning_rate": 3.017508190680163e-07,
      "loss": 0.0627,
      "step": 30170
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3600759638384058,
      "learning_rate": 3.0150904315787045e-07,
      "loss": 0.2529,
      "step": 30171
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.33902870091020176,
      "learning_rate": 3.0126736266467116e-07,
      "loss": 0.1516,
      "step": 30172
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5678873211453384,
      "learning_rate": 3.0102577759079433e-07,
      "loss": 0.3459,
      "step": 30173
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6855200122111571,
      "learning_rate": 3.007842879386158e-07,
      "loss": 0.2756,
      "step": 30174
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.3657732365919197,
      "learning_rate": 3.005428937105137e-07,
      "loss": 0.7121,
      "step": 30175
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.1521799595636231,
      "learning_rate": 3.0030159490886277e-07,
      "loss": 0.067,
      "step": 30176
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5142852281748689,
      "learning_rate": 3.0006039153603674e-07,
      "loss": 0.2855,
      "step": 30177
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.27025520951095117,
      "learning_rate": 2.9981928359440583e-07,
      "loss": 0.207,
      "step": 30178
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.36974036778903,
      "learning_rate": 2.995782710863471e-07,
      "loss": 0.4871,
      "step": 30179
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8824387696457541,
      "learning_rate": 2.993373540142286e-07,
      "loss": 0.2344,
      "step": 30180
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7413321702961071,
      "learning_rate": 2.990965323804196e-07,
      "loss": 0.3388,
      "step": 30181
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.34705691175094233,
      "learning_rate": 2.988558061872937e-07,
      "loss": 0.2012,
      "step": 30182
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.32184038587249414,
      "learning_rate": 2.986151754372146e-07,
      "loss": 0.2378,
      "step": 30183
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.5941847662112263,
      "learning_rate": 2.983746401325516e-07,
      "loss": 0.7356,
      "step": 30184
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.46433309771121234,
      "learning_rate": 2.981342002756715e-07,
      "loss": 0.2033,
      "step": 30185
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36290737013230745,
      "learning_rate": 2.978938558689404e-07,
      "loss": 0.2466,
      "step": 30186
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.21453313168291901,
      "learning_rate": 2.9765360691472065e-07,
      "loss": 0.0645,
      "step": 30187
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.9225504286973565,
      "learning_rate": 2.974134534153772e-07,
      "loss": 0.8564,
      "step": 30188
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.314879988567056,
      "learning_rate": 2.971733953732747e-07,
      "loss": 0.2193,
      "step": 30189
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.6585161229173807,
      "learning_rate": 2.969334327907703e-07,
      "loss": 0.2972,
      "step": 30190
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.36177022567151595,
      "learning_rate": 2.9669356567022854e-07,
      "loss": 0.2145,
      "step": 30191
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.6126457649664074,
      "learning_rate": 2.9645379401400885e-07,
      "loss": 0.6073,
      "step": 30192
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.9983875248766688,
      "learning_rate": 2.962141178244693e-07,
      "loss": 0.5491,
      "step": 30193
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.33561563003633904,
      "learning_rate": 2.959745371039691e-07,
      "loss": 0.2153,
      "step": 30194
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.2689319038421426,
      "learning_rate": 2.9573505185486294e-07,
      "loss": 0.1598,
      "step": 30195
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.278615593612096,
      "learning_rate": 2.954956620795091e-07,
      "loss": 0.1474,
      "step": 30196
    },
    {
      "epoch": 0.92,
      "grad_norm": 1.8298191482589248,
      "learning_rate": 2.9525636778026114e-07,
      "loss": 0.7628,
      "step": 30197
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5880530832763874,
      "learning_rate": 2.950171689594761e-07,
      "loss": 0.2619,
      "step": 30198
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5103696932828071,
      "learning_rate": 2.947780656195043e-07,
      "loss": 0.2567,
      "step": 30199
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.7669247473800079,
      "learning_rate": 2.9453905776269944e-07,
      "loss": 0.2002,
      "step": 30200
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.5311462851720413,
      "learning_rate": 2.9430014539141295e-07,
      "loss": 0.3408,
      "step": 30201
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.3994413374070547,
      "learning_rate": 2.9406132850799517e-07,
      "loss": 0.2563,
      "step": 30202
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.8146996917393197,
      "learning_rate": 2.9382260711479647e-07,
      "loss": 0.2759,
      "step": 30203
    },
    {
      "epoch": 0.92,
      "grad_norm": 0.39264734460181977,
      "learning_rate": 2.93583981214165e-07,
      "loss": 0.1794,
      "step": 30204
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.25359489420896425,
      "learning_rate": 2.9334545080844655e-07,
      "loss": 0.1785,
      "step": 30205
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8112480187792149,
      "learning_rate": 2.931070158999916e-07,
      "loss": 0.2501,
      "step": 30206
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.41393010191680807,
      "learning_rate": 2.9286867649114037e-07,
      "loss": 0.2932,
      "step": 30207
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6353208186128152,
      "learning_rate": 2.926304325842444e-07,
      "loss": 0.2035,
      "step": 30208
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34617772390169094,
      "learning_rate": 2.923922841816429e-07,
      "loss": 0.1931,
      "step": 30209
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5578579901608993,
      "learning_rate": 2.921542312856807e-07,
      "loss": 0.6193,
      "step": 30210
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2289609732958167,
      "learning_rate": 2.919162738987003e-07,
      "loss": 0.4405,
      "step": 30211
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5190710140373512,
      "learning_rate": 2.916784120230409e-07,
      "loss": 0.2829,
      "step": 30212
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2898700519564641,
      "learning_rate": 2.91440645661043e-07,
      "loss": 0.1816,
      "step": 30213
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3590142133651739,
      "learning_rate": 2.91202974815048e-07,
      "loss": 0.2391,
      "step": 30214
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5458820067087489,
      "learning_rate": 2.9096539948739287e-07,
      "loss": 0.1206,
      "step": 30215
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7647589329563178,
      "learning_rate": 2.907279196804125e-07,
      "loss": 0.3075,
      "step": 30216
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34136715162072195,
      "learning_rate": 2.9049053539644825e-07,
      "loss": 0.0701,
      "step": 30217
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3961454000738529,
      "learning_rate": 2.902532466378327e-07,
      "loss": 0.2596,
      "step": 30218
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43022838262761315,
      "learning_rate": 2.900160534068996e-07,
      "loss": 0.2223,
      "step": 30219
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5070530696667565,
      "learning_rate": 2.8977895570598493e-07,
      "loss": 0.3617,
      "step": 30220
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.042898805643786,
      "learning_rate": 2.895419535374189e-07,
      "loss": 0.1313,
      "step": 30221
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40894115991728003,
      "learning_rate": 2.8930504690353414e-07,
      "loss": 0.2518,
      "step": 30222
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.28589044103266564,
      "learning_rate": 2.89068235806661e-07,
      "loss": 0.1354,
      "step": 30223
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7533560587404755,
      "learning_rate": 2.8883152024913206e-07,
      "loss": 0.2715,
      "step": 30224
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2555923820838925,
      "learning_rate": 2.885949002332722e-07,
      "loss": 0.2258,
      "step": 30225
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5485718807554946,
      "learning_rate": 2.883583757614117e-07,
      "loss": 0.1038,
      "step": 30226
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5823096275882864,
      "learning_rate": 2.8812194683587644e-07,
      "loss": 0.3289,
      "step": 30227
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4651823998235743,
      "learning_rate": 2.878856134589936e-07,
      "loss": 0.2202,
      "step": 30228
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4737076311861486,
      "learning_rate": 2.876493756330889e-07,
      "loss": 0.743,
      "step": 30229
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2903111437747677,
      "learning_rate": 2.8741323336048397e-07,
      "loss": 0.2059,
      "step": 30230
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9938950648534809,
      "learning_rate": 2.871771866435036e-07,
      "loss": 0.3912,
      "step": 30231
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.362235467612256,
      "learning_rate": 2.869412354844714e-07,
      "loss": 0.2059,
      "step": 30232
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6468981690045689,
      "learning_rate": 2.867053798857056e-07,
      "loss": 0.3351,
      "step": 30233
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.28415683237659034,
      "learning_rate": 2.8646961984952983e-07,
      "loss": 0.1071,
      "step": 30234
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1718748908039347,
      "learning_rate": 2.8623395537826113e-07,
      "loss": 0.4978,
      "step": 30235
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3309369307580931,
      "learning_rate": 2.859983864742188e-07,
      "loss": 0.1763,
      "step": 30236
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.310523739482494,
      "learning_rate": 2.857629131397222e-07,
      "loss": 0.2282,
      "step": 30237
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3560697550680942,
      "learning_rate": 2.8552753537708477e-07,
      "loss": 0.788,
      "step": 30238
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2833686856291084,
      "learning_rate": 2.8529225318862487e-07,
      "loss": 0.0723,
      "step": 30239
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6147825162197933,
      "learning_rate": 2.85057066576655e-07,
      "loss": 0.2652,
      "step": 30240
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3305352585608052,
      "learning_rate": 2.8482197554349223e-07,
      "loss": 0.2176,
      "step": 30241
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9237813631068928,
      "learning_rate": 2.8458698009144693e-07,
      "loss": 0.5024,
      "step": 30242
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2400570048984614,
      "learning_rate": 2.8435208022283166e-07,
      "loss": 0.1823,
      "step": 30243
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5031400170109834,
      "learning_rate": 2.8411727593995796e-07,
      "loss": 0.2671,
      "step": 30244
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3037689055134603,
      "learning_rate": 2.8388256724513395e-07,
      "loss": 0.1699,
      "step": 30245
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5468402264645704,
      "learning_rate": 2.836479541406711e-07,
      "loss": 0.7932,
      "step": 30246
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.098287520142745,
      "learning_rate": 2.8341343662887764e-07,
      "loss": 0.4983,
      "step": 30247
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.450338814222105,
      "learning_rate": 2.8317901471205835e-07,
      "loss": 0.2978,
      "step": 30248
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2706145765654494,
      "learning_rate": 2.829446883925224e-07,
      "loss": 0.1759,
      "step": 30249
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5437966595170938,
      "learning_rate": 2.8271045767257363e-07,
      "loss": 0.2213,
      "step": 30250
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.406181224356772,
      "learning_rate": 2.8247632255451687e-07,
      "loss": 0.3635,
      "step": 30251
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3249360435616713,
      "learning_rate": 2.822422830406546e-07,
      "loss": 0.0748,
      "step": 30252
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9459888113481866,
      "learning_rate": 2.8200833913329173e-07,
      "loss": 0.4393,
      "step": 30253
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43950903241786554,
      "learning_rate": 2.817744908347275e-07,
      "loss": 0.1704,
      "step": 30254
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.28903036714543834,
      "learning_rate": 2.8154073814726456e-07,
      "loss": 0.2456,
      "step": 30255
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.110165576470094,
      "learning_rate": 2.8130708107320103e-07,
      "loss": 0.4955,
      "step": 30256
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8411788890347974,
      "learning_rate": 2.810735196148351e-07,
      "loss": 0.3279,
      "step": 30257
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5934109314193702,
      "learning_rate": 2.808400537744682e-07,
      "loss": 0.1675,
      "step": 30258
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34257956177825993,
      "learning_rate": 2.806066835543919e-07,
      "loss": 0.239,
      "step": 30259
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43561381339204236,
      "learning_rate": 2.8037340895690767e-07,
      "loss": 0.1987,
      "step": 30260
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.500478202806924,
      "learning_rate": 2.8014022998430703e-07,
      "loss": 0.316,
      "step": 30261
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.146661568853278,
      "learning_rate": 2.7990714663888587e-07,
      "loss": 0.0679,
      "step": 30262
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2725226746962857,
      "learning_rate": 2.796741589229368e-07,
      "loss": 0.166,
      "step": 30263
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5877301441657805,
      "learning_rate": 2.794412668387503e-07,
      "loss": 0.7752,
      "step": 30264
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1234411135413769,
      "learning_rate": 2.7920847038861997e-07,
      "loss": 0.2653,
      "step": 30265
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49285704148853077,
      "learning_rate": 2.7897576957483407e-07,
      "loss": 0.3124,
      "step": 30266
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3530689654386583,
      "learning_rate": 2.7874316439968516e-07,
      "loss": 0.1777,
      "step": 30267
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39248947503982257,
      "learning_rate": 2.7851065486545815e-07,
      "loss": 0.2649,
      "step": 30268
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6896559355921376,
      "learning_rate": 2.782782409744422e-07,
      "loss": 0.0216,
      "step": 30269
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2577031747712937,
      "learning_rate": 2.780459227289256e-07,
      "loss": 0.4476,
      "step": 30270
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20509228022150494,
      "learning_rate": 2.778137001311898e-07,
      "loss": 0.1354,
      "step": 30271
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43692642455876785,
      "learning_rate": 2.77581573183523e-07,
      "loss": 0.3,
      "step": 30272
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3427907017666021,
      "learning_rate": 2.773495418882077e-07,
      "loss": 0.2082,
      "step": 30273
    },
    {
      "epoch": 0.93,
      "grad_norm": 3.6094718229746747,
      "learning_rate": 2.771176062475267e-07,
      "loss": 0.7739,
      "step": 30274
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6566220179098688,
      "learning_rate": 2.768857662637625e-07,
      "loss": 0.2595,
      "step": 30275
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5486224632787515,
      "learning_rate": 2.7665402193919443e-07,
      "loss": 0.0728,
      "step": 30276
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5706121128836582,
      "learning_rate": 2.76422373276104e-07,
      "loss": 0.3499,
      "step": 30277
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3519094426247401,
      "learning_rate": 2.7619082027676934e-07,
      "loss": 0.2128,
      "step": 30278
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5257291240264732,
      "learning_rate": 2.759593629434687e-07,
      "loss": 0.3355,
      "step": 30279
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.20697253408040375,
      "learning_rate": 2.757280012784813e-07,
      "loss": 0.0889,
      "step": 30280
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9181465893548646,
      "learning_rate": 2.7549673528407983e-07,
      "loss": 0.4387,
      "step": 30281
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3601009377656692,
      "learning_rate": 2.752655649625413e-07,
      "loss": 0.2106,
      "step": 30282
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7716279949737684,
      "learning_rate": 2.750344903161406e-07,
      "loss": 0.3738,
      "step": 30283
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3128165581436527,
      "learning_rate": 2.7480351134715145e-07,
      "loss": 0.2291,
      "step": 30284
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4928998843290615,
      "learning_rate": 2.7457262805784315e-07,
      "loss": 0.1985,
      "step": 30285
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30135086279867346,
      "learning_rate": 2.743418404504905e-07,
      "loss": 0.1745,
      "step": 30286
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1792510406417887,
      "learning_rate": 2.7411114852736286e-07,
      "loss": 0.3393,
      "step": 30287
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4612074590012429,
      "learning_rate": 2.738805522907306e-07,
      "loss": 0.077,
      "step": 30288
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.111684957375529,
      "learning_rate": 2.7365005174286197e-07,
      "loss": 0.4337,
      "step": 30289
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3523792827657487,
      "learning_rate": 2.7341964688602395e-07,
      "loss": 0.2301,
      "step": 30290
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2609077718141862,
      "learning_rate": 2.731893377224837e-07,
      "loss": 0.2042,
      "step": 30291
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.663216528637633,
      "learning_rate": 2.729591242545071e-07,
      "loss": 0.3283,
      "step": 30292
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0941239912845764,
      "learning_rate": 2.727290064843602e-07,
      "loss": 0.2675,
      "step": 30293
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3401052230620596,
      "learning_rate": 2.724989844143056e-07,
      "loss": 0.1572,
      "step": 30294
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.314632127303931,
      "learning_rate": 2.722690580466059e-07,
      "loss": 0.1863,
      "step": 30295
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5368528904601071,
      "learning_rate": 2.720392273835259e-07,
      "loss": 0.3402,
      "step": 30296
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4198755706234401,
      "learning_rate": 2.718094924273218e-07,
      "loss": 0.2035,
      "step": 30297
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2106901998264097,
      "learning_rate": 2.7157985318026047e-07,
      "loss": 0.5093,
      "step": 30298
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.352488447063097,
      "learning_rate": 2.7135030964459574e-07,
      "loss": 0.166,
      "step": 30299
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5513511232793523,
      "learning_rate": 2.71120861822588e-07,
      "loss": 0.2965,
      "step": 30300
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8161237834981476,
      "learning_rate": 2.7089150971649546e-07,
      "loss": 0.2614,
      "step": 30301
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3119717771320827,
      "learning_rate": 2.706622533285708e-07,
      "loss": 0.2379,
      "step": 30302
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2639283295537252,
      "learning_rate": 2.704330926610754e-07,
      "loss": 0.1652,
      "step": 30303
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4400741329287332,
      "learning_rate": 2.7020402771625876e-07,
      "loss": 0.1651,
      "step": 30304
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5298954478464774,
      "learning_rate": 2.699750584963767e-07,
      "loss": 0.273,
      "step": 30305
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0085428034897732,
      "learning_rate": 2.697461850036831e-07,
      "loss": 0.114,
      "step": 30306
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4520267969828683,
      "learning_rate": 2.695174072404261e-07,
      "loss": 0.3257,
      "step": 30307
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3998643534504276,
      "learning_rate": 2.692887252088594e-07,
      "loss": 0.2034,
      "step": 30308
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36508702601269777,
      "learning_rate": 2.690601389112324e-07,
      "loss": 0.2797,
      "step": 30309
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6067281064733469,
      "learning_rate": 2.688316483497944e-07,
      "loss": 0.2686,
      "step": 30310
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6894730178903599,
      "learning_rate": 2.686032535267913e-07,
      "loss": 0.7052,
      "step": 30311
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.15112982509684372,
      "learning_rate": 2.6837495444447137e-07,
      "loss": 0.069,
      "step": 30312
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4297901351501071,
      "learning_rate": 2.681467511050828e-07,
      "loss": 0.2661,
      "step": 30313
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.27839533819166334,
      "learning_rate": 2.6791864351086493e-07,
      "loss": 0.2187,
      "step": 30314
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2390560076006139,
      "learning_rate": 2.676906316640693e-07,
      "loss": 0.0362,
      "step": 30315
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7363338381714692,
      "learning_rate": 2.67462715566934e-07,
      "loss": 0.7719,
      "step": 30316
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5960472117587032,
      "learning_rate": 2.67234895221703e-07,
      "loss": 0.2044,
      "step": 30317
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.44425074060116226,
      "learning_rate": 2.6700717063061763e-07,
      "loss": 0.2685,
      "step": 30318
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4583331353925263,
      "learning_rate": 2.6677954179591847e-07,
      "loss": 0.2513,
      "step": 30319
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48790360677682587,
      "learning_rate": 2.6655200871984475e-07,
      "loss": 0.326,
      "step": 30320
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18199246510903005,
      "learning_rate": 2.6632457140463476e-07,
      "loss": 0.0803,
      "step": 30321
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4348836919824216,
      "learning_rate": 2.6609722985252774e-07,
      "loss": 0.2658,
      "step": 30322
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48926070228600244,
      "learning_rate": 2.658699840657575e-07,
      "loss": 0.1837,
      "step": 30323
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4802970700306224,
      "learning_rate": 2.656428340465622e-07,
      "loss": 0.4851,
      "step": 30324
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4305176044245818,
      "learning_rate": 2.654157797971757e-07,
      "loss": 0.2351,
      "step": 30325
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3759303135990249,
      "learning_rate": 2.651888213198306e-07,
      "loss": 0.2662,
      "step": 30326
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.367123755968671,
      "learning_rate": 2.649619586167629e-07,
      "loss": 0.2121,
      "step": 30327
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9685134175284503,
      "learning_rate": 2.6473519169020077e-07,
      "loss": 0.3334,
      "step": 30328
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4943025764098112,
      "learning_rate": 2.645085205423792e-07,
      "loss": 0.7141,
      "step": 30329
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.15106729705397975,
      "learning_rate": 2.642819451755252e-07,
      "loss": 0.0692,
      "step": 30330
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5068171746405511,
      "learning_rate": 2.6405546559186815e-07,
      "loss": 0.3016,
      "step": 30331
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3530129263288576,
      "learning_rate": 2.6382908179363843e-07,
      "loss": 0.213,
      "step": 30332
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49521082249946663,
      "learning_rate": 2.6360279378306097e-07,
      "loss": 0.2912,
      "step": 30333
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6184268494252804,
      "learning_rate": 2.633766015623629e-07,
      "loss": 0.2397,
      "step": 30334
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0935094059228794,
      "learning_rate": 2.6315050513377017e-07,
      "loss": 0.4709,
      "step": 30335
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2970661769046898,
      "learning_rate": 2.6292450449950656e-07,
      "loss": 0.1745,
      "step": 30336
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5274650711014854,
      "learning_rate": 2.6269859966179477e-07,
      "loss": 0.8046,
      "step": 30337
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2999314982423629,
      "learning_rate": 2.624727906228586e-07,
      "loss": 0.2187,
      "step": 30338
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9752206973594344,
      "learning_rate": 2.622470773849206e-07,
      "loss": 0.7676,
      "step": 30339
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24384788119730108,
      "learning_rate": 2.620214599501969e-07,
      "loss": 0.1209,
      "step": 30340
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3754408768745562,
      "learning_rate": 2.617959383209134e-07,
      "loss": 0.1414,
      "step": 30341
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7805236059944883,
      "learning_rate": 2.61570512499284e-07,
      "loss": 0.3422,
      "step": 30342
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4069932759303942,
      "learning_rate": 2.6134518248752796e-07,
      "loss": 0.2171,
      "step": 30343
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42963583980617764,
      "learning_rate": 2.611199482878646e-07,
      "loss": 0.2727,
      "step": 30344
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3149313096637438,
      "learning_rate": 2.608948099025055e-07,
      "loss": 0.1792,
      "step": 30345
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3224733581524868,
      "learning_rate": 2.606697673336689e-07,
      "loss": 0.6852,
      "step": 30346
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0667395275726497,
      "learning_rate": 2.6044482058356746e-07,
      "loss": 0.2871,
      "step": 30347
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4558300561358924,
      "learning_rate": 2.60219969654415e-07,
      "loss": 0.2865,
      "step": 30348
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.218460759245817,
      "learning_rate": 2.599952145484219e-07,
      "loss": 0.1243,
      "step": 30349
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.29657901908664114,
      "learning_rate": 2.59770555267802e-07,
      "loss": 0.2612,
      "step": 30350
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0471598336465255,
      "learning_rate": 2.595459918147658e-07,
      "loss": 0.2127,
      "step": 30351
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7413848217340584,
      "learning_rate": 2.593215241915181e-07,
      "loss": 0.347,
      "step": 30352
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7507129202973599,
      "learning_rate": 2.5909715240027275e-07,
      "loss": 0.1194,
      "step": 30353
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.31786408529409776,
      "learning_rate": 2.5887287644323243e-07,
      "loss": 0.1804,
      "step": 30354
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5625010510993875,
      "learning_rate": 2.5864869632260757e-07,
      "loss": 0.6906,
      "step": 30355
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.31726556296387176,
      "learning_rate": 2.58424612040602e-07,
      "loss": 0.2164,
      "step": 30356
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1081373555641822,
      "learning_rate": 2.5820062359942055e-07,
      "loss": 0.3333,
      "step": 30357
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34584215602387586,
      "learning_rate": 2.579767310012671e-07,
      "loss": 0.1622,
      "step": 30358
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3495187704285478,
      "learning_rate": 2.5775293424834423e-07,
      "loss": 0.2182,
      "step": 30359
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4362869858245287,
      "learning_rate": 2.575292333428525e-07,
      "loss": 0.1934,
      "step": 30360
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3589997630847122,
      "learning_rate": 2.5730562828699455e-07,
      "loss": 0.2775,
      "step": 30361
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5424532953598333,
      "learning_rate": 2.5708211908296977e-07,
      "loss": 0.0248,
      "step": 30362
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.39069294551261113,
      "learning_rate": 2.568587057329774e-07,
      "loss": 0.2386,
      "step": 30363
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43944311187880025,
      "learning_rate": 2.5663538823921364e-07,
      "loss": 0.2315,
      "step": 30364
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5381952464633812,
      "learning_rate": 2.5641216660387877e-07,
      "loss": 0.7689,
      "step": 30365
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.688539672476944,
      "learning_rate": 2.561890408291645e-07,
      "loss": 0.1272,
      "step": 30366
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23198031499028357,
      "learning_rate": 2.5596601091727126e-07,
      "loss": 0.1363,
      "step": 30367
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30437057459843,
      "learning_rate": 2.5574307687038837e-07,
      "loss": 0.2553,
      "step": 30368
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.626265727765347,
      "learning_rate": 2.55520238690713e-07,
      "loss": 0.2542,
      "step": 30369
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.455634197495201,
      "learning_rate": 2.5529749638043557e-07,
      "loss": 0.1904,
      "step": 30370
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6188368979992729,
      "learning_rate": 2.550748499417477e-07,
      "loss": 0.0591,
      "step": 30371
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3446887922814425,
      "learning_rate": 2.548522993768388e-07,
      "loss": 0.2369,
      "step": 30372
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.420432137812436,
      "learning_rate": 2.546298446879003e-07,
      "loss": 0.2667,
      "step": 30373
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4880988648006637,
      "learning_rate": 2.544074858771217e-07,
      "loss": 0.306,
      "step": 30374
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2944287188937456,
      "learning_rate": 2.541852229466868e-07,
      "loss": 0.1384,
      "step": 30375
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5607968045705702,
      "learning_rate": 2.539630558987849e-07,
      "loss": 0.2415,
      "step": 30376
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.34330067320087243,
      "learning_rate": 2.5374098473560313e-07,
      "loss": 0.2001,
      "step": 30377
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9811174167763119,
      "learning_rate": 2.5351900945932094e-07,
      "loss": 0.5153,
      "step": 30378
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2382136002679322,
      "learning_rate": 2.5329713007212876e-07,
      "loss": 0.1712,
      "step": 30379
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2081825841922836,
      "learning_rate": 2.530753465762059e-07,
      "loss": 0.0696,
      "step": 30380
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36024951527951105,
      "learning_rate": 2.52853658973734e-07,
      "loss": 0.2708,
      "step": 30381
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1629679518873872,
      "learning_rate": 2.5263206726689695e-07,
      "loss": 0.4431,
      "step": 30382
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.472414615882227,
      "learning_rate": 2.524105714578695e-07,
      "loss": 0.6437,
      "step": 30383
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5271969556099337,
      "learning_rate": 2.521891715488378e-07,
      "loss": 0.2479,
      "step": 30384
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43139831999933265,
      "learning_rate": 2.5196786754197455e-07,
      "loss": 0.2925,
      "step": 30385
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.33536746964330993,
      "learning_rate": 2.517466594394591e-07,
      "loss": 0.1933,
      "step": 30386
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7906605989048774,
      "learning_rate": 2.5152554724346747e-07,
      "loss": 0.3405,
      "step": 30387
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2642468894874124,
      "learning_rate": 2.513045309561768e-07,
      "loss": 0.0701,
      "step": 30388
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.409387524066518,
      "learning_rate": 2.510836105797587e-07,
      "loss": 0.5355,
      "step": 30389
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.32208837716555677,
      "learning_rate": 2.5086278611638816e-07,
      "loss": 0.1381,
      "step": 30390
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5827646568600984,
      "learning_rate": 2.506420575682378e-07,
      "loss": 0.3587,
      "step": 30391
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2899378980894299,
      "learning_rate": 2.5042142493747703e-07,
      "loss": 0.2113,
      "step": 30392
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9737124746570097,
      "learning_rate": 2.5020088822627966e-07,
      "loss": 0.2457,
      "step": 30393
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6477274030317076,
      "learning_rate": 2.4998044743681394e-07,
      "loss": 0.3171,
      "step": 30394
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3550047692224332,
      "learning_rate": 2.4976010257124816e-07,
      "loss": 0.2083,
      "step": 30395
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.7928378522106077,
      "learning_rate": 2.495398536317528e-07,
      "loss": 0.6884,
      "step": 30396
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.32391818730040517,
      "learning_rate": 2.493197006204906e-07,
      "loss": 0.2226,
      "step": 30397
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4145887630574933,
      "learning_rate": 2.4909964353962976e-07,
      "loss": 0.1851,
      "step": 30398
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.33947664407615363,
      "learning_rate": 2.488796823913353e-07,
      "loss": 0.1754,
      "step": 30399
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5422507126404829,
      "learning_rate": 2.486598171777721e-07,
      "loss": 0.3731,
      "step": 30400
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0247693761515488,
      "learning_rate": 2.484400479011007e-07,
      "loss": 0.2899,
      "step": 30401
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7006906071323088,
      "learning_rate": 2.4822037456348593e-07,
      "loss": 0.3323,
      "step": 30402
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.26184813222514064,
      "learning_rate": 2.4800079716708834e-07,
      "loss": 0.1902,
      "step": 30403
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37307769688667275,
      "learning_rate": 2.4778131571406515e-07,
      "loss": 0.2586,
      "step": 30404
    },
    {
      "epoch": 0.93,
      "grad_norm": 2.097088593408419,
      "learning_rate": 2.4756193020658127e-07,
      "loss": 0.1069,
      "step": 30405
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3862290226784815,
      "learning_rate": 2.473426406467905e-07,
      "loss": 0.4757,
      "step": 30406
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.17629804070897345,
      "learning_rate": 2.471234470368522e-07,
      "loss": 0.094,
      "step": 30407
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.28075400614908774,
      "learning_rate": 2.469043493789225e-07,
      "loss": 0.1731,
      "step": 30408
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.583315160822035,
      "learning_rate": 2.466853476751563e-07,
      "loss": 0.7972,
      "step": 30409
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30238513226428604,
      "learning_rate": 2.4646644192771073e-07,
      "loss": 0.2244,
      "step": 30410
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6945265361826399,
      "learning_rate": 2.462476321387364e-07,
      "loss": 0.3438,
      "step": 30411
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7544354005708231,
      "learning_rate": 2.46028918310387e-07,
      "loss": 0.1812,
      "step": 30412
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.489703670040294,
      "learning_rate": 2.4581030044481536e-07,
      "loss": 0.3197,
      "step": 30413
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5059862044996775,
      "learning_rate": 2.455917785441708e-07,
      "loss": 0.1741,
      "step": 30414
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37340648887416966,
      "learning_rate": 2.453733526106039e-07,
      "loss": 0.2589,
      "step": 30415
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.16556582354338048,
      "learning_rate": 2.45155022646264e-07,
      "loss": 0.0833,
      "step": 30416
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8852764804870588,
      "learning_rate": 2.449367886533005e-07,
      "loss": 0.4275,
      "step": 30417
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3196816821999902,
      "learning_rate": 2.4471865063385504e-07,
      "loss": 0.1938,
      "step": 30418
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.80193682550146,
      "learning_rate": 2.445006085900803e-07,
      "loss": 0.3938,
      "step": 30419
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4463940323453333,
      "learning_rate": 2.442826625241179e-07,
      "loss": 0.2333,
      "step": 30420
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37084624859248533,
      "learning_rate": 2.4406481243811177e-07,
      "loss": 0.1953,
      "step": 30421
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.35996561750092,
      "learning_rate": 2.4384705833420783e-07,
      "loss": 0.2527,
      "step": 30422
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.3043700049075229,
      "learning_rate": 2.436294002145456e-07,
      "loss": 0.0715,
      "step": 30423
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.8918628377340234,
      "learning_rate": 2.4341183808126776e-07,
      "loss": 0.7549,
      "step": 30424
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.23365698619788006,
      "learning_rate": 2.4319437193651374e-07,
      "loss": 0.1104,
      "step": 30425
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7840614134765213,
      "learning_rate": 2.4297700178242625e-07,
      "loss": 0.3902,
      "step": 30426
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22596300588760082,
      "learning_rate": 2.427597276211391e-07,
      "loss": 0.2,
      "step": 30427
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.176670028284548,
      "learning_rate": 2.4254254945479394e-07,
      "loss": 0.4512,
      "step": 30428
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6426975005583216,
      "learning_rate": 2.4232546728552464e-07,
      "loss": 0.2676,
      "step": 30429
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5357016178395297,
      "learning_rate": 2.4210848111546836e-07,
      "loss": 0.2271,
      "step": 30430
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2717016308954132,
      "learning_rate": 2.4189159094676007e-07,
      "loss": 0.1692,
      "step": 30431
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.579021021049383,
      "learning_rate": 2.4167479678153247e-07,
      "loss": 0.5684,
      "step": 30432
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3265826144032088,
      "learning_rate": 2.4145809862191946e-07,
      "loss": 0.2102,
      "step": 30433
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3906701052725804,
      "learning_rate": 2.4124149647005377e-07,
      "loss": 0.1282,
      "step": 30434
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4057679177166432,
      "learning_rate": 2.410249903280637e-07,
      "loss": 0.256,
      "step": 30435
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4754579386565392,
      "learning_rate": 2.4080858019808197e-07,
      "loss": 0.2181,
      "step": 30436
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6836613807466704,
      "learning_rate": 2.4059226608223576e-07,
      "loss": 0.3368,
      "step": 30437
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3070727055702775,
      "learning_rate": 2.4037604798265446e-07,
      "loss": 0.2273,
      "step": 30438
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.409619135180222,
      "learning_rate": 2.401599259014653e-07,
      "loss": 0.1841,
      "step": 30439
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3941537572469386,
      "learning_rate": 2.3994389984079325e-07,
      "loss": 0.1611,
      "step": 30440
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5983767497209928,
      "learning_rate": 2.3972796980276545e-07,
      "loss": 0.2976,
      "step": 30441
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8092603008266287,
      "learning_rate": 2.395121357895047e-07,
      "loss": 0.027,
      "step": 30442
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8847549245412554,
      "learning_rate": 2.392963978031371e-07,
      "loss": 0.4955,
      "step": 30443
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.42276394732527645,
      "learning_rate": 2.3908075584578085e-07,
      "loss": 0.1872,
      "step": 30444
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37490885235825355,
      "learning_rate": 2.38865209919561e-07,
      "loss": 0.2507,
      "step": 30445
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5013461705565988,
      "learning_rate": 2.38649760026598e-07,
      "loss": 0.2469,
      "step": 30446
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.24176134229139767,
      "learning_rate": 2.3843440616900805e-07,
      "loss": 0.1113,
      "step": 30447
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3661146858093598,
      "learning_rate": 2.382191483489149e-07,
      "loss": 0.1585,
      "step": 30448
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3843319257932382,
      "learning_rate": 2.3800398656843248e-07,
      "loss": 0.1422,
      "step": 30449
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5806370755726206,
      "learning_rate": 2.3778892082967908e-07,
      "loss": 0.3163,
      "step": 30450
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3342449289906806,
      "learning_rate": 2.3757395113477077e-07,
      "loss": 0.2212,
      "step": 30451
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2946191275984906,
      "learning_rate": 2.3735907748582254e-07,
      "loss": 0.6129,
      "step": 30452
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5095987795958627,
      "learning_rate": 2.3714429988494714e-07,
      "loss": 0.1846,
      "step": 30453
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3767617216949262,
      "learning_rate": 2.3692961833425843e-07,
      "loss": 0.2665,
      "step": 30454
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0297973457043503,
      "learning_rate": 2.3671503283586915e-07,
      "loss": 0.2854,
      "step": 30455
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.5907671549865876,
      "learning_rate": 2.3650054339188876e-07,
      "loss": 0.8488,
      "step": 30456
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18523739463692027,
      "learning_rate": 2.362861500044311e-07,
      "loss": 0.1467,
      "step": 30457
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4120300883334942,
      "learning_rate": 2.3607185267560117e-07,
      "loss": 0.258,
      "step": 30458
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.49251905098054805,
      "learning_rate": 2.3585765140750948e-07,
      "loss": 0.1868,
      "step": 30459
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.2570175873226803,
      "learning_rate": 2.3564354620226438e-07,
      "loss": 0.0684,
      "step": 30460
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7789097866967688,
      "learning_rate": 2.3542953706196859e-07,
      "loss": 0.3762,
      "step": 30461
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.27245847698896597,
      "learning_rate": 2.3521562398873154e-07,
      "loss": 0.1989,
      "step": 30462
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5366942960353981,
      "learning_rate": 2.350018069846549e-07,
      "loss": 0.2956,
      "step": 30463
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4905028350086695,
      "learning_rate": 2.347880860518448e-07,
      "loss": 0.2418,
      "step": 30464
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.805886722728098,
      "learning_rate": 2.345744611924028e-07,
      "loss": 0.5805,
      "step": 30465
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.18816138812784747,
      "learning_rate": 2.343609324084306e-07,
      "loss": 0.0731,
      "step": 30466
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1329439075397152,
      "learning_rate": 2.3414749970202765e-07,
      "loss": 0.4039,
      "step": 30467
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3379829712503804,
      "learning_rate": 2.3393416307529558e-07,
      "loss": 0.1909,
      "step": 30468
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3862272914395656,
      "learning_rate": 2.3372092253033385e-07,
      "loss": 0.2873,
      "step": 30469
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7483524039848657,
      "learning_rate": 2.3350777806923742e-07,
      "loss": 0.246,
      "step": 30470
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9420497114445413,
      "learning_rate": 2.3329472969410572e-07,
      "loss": 0.4365,
      "step": 30471
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3042999277703172,
      "learning_rate": 2.3308177740703486e-07,
      "loss": 0.1693,
      "step": 30472
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0723769674668102,
      "learning_rate": 2.328689212101176e-07,
      "loss": 0.5657,
      "step": 30473
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5233217281978496,
      "learning_rate": 2.326561611054512e-07,
      "loss": 0.3011,
      "step": 30474
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2731890050840179,
      "learning_rate": 2.3244349709512616e-07,
      "loss": 0.1481,
      "step": 30475
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.29467514284288,
      "learning_rate": 2.3223092918123635e-07,
      "loss": 0.1461,
      "step": 30476
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3446711960591907,
      "learning_rate": 2.3201845736587458e-07,
      "loss": 0.1863,
      "step": 30477
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0029794078180423,
      "learning_rate": 2.3180608165112694e-07,
      "loss": 0.4005,
      "step": 30478
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6869418317555578,
      "learning_rate": 2.315938020390862e-07,
      "loss": 0.2576,
      "step": 30479
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3232115100653363,
      "learning_rate": 2.313816185318396e-07,
      "loss": 0.2483,
      "step": 30480
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.30894809656087246,
      "learning_rate": 2.3116953113147433e-07,
      "loss": 0.1744,
      "step": 30481
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.6115393311043582,
      "learning_rate": 2.3095753984007875e-07,
      "loss": 0.799,
      "step": 30482
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.42013569411939,
      "learning_rate": 2.307456446597367e-07,
      "loss": 0.2753,
      "step": 30483
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2950091015364498,
      "learning_rate": 2.3053384559253323e-07,
      "loss": 0.158,
      "step": 30484
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.33299531145169153,
      "learning_rate": 2.3032214264055332e-07,
      "loss": 0.1433,
      "step": 30485
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6674618589376422,
      "learning_rate": 2.301105358058786e-07,
      "loss": 0.1925,
      "step": 30486
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.37939666681151024,
      "learning_rate": 2.298990250905897e-07,
      "loss": 0.2444,
      "step": 30487
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7525262301459216,
      "learning_rate": 2.2968761049677045e-07,
      "loss": 0.2714,
      "step": 30488
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8639852664629857,
      "learning_rate": 2.2947629202649922e-07,
      "loss": 0.4268,
      "step": 30489
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4064778228829329,
      "learning_rate": 2.2926506968185437e-07,
      "loss": 0.1811,
      "step": 30490
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5459581539886285,
      "learning_rate": 2.2905394346491639e-07,
      "loss": 0.3177,
      "step": 30491
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.31956681226003025,
      "learning_rate": 2.288429133777592e-07,
      "loss": 0.2296,
      "step": 30492
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.40132536849208117,
      "learning_rate": 2.2863197942246119e-07,
      "loss": 0.1974,
      "step": 30493
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.19516477204311483,
      "learning_rate": 2.2842114160109618e-07,
      "loss": 0.07,
      "step": 30494
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3477860725104287,
      "learning_rate": 2.2821039991574034e-07,
      "loss": 0.2365,
      "step": 30495
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.7105807650869587,
      "learning_rate": 2.2799975436846533e-07,
      "loss": 0.2532,
      "step": 30496
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8692410702487399,
      "learning_rate": 2.2778920496134505e-07,
      "loss": 0.3648,
      "step": 30497
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3410949327095116,
      "learning_rate": 2.2757875169645005e-07,
      "loss": 0.212,
      "step": 30498
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.29114888235813585,
      "learning_rate": 2.273683945758487e-07,
      "loss": 0.1694,
      "step": 30499
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.4011003633944208,
      "learning_rate": 2.2715813360161487e-07,
      "loss": 0.7739,
      "step": 30500
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0130624500570666,
      "learning_rate": 2.269479687758136e-07,
      "loss": 0.1546,
      "step": 30501
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4407113217559135,
      "learning_rate": 2.2673790010051434e-07,
      "loss": 0.1991,
      "step": 30502
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3623562615903066,
      "learning_rate": 2.265279275777843e-07,
      "loss": 0.1211,
      "step": 30503
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3025649270304106,
      "learning_rate": 2.2631805120968854e-07,
      "loss": 0.2377,
      "step": 30504
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.476852483394315,
      "learning_rate": 2.261082709982909e-07,
      "loss": 0.2334,
      "step": 30505
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.0096641141968878,
      "learning_rate": 2.2589858694565536e-07,
      "loss": 0.4893,
      "step": 30506
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.138793641209685,
      "learning_rate": 2.256889990538469e-07,
      "loss": 0.2477,
      "step": 30507
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.43637685156393863,
      "learning_rate": 2.2547950732492606e-07,
      "loss": 0.2633,
      "step": 30508
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.47664086872762756,
      "learning_rate": 2.2527011176095347e-07,
      "loss": 0.2214,
      "step": 30509
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.4827225642948241,
      "learning_rate": 2.250608123639908e-07,
      "loss": 0.2903,
      "step": 30510
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.44301383777347425,
      "learning_rate": 2.2485160913609526e-07,
      "loss": 0.198,
      "step": 30511
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.22642843518778535,
      "learning_rate": 2.2464250207932747e-07,
      "loss": 0.1175,
      "step": 30512
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5480462473692261,
      "learning_rate": 2.244334911957413e-07,
      "loss": 0.3208,
      "step": 30513
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6556621367866727,
      "learning_rate": 2.2422457648739627e-07,
      "loss": 0.2653,
      "step": 30514
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.48899040977740776,
      "learning_rate": 2.2401575795634734e-07,
      "loss": 0.2319,
      "step": 30515
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.29892608004891524,
      "learning_rate": 2.238070356046451e-07,
      "loss": 0.2181,
      "step": 30516
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.9825124634531229,
      "learning_rate": 2.2359840943434906e-07,
      "loss": 0.455,
      "step": 30517
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3602896312516641,
      "learning_rate": 2.2338987944750757e-07,
      "loss": 0.221,
      "step": 30518
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.1863292485897943,
      "learning_rate": 2.231814456461734e-07,
      "loss": 0.4265,
      "step": 30519
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.8247894651870937,
      "learning_rate": 2.2297310803239715e-07,
      "loss": 0.2424,
      "step": 30520
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5434438973752601,
      "learning_rate": 2.2276486660822939e-07,
      "loss": 0.2976,
      "step": 30521
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.2830831600585182,
      "learning_rate": 2.2255672137571628e-07,
      "loss": 0.1628,
      "step": 30522
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.5272178321443354,
      "learning_rate": 2.223486723369084e-07,
      "loss": 0.3245,
      "step": 30523
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.1965331969270729,
      "learning_rate": 2.2214071949385186e-07,
      "loss": 0.0697,
      "step": 30524
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.079899920561095,
      "learning_rate": 2.2193286284859062e-07,
      "loss": 0.4271,
      "step": 30525
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6182798747989376,
      "learning_rate": 2.2172510240317303e-07,
      "loss": 0.3184,
      "step": 30526
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.3621362992751987,
      "learning_rate": 2.2151743815964076e-07,
      "loss": 0.2132,
      "step": 30527
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.36974038826187966,
      "learning_rate": 2.213098701200378e-07,
      "loss": 0.284,
      "step": 30528
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.404757072622505,
      "learning_rate": 2.2110239828640577e-07,
      "loss": 0.2548,
      "step": 30529
    },
    {
      "epoch": 0.93,
      "grad_norm": 0.6651226387769685,
      "learning_rate": 2.2089502266078644e-07,
      "loss": 0.3306,
      "step": 30530
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3372970859419986,
      "learning_rate": 2.2068774324521923e-07,
      "loss": 0.204,
      "step": 30531
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46594777265005216,
      "learning_rate": 2.2048056004174257e-07,
      "loss": 0.2176,
      "step": 30532
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37084994131426224,
      "learning_rate": 2.2027347305239698e-07,
      "loss": 0.0701,
      "step": 30533
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37049951286664046,
      "learning_rate": 2.200664822792198e-07,
      "loss": 0.2921,
      "step": 30534
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36162451559929365,
      "learning_rate": 2.19859587724246e-07,
      "loss": 0.1762,
      "step": 30535
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5609148242453194,
      "learning_rate": 2.1965278938951174e-07,
      "loss": 0.3558,
      "step": 30536
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7572988335338462,
      "learning_rate": 2.1944608727705097e-07,
      "loss": 0.1144,
      "step": 30537
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7093319511180216,
      "learning_rate": 2.1923948138889873e-07,
      "loss": 0.2638,
      "step": 30538
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.45178661485808436,
      "learning_rate": 2.190329717270867e-07,
      "loss": 0.2897,
      "step": 30539
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3975803485381739,
      "learning_rate": 2.188265582936455e-07,
      "loss": 0.1606,
      "step": 30540
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36797563138874056,
      "learning_rate": 2.1862024109060908e-07,
      "loss": 0.2693,
      "step": 30541
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2693970672148938,
      "learning_rate": 2.1841402012000245e-07,
      "loss": 0.0787,
      "step": 30542
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.249344271428956,
      "learning_rate": 2.1820789538385956e-07,
      "loss": 0.8262,
      "step": 30543
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3071227189605222,
      "learning_rate": 2.1800186688420434e-07,
      "loss": 0.0725,
      "step": 30544
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40249909715286175,
      "learning_rate": 2.1779593462306626e-07,
      "loss": 0.286,
      "step": 30545
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3359622254573512,
      "learning_rate": 2.175900986024715e-07,
      "loss": 0.2164,
      "step": 30546
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8330834426811661,
      "learning_rate": 2.1738435882444285e-07,
      "loss": 0.3537,
      "step": 30547
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.866718157301376,
      "learning_rate": 2.1717871529100655e-07,
      "loss": 0.2036,
      "step": 30548
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41366454180568685,
      "learning_rate": 2.1697316800418423e-07,
      "loss": 0.2518,
      "step": 30549
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5370840107464598,
      "learning_rate": 2.1676771696599986e-07,
      "loss": 0.1728,
      "step": 30550
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19486042909128548,
      "learning_rate": 2.1656236217847293e-07,
      "loss": 0.1536,
      "step": 30551
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3580420046189106,
      "learning_rate": 2.1635710364362517e-07,
      "loss": 0.6613,
      "step": 30552
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34870880301505414,
      "learning_rate": 2.1615194136347607e-07,
      "loss": 0.1544,
      "step": 30553
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41180481049788714,
      "learning_rate": 2.1594687534004176e-07,
      "loss": 0.277,
      "step": 30554
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.668560766052097,
      "learning_rate": 2.1574190557534292e-07,
      "loss": 0.2478,
      "step": 30555
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1178332920102383,
      "learning_rate": 2.1553703207139343e-07,
      "loss": 0.4609,
      "step": 30556
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.297790518787899,
      "learning_rate": 2.153322548302117e-07,
      "loss": 0.2072,
      "step": 30557
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38844668221499123,
      "learning_rate": 2.1512757385380945e-07,
      "loss": 0.2459,
      "step": 30558
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5311097741988425,
      "learning_rate": 2.1492298914420173e-07,
      "loss": 0.1796,
      "step": 30559
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4175703365181895,
      "learning_rate": 2.147185007034036e-07,
      "loss": 0.5142,
      "step": 30560
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2678604505360787,
      "learning_rate": 2.1451410853342237e-07,
      "loss": 0.139,
      "step": 30561
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6492295546716653,
      "learning_rate": 2.1430981263627193e-07,
      "loss": 0.304,
      "step": 30562
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2403670272086177,
      "learning_rate": 2.1410561301396071e-07,
      "loss": 0.153,
      "step": 30563
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32464512195735307,
      "learning_rate": 2.1390150966850042e-07,
      "loss": 0.2265,
      "step": 30564
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6599101755108779,
      "learning_rate": 2.1369750260189503e-07,
      "loss": 0.3197,
      "step": 30565
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7842014377788304,
      "learning_rate": 2.1349359181615403e-07,
      "loss": 0.0851,
      "step": 30566
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4855615583245773,
      "learning_rate": 2.132897773132847e-07,
      "loss": 0.2588,
      "step": 30567
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5021517191972491,
      "learning_rate": 2.1308605909528878e-07,
      "loss": 0.1635,
      "step": 30568
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6291561920641758,
      "learning_rate": 2.1288243716417466e-07,
      "loss": 0.3245,
      "step": 30569
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4360996402343048,
      "learning_rate": 2.1267891152194188e-07,
      "loss": 0.2773,
      "step": 30570
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.931130073053999,
      "learning_rate": 2.1247548217059543e-07,
      "loss": 0.433,
      "step": 30571
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3150823648678811,
      "learning_rate": 2.12272149112136e-07,
      "loss": 0.1815,
      "step": 30572
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6855690504886391,
      "learning_rate": 2.1206891234856196e-07,
      "loss": 0.3338,
      "step": 30573
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27593145060592517,
      "learning_rate": 2.118657718818762e-07,
      "loss": 0.1139,
      "step": 30574
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.49610373220380766,
      "learning_rate": 2.1166272771407482e-07,
      "loss": 0.2947,
      "step": 30575
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.26240685454349333,
      "learning_rate": 2.1145977984715736e-07,
      "loss": 0.182,
      "step": 30576
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4986318575987269,
      "learning_rate": 2.1125692828311894e-07,
      "loss": 0.1916,
      "step": 30577
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.114349892839488,
      "learning_rate": 2.110541730239546e-07,
      "loss": 0.4406,
      "step": 30578
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1240616825307472,
      "learning_rate": 2.1085151407166272e-07,
      "loss": 0.4227,
      "step": 30579
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6463275071249324,
      "learning_rate": 2.1064895142823172e-07,
      "loss": 0.3108,
      "step": 30580
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32085059635257307,
      "learning_rate": 2.1044648509566003e-07,
      "loss": 0.1847,
      "step": 30581
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38154193811549486,
      "learning_rate": 2.10244115075936e-07,
      "loss": 0.2903,
      "step": 30582
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23104944664702642,
      "learning_rate": 2.1004184137105144e-07,
      "loss": 0.0867,
      "step": 30583
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4569097099259898,
      "learning_rate": 2.0983966398299693e-07,
      "loss": 0.2108,
      "step": 30584
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33566303151421223,
      "learning_rate": 2.096375829137598e-07,
      "loss": 0.1316,
      "step": 30585
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6163520813718896,
      "learning_rate": 2.0943559816533177e-07,
      "loss": 0.344,
      "step": 30586
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.41039008280286915,
      "learning_rate": 2.092337097396968e-07,
      "loss": 0.206,
      "step": 30587
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4747275002352524,
      "learning_rate": 2.0903191763884223e-07,
      "loss": 0.3476,
      "step": 30588
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.665469640359462,
      "learning_rate": 2.088302218647531e-07,
      "loss": 0.2345,
      "step": 30589
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.355137053040065,
      "learning_rate": 2.0862862241941452e-07,
      "loss": 0.2277,
      "step": 30590
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7648219986209561,
      "learning_rate": 2.084271193048093e-07,
      "loss": 0.2552,
      "step": 30591
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22056740641568648,
      "learning_rate": 2.0822571252291922e-07,
      "loss": 0.0863,
      "step": 30592
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34300153719685955,
      "learning_rate": 2.0802440207572717e-07,
      "loss": 0.2886,
      "step": 30593
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3327422243981375,
      "learning_rate": 2.078231879652126e-07,
      "loss": 0.1477,
      "step": 30594
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5963967364829681,
      "learning_rate": 2.0762207019335623e-07,
      "loss": 0.3038,
      "step": 30595
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0783231158211812,
      "learning_rate": 2.0742104876213532e-07,
      "loss": 0.1345,
      "step": 30596
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7693558399496508,
      "learning_rate": 2.0722012367352827e-07,
      "loss": 0.3576,
      "step": 30597
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6564991257792271,
      "learning_rate": 2.0701929492951245e-07,
      "loss": 0.2168,
      "step": 30598
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4644937088264058,
      "learning_rate": 2.068185625320629e-07,
      "loss": 0.3003,
      "step": 30599
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.26514606155535003,
      "learning_rate": 2.0661792648315471e-07,
      "loss": 0.2045,
      "step": 30600
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4762579122392751,
      "learning_rate": 2.0641738678476188e-07,
      "loss": 0.2928,
      "step": 30601
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44143055031686784,
      "learning_rate": 2.0621694343885722e-07,
      "loss": 0.0797,
      "step": 30602
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8327779007485127,
      "learning_rate": 2.0601659644741257e-07,
      "loss": 0.2471,
      "step": 30603
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2938187320586557,
      "learning_rate": 2.0581634581239963e-07,
      "loss": 0.1788,
      "step": 30604
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2990830344954787,
      "learning_rate": 2.0561619153578903e-07,
      "loss": 0.2212,
      "step": 30605
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.097634199466263,
      "learning_rate": 2.0541613361954705e-07,
      "loss": 0.4436,
      "step": 30606
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6082379033880589,
      "learning_rate": 2.0521617206564647e-07,
      "loss": 0.2163,
      "step": 30607
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37039419877749685,
      "learning_rate": 2.0501630687605024e-07,
      "loss": 0.2445,
      "step": 30608
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1974131625757443,
      "learning_rate": 2.0481653805272673e-07,
      "loss": 0.5146,
      "step": 30609
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9301170723337218,
      "learning_rate": 2.0461686559764215e-07,
      "loss": 0.7445,
      "step": 30610
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20495803957141775,
      "learning_rate": 2.044172895127583e-07,
      "loss": 0.1443,
      "step": 30611
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5834213899293932,
      "learning_rate": 2.0421780980004248e-07,
      "loss": 0.2384,
      "step": 30612
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3263663582922654,
      "learning_rate": 2.040184264614531e-07,
      "loss": 0.1807,
      "step": 30613
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.394576569289921,
      "learning_rate": 2.0381913949895527e-07,
      "loss": 0.504,
      "step": 30614
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7174321901187025,
      "learning_rate": 2.0361994891450742e-07,
      "loss": 0.2384,
      "step": 30615
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.566960575117718,
      "learning_rate": 2.034208547100702e-07,
      "loss": 0.3226,
      "step": 30616
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.31759801014067873,
      "learning_rate": 2.032218568876021e-07,
      "loss": 0.1783,
      "step": 30617
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46911949484367627,
      "learning_rate": 2.0302295544906037e-07,
      "loss": 0.2262,
      "step": 30618
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1744463968058552,
      "learning_rate": 2.0282415039640346e-07,
      "loss": 0.5747,
      "step": 30619
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2616321970300674,
      "learning_rate": 2.0262544173158428e-07,
      "loss": 0.0757,
      "step": 30620
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4227519376187744,
      "learning_rate": 2.0242682945656233e-07,
      "loss": 0.2607,
      "step": 30621
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4976828125609365,
      "learning_rate": 2.022283135732883e-07,
      "loss": 0.1954,
      "step": 30622
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5115028103136579,
      "learning_rate": 2.02029894083714e-07,
      "loss": 0.3036,
      "step": 30623
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.49034142762389393,
      "learning_rate": 2.0183157098979557e-07,
      "loss": 0.2368,
      "step": 30624
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9153748723093867,
      "learning_rate": 2.0163334429348148e-07,
      "loss": 0.4372,
      "step": 30625
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3971725171491564,
      "learning_rate": 2.014352139967235e-07,
      "loss": 0.173,
      "step": 30626
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5513945142432591,
      "learning_rate": 2.0123718010146897e-07,
      "loss": 0.358,
      "step": 30627
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4370829467632429,
      "learning_rate": 2.0103924260966857e-07,
      "loss": 0.2263,
      "step": 30628
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3297701520128804,
      "learning_rate": 2.0084140152326735e-07,
      "loss": 0.2232,
      "step": 30629
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23817683061517897,
      "learning_rate": 2.0064365684421273e-07,
      "loss": 0.0624,
      "step": 30630
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.30530220420071613,
      "learning_rate": 2.0044600857445083e-07,
      "loss": 0.1693,
      "step": 30631
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7876940319176523,
      "learning_rate": 2.0024845671592352e-07,
      "loss": 0.3992,
      "step": 30632
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1912175186660579,
      "learning_rate": 2.000510012705792e-07,
      "loss": 0.2328,
      "step": 30633
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4819907486817285,
      "learning_rate": 1.998536422403563e-07,
      "loss": 0.3289,
      "step": 30634
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2855153460031618,
      "learning_rate": 1.9965637962719774e-07,
      "loss": 0.1921,
      "step": 30635
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5988623830671975,
      "learning_rate": 1.994592134330464e-07,
      "loss": 0.3668,
      "step": 30636
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2342822913900997,
      "learning_rate": 1.9926214365983744e-07,
      "loss": 0.1811,
      "step": 30637
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4811405541040326,
      "learning_rate": 1.9906517030951478e-07,
      "loss": 0.1972,
      "step": 30638
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.15686750339584715,
      "learning_rate": 1.9886829338401248e-07,
      "loss": 0.069,
      "step": 30639
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40807698843842394,
      "learning_rate": 1.98671512885269e-07,
      "loss": 0.2881,
      "step": 30640
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4917834848699332,
      "learning_rate": 1.984748288152205e-07,
      "loss": 0.2128,
      "step": 30641
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46524926395054317,
      "learning_rate": 1.982782411758022e-07,
      "loss": 0.29,
      "step": 30642
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5521778565917,
      "learning_rate": 1.980817499689469e-07,
      "loss": 0.3161,
      "step": 30643
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.28694021472149084,
      "learning_rate": 1.978853551965887e-07,
      "loss": 0.1718,
      "step": 30644
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5705281557419013,
      "learning_rate": 1.9768905686066042e-07,
      "loss": 0.8612,
      "step": 30645
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4154034330424388,
      "learning_rate": 1.9749285496309166e-07,
      "loss": 0.2192,
      "step": 30646
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3767647326966952,
      "learning_rate": 1.9729674950581423e-07,
      "loss": 0.2382,
      "step": 30647
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3034426017261143,
      "learning_rate": 1.9710074049075768e-07,
      "loss": 0.117,
      "step": 30648
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6349263728169487,
      "learning_rate": 1.9690482791984822e-07,
      "loss": 0.3357,
      "step": 30649
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5230211988597686,
      "learning_rate": 1.9670901179501544e-07,
      "loss": 0.2476,
      "step": 30650
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1523150802852036,
      "learning_rate": 1.9651329211818448e-07,
      "loss": 0.3001,
      "step": 30651
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33904957828725274,
      "learning_rate": 1.9631766889128156e-07,
      "loss": 0.2452,
      "step": 30652
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6621477470807485,
      "learning_rate": 1.9612214211623293e-07,
      "loss": 0.2953,
      "step": 30653
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3621940359126611,
      "learning_rate": 1.9592671179495926e-07,
      "loss": 0.2252,
      "step": 30654
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.541091214318834,
      "learning_rate": 1.9573137792938346e-07,
      "loss": 0.5764,
      "step": 30655
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0903280887729117,
      "learning_rate": 1.9553614052142955e-07,
      "loss": 0.2402,
      "step": 30656
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6777055621342223,
      "learning_rate": 1.953409995730182e-07,
      "loss": 0.2232,
      "step": 30657
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.364321824855691,
      "learning_rate": 1.951459550860657e-07,
      "loss": 0.2627,
      "step": 30658
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2793520346763597,
      "learning_rate": 1.9495100706249603e-07,
      "loss": 0.2011,
      "step": 30659
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5172978485917658,
      "learning_rate": 1.947561555042232e-07,
      "loss": 0.2314,
      "step": 30660
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1104455502463633,
      "learning_rate": 1.9456140041316574e-07,
      "loss": 0.4059,
      "step": 30661
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6798350908896958,
      "learning_rate": 1.9436674179124092e-07,
      "loss": 0.3328,
      "step": 30662
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33579058701295655,
      "learning_rate": 1.941721796403606e-07,
      "loss": 0.1982,
      "step": 30663
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5365892436326711,
      "learning_rate": 1.9397771396244326e-07,
      "loss": 0.5444,
      "step": 30664
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34572242222287786,
      "learning_rate": 1.9378334475939732e-07,
      "loss": 0.2172,
      "step": 30665
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.714375769792839,
      "learning_rate": 1.9358907203313902e-07,
      "loss": 0.3788,
      "step": 30666
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32182010486726176,
      "learning_rate": 1.9339489578557802e-07,
      "loss": 0.1853,
      "step": 30667
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.19350236033969,
      "learning_rate": 1.9320081601862494e-07,
      "loss": 0.5011,
      "step": 30668
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17535495863549516,
      "learning_rate": 1.930068327341883e-07,
      "loss": 0.0687,
      "step": 30669
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3272084679379656,
      "learning_rate": 1.9281294593417766e-07,
      "loss": 0.2441,
      "step": 30670
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.43970513900168495,
      "learning_rate": 1.926191556205015e-07,
      "loss": 0.2672,
      "step": 30671
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.48764290190492715,
      "learning_rate": 1.9242546179506383e-07,
      "loss": 0.2128,
      "step": 30672
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0754928994416122,
      "learning_rate": 1.9223186445977203e-07,
      "loss": 0.4204,
      "step": 30673
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6944452494923133,
      "learning_rate": 1.9203836361653017e-07,
      "loss": 0.2395,
      "step": 30674
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7889945699837227,
      "learning_rate": 1.918449592672411e-07,
      "loss": 0.3232,
      "step": 30675
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.395086752183721,
      "learning_rate": 1.9165165141381004e-07,
      "loss": 0.1693,
      "step": 30676
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2795334709756177,
      "learning_rate": 1.9145844005813762e-07,
      "loss": 0.2447,
      "step": 30677
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.23458654836830944,
      "learning_rate": 1.9126532520212348e-07,
      "loss": 0.0782,
      "step": 30678
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5137421468780503,
      "learning_rate": 1.9107230684766942e-07,
      "loss": 0.8436,
      "step": 30679
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.46601149038951317,
      "learning_rate": 1.908793849966728e-07,
      "loss": 0.1409,
      "step": 30680
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3569823581332563,
      "learning_rate": 1.906865596510321e-07,
      "loss": 0.2589,
      "step": 30681
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4307243004496941,
      "learning_rate": 1.9049383081264582e-07,
      "loss": 0.214,
      "step": 30682
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4159191340471889,
      "learning_rate": 1.90301198483408e-07,
      "loss": 0.2329,
      "step": 30683
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.036048755966017,
      "learning_rate": 1.90108662665216e-07,
      "loss": 0.3395,
      "step": 30684
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.29462715187880517,
      "learning_rate": 1.8991622335996272e-07,
      "loss": 0.1784,
      "step": 30685
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.111343447080582,
      "learning_rate": 1.8972388056954006e-07,
      "loss": 0.341,
      "step": 30686
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16606246561002044,
      "learning_rate": 1.895316342958431e-07,
      "loss": 0.0837,
      "step": 30687
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3472677763806053,
      "learning_rate": 1.8933948454076257e-07,
      "loss": 0.2867,
      "step": 30688
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.33557215773609506,
      "learning_rate": 1.8914743130618584e-07,
      "loss": 0.148,
      "step": 30689
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4074063652854754,
      "learning_rate": 1.8895547459400699e-07,
      "loss": 0.292,
      "step": 30690
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8183451960518396,
      "learning_rate": 1.8876361440611223e-07,
      "loss": 0.0524,
      "step": 30691
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6938601870504412,
      "learning_rate": 1.88571850744389e-07,
      "loss": 0.331,
      "step": 30692
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42914849900858426,
      "learning_rate": 1.8838018361072462e-07,
      "loss": 0.2132,
      "step": 30693
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.402084904288921,
      "learning_rate": 1.8818861300700432e-07,
      "loss": 0.2674,
      "step": 30694
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34467137053142266,
      "learning_rate": 1.879971389351132e-07,
      "loss": 0.2016,
      "step": 30695
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.17381726194806904,
      "learning_rate": 1.8780576139693419e-07,
      "loss": 0.0686,
      "step": 30696
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.672778166522099,
      "learning_rate": 1.8761448039435138e-07,
      "loss": 0.7172,
      "step": 30697
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5394251028955467,
      "learning_rate": 1.8742329592924547e-07,
      "loss": 0.1244,
      "step": 30698
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.604371513432656,
      "learning_rate": 1.8723220800349828e-07,
      "loss": 0.3391,
      "step": 30699
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3321512425202793,
      "learning_rate": 1.8704121661898945e-07,
      "loss": 0.2207,
      "step": 30700
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.47066799530761533,
      "learning_rate": 1.8685032177759743e-07,
      "loss": 0.3018,
      "step": 30701
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8442927966952221,
      "learning_rate": 1.8665952348120187e-07,
      "loss": 0.072,
      "step": 30702
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8688981507984889,
      "learning_rate": 1.864688217316779e-07,
      "loss": 0.354,
      "step": 30703
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.38071101195694035,
      "learning_rate": 1.8627821653090295e-07,
      "loss": 0.1975,
      "step": 30704
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40948839227774364,
      "learning_rate": 1.8608770788075326e-07,
      "loss": 0.2171,
      "step": 30705
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.30615215039026294,
      "learning_rate": 1.8589729578310067e-07,
      "loss": 0.2228,
      "step": 30706
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.66311672038735,
      "learning_rate": 1.8570698023982038e-07,
      "loss": 0.7556,
      "step": 30707
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.29076221752100223,
      "learning_rate": 1.8551676125278307e-07,
      "loss": 0.1602,
      "step": 30708
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6741439943443162,
      "learning_rate": 1.853266388238617e-07,
      "loss": 0.249,
      "step": 30709
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.48423376321569456,
      "learning_rate": 1.8513661295492702e-07,
      "loss": 0.239,
      "step": 30710
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5228641957204222,
      "learning_rate": 1.8494668364784752e-07,
      "loss": 0.2198,
      "step": 30711
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3930824987194172,
      "learning_rate": 1.847568509044917e-07,
      "loss": 0.246,
      "step": 30712
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3589915565790435,
      "learning_rate": 1.8456711472672807e-07,
      "loss": 0.1929,
      "step": 30713
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6072500955057998,
      "learning_rate": 1.84377475116424e-07,
      "loss": 0.5804,
      "step": 30714
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1384970915531185,
      "learning_rate": 1.841879320754425e-07,
      "loss": 0.4993,
      "step": 30715
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7277235824512944,
      "learning_rate": 1.839984856056498e-07,
      "loss": 0.4034,
      "step": 30716
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.36685903054679436,
      "learning_rate": 1.838091357089111e-07,
      "loss": 0.1997,
      "step": 30717
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.527872402380654,
      "learning_rate": 1.8361988238708606e-07,
      "loss": 0.3299,
      "step": 30718
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3239116628867282,
      "learning_rate": 1.834307256420409e-07,
      "loss": 0.1823,
      "step": 30719
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5213925294304906,
      "learning_rate": 1.8324166547563306e-07,
      "loss": 0.2377,
      "step": 30720
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4187858234049443,
      "learning_rate": 1.8305270188972435e-07,
      "loss": 0.1439,
      "step": 30721
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.524150695634691,
      "learning_rate": 1.8286383488617444e-07,
      "loss": 0.192,
      "step": 30722
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6705955080733623,
      "learning_rate": 1.8267506446683846e-07,
      "loss": 0.5158,
      "step": 30723
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3133208689350957,
      "learning_rate": 1.824863906335761e-07,
      "loss": 0.2234,
      "step": 30724
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6355167866757342,
      "learning_rate": 1.822978133882425e-07,
      "loss": 0.3048,
      "step": 30725
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4196507747186108,
      "learning_rate": 1.8210933273269503e-07,
      "loss": 0.1801,
      "step": 30726
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5442946860533375,
      "learning_rate": 1.8192094866878453e-07,
      "loss": 0.2909,
      "step": 30727
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.20997260380345764,
      "learning_rate": 1.817326611983683e-07,
      "loss": 0.0896,
      "step": 30728
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3686670577235796,
      "learning_rate": 1.8154447032329602e-07,
      "loss": 0.2783,
      "step": 30729
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3312943976743065,
      "learning_rate": 1.8135637604541845e-07,
      "loss": 0.0665,
      "step": 30730
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3374185354605888,
      "learning_rate": 1.8116837836658963e-07,
      "loss": 0.2474,
      "step": 30731
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.283600813010514,
      "learning_rate": 1.8098047728865698e-07,
      "loss": 0.1792,
      "step": 30732
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7952879019721842,
      "learning_rate": 1.8079267281346791e-07,
      "loss": 0.4042,
      "step": 30733
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.890968824804974,
      "learning_rate": 1.8060496494287205e-07,
      "loss": 0.2591,
      "step": 30734
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3000286106662116,
      "learning_rate": 1.8041735367871573e-07,
      "loss": 0.1735,
      "step": 30735
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3390193799907671,
      "learning_rate": 1.8022983902284518e-07,
      "loss": 0.2926,
      "step": 30736
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24657440774407133,
      "learning_rate": 1.8004242097710455e-07,
      "loss": 0.0822,
      "step": 30737
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2897955008136381,
      "learning_rate": 1.798550995433368e-07,
      "loss": 0.5785,
      "step": 30738
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27788029014960636,
      "learning_rate": 1.7966787472338597e-07,
      "loss": 0.0662,
      "step": 30739
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3935978360443459,
      "learning_rate": 1.7948074651909509e-07,
      "loss": 0.2546,
      "step": 30740
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3136646931497151,
      "learning_rate": 1.7929371493230263e-07,
      "loss": 0.4624,
      "step": 30741
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.34904543568765406,
      "learning_rate": 1.7910677996485047e-07,
      "loss": 0.2743,
      "step": 30742
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5959760799673596,
      "learning_rate": 1.7891994161857828e-07,
      "loss": 0.2558,
      "step": 30743
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40946303019669783,
      "learning_rate": 1.7873319989532123e-07,
      "loss": 0.2651,
      "step": 30744
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4958289252634475,
      "learning_rate": 1.7854655479692006e-07,
      "loss": 0.237,
      "step": 30745
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44646504549145966,
      "learning_rate": 1.783600063252089e-07,
      "loss": 0.1897,
      "step": 30746
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3064021679396094,
      "learning_rate": 1.78173554482024e-07,
      "loss": 0.1555,
      "step": 30747
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2673924169005802,
      "learning_rate": 1.779871992692006e-07,
      "loss": 0.1771,
      "step": 30748
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5584646727377579,
      "learning_rate": 1.7780094068856945e-07,
      "loss": 0.3163,
      "step": 30749
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0762811443462041,
      "learning_rate": 1.7761477874196463e-07,
      "loss": 0.1558,
      "step": 30750
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7048606371478375,
      "learning_rate": 1.7742871343121803e-07,
      "loss": 0.3422,
      "step": 30751
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7088217302089008,
      "learning_rate": 1.772427447581604e-07,
      "loss": 0.253,
      "step": 30752
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.9349567751777708,
      "learning_rate": 1.7705687272462026e-07,
      "loss": 0.3257,
      "step": 30753
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2508372585058128,
      "learning_rate": 1.7687109733242613e-07,
      "loss": 0.214,
      "step": 30754
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5144252544433423,
      "learning_rate": 1.766854185834077e-07,
      "loss": 0.5889,
      "step": 30755
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16678442627760992,
      "learning_rate": 1.7649983647938906e-07,
      "loss": 0.0669,
      "step": 30756
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7420112132312287,
      "learning_rate": 1.7631435102219875e-07,
      "loss": 0.5739,
      "step": 30757
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2693802319031216,
      "learning_rate": 1.761289622136586e-07,
      "loss": 0.1601,
      "step": 30758
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7880245704311227,
      "learning_rate": 1.7594367005559498e-07,
      "loss": 0.3565,
      "step": 30759
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3354953107450906,
      "learning_rate": 1.7575847454982976e-07,
      "loss": 0.2341,
      "step": 30760
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3659563171429439,
      "learning_rate": 1.7557337569818588e-07,
      "loss": 0.4475,
      "step": 30761
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4067757103295435,
      "learning_rate": 1.7538837350248418e-07,
      "loss": 0.2622,
      "step": 30762
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4782021084707321,
      "learning_rate": 1.7520346796454425e-07,
      "loss": 0.2276,
      "step": 30763
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5114124080523536,
      "learning_rate": 1.7501865908618464e-07,
      "loss": 0.2629,
      "step": 30764
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.27682999050745577,
      "learning_rate": 1.7483394686922394e-07,
      "loss": 0.1531,
      "step": 30765
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44165574439717403,
      "learning_rate": 1.746493313154818e-07,
      "loss": 0.3045,
      "step": 30766
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3455167369008436,
      "learning_rate": 1.7446481242677115e-07,
      "loss": 0.1854,
      "step": 30767
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7505987805949024,
      "learning_rate": 1.742803902049084e-07,
      "loss": 0.3172,
      "step": 30768
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7936958006406021,
      "learning_rate": 1.740960646517087e-07,
      "loss": 0.2535,
      "step": 30769
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4413229525434328,
      "learning_rate": 1.7391183576898396e-07,
      "loss": 0.8171,
      "step": 30770
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.288297524901919,
      "learning_rate": 1.737277035585494e-07,
      "loss": 0.186,
      "step": 30771
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37594606924610197,
      "learning_rate": 1.7354366802221468e-07,
      "loss": 0.295,
      "step": 30772
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.825368638304931,
      "learning_rate": 1.7335972916179055e-07,
      "loss": 0.0823,
      "step": 30773
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.16792721932261456,
      "learning_rate": 1.731758869790867e-07,
      "loss": 0.0686,
      "step": 30774
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7257942381734486,
      "learning_rate": 1.729921414759117e-07,
      "loss": 0.3218,
      "step": 30775
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32033989171205113,
      "learning_rate": 1.7280849265407295e-07,
      "loss": 0.1839,
      "step": 30776
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7005405025626841,
      "learning_rate": 1.7262494051537904e-07,
      "loss": 0.3797,
      "step": 30777
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32600364804030896,
      "learning_rate": 1.7244148506163516e-07,
      "loss": 0.2239,
      "step": 30778
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.4033667147430915,
      "learning_rate": 1.7225812629464434e-07,
      "loss": 0.7628,
      "step": 30779
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.37720193624167586,
      "learning_rate": 1.7207486421621177e-07,
      "loss": 0.0668,
      "step": 30780
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4488781901950323,
      "learning_rate": 1.718916988281416e-07,
      "loss": 0.285,
      "step": 30781
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3588837264570828,
      "learning_rate": 1.7170863013223238e-07,
      "loss": 0.1551,
      "step": 30782
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.26578521487450835,
      "learning_rate": 1.7152565813029042e-07,
      "loss": 0.2161,
      "step": 30783
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4522566081291897,
      "learning_rate": 1.71342782824111e-07,
      "loss": 0.1456,
      "step": 30784
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4201572716251784,
      "learning_rate": 1.7116000421549595e-07,
      "loss": 0.2277,
      "step": 30785
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.474391902395085,
      "learning_rate": 1.7097732230624388e-07,
      "loss": 0.2456,
      "step": 30786
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.457353171069155,
      "learning_rate": 1.7079473709814887e-07,
      "loss": 0.5738,
      "step": 30787
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3175239362360045,
      "learning_rate": 1.7061224859301173e-07,
      "loss": 0.4536,
      "step": 30788
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.31109196940648265,
      "learning_rate": 1.7042985679262435e-07,
      "loss": 0.1895,
      "step": 30789
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.29253726879935654,
      "learning_rate": 1.702475616987831e-07,
      "loss": 0.2475,
      "step": 30790
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3734331286411847,
      "learning_rate": 1.7006536331327982e-07,
      "loss": 0.0526,
      "step": 30791
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.39832296953468666,
      "learning_rate": 1.698832616379098e-07,
      "loss": 0.1762,
      "step": 30792
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.40809348572493576,
      "learning_rate": 1.6970125667446157e-07,
      "loss": 0.1625,
      "step": 30793
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3650390489171527,
      "learning_rate": 1.6951934842472595e-07,
      "loss": 0.2532,
      "step": 30794
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3405245385795027,
      "learning_rate": 1.6933753689049592e-07,
      "loss": 0.2207,
      "step": 30795
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.362829551606727,
      "learning_rate": 1.6915582207355563e-07,
      "loss": 0.501,
      "step": 30796
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.2455550112872278,
      "learning_rate": 1.6897420397569697e-07,
      "loss": 0.5321,
      "step": 30797
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6874641961855837,
      "learning_rate": 1.6879268259870407e-07,
      "loss": 0.2835,
      "step": 30798
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3690362665717323,
      "learning_rate": 1.6861125794436329e-07,
      "loss": 0.1819,
      "step": 30799
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.9037404995824707,
      "learning_rate": 1.6842993001446096e-07,
      "loss": 0.7026,
      "step": 30800
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3314601980749241,
      "learning_rate": 1.68248698810779e-07,
      "loss": 0.2009,
      "step": 30801
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5900782793573214,
      "learning_rate": 1.680675643351015e-07,
      "loss": 0.2236,
      "step": 30802
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5172439637214271,
      "learning_rate": 1.6788652658921044e-07,
      "loss": 0.3042,
      "step": 30803
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3339323986816538,
      "learning_rate": 1.6770558557488881e-07,
      "loss": 0.1524,
      "step": 30804
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.503883980339375,
      "learning_rate": 1.6752474129391294e-07,
      "loss": 0.2381,
      "step": 30805
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4191547008384424,
      "learning_rate": 1.6734399374806365e-07,
      "loss": 0.2592,
      "step": 30806
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.43514235605633894,
      "learning_rate": 1.6716334293912062e-07,
      "loss": 0.2885,
      "step": 30807
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3167991288587572,
      "learning_rate": 1.669827888688591e-07,
      "loss": 0.1801,
      "step": 30808
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1361563440195261,
      "learning_rate": 1.6680233153905767e-07,
      "loss": 0.3691,
      "step": 30809
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7035999705962125,
      "learning_rate": 1.6662197095148936e-07,
      "loss": 0.2369,
      "step": 30810
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.136620072734799,
      "learning_rate": 1.6644170710792939e-07,
      "loss": 0.4454,
      "step": 30811
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32356325184286655,
      "learning_rate": 1.6626154001015305e-07,
      "loss": 0.151,
      "step": 30812
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3578554025475828,
      "learning_rate": 1.6608146965992998e-07,
      "loss": 0.2618,
      "step": 30813
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.1863104186526584,
      "learning_rate": 1.6590149605903438e-07,
      "loss": 0.126,
      "step": 30814
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3539651289158314,
      "learning_rate": 1.657216192092348e-07,
      "loss": 0.4532,
      "step": 30815
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7795205329459929,
      "learning_rate": 1.6554183911230205e-07,
      "loss": 0.3956,
      "step": 30816
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2996477668398384,
      "learning_rate": 1.653621557700058e-07,
      "loss": 0.1796,
      "step": 30817
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8490124771346421,
      "learning_rate": 1.6518256918411135e-07,
      "loss": 0.4039,
      "step": 30818
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32133324713687766,
      "learning_rate": 1.6500307935638616e-07,
      "loss": 0.2237,
      "step": 30819
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8178453533333904,
      "learning_rate": 1.6482368628859767e-07,
      "loss": 0.3378,
      "step": 30820
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3323434402889244,
      "learning_rate": 1.646443899825112e-07,
      "loss": 0.1486,
      "step": 30821
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3090703726295278,
      "learning_rate": 1.6446519043988863e-07,
      "loss": 0.2068,
      "step": 30822
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.25776959826936974,
      "learning_rate": 1.64286087662493e-07,
      "loss": 0.0957,
      "step": 30823
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3818885931298908,
      "learning_rate": 1.6410708165208845e-07,
      "loss": 0.6939,
      "step": 30824
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2518284094448306,
      "learning_rate": 1.6392817241043357e-07,
      "loss": 0.1812,
      "step": 30825
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.388509656179382,
      "learning_rate": 1.6374935993929032e-07,
      "loss": 0.2614,
      "step": 30826
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.5297242126821202,
      "learning_rate": 1.6357064424041835e-07,
      "loss": 0.0675,
      "step": 30827
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.7847146808872307,
      "learning_rate": 1.6339202531557409e-07,
      "loss": 0.2382,
      "step": 30828
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.1734195863787908,
      "learning_rate": 1.6321350316651497e-07,
      "loss": 0.387,
      "step": 30829
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.32235299888348645,
      "learning_rate": 1.630350777949996e-07,
      "loss": 0.1475,
      "step": 30830
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.5808317255341785,
      "learning_rate": 1.6285674920278215e-07,
      "loss": 0.2724,
      "step": 30831
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2412289321270468,
      "learning_rate": 1.6267851739161567e-07,
      "loss": 0.1744,
      "step": 30832
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.33624074855127,
      "learning_rate": 1.625003823632554e-07,
      "loss": 0.6874,
      "step": 30833
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.19708578092391407,
      "learning_rate": 1.6232234411945213e-07,
      "loss": 0.0684,
      "step": 30834
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.49004155212151496,
      "learning_rate": 1.621444026619612e-07,
      "loss": 0.3261,
      "step": 30835
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.44530671510678377,
      "learning_rate": 1.6196655799252893e-07,
      "loss": 0.2305,
      "step": 30836
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3563466760151994,
      "learning_rate": 1.6178881011290615e-07,
      "loss": 0.2668,
      "step": 30837
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8950765738911837,
      "learning_rate": 1.616111590248437e-07,
      "loss": 0.0422,
      "step": 30838
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8840798858659638,
      "learning_rate": 1.6143360473008685e-07,
      "loss": 0.3557,
      "step": 30839
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.30815294820381495,
      "learning_rate": 1.612561472303853e-07,
      "loss": 0.18,
      "step": 30840
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2742147575350643,
      "learning_rate": 1.6107878652748098e-07,
      "loss": 0.1406,
      "step": 30841
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.7897259679025197,
      "learning_rate": 1.6090152262312252e-07,
      "loss": 0.6287,
      "step": 30842
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.22291601270462091,
      "learning_rate": 1.6072435551905186e-07,
      "loss": 0.1653,
      "step": 30843
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.3510770528161886,
      "learning_rate": 1.6054728521701314e-07,
      "loss": 0.2466,
      "step": 30844
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.6668538288832049,
      "learning_rate": 1.6037031171874606e-07,
      "loss": 0.2463,
      "step": 30845
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.0789583707226704,
      "learning_rate": 1.6019343502599483e-07,
      "loss": 0.4294,
      "step": 30846
    },
    {
      "epoch": 0.94,
      "grad_norm": 2.696824784818188,
      "learning_rate": 1.6001665514049914e-07,
      "loss": 0.0709,
      "step": 30847
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8215370489858672,
      "learning_rate": 1.5983997206399647e-07,
      "loss": 0.2682,
      "step": 30848
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.24121352436105226,
      "learning_rate": 1.596633857982255e-07,
      "loss": 0.2053,
      "step": 30849
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.6321182077125151,
      "learning_rate": 1.5948689634492476e-07,
      "loss": 0.5659,
      "step": 30850
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.024236748554698,
      "learning_rate": 1.5931050370582847e-07,
      "loss": 0.5417,
      "step": 30851
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.4029183922498416,
      "learning_rate": 1.591342078826752e-07,
      "loss": 0.2335,
      "step": 30852
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2906139329735006,
      "learning_rate": 1.5895800887719692e-07,
      "loss": 0.1749,
      "step": 30853
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.8863380350439976,
      "learning_rate": 1.587819066911278e-07,
      "loss": 0.2589,
      "step": 30854
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.2792622975752,
      "learning_rate": 1.586059013262009e-07,
      "loss": 0.2264,
      "step": 30855
    },
    {
      "epoch": 0.94,
      "grad_norm": 1.3174916314919414,
      "learning_rate": 1.5842999278414704e-07,
      "loss": 0.0789,
      "step": 30856
    },
    {
      "epoch": 0.94,
      "grad_norm": 0.42121966792274773,
      "learning_rate": 1.582541810666971e-07,
      "loss": 0.2542,
      "step": 30857
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4280203147560213,
      "learning_rate": 1.5807846617558187e-07,
      "loss": 0.1763,
      "step": 30858
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5309782384527038,
      "learning_rate": 1.5790284811252887e-07,
      "loss": 0.7909,
      "step": 30859
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3249839549708033,
      "learning_rate": 1.5772732687926452e-07,
      "loss": 0.2425,
      "step": 30860
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6936563392695697,
      "learning_rate": 1.575519024775185e-07,
      "loss": 0.3423,
      "step": 30861
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44699105348528717,
      "learning_rate": 1.5737657490901504e-07,
      "loss": 0.1768,
      "step": 30862
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4913497093681377,
      "learning_rate": 1.572013441754805e-07,
      "loss": 0.2998,
      "step": 30863
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.32936036987998035,
      "learning_rate": 1.5702621027863685e-07,
      "loss": 0.0772,
      "step": 30864
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3663430612051368,
      "learning_rate": 1.5685117322020826e-07,
      "loss": 0.5945,
      "step": 30865
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3346375693444424,
      "learning_rate": 1.566762330019167e-07,
      "loss": 0.1721,
      "step": 30866
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2731557944725377,
      "learning_rate": 1.56501389625483e-07,
      "loss": 0.2037,
      "step": 30867
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4687910295120659,
      "learning_rate": 1.5632664309262803e-07,
      "loss": 0.8031,
      "step": 30868
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3861705495570593,
      "learning_rate": 1.5615199340507147e-07,
      "loss": 0.4593,
      "step": 30869
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6482877925987285,
      "learning_rate": 1.559774405645287e-07,
      "loss": 0.3237,
      "step": 30870
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3334123335177009,
      "learning_rate": 1.5580298457271937e-07,
      "loss": 0.1688,
      "step": 30871
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31332812725414105,
      "learning_rate": 1.5562862543135993e-07,
      "loss": 0.187,
      "step": 30872
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3238458784953193,
      "learning_rate": 1.5545436314216567e-07,
      "loss": 0.2275,
      "step": 30873
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5269127115809752,
      "learning_rate": 1.5528019770685077e-07,
      "loss": 0.1942,
      "step": 30874
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5495843467350192,
      "learning_rate": 1.551061291271283e-07,
      "loss": 0.069,
      "step": 30875
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4157286579834749,
      "learning_rate": 1.5493215740471135e-07,
      "loss": 0.2607,
      "step": 30876
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5744441191467706,
      "learning_rate": 1.5475828254130965e-07,
      "loss": 0.196,
      "step": 30877
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4759420904038198,
      "learning_rate": 1.545845045386374e-07,
      "loss": 0.321,
      "step": 30878
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4313080802710111,
      "learning_rate": 1.5441082339840206e-07,
      "loss": 0.239,
      "step": 30879
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2853480032882373,
      "learning_rate": 1.5423723912231127e-07,
      "loss": 0.1739,
      "step": 30880
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6215623152502354,
      "learning_rate": 1.5406375171207576e-07,
      "loss": 0.7427,
      "step": 30881
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.26282232160703667,
      "learning_rate": 1.5389036116940092e-07,
      "loss": 0.0883,
      "step": 30882
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6787239249065993,
      "learning_rate": 1.5371706749599092e-07,
      "loss": 0.5549,
      "step": 30883
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.27074554539620255,
      "learning_rate": 1.5354387069355324e-07,
      "loss": 0.1882,
      "step": 30884
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.58664747318802,
      "learning_rate": 1.5337077076379215e-07,
      "loss": 0.3341,
      "step": 30885
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4774629146958937,
      "learning_rate": 1.5319776770840732e-07,
      "loss": 0.2204,
      "step": 30886
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6447566718898404,
      "learning_rate": 1.5302486152910411e-07,
      "loss": 0.316,
      "step": 30887
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6472273600040955,
      "learning_rate": 1.5285205222758116e-07,
      "loss": 0.2554,
      "step": 30888
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0277055491064715,
      "learning_rate": 1.526793398055404e-07,
      "loss": 0.4505,
      "step": 30889
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3416936680798156,
      "learning_rate": 1.525067242646816e-07,
      "loss": 0.2105,
      "step": 30890
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.28678611318216185,
      "learning_rate": 1.523342056067001e-07,
      "loss": 0.2353,
      "step": 30891
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4785755652856132,
      "learning_rate": 1.521617838332967e-07,
      "loss": 0.0774,
      "step": 30892
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30500968905135145,
      "learning_rate": 1.5198945894616457e-07,
      "loss": 0.0596,
      "step": 30893
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37835805172161996,
      "learning_rate": 1.5181723094700118e-07,
      "loss": 0.2744,
      "step": 30894
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6228674524164097,
      "learning_rate": 1.5164509983750186e-07,
      "loss": 0.2314,
      "step": 30895
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3489077049010633,
      "learning_rate": 1.514730656193564e-07,
      "loss": 0.2881,
      "step": 30896
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9901803717715904,
      "learning_rate": 1.5130112829426001e-07,
      "loss": 0.2639,
      "step": 30897
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5944887849409619,
      "learning_rate": 1.5112928786390368e-07,
      "loss": 0.272,
      "step": 30898
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3388728327877258,
      "learning_rate": 1.5095754432997932e-07,
      "loss": 0.1999,
      "step": 30899
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4024173873079172,
      "learning_rate": 1.507858976941745e-07,
      "loss": 0.1862,
      "step": 30900
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.411101215038705,
      "learning_rate": 1.5061434795817899e-07,
      "loss": 0.0897,
      "step": 30901
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3645637006294839,
      "learning_rate": 1.5044289512368027e-07,
      "loss": 0.2901,
      "step": 30902
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2821398575692184,
      "learning_rate": 1.5027153919236481e-07,
      "loss": 0.1667,
      "step": 30903
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0278343374197576,
      "learning_rate": 1.5010028016592016e-07,
      "loss": 0.4315,
      "step": 30904
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7716789622889659,
      "learning_rate": 1.4992911804602938e-07,
      "loss": 0.2516,
      "step": 30905
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.092371871678026,
      "learning_rate": 1.4975805283437673e-07,
      "loss": 0.5671,
      "step": 30906
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3884956107944375,
      "learning_rate": 1.4958708453264636e-07,
      "loss": 0.2344,
      "step": 30907
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5152898238907275,
      "learning_rate": 1.4941621314251808e-07,
      "loss": 0.1968,
      "step": 30908
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4177888043989014,
      "learning_rate": 1.4924543866567497e-07,
      "loss": 0.2854,
      "step": 30909
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3346421731866734,
      "learning_rate": 1.4907476110379683e-07,
      "loss": 0.0702,
      "step": 30910
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6824740033542631,
      "learning_rate": 1.4890418045856337e-07,
      "loss": 0.3145,
      "step": 30911
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3719188646645851,
      "learning_rate": 1.4873369673164996e-07,
      "loss": 0.1601,
      "step": 30912
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4999568118840692,
      "learning_rate": 1.4856330992473743e-07,
      "loss": 0.3036,
      "step": 30913
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34007311153321446,
      "learning_rate": 1.4839302003950008e-07,
      "loss": 0.2224,
      "step": 30914
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3040647916452917,
      "learning_rate": 1.4822282707761426e-07,
      "loss": 0.7382,
      "step": 30915
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4641666620512226,
      "learning_rate": 1.4805273104075424e-07,
      "loss": 0.1312,
      "step": 30916
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.39232305129651635,
      "learning_rate": 1.4788273193059198e-07,
      "loss": 0.2826,
      "step": 30917
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.1730181011621351,
      "learning_rate": 1.4771282974880175e-07,
      "loss": 0.0663,
      "step": 30918
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.28937812512622757,
      "learning_rate": 1.4754302449705439e-07,
      "loss": 0.1515,
      "step": 30919
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5202447798672383,
      "learning_rate": 1.4737331617702077e-07,
      "loss": 0.2983,
      "step": 30920
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29474261261322876,
      "learning_rate": 1.4720370479037072e-07,
      "loss": 0.1764,
      "step": 30921
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6617362421050996,
      "learning_rate": 1.4703419033877175e-07,
      "loss": 0.3419,
      "step": 30922
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9998736929484897,
      "learning_rate": 1.4686477282389365e-07,
      "loss": 0.2644,
      "step": 30923
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5628023996024931,
      "learning_rate": 1.4669545224740178e-07,
      "loss": 0.7238,
      "step": 30924
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6467457613306915,
      "learning_rate": 1.4652622861096143e-07,
      "loss": 0.138,
      "step": 30925
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29974501087677113,
      "learning_rate": 1.4635710191623907e-07,
      "loss": 0.2418,
      "step": 30926
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4505993217448141,
      "learning_rate": 1.4618807216489673e-07,
      "loss": 0.1969,
      "step": 30927
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5890245621761754,
      "learning_rate": 1.460191393585997e-07,
      "loss": 0.2912,
      "step": 30928
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45772496200195534,
      "learning_rate": 1.4585030349900774e-07,
      "loss": 0.1841,
      "step": 30929
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3935759763568709,
      "learning_rate": 1.4568156458778404e-07,
      "loss": 0.2327,
      "step": 30930
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.535070930400355,
      "learning_rate": 1.4551292262658722e-07,
      "loss": 0.2204,
      "step": 30931
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6147797493390863,
      "learning_rate": 1.45344377617076e-07,
      "loss": 0.2128,
      "step": 30932
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.341613143749921,
      "learning_rate": 1.451759295609112e-07,
      "loss": 0.7259,
      "step": 30933
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34425331527537817,
      "learning_rate": 1.45007578459746e-07,
      "loss": 0.1693,
      "step": 30934
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6348838024258547,
      "learning_rate": 1.4483932431524017e-07,
      "loss": 0.3483,
      "step": 30935
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.344788876105892,
      "learning_rate": 1.4467116712904794e-07,
      "loss": 0.1196,
      "step": 30936
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3442713039467387,
      "learning_rate": 1.4450310690282353e-07,
      "loss": 0.2505,
      "step": 30937
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45100387782890966,
      "learning_rate": 1.443351436382201e-07,
      "loss": 0.2256,
      "step": 30938
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0418551297833865,
      "learning_rate": 1.4416727733689073e-07,
      "loss": 0.4448,
      "step": 30939
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.27835693134747025,
      "learning_rate": 1.4399950800048634e-07,
      "loss": 0.1781,
      "step": 30940
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.491297061745911,
      "learning_rate": 1.4383183563065782e-07,
      "loss": 0.6841,
      "step": 30941
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1594609068175534,
      "learning_rate": 1.436642602290561e-07,
      "loss": 0.5274,
      "step": 30942
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5288779935256315,
      "learning_rate": 1.4349678179732763e-07,
      "loss": 0.3195,
      "step": 30943
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2747502071333521,
      "learning_rate": 1.4332940033712107e-07,
      "loss": 0.1664,
      "step": 30944
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4199754623787576,
      "learning_rate": 1.431621158500829e-07,
      "loss": 0.162,
      "step": 30945
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7674636072476563,
      "learning_rate": 1.4299492833785843e-07,
      "loss": 0.3721,
      "step": 30946
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7839775687993001,
      "learning_rate": 1.428278378020953e-07,
      "loss": 0.2709,
      "step": 30947
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30174808851552004,
      "learning_rate": 1.4266084424443438e-07,
      "loss": 0.1633,
      "step": 30948
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.26512725642320567,
      "learning_rate": 1.4249394766651882e-07,
      "loss": 0.1642,
      "step": 30949
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3449890702709499,
      "learning_rate": 1.423271480699928e-07,
      "loss": 0.2764,
      "step": 30950
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1657174375949013,
      "learning_rate": 1.4216044545649398e-07,
      "loss": 0.4407,
      "step": 30951
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9782716517008374,
      "learning_rate": 1.4199383982766545e-07,
      "loss": 0.4424,
      "step": 30952
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31232749683327965,
      "learning_rate": 1.418273311851448e-07,
      "loss": 0.1818,
      "step": 30953
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0345931473694503,
      "learning_rate": 1.4166091953057181e-07,
      "loss": 0.4737,
      "step": 30954
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8876165626646821,
      "learning_rate": 1.4149460486558186e-07,
      "loss": 0.2657,
      "step": 30955
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.33261695768884614,
      "learning_rate": 1.413283871918114e-07,
      "loss": 0.2808,
      "step": 30956
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3476298752967748,
      "learning_rate": 1.4116226651089694e-07,
      "loss": 0.1525,
      "step": 30957
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5673776621789901,
      "learning_rate": 1.4099624282447043e-07,
      "loss": 0.207,
      "step": 30958
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29502440446752715,
      "learning_rate": 1.4083031613416952e-07,
      "loss": 0.165,
      "step": 30959
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3424469741603766,
      "learning_rate": 1.406644864416218e-07,
      "loss": 0.3389,
      "step": 30960
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3544759426210509,
      "learning_rate": 1.4049875374846256e-07,
      "loss": 0.2545,
      "step": 30961
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3835116629369661,
      "learning_rate": 1.4033311805631943e-07,
      "loss": 0.1615,
      "step": 30962
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.54684009095553,
      "learning_rate": 1.4016757936682445e-07,
      "loss": 0.28,
      "step": 30963
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5963864610315404,
      "learning_rate": 1.400021376816052e-07,
      "loss": 0.2495,
      "step": 30964
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3394774802848726,
      "learning_rate": 1.3983679300228813e-07,
      "loss": 0.388,
      "step": 30965
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.24743902391442807,
      "learning_rate": 1.3967154533050198e-07,
      "loss": 0.0663,
      "step": 30966
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2721863923918579,
      "learning_rate": 1.3950639466787098e-07,
      "loss": 0.1915,
      "step": 30967
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.25190940578796794,
      "learning_rate": 1.393413410160216e-07,
      "loss": 0.1924,
      "step": 30968
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3620364142552712,
      "learning_rate": 1.3917638437657588e-07,
      "loss": 0.7972,
      "step": 30969
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7061365794352947,
      "learning_rate": 1.3901152475115808e-07,
      "loss": 0.141,
      "step": 30970
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42155789296096857,
      "learning_rate": 1.3884676214139027e-07,
      "loss": 0.2469,
      "step": 30971
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45787882120394396,
      "learning_rate": 1.3868209654888998e-07,
      "loss": 0.2263,
      "step": 30972
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.33747058571052624,
      "learning_rate": 1.385175279752826e-07,
      "loss": 0.2354,
      "step": 30973
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6082764618631873,
      "learning_rate": 1.3835305642218355e-07,
      "loss": 0.7692,
      "step": 30974
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4827273708824777,
      "learning_rate": 1.381886818912126e-07,
      "loss": 0.0701,
      "step": 30975
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3962073906282433,
      "learning_rate": 1.3802440438398735e-07,
      "loss": 0.2909,
      "step": 30976
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2511814237360956,
      "learning_rate": 1.3786022390212096e-07,
      "loss": 0.1414,
      "step": 30977
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6465059611708783,
      "learning_rate": 1.3769614044723212e-07,
      "loss": 0.7355,
      "step": 30978
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22397436925082895,
      "learning_rate": 1.375321540209329e-07,
      "loss": 0.18,
      "step": 30979
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4165997241030588,
      "learning_rate": 1.3736826462483865e-07,
      "loss": 0.2548,
      "step": 30980
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4807937273987863,
      "learning_rate": 1.3720447226055923e-07,
      "loss": 0.2265,
      "step": 30981
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7669683638730402,
      "learning_rate": 1.3704077692970775e-07,
      "loss": 0.3607,
      "step": 30982
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2315117810570113,
      "learning_rate": 1.3687717863389627e-07,
      "loss": 0.0783,
      "step": 30983
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9129280547521019,
      "learning_rate": 1.367136773747302e-07,
      "loss": 0.2682,
      "step": 30984
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2375344924928931,
      "learning_rate": 1.3655027315382153e-07,
      "loss": 0.1451,
      "step": 30985
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30022490859738027,
      "learning_rate": 1.3638696597277678e-07,
      "loss": 0.2041,
      "step": 30986
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.436178223443757,
      "learning_rate": 1.3622375583320246e-07,
      "loss": 0.8134,
      "step": 30987
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3859163683229257,
      "learning_rate": 1.3606064273670504e-07,
      "loss": 0.1477,
      "step": 30988
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44572310705563306,
      "learning_rate": 1.358976266848866e-07,
      "loss": 0.2463,
      "step": 30989
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5517798385291136,
      "learning_rate": 1.3573470767935472e-07,
      "loss": 0.2367,
      "step": 30990
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47410869588815274,
      "learning_rate": 1.3557188572171032e-07,
      "loss": 0.262,
      "step": 30991
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4780357029561184,
      "learning_rate": 1.354091608135555e-07,
      "loss": 0.2093,
      "step": 30992
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3921206974118736,
      "learning_rate": 1.3524653295649114e-07,
      "loss": 0.5648,
      "step": 30993
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2684146723094194,
      "learning_rate": 1.3508400215211825e-07,
      "loss": 0.1683,
      "step": 30994
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1983804071885082,
      "learning_rate": 1.349215684020333e-07,
      "loss": 0.5947,
      "step": 30995
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3969121640393097,
      "learning_rate": 1.3475923170783723e-07,
      "loss": 0.169,
      "step": 30996
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5112189886488879,
      "learning_rate": 1.3459699207112654e-07,
      "loss": 0.3259,
      "step": 30997
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.28263609125678574,
      "learning_rate": 1.344348494934944e-07,
      "loss": 0.1742,
      "step": 30998
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.48211322146459457,
      "learning_rate": 1.3427280397654063e-07,
      "loss": 0.2313,
      "step": 30999
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5257014177654379,
      "learning_rate": 1.341108555218562e-07,
      "loss": 0.1953,
      "step": 31000
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7339907306484246,
      "learning_rate": 1.339490041310354e-07,
      "loss": 0.02,
      "step": 31001
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9538564657668855,
      "learning_rate": 1.3378724980567136e-07,
      "loss": 0.3997,
      "step": 31002
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.27123347024390665,
      "learning_rate": 1.3362559254735397e-07,
      "loss": 0.1938,
      "step": 31003
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.493296235766338,
      "learning_rate": 1.3346403235767525e-07,
      "loss": 0.3019,
      "step": 31004
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1069788430246545,
      "learning_rate": 1.333025692382228e-07,
      "loss": 0.4243,
      "step": 31005
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6414680123857588,
      "learning_rate": 1.3314120319058766e-07,
      "loss": 0.3473,
      "step": 31006
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3920199956197747,
      "learning_rate": 1.3297993421635513e-07,
      "loss": 0.1831,
      "step": 31007
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3845558104082905,
      "learning_rate": 1.3281876231711289e-07,
      "loss": 0.2582,
      "step": 31008
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29709081954636307,
      "learning_rate": 1.326576874944474e-07,
      "loss": 0.1703,
      "step": 31009
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5936182581544569,
      "learning_rate": 1.3249670974993968e-07,
      "loss": 0.3179,
      "step": 31010
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7779033235980554,
      "learning_rate": 1.323358290851784e-07,
      "loss": 0.0571,
      "step": 31011
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31253311271629486,
      "learning_rate": 1.3217504550174342e-07,
      "loss": 0.1821,
      "step": 31012
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.608232927164999,
      "learning_rate": 1.3201435900121794e-07,
      "loss": 0.7298,
      "step": 31013
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6704949991675763,
      "learning_rate": 1.318537695851818e-07,
      "loss": 0.2663,
      "step": 31014
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3549197555352764,
      "learning_rate": 1.3169327725521485e-07,
      "loss": 0.2642,
      "step": 31015
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3391559157358529,
      "learning_rate": 1.3153288201289694e-07,
      "loss": 0.1525,
      "step": 31016
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5869304745618924,
      "learning_rate": 1.3137258385980567e-07,
      "loss": 0.2964,
      "step": 31017
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29539334419782937,
      "learning_rate": 1.3121238279751758e-07,
      "loss": 0.0677,
      "step": 31018
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.3887759001418054,
      "learning_rate": 1.310522788276103e-07,
      "loss": 0.6083,
      "step": 31019
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4441272589992601,
      "learning_rate": 1.308922719516581e-07,
      "loss": 0.2063,
      "step": 31020
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4856991258644412,
      "learning_rate": 1.307323621712342e-07,
      "loss": 0.2425,
      "step": 31021
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3213635433519513,
      "learning_rate": 1.3057254948791286e-07,
      "loss": 0.2436,
      "step": 31022
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7491251950475846,
      "learning_rate": 1.304128339032673e-07,
      "loss": 0.3564,
      "step": 31023
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8465142920261947,
      "learning_rate": 1.3025321541886627e-07,
      "loss": 0.2377,
      "step": 31024
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3183564997769044,
      "learning_rate": 1.3009369403628292e-07,
      "loss": 0.1128,
      "step": 31025
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4174969095309344,
      "learning_rate": 1.299342697570849e-07,
      "loss": 0.2846,
      "step": 31026
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20888064721809518,
      "learning_rate": 1.2977494258283983e-07,
      "loss": 0.158,
      "step": 31027
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6316173427430647,
      "learning_rate": 1.2961571251511872e-07,
      "loss": 0.5401,
      "step": 31028
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8056856693886372,
      "learning_rate": 1.2945657955548585e-07,
      "loss": 0.0868,
      "step": 31029
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34589399620780403,
      "learning_rate": 1.292975437055055e-07,
      "loss": 0.2419,
      "step": 31030
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7593288709630241,
      "learning_rate": 1.2913860496674535e-07,
      "loss": 0.3021,
      "step": 31031
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1068922318228602,
      "learning_rate": 1.2897976334076744e-07,
      "loss": 0.4316,
      "step": 31032
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.35174760883617473,
      "learning_rate": 1.288210188291339e-07,
      "loss": 0.2341,
      "step": 31033
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6518586774502227,
      "learning_rate": 1.2866237143340676e-07,
      "loss": 0.323,
      "step": 31034
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.35229129947874605,
      "learning_rate": 1.2850382115514926e-07,
      "loss": 0.1855,
      "step": 31035
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.43953149488667026,
      "learning_rate": 1.2834536799591679e-07,
      "loss": 0.2078,
      "step": 31036
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22685929949357655,
      "learning_rate": 1.2818701195727256e-07,
      "loss": 0.0691,
      "step": 31037
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2743289547862786,
      "learning_rate": 1.28028753040772e-07,
      "loss": 0.1868,
      "step": 31038
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.33543626706711427,
      "learning_rate": 1.2787059124797275e-07,
      "loss": 0.2488,
      "step": 31039
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9995787716114951,
      "learning_rate": 1.2771252658043243e-07,
      "loss": 0.2524,
      "step": 31040
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.723909932529711,
      "learning_rate": 1.2755455903970204e-07,
      "loss": 0.3644,
      "step": 31041
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4431460787399741,
      "learning_rate": 1.2739668862734145e-07,
      "loss": 0.3809,
      "step": 31042
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6979766278338512,
      "learning_rate": 1.272389153448983e-07,
      "loss": 0.3204,
      "step": 31043
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3436182742402594,
      "learning_rate": 1.27081239193928e-07,
      "loss": 0.202,
      "step": 31044
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3457292982618883,
      "learning_rate": 1.2692366017598157e-07,
      "loss": 0.2645,
      "step": 31045
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2092413164877967,
      "learning_rate": 1.2676617829260885e-07,
      "loss": 0.0665,
      "step": 31046
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9022098467981867,
      "learning_rate": 1.266087935453586e-07,
      "loss": 0.7115,
      "step": 31047
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3462060976875959,
      "learning_rate": 1.264515059357807e-07,
      "loss": 0.1534,
      "step": 31048
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5803131804987051,
      "learning_rate": 1.262943154654217e-07,
      "loss": 0.3244,
      "step": 31049
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.48167072731773825,
      "learning_rate": 1.2613722213582702e-07,
      "loss": 0.2426,
      "step": 31050
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46945323973014846,
      "learning_rate": 1.259802259485432e-07,
      "loss": 0.2385,
      "step": 31051
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7967159703890405,
      "learning_rate": 1.258233269051168e-07,
      "loss": 0.3784,
      "step": 31052
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31221895032278346,
      "learning_rate": 1.2566652500708655e-07,
      "loss": 0.177,
      "step": 31053
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30483336234052544,
      "learning_rate": 1.2550982025600123e-07,
      "loss": 0.1971,
      "step": 31054
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9611941745151085,
      "learning_rate": 1.253532126533974e-07,
      "loss": 0.0885,
      "step": 31055
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5141088220704786,
      "learning_rate": 1.2519670220081826e-07,
      "loss": 0.3123,
      "step": 31056
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.264270784078767,
      "learning_rate": 1.2504028889980368e-07,
      "loss": 0.1791,
      "step": 31057
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5084836043448567,
      "learning_rate": 1.2488397275189135e-07,
      "loss": 0.3066,
      "step": 31058
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1197401330899435,
      "learning_rate": 1.2472775375862e-07,
      "loss": 0.2793,
      "step": 31059
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.4282804481431934,
      "learning_rate": 1.2457163192152732e-07,
      "loss": 0.755,
      "step": 31060
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31071710266881003,
      "learning_rate": 1.2441560724214763e-07,
      "loss": 0.0703,
      "step": 31061
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7474491066697664,
      "learning_rate": 1.2425967972201637e-07,
      "loss": 0.2637,
      "step": 31062
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3185389124504415,
      "learning_rate": 1.2410384936266895e-07,
      "loss": 0.2286,
      "step": 31063
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.20183339607420095,
      "learning_rate": 1.239481161656364e-07,
      "loss": 0.0653,
      "step": 31064
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7961788186626707,
      "learning_rate": 1.2379248013245194e-07,
      "loss": 0.346,
      "step": 31065
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2788667800340588,
      "learning_rate": 1.2363694126464764e-07,
      "loss": 0.1682,
      "step": 31066
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.153626156248342,
      "learning_rate": 1.2348149956375234e-07,
      "loss": 0.4689,
      "step": 31067
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4218889924751761,
      "learning_rate": 1.2332615503129586e-07,
      "loss": 0.2333,
      "step": 31068
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.48823684489515695,
      "learning_rate": 1.2317090766880702e-07,
      "loss": 0.3406,
      "step": 31069
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.46501460866148475,
      "learning_rate": 1.2301575747781125e-07,
      "loss": 0.0663,
      "step": 31070
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47451102090726943,
      "learning_rate": 1.2286070445983845e-07,
      "loss": 0.2799,
      "step": 31071
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47334947891741525,
      "learning_rate": 1.227057486164107e-07,
      "loss": 0.1673,
      "step": 31072
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.33128034007991697,
      "learning_rate": 1.2255088994905462e-07,
      "loss": 0.1842,
      "step": 31073
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3549283950266127,
      "learning_rate": 1.2239612845929228e-07,
      "loss": 0.2325,
      "step": 31074
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5915892331456521,
      "learning_rate": 1.222414641486469e-07,
      "loss": 0.262,
      "step": 31075
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3973187563448532,
      "learning_rate": 1.2208689701864062e-07,
      "loss": 0.2074,
      "step": 31076
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.119438014549581,
      "learning_rate": 1.2193242707079334e-07,
      "loss": 0.5052,
      "step": 31077
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7485779394824301,
      "learning_rate": 1.2177805430662604e-07,
      "loss": 0.7944,
      "step": 31078
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6415097356959238,
      "learning_rate": 1.216237787276553e-07,
      "loss": 0.0587,
      "step": 31079
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.37458375608235345,
      "learning_rate": 1.2146960033540212e-07,
      "loss": 0.2629,
      "step": 31080
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31316517871521465,
      "learning_rate": 1.2131551913137973e-07,
      "loss": 0.2061,
      "step": 31081
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4455123845350338,
      "learning_rate": 1.2116153511710582e-07,
      "loss": 0.2121,
      "step": 31082
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6155218598172475,
      "learning_rate": 1.2100764829409695e-07,
      "loss": 0.2609,
      "step": 31083
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.35032823701689925,
      "learning_rate": 1.2085385866386413e-07,
      "loss": 0.1763,
      "step": 31084
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3498899025557475,
      "learning_rate": 1.207001662279217e-07,
      "loss": 0.2005,
      "step": 31085
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6229246421017147,
      "learning_rate": 1.2054657098778177e-07,
      "loss": 0.7084,
      "step": 31086
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.30860871709069093,
      "learning_rate": 1.203930729449565e-07,
      "loss": 0.2385,
      "step": 31087
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.5370019732169529,
      "learning_rate": 1.2023967210095354e-07,
      "loss": 0.4021,
      "step": 31088
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2860001801236563,
      "learning_rate": 1.2008636845728394e-07,
      "loss": 0.1678,
      "step": 31089
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7181596953181184,
      "learning_rate": 1.1993316201545647e-07,
      "loss": 0.1565,
      "step": 31090
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6498167012123012,
      "learning_rate": 1.1978005277697658e-07,
      "loss": 0.3359,
      "step": 31091
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3239401509274861,
      "learning_rate": 1.1962704074335196e-07,
      "loss": 0.2182,
      "step": 31092
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4282177922504922,
      "learning_rate": 1.1947412591608697e-07,
      "loss": 0.2514,
      "step": 31093
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.25911340019873885,
      "learning_rate": 1.193213082966871e-07,
      "loss": 0.1292,
      "step": 31094
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4550363904486031,
      "learning_rate": 1.1916858788665664e-07,
      "loss": 0.2529,
      "step": 31095
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.188681356433545,
      "learning_rate": 1.1901596468749444e-07,
      "loss": 0.4364,
      "step": 31096
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.750317317965833,
      "learning_rate": 1.1886343870070594e-07,
      "loss": 0.3364,
      "step": 31097
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4022135325658819,
      "learning_rate": 1.1871100992778994e-07,
      "loss": 0.1746,
      "step": 31098
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2883486983763328,
      "learning_rate": 1.1855867837024637e-07,
      "loss": 0.2575,
      "step": 31099
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6764623588545194,
      "learning_rate": 1.1840644402957401e-07,
      "loss": 0.2651,
      "step": 31100
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5832103042222363,
      "learning_rate": 1.1825430690727058e-07,
      "loss": 0.2038,
      "step": 31101
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2087950351858752,
      "learning_rate": 1.1810226700483262e-07,
      "loss": 0.0705,
      "step": 31102
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3153088732485138,
      "learning_rate": 1.1795032432375675e-07,
      "loss": 0.1968,
      "step": 31103
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3303054065623063,
      "learning_rate": 1.177984788655373e-07,
      "loss": 0.2588,
      "step": 31104
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.355716856866934,
      "learning_rate": 1.1764673063166642e-07,
      "loss": 0.2523,
      "step": 31105
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2739953963491633,
      "learning_rate": 1.174950796236407e-07,
      "loss": 0.3483,
      "step": 31106
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2785315051490523,
      "learning_rate": 1.1734352584294894e-07,
      "loss": 0.1663,
      "step": 31107
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8007916247323524,
      "learning_rate": 1.1719206929108329e-07,
      "loss": 0.4001,
      "step": 31108
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8402753026747194,
      "learning_rate": 1.1704070996953476e-07,
      "loss": 0.2476,
      "step": 31109
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34834187830810953,
      "learning_rate": 1.1688944787979106e-07,
      "loss": 0.2631,
      "step": 31110
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5064539552327565,
      "learning_rate": 1.16738283023341e-07,
      "loss": 0.0866,
      "step": 31111
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.29431124495912675,
      "learning_rate": 1.1658721540167228e-07,
      "loss": 0.225,
      "step": 31112
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2241420906312994,
      "learning_rate": 1.1643624501627038e-07,
      "loss": 0.443,
      "step": 31113
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4256585031800294,
      "learning_rate": 1.1628537186862077e-07,
      "loss": 0.238,
      "step": 31114
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5046601993828651,
      "learning_rate": 1.1613459596020671e-07,
      "loss": 0.217,
      "step": 31115
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2845012320325971,
      "learning_rate": 1.1598391729251369e-07,
      "loss": 0.1748,
      "step": 31116
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4800828570266963,
      "learning_rate": 1.1583333586702384e-07,
      "loss": 0.2928,
      "step": 31117
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.7121059970172345,
      "learning_rate": 1.1568285168521709e-07,
      "loss": 0.2644,
      "step": 31118
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.9853856698545627,
      "learning_rate": 1.155324647485756e-07,
      "loss": 0.911,
      "step": 31119
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.44984265148018154,
      "learning_rate": 1.1538217505857707e-07,
      "loss": 0.0992,
      "step": 31120
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.572228927375954,
      "learning_rate": 1.1523198261670254e-07,
      "loss": 0.3355,
      "step": 31121
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2808158983976052,
      "learning_rate": 1.1508188742442638e-07,
      "loss": 0.2261,
      "step": 31122
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4151918812149011,
      "learning_rate": 1.1493188948322853e-07,
      "loss": 0.2364,
      "step": 31123
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3614649569103121,
      "learning_rate": 1.1478198879458335e-07,
      "loss": 0.1209,
      "step": 31124
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0713278134922315,
      "learning_rate": 1.1463218535996412e-07,
      "loss": 0.4713,
      "step": 31125
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.403736125030302,
      "learning_rate": 1.1448247918084853e-07,
      "loss": 0.2178,
      "step": 31126
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8671934082084544,
      "learning_rate": 1.1433287025870544e-07,
      "loss": 0.4064,
      "step": 31127
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3191301149133522,
      "learning_rate": 1.1418335859500917e-07,
      "loss": 0.21,
      "step": 31128
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3402009589647499,
      "learning_rate": 1.140339441912297e-07,
      "loss": 0.0587,
      "step": 31129
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3710357545497585,
      "learning_rate": 1.1388462704883696e-07,
      "loss": 0.2472,
      "step": 31130
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.28497527198409867,
      "learning_rate": 1.1373540716929976e-07,
      "loss": 0.1431,
      "step": 31131
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.47614551459709287,
      "learning_rate": 1.1358628455408804e-07,
      "loss": 0.2569,
      "step": 31132
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42117619654835614,
      "learning_rate": 1.134372592046662e-07,
      "loss": 0.2072,
      "step": 31133
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3892517860298314,
      "learning_rate": 1.1328833112250193e-07,
      "loss": 0.2763,
      "step": 31134
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3551910677517689,
      "learning_rate": 1.1313950030906073e-07,
      "loss": 0.2196,
      "step": 31135
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6452257449146432,
      "learning_rate": 1.129907667658059e-07,
      "loss": 0.6694,
      "step": 31136
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1051561804924803,
      "learning_rate": 1.1284213049420179e-07,
      "loss": 0.0704,
      "step": 31137
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.9583117588944997,
      "learning_rate": 1.1269359149570946e-07,
      "loss": 0.3102,
      "step": 31138
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31168450489302485,
      "learning_rate": 1.125451497717911e-07,
      "loss": 0.1935,
      "step": 31139
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5197811894654835,
      "learning_rate": 1.1239680532390773e-07,
      "loss": 0.3422,
      "step": 31140
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.42048359270691005,
      "learning_rate": 1.122485581535171e-07,
      "loss": 0.2337,
      "step": 31141
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.6743485868018325,
      "learning_rate": 1.1210040826207802e-07,
      "loss": 0.249,
      "step": 31142
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3969801101945107,
      "learning_rate": 1.1195235565104934e-07,
      "loss": 0.2589,
      "step": 31143
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2677527172432703,
      "learning_rate": 1.1180440032188767e-07,
      "loss": 0.1303,
      "step": 31144
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.45049747045638877,
      "learning_rate": 1.1165654227604739e-07,
      "loss": 0.2184,
      "step": 31145
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3628988707321728,
      "learning_rate": 1.1150878151498401e-07,
      "loss": 0.2288,
      "step": 31146
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8795267136699733,
      "learning_rate": 1.1136111804015082e-07,
      "loss": 0.3447,
      "step": 31147
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.41945280860054635,
      "learning_rate": 1.1121355185299998e-07,
      "loss": 0.1762,
      "step": 31148
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5751705762226189,
      "learning_rate": 1.1106608295498588e-07,
      "loss": 0.3718,
      "step": 31149
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.800099152742316,
      "learning_rate": 1.1091871134755628e-07,
      "loss": 0.301,
      "step": 31150
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3244668167265625,
      "learning_rate": 1.107714370321622e-07,
      "loss": 0.2666,
      "step": 31151
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.688433602207875,
      "learning_rate": 1.1062426001025361e-07,
      "loss": 0.1723,
      "step": 31152
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3559174435404416,
      "learning_rate": 1.1047718028327714e-07,
      "loss": 0.2532,
      "step": 31153
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2269922392725048,
      "learning_rate": 1.103301978526794e-07,
      "loss": 0.0707,
      "step": 31154
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.6022452851790907,
      "learning_rate": 1.1018331271990812e-07,
      "loss": 0.5295,
      "step": 31155
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.2699325312453726,
      "learning_rate": 1.100365248864077e-07,
      "loss": 0.1807,
      "step": 31156
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.28479164500206783,
      "learning_rate": 1.0988983435362144e-07,
      "loss": 0.167,
      "step": 31157
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3370580388639614,
      "learning_rate": 1.0974324112299373e-07,
      "loss": 0.2724,
      "step": 31158
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.755431481316333,
      "learning_rate": 1.0959674519596675e-07,
      "loss": 0.2855,
      "step": 31159
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.282014273777268,
      "learning_rate": 1.0945034657398046e-07,
      "loss": 0.4211,
      "step": 31160
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.32016475902086766,
      "learning_rate": 1.0930404525847704e-07,
      "loss": 0.1068,
      "step": 31161
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2737739564118493,
      "learning_rate": 1.0915784125089424e-07,
      "loss": 0.2054,
      "step": 31162
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2956510655603542,
      "learning_rate": 1.09011734552672e-07,
      "loss": 0.1443,
      "step": 31163
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5131666984073353,
      "learning_rate": 1.0886572516524586e-07,
      "loss": 0.3275,
      "step": 31164
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.172048960129367,
      "learning_rate": 1.0871981309005353e-07,
      "loss": 0.2809,
      "step": 31165
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.3919884180435022,
      "learning_rate": 1.0857399832853055e-07,
      "loss": 0.2438,
      "step": 31166
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.4798016034521532,
      "learning_rate": 1.0842828088211133e-07,
      "loss": 0.2338,
      "step": 31167
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8814032980277874,
      "learning_rate": 1.0828266075222916e-07,
      "loss": 0.3653,
      "step": 31168
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.34609327967190684,
      "learning_rate": 1.0813713794031733e-07,
      "loss": 0.2387,
      "step": 31169
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31865576562765235,
      "learning_rate": 1.0799171244780692e-07,
      "loss": 0.1632,
      "step": 31170
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5954577866667496,
      "learning_rate": 1.0784638427612904e-07,
      "loss": 0.3276,
      "step": 31171
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.22220192331224983,
      "learning_rate": 1.0770115342671362e-07,
      "loss": 0.0701,
      "step": 31172
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.1248321493742008,
      "learning_rate": 1.0755601990098952e-07,
      "loss": 0.4314,
      "step": 31173
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.5056842719001039,
      "learning_rate": 1.0741098370038228e-07,
      "loss": 0.2086,
      "step": 31174
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.8986576468019546,
      "learning_rate": 1.0726604482632297e-07,
      "loss": 0.4049,
      "step": 31175
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.23934902755093285,
      "learning_rate": 1.071212032802349e-07,
      "loss": 0.2046,
      "step": 31176
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.71649622581289,
      "learning_rate": 1.0697645906354359e-07,
      "loss": 0.3348,
      "step": 31177
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0813779933997325,
      "learning_rate": 1.0683181217767347e-07,
      "loss": 0.3098,
      "step": 31178
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.7479746190793104,
      "learning_rate": 1.0668726262404671e-07,
      "loss": 0.6778,
      "step": 31179
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.2831497938902455,
      "learning_rate": 1.0654281040408665e-07,
      "loss": 0.1714,
      "step": 31180
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.26588656058124394,
      "learning_rate": 1.0639845551921324e-07,
      "loss": 0.1537,
      "step": 31181
    },
    {
      "epoch": 0.95,
      "grad_norm": 0.31428841332850255,
      "learning_rate": 1.0625419797084757e-07,
      "loss": 0.2271,
      "step": 31182
    },
    {
      "epoch": 0.95,
      "grad_norm": 1.0396547329097194,
      "learning_rate": 1.0611003776040852e-07,
      "loss": 0.285,
      "step": 31183
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.637971112944291,
      "learning_rate": 1.0596597488931381e-07,
      "loss": 0.2613,
      "step": 31184
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37930870946906775,
      "learning_rate": 1.0582200935898234e-07,
      "loss": 0.2137,
      "step": 31185
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9677828497481532,
      "learning_rate": 1.0567814117082853e-07,
      "loss": 0.4434,
      "step": 31186
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.31768939587294137,
      "learning_rate": 1.0553437032627012e-07,
      "loss": 0.224,
      "step": 31187
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8922236323273961,
      "learning_rate": 1.0539069682671931e-07,
      "loss": 0.4316,
      "step": 31188
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3225764959493482,
      "learning_rate": 1.0524712067359055e-07,
      "loss": 0.1766,
      "step": 31189
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2835809133826667,
      "learning_rate": 1.0510364186829713e-07,
      "loss": 0.16,
      "step": 31190
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5574360213206384,
      "learning_rate": 1.0496026041225015e-07,
      "loss": 0.065,
      "step": 31191
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7196167582777245,
      "learning_rate": 1.048169763068585e-07,
      "loss": 0.3157,
      "step": 31192
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.27103650195197865,
      "learning_rate": 1.0467378955353435e-07,
      "loss": 0.1757,
      "step": 31193
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3557040414035961,
      "learning_rate": 1.045307001536855e-07,
      "loss": 0.2543,
      "step": 31194
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9640692683185706,
      "learning_rate": 1.0438770810871968e-07,
      "loss": 0.2577,
      "step": 31195
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5421118149850566,
      "learning_rate": 1.0424481342004356e-07,
      "loss": 0.3992,
      "step": 31196
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0391371070539372,
      "learning_rate": 1.0410201608906268e-07,
      "loss": 0.4182,
      "step": 31197
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45781237086029875,
      "learning_rate": 1.0395931611718257e-07,
      "loss": 0.1655,
      "step": 31198
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24892595417779687,
      "learning_rate": 1.0381671350580768e-07,
      "loss": 0.1673,
      "step": 31199
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.30587678228554827,
      "learning_rate": 1.0367420825633911e-07,
      "loss": 0.2127,
      "step": 31200
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7817032921362328,
      "learning_rate": 1.0353180037018129e-07,
      "loss": 0.371,
      "step": 31201
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2645225201765364,
      "learning_rate": 1.0338948984873309e-07,
      "loss": 0.0667,
      "step": 31202
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4127336354775735,
      "learning_rate": 1.0324727669339563e-07,
      "loss": 0.2833,
      "step": 31203
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.963207928072987,
      "learning_rate": 1.0310516090556777e-07,
      "loss": 0.3253,
      "step": 31204
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3374810502769126,
      "learning_rate": 1.029631424866484e-07,
      "loss": 0.2646,
      "step": 31205
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3109246778755186,
      "learning_rate": 1.028212214380342e-07,
      "loss": 0.0367,
      "step": 31206
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38885689449514943,
      "learning_rate": 1.0267939776112069e-07,
      "loss": 0.224,
      "step": 31207
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33113619822498797,
      "learning_rate": 1.0253767145730454e-07,
      "loss": 0.1262,
      "step": 31208
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45822068223266293,
      "learning_rate": 1.0239604252797908e-07,
      "loss": 0.1569,
      "step": 31209
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9992710176291173,
      "learning_rate": 1.0225451097453764e-07,
      "loss": 0.4347,
      "step": 31210
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2831387872519398,
      "learning_rate": 1.0211307679837468e-07,
      "loss": 0.2009,
      "step": 31211
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40123493240218083,
      "learning_rate": 1.0197174000087795e-07,
      "loss": 0.2919,
      "step": 31212
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2874981304522357,
      "learning_rate": 1.018305005834419e-07,
      "loss": 0.4971,
      "step": 31213
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6328191047786111,
      "learning_rate": 1.0168935854745431e-07,
      "loss": 0.8041,
      "step": 31214
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7277451047102081,
      "learning_rate": 1.0154831389430297e-07,
      "loss": 0.0827,
      "step": 31215
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35243727080086945,
      "learning_rate": 1.0140736662537675e-07,
      "loss": 0.2293,
      "step": 31216
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3144825362978622,
      "learning_rate": 1.0126651674206123e-07,
      "loss": 0.2173,
      "step": 31217
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5126259333002864,
      "learning_rate": 1.0112576424574306e-07,
      "loss": 0.222,
      "step": 31218
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.619279323338027,
      "learning_rate": 1.0098510913780779e-07,
      "loss": 0.2499,
      "step": 31219
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3598118691618828,
      "learning_rate": 1.0084455141963656e-07,
      "loss": 0.1986,
      "step": 31220
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41846614353898587,
      "learning_rate": 1.0070409109261602e-07,
      "loss": 0.2115,
      "step": 31221
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0906658558358,
      "learning_rate": 1.0056372815812398e-07,
      "loss": 0.5027,
      "step": 31222
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3334090906042515,
      "learning_rate": 1.0042346261754376e-07,
      "loss": 0.2745,
      "step": 31223
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7349220841399406,
      "learning_rate": 1.0028329447225538e-07,
      "loss": 0.0686,
      "step": 31224
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5156872571837163,
      "learning_rate": 1.0014322372363772e-07,
      "loss": 0.2726,
      "step": 31225
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4604104017950706,
      "learning_rate": 1.0000325037306746e-07,
      "loss": 0.1872,
      "step": 31226
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.524836004647525,
      "learning_rate": 9.986337442192351e-08,
      "loss": 0.282,
      "step": 31227
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3394724419901141,
      "learning_rate": 9.972359587158142e-08,
      "loss": 0.2404,
      "step": 31228
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.32693863307900967,
      "learning_rate": 9.958391472341567e-08,
      "loss": 0.1497,
      "step": 31229
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3208181127249646,
      "learning_rate": 9.944433097880179e-08,
      "loss": 0.1833,
      "step": 31230
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.460201264388004,
      "learning_rate": 9.930484463911094e-08,
      "loss": 0.527,
      "step": 31231
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4839805424314545,
      "learning_rate": 9.916545570571757e-08,
      "loss": 0.4345,
      "step": 31232
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9849287797358166,
      "learning_rate": 9.902616417999167e-08,
      "loss": 0.7359,
      "step": 31233
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34511763491425673,
      "learning_rate": 9.88869700633055e-08,
      "loss": 0.1514,
      "step": 31234
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2655727542062488,
      "learning_rate": 9.874787335702684e-08,
      "loss": 0.1994,
      "step": 31235
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6983084478973607,
      "learning_rate": 9.860887406252351e-08,
      "loss": 0.3526,
      "step": 31236
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.952597187186485,
      "learning_rate": 9.84699721811655e-08,
      "loss": 0.2446,
      "step": 31237
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.245997623011013,
      "learning_rate": 9.833116771431616e-08,
      "loss": 0.1488,
      "step": 31238
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33537662185877365,
      "learning_rate": 9.819246066334443e-08,
      "loss": 0.1946,
      "step": 31239
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.429809266881979,
      "learning_rate": 9.805385102961362e-08,
      "loss": 0.7973,
      "step": 31240
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.31719592804909463,
      "learning_rate": 9.7915338814486e-08,
      "loss": 0.2221,
      "step": 31241
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.138126266208712,
      "learning_rate": 9.777692401932604e-08,
      "loss": 0.42,
      "step": 31242
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3612246549392581,
      "learning_rate": 9.763860664549484e-08,
      "loss": 0.1477,
      "step": 31243
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45354344463024243,
      "learning_rate": 9.750038669435358e-08,
      "loss": 0.2541,
      "step": 31244
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9527162174742463,
      "learning_rate": 9.736226416726113e-08,
      "loss": 0.2552,
      "step": 31245
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5173270565228586,
      "learning_rate": 9.722423906557755e-08,
      "loss": 0.2642,
      "step": 31246
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.233042060009211,
      "learning_rate": 9.70863113906606e-08,
      "loss": 0.1343,
      "step": 31247
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.24563921988777151,
      "learning_rate": 9.694848114386701e-08,
      "loss": 0.158,
      "step": 31248
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5450705015660366,
      "learning_rate": 9.681074832655236e-08,
      "loss": 0.789,
      "step": 31249
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0362233760766226,
      "learning_rate": 9.66731129400722e-08,
      "loss": 0.396,
      "step": 31250
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9191956070030635,
      "learning_rate": 9.653557498578215e-08,
      "loss": 0.3679,
      "step": 31251
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3355500069557637,
      "learning_rate": 9.639813446503222e-08,
      "loss": 0.1605,
      "step": 31252
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35528081890052204,
      "learning_rate": 9.626079137917577e-08,
      "loss": 0.2585,
      "step": 31253
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41176112486897515,
      "learning_rate": 9.612354572956616e-08,
      "loss": 0.2371,
      "step": 31254
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.2099285780468927,
      "learning_rate": 9.598639751755012e-08,
      "loss": 0.4073,
      "step": 31255
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2275280546951648,
      "learning_rate": 9.584934674447987e-08,
      "loss": 0.0956,
      "step": 31256
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4995571038291137,
      "learning_rate": 9.571239341170214e-08,
      "loss": 0.2691,
      "step": 31257
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4174053492260717,
      "learning_rate": 9.55755375205647e-08,
      "loss": 0.2399,
      "step": 31258
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2691599528622285,
      "learning_rate": 9.543877907241428e-08,
      "loss": 0.2205,
      "step": 31259
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.149601410813594,
      "learning_rate": 9.530211806859647e-08,
      "loss": 0.06,
      "step": 31260
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5692894610428577,
      "learning_rate": 9.516555451045462e-08,
      "loss": 0.1702,
      "step": 31261
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4032397902195499,
      "learning_rate": 9.502908839933322e-08,
      "loss": 0.2736,
      "step": 31262
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0678675512699403,
      "learning_rate": 9.489271973657677e-08,
      "loss": 0.2221,
      "step": 31263
    },
    {
      "epoch": 0.96,
      "grad_norm": 6.686811654569086,
      "learning_rate": 9.475644852352195e-08,
      "loss": 0.6579,
      "step": 31264
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2670568609229876,
      "learning_rate": 9.462027476151437e-08,
      "loss": 0.2001,
      "step": 31265
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38048238593972555,
      "learning_rate": 9.448419845189294e-08,
      "loss": 0.2681,
      "step": 31266
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0845492251068458,
      "learning_rate": 9.434821959599216e-08,
      "loss": 0.5563,
      "step": 31267
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4091511361303328,
      "learning_rate": 9.42123381951554e-08,
      "loss": 0.2063,
      "step": 31268
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4056285234844382,
      "learning_rate": 9.407655425071605e-08,
      "loss": 0.182,
      "step": 31269
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.39307671623536483,
      "learning_rate": 9.39408677640119e-08,
      "loss": 0.2611,
      "step": 31270
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3053205643819058,
      "learning_rate": 9.380527873637635e-08,
      "loss": 0.223,
      "step": 31271
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5963181145048997,
      "learning_rate": 9.366978716914388e-08,
      "loss": 0.5161,
      "step": 31272
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7086274345366312,
      "learning_rate": 9.353439306364897e-08,
      "loss": 0.1542,
      "step": 31273
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6786569223098665,
      "learning_rate": 9.339909642122058e-08,
      "loss": 0.1677,
      "step": 31274
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38518558023833527,
      "learning_rate": 9.326389724319207e-08,
      "loss": 0.2605,
      "step": 31275
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1297379042188551,
      "learning_rate": 9.312879553089349e-08,
      "loss": 0.4205,
      "step": 31276
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2621550256833759,
      "learning_rate": 9.299379128565378e-08,
      "loss": 0.2076,
      "step": 31277
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.411911141869798,
      "learning_rate": 9.285888450880077e-08,
      "loss": 0.177,
      "step": 31278
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3731915369085593,
      "learning_rate": 9.272407520166115e-08,
      "loss": 0.2277,
      "step": 31279
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.47837254125476336,
      "learning_rate": 9.25893633655639e-08,
      "loss": 0.2155,
      "step": 31280
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.209891289699327,
      "learning_rate": 9.245474900183015e-08,
      "loss": 0.3029,
      "step": 31281
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.30512319261821147,
      "learning_rate": 9.232023211178887e-08,
      "loss": 0.2176,
      "step": 31282
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9351017602692014,
      "learning_rate": 9.218581269675897e-08,
      "loss": 0.4537,
      "step": 31283
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42427172246437533,
      "learning_rate": 9.205149075806607e-08,
      "loss": 0.177,
      "step": 31284
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5768934297434927,
      "learning_rate": 9.191726629703135e-08,
      "loss": 0.3441,
      "step": 31285
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7859746553596897,
      "learning_rate": 9.178313931497485e-08,
      "loss": 0.2699,
      "step": 31286
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8069070869478288,
      "learning_rate": 9.164910981321551e-08,
      "loss": 0.2412,
      "step": 31287
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3370391940836369,
      "learning_rate": 9.151517779307228e-08,
      "loss": 0.2766,
      "step": 31288
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22800269265161466,
      "learning_rate": 9.138134325586301e-08,
      "loss": 0.1512,
      "step": 31289
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.521595133375761,
      "learning_rate": 9.124760620290551e-08,
      "loss": 0.2457,
      "step": 31290
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4254366162866774,
      "learning_rate": 9.111396663551319e-08,
      "loss": 0.0623,
      "step": 31291
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.173663262082362,
      "learning_rate": 9.098042455500278e-08,
      "loss": 0.5214,
      "step": 31292
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3379432749421552,
      "learning_rate": 9.084697996268654e-08,
      "loss": 0.165,
      "step": 31293
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34477083387536395,
      "learning_rate": 9.071363285988011e-08,
      "loss": 0.2679,
      "step": 31294
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4922524633242988,
      "learning_rate": 9.058038324789242e-08,
      "loss": 0.2403,
      "step": 31295
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7251539046178908,
      "learning_rate": 9.044723112803466e-08,
      "loss": 0.3504,
      "step": 31296
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1936816960508409,
      "learning_rate": 9.03141765016191e-08,
      "loss": 0.0653,
      "step": 31297
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.29699055206649827,
      "learning_rate": 9.01812193699525e-08,
      "loss": 0.2154,
      "step": 31298
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6486008255695582,
      "learning_rate": 9.00483597343449e-08,
      "loss": 0.0198,
      "step": 31299
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.32396952265479345,
      "learning_rate": 8.991559759610191e-08,
      "loss": 0.2297,
      "step": 31300
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.460162882983859,
      "learning_rate": 8.978293295652918e-08,
      "loss": 0.6029,
      "step": 31301
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.28309647158477025,
      "learning_rate": 8.965036581693343e-08,
      "loss": 0.1618,
      "step": 31302
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5700893822105304,
      "learning_rate": 8.951789617861917e-08,
      "loss": 0.7507,
      "step": 31303
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6784342264068975,
      "learning_rate": 8.938552404288758e-08,
      "loss": 0.2711,
      "step": 31304
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9534780248477218,
      "learning_rate": 8.925324941104319e-08,
      "loss": 0.3927,
      "step": 31305
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2621186440464737,
      "learning_rate": 8.912107228438605e-08,
      "loss": 0.1874,
      "step": 31306
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5770610822833236,
      "learning_rate": 8.898899266421624e-08,
      "loss": 0.3492,
      "step": 31307
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22812630835610154,
      "learning_rate": 8.885701055183493e-08,
      "loss": 0.1095,
      "step": 31308
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3917800462039018,
      "learning_rate": 8.872512594853889e-08,
      "loss": 0.4318,
      "step": 31309
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1045071625292577,
      "learning_rate": 8.859333885562594e-08,
      "loss": 0.5041,
      "step": 31310
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6264734315226705,
      "learning_rate": 8.846164927439394e-08,
      "loss": 0.275,
      "step": 31311
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2992160601065139,
      "learning_rate": 8.833005720613741e-08,
      "loss": 0.2153,
      "step": 31312
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4056876292946566,
      "learning_rate": 8.819856265215088e-08,
      "loss": 0.2471,
      "step": 31313
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8295694714988289,
      "learning_rate": 8.806716561372775e-08,
      "loss": 0.5378,
      "step": 31314
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48738841965264157,
      "learning_rate": 8.793586609216254e-08,
      "loss": 0.0721,
      "step": 31315
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37475147944452786,
      "learning_rate": 8.780466408874533e-08,
      "loss": 0.2641,
      "step": 31316
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23224729400844304,
      "learning_rate": 8.76735596047673e-08,
      "loss": 0.0712,
      "step": 31317
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3472506283380045,
      "learning_rate": 8.754255264151857e-08,
      "loss": 0.2786,
      "step": 31318
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1385842300647477,
      "learning_rate": 8.741164320028694e-08,
      "loss": 0.439,
      "step": 31319
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5967534819032965,
      "learning_rate": 8.728083128236142e-08,
      "loss": 0.2797,
      "step": 31320
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42184827009276316,
      "learning_rate": 8.715011688902985e-08,
      "loss": 0.2235,
      "step": 31321
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0445756088694595,
      "learning_rate": 8.701950002157567e-08,
      "loss": 0.5197,
      "step": 31322
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.740717057876132,
      "learning_rate": 8.68889806812867e-08,
      "loss": 0.2536,
      "step": 31323
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33223704852476316,
      "learning_rate": 8.675855886944528e-08,
      "loss": 0.2821,
      "step": 31324
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.30893854869857446,
      "learning_rate": 8.662823458733593e-08,
      "loss": 0.1738,
      "step": 31325
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.1410042349609756,
      "learning_rate": 8.649800783623874e-08,
      "loss": 0.0672,
      "step": 31326
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1807849172682217,
      "learning_rate": 8.6367878617436e-08,
      "loss": 0.3183,
      "step": 31327
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8305631999802009,
      "learning_rate": 8.623784693220893e-08,
      "loss": 0.2851,
      "step": 31328
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.403684861091579,
      "learning_rate": 8.61079127818365e-08,
      "loss": 0.2343,
      "step": 31329
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33602598575803005,
      "learning_rate": 8.597807616759546e-08,
      "loss": 0.212,
      "step": 31330
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4714965984179399,
      "learning_rate": 8.584833709076368e-08,
      "loss": 0.3013,
      "step": 31331
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0621370717027745,
      "learning_rate": 8.571869555262013e-08,
      "loss": 0.4162,
      "step": 31332
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9417646070738954,
      "learning_rate": 8.558915155443603e-08,
      "loss": 0.3959,
      "step": 31333
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3779521160877876,
      "learning_rate": 8.545970509748924e-08,
      "loss": 0.1623,
      "step": 31334
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2622589922665971,
      "learning_rate": 8.533035618305207e-08,
      "loss": 0.1608,
      "step": 31335
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35428979116394865,
      "learning_rate": 8.520110481239796e-08,
      "loss": 0.1944,
      "step": 31336
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8662415377856082,
      "learning_rate": 8.5071950986797e-08,
      "loss": 0.4341,
      "step": 31337
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7140759804032683,
      "learning_rate": 8.494289470752037e-08,
      "loss": 0.1926,
      "step": 31338
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.33316748698856224,
      "learning_rate": 8.48139359758382e-08,
      "loss": 0.1899,
      "step": 31339
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4196757277202559,
      "learning_rate": 8.468507479301947e-08,
      "loss": 0.742,
      "step": 31340
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5208715024880689,
      "learning_rate": 8.455631116033203e-08,
      "loss": 0.212,
      "step": 31341
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5282147481342951,
      "learning_rate": 8.442764507904156e-08,
      "loss": 0.3286,
      "step": 31342
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2873999049913868,
      "learning_rate": 8.42990765504137e-08,
      "loss": 0.1731,
      "step": 31343
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3085168809230531,
      "learning_rate": 8.417060557571521e-08,
      "loss": 0.1542,
      "step": 31344
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0486898549763604,
      "learning_rate": 8.404223215620955e-08,
      "loss": 0.2159,
      "step": 31345
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8747313441890477,
      "learning_rate": 8.391395629315902e-08,
      "loss": 0.3875,
      "step": 31346
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46893574371047275,
      "learning_rate": 8.378577798782484e-08,
      "loss": 0.1651,
      "step": 31347
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.29062071694533864,
      "learning_rate": 8.365769724146933e-08,
      "loss": 0.2507,
      "step": 31348
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4137780128501106,
      "learning_rate": 8.35297140553526e-08,
      "loss": 0.2167,
      "step": 31349
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5799254394007398,
      "learning_rate": 8.340182843073253e-08,
      "loss": 0.7316,
      "step": 31350
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4366878981096272,
      "learning_rate": 8.327404036887032e-08,
      "loss": 0.1409,
      "step": 31351
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41939826811858605,
      "learning_rate": 8.314634987101944e-08,
      "loss": 0.241,
      "step": 31352
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.26808758265020627,
      "learning_rate": 8.301875693843775e-08,
      "loss": 0.1317,
      "step": 31353
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.25002565795013026,
      "learning_rate": 8.289126157238203e-08,
      "loss": 0.1757,
      "step": 31354
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7312139508748776,
      "learning_rate": 8.276386377410462e-08,
      "loss": 0.3488,
      "step": 31355
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.42442901914602776,
      "learning_rate": 8.263656354485894e-08,
      "loss": 0.1787,
      "step": 31356
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5614389767840221,
      "learning_rate": 8.250936088589844e-08,
      "loss": 0.2952,
      "step": 31357
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0586242957784462,
      "learning_rate": 8.238225579847436e-08,
      "loss": 0.5621,
      "step": 31358
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48668227251312995,
      "learning_rate": 8.22552482838368e-08,
      "loss": 0.3415,
      "step": 31359
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3730950852107492,
      "learning_rate": 8.212833834323474e-08,
      "loss": 0.1653,
      "step": 31360
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9066622158404716,
      "learning_rate": 8.200152597791944e-08,
      "loss": 0.4,
      "step": 31361
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3468923168265379,
      "learning_rate": 8.187481118913432e-08,
      "loss": 0.1942,
      "step": 31362
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5061703998259636,
      "learning_rate": 8.174819397812949e-08,
      "loss": 0.2307,
      "step": 31363
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4465265064095106,
      "learning_rate": 8.162167434614954e-08,
      "loss": 0.1812,
      "step": 31364
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48105198917495634,
      "learning_rate": 8.149525229443899e-08,
      "loss": 0.301,
      "step": 31365
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.29056914561481806,
      "learning_rate": 8.136892782424133e-08,
      "loss": 0.1788,
      "step": 31366
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45917291144730565,
      "learning_rate": 8.124270093679998e-08,
      "loss": 0.2572,
      "step": 31367
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4949321904078032,
      "learning_rate": 8.111657163335728e-08,
      "loss": 0.4087,
      "step": 31368
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5399562516657371,
      "learning_rate": 8.099053991515227e-08,
      "loss": 0.0151,
      "step": 31369
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6752743455935293,
      "learning_rate": 8.086460578342726e-08,
      "loss": 0.3003,
      "step": 31370
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.28975650145113363,
      "learning_rate": 8.073876923941904e-08,
      "loss": 0.2117,
      "step": 31371
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4897009516800008,
      "learning_rate": 8.061303028436774e-08,
      "loss": 0.2973,
      "step": 31372
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4409912884774642,
      "learning_rate": 8.048738891950902e-08,
      "loss": 0.1398,
      "step": 31373
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3101239636866351,
      "learning_rate": 8.036184514607858e-08,
      "loss": 0.1773,
      "step": 31374
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2889857379905117,
      "learning_rate": 8.023639896531433e-08,
      "loss": 0.181,
      "step": 31375
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7522313234415794,
      "learning_rate": 8.011105037844525e-08,
      "loss": 0.7703,
      "step": 31376
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5974710083581727,
      "learning_rate": 7.998579938671036e-08,
      "loss": 0.1997,
      "step": 31377
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4757062156131133,
      "learning_rate": 7.98606459913387e-08,
      "loss": 0.2918,
      "step": 31378
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34963571879903,
      "learning_rate": 7.973559019356147e-08,
      "loss": 0.1591,
      "step": 31379
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5103771282663365,
      "learning_rate": 7.961063199461106e-08,
      "loss": 0.206,
      "step": 31380
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6830639624605379,
      "learning_rate": 7.948577139571422e-08,
      "loss": 0.3517,
      "step": 31381
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2806142630398415,
      "learning_rate": 7.936100839810113e-08,
      "loss": 0.1143,
      "step": 31382
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3256593024218895,
      "learning_rate": 7.923634300299743e-08,
      "loss": 0.2599,
      "step": 31383
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3767616957664967,
      "learning_rate": 7.911177521163326e-08,
      "loss": 0.1583,
      "step": 31384
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5922284972294397,
      "learning_rate": 7.898730502522988e-08,
      "loss": 0.3941,
      "step": 31385
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1923608172101214,
      "learning_rate": 7.886293244501519e-08,
      "loss": 0.1971,
      "step": 31386
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9555966657207711,
      "learning_rate": 7.873865747221154e-08,
      "loss": 0.4051,
      "step": 31387
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.28438617492704094,
      "learning_rate": 7.861448010804018e-08,
      "loss": 0.0672,
      "step": 31388
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3854835626137082,
      "learning_rate": 7.849040035372458e-08,
      "loss": 0.2583,
      "step": 31389
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3451993112780352,
      "learning_rate": 7.836641821048596e-08,
      "loss": 0.2305,
      "step": 31390
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5239723085496075,
      "learning_rate": 7.824253367954226e-08,
      "loss": 0.2085,
      "step": 31391
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2612971452522507,
      "learning_rate": 7.811874676211362e-08,
      "loss": 0.0955,
      "step": 31392
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.30205858583935147,
      "learning_rate": 7.799505745941683e-08,
      "loss": 0.1718,
      "step": 31393
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.336833829350822,
      "learning_rate": 7.78714657726698e-08,
      "loss": 0.7048,
      "step": 31394
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.31300102613924763,
      "learning_rate": 7.774797170308824e-08,
      "loss": 0.2051,
      "step": 31395
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.6645619612655995,
      "learning_rate": 7.762457525188672e-08,
      "loss": 0.6725,
      "step": 31396
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5632991660849594,
      "learning_rate": 7.750127642028094e-08,
      "loss": 0.1906,
      "step": 31397
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.41555161149277325,
      "learning_rate": 7.737807520948104e-08,
      "loss": 0.271,
      "step": 31398
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9550477532047332,
      "learning_rate": 7.72549716207005e-08,
      "loss": 0.2253,
      "step": 31399
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1851058438637263,
      "learning_rate": 7.713196565515169e-08,
      "loss": 0.3829,
      "step": 31400
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.22488012444332403,
      "learning_rate": 7.700905731404251e-08,
      "loss": 0.1774,
      "step": 31401
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4467773688210067,
      "learning_rate": 7.688624659858312e-08,
      "loss": 0.2824,
      "step": 31402
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4930078884862261,
      "learning_rate": 7.676353350998367e-08,
      "loss": 0.2404,
      "step": 31403
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46909376919819284,
      "learning_rate": 7.664091804944874e-08,
      "loss": 0.2499,
      "step": 31404
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7448349121094054,
      "learning_rate": 7.651840021818513e-08,
      "loss": 0.2562,
      "step": 31405
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38749426097082923,
      "learning_rate": 7.639598001739967e-08,
      "loss": 0.1765,
      "step": 31406
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6088378054095487,
      "learning_rate": 7.627365744829585e-08,
      "loss": 0.2765,
      "step": 31407
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34295174452446736,
      "learning_rate": 7.615143251207602e-08,
      "loss": 0.2183,
      "step": 31408
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3623980017840085,
      "learning_rate": 7.602930520994479e-08,
      "loss": 0.5096,
      "step": 31409
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5224531844445429,
      "learning_rate": 7.590727554310229e-08,
      "loss": 0.2026,
      "step": 31410
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44487221920228776,
      "learning_rate": 7.578534351274979e-08,
      "loss": 0.2593,
      "step": 31411
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4643714031325716,
      "learning_rate": 7.566350912008635e-08,
      "loss": 0.2317,
      "step": 31412
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.28191682264474516,
      "learning_rate": 7.5541772366311e-08,
      "loss": 0.2388,
      "step": 31413
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44567344157958055,
      "learning_rate": 7.542013325262054e-08,
      "loss": 0.1869,
      "step": 31414
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9865059667731865,
      "learning_rate": 7.529859178021293e-08,
      "loss": 0.4807,
      "step": 31415
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2960094466157756,
      "learning_rate": 7.517714795028386e-08,
      "loss": 0.1698,
      "step": 31416
    },
    {
      "epoch": 0.96,
      "grad_norm": 2.047483691319593,
      "learning_rate": 7.505580176402683e-08,
      "loss": 0.5739,
      "step": 31417
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.477388826467573,
      "learning_rate": 7.493455322263754e-08,
      "loss": 0.1869,
      "step": 31418
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44734390838629257,
      "learning_rate": 7.481340232730727e-08,
      "loss": 0.2419,
      "step": 31419
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36781444266679364,
      "learning_rate": 7.46923490792284e-08,
      "loss": 0.2158,
      "step": 31420
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5010902357510294,
      "learning_rate": 7.45713934795933e-08,
      "loss": 0.2552,
      "step": 31421
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4263536078648552,
      "learning_rate": 7.445053552958881e-08,
      "loss": 0.2292,
      "step": 31422
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8093689242130637,
      "learning_rate": 7.432977523040729e-08,
      "loss": 0.2403,
      "step": 31423
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4325459675272671,
      "learning_rate": 7.42091125832356e-08,
      "loss": 0.1609,
      "step": 31424
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23710400709703117,
      "learning_rate": 7.408854758925943e-08,
      "loss": 0.1893,
      "step": 31425
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5638994894677025,
      "learning_rate": 7.396808024966672e-08,
      "loss": 0.3091,
      "step": 31426
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5147701440182756,
      "learning_rate": 7.384771056564321e-08,
      "loss": 0.015,
      "step": 31427
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5602851784383844,
      "learning_rate": 7.372743853837128e-08,
      "loss": 0.7556,
      "step": 31428
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.333151632070787,
      "learning_rate": 7.360726416903441e-08,
      "loss": 0.1628,
      "step": 31429
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5787825731159347,
      "learning_rate": 7.348718745881611e-08,
      "loss": 0.3705,
      "step": 31430
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4384944643294101,
      "learning_rate": 7.336720840889655e-08,
      "loss": 0.2133,
      "step": 31431
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46854308873573447,
      "learning_rate": 7.3247327020457e-08,
      "loss": 0.2484,
      "step": 31432
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.23871034240088504,
      "learning_rate": 7.312754329467653e-08,
      "loss": 0.14,
      "step": 31433
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3366429768600962,
      "learning_rate": 7.300785723273419e-08,
      "loss": 0.1807,
      "step": 31434
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.239521069078991,
      "learning_rate": 7.288826883580569e-08,
      "loss": 0.512,
      "step": 31435
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.45878448390797,
      "learning_rate": 7.27687781050701e-08,
      "loss": 0.1942,
      "step": 31436
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.47919631531679674,
      "learning_rate": 7.264938504170205e-08,
      "loss": 0.3279,
      "step": 31437
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.27216092792735513,
      "learning_rate": 7.253008964687502e-08,
      "loss": 0.0676,
      "step": 31438
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4337892423826876,
      "learning_rate": 7.241089192176365e-08,
      "loss": 0.2601,
      "step": 31439
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.901378493596788,
      "learning_rate": 7.229179186754032e-08,
      "loss": 0.278,
      "step": 31440
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.502384762351275,
      "learning_rate": 7.217278948537743e-08,
      "loss": 0.2388,
      "step": 31441
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2804768493693863,
      "learning_rate": 7.205388477644515e-08,
      "loss": 0.1605,
      "step": 31442
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3928183342176327,
      "learning_rate": 7.193507774191366e-08,
      "loss": 0.2245,
      "step": 31443
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34271477886900265,
      "learning_rate": 7.181636838295091e-08,
      "loss": 0.2135,
      "step": 31444
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1486967218486466,
      "learning_rate": 7.169775670072598e-08,
      "loss": 0.1676,
      "step": 31445
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7734307829664056,
      "learning_rate": 7.15792426964057e-08,
      "loss": 0.7584,
      "step": 31446
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48738251533768473,
      "learning_rate": 7.14608263711558e-08,
      "loss": 0.175,
      "step": 31447
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.38410154600744534,
      "learning_rate": 7.134250772614093e-08,
      "loss": 0.2569,
      "step": 31448
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.32090865461564655,
      "learning_rate": 7.122428676252568e-08,
      "loss": 0.2197,
      "step": 31449
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5857507586598025,
      "learning_rate": 7.110616348147248e-08,
      "loss": 0.78,
      "step": 31450
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3391078151790091,
      "learning_rate": 7.098813788414372e-08,
      "loss": 0.0783,
      "step": 31451
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3620350042554466,
      "learning_rate": 7.087020997170069e-08,
      "loss": 0.2714,
      "step": 31452
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3736961377096404,
      "learning_rate": 7.075237974530468e-08,
      "loss": 0.0617,
      "step": 31453
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.582210903980481,
      "learning_rate": 7.063464720611258e-08,
      "loss": 0.6082,
      "step": 31454
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3066881446166088,
      "learning_rate": 7.051701235528451e-08,
      "loss": 0.234,
      "step": 31455
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.6752033059837228,
      "learning_rate": 7.03994751939785e-08,
      "loss": 0.3372,
      "step": 31456
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3868814389255147,
      "learning_rate": 7.028203572334802e-08,
      "loss": 0.2245,
      "step": 31457
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7967649841351265,
      "learning_rate": 7.016469394455105e-08,
      "loss": 0.2415,
      "step": 31458
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5731547577224907,
      "learning_rate": 7.004744985874112e-08,
      "loss": 0.5906,
      "step": 31459
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3084424592187974,
      "learning_rate": 6.993030346707064e-08,
      "loss": 0.2003,
      "step": 31460
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.37502666629042897,
      "learning_rate": 6.981325477069533e-08,
      "loss": 0.2561,
      "step": 31461
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.14757997280196472,
      "learning_rate": 6.96963037707632e-08,
      "loss": 0.0718,
      "step": 31462
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.8453761038209857,
      "learning_rate": 6.957945046842662e-08,
      "loss": 0.6997,
      "step": 31463
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7093311417327745,
      "learning_rate": 6.94626948648347e-08,
      "loss": 0.2899,
      "step": 31464
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8334072744774446,
      "learning_rate": 6.934603696113762e-08,
      "loss": 0.36,
      "step": 31465
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.32481165798154055,
      "learning_rate": 6.922947675848002e-08,
      "loss": 0.1814,
      "step": 31466
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.36060132600523104,
      "learning_rate": 6.911301425801098e-08,
      "loss": 0.2837,
      "step": 31467
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0180894233919047,
      "learning_rate": 6.899664946087737e-08,
      "loss": 0.1922,
      "step": 31468
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4232990723560551,
      "learning_rate": 6.888038236822048e-08,
      "loss": 0.2105,
      "step": 31469
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3399499212583027,
      "learning_rate": 6.876421298118718e-08,
      "loss": 0.1501,
      "step": 31470
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.31551420520691487,
      "learning_rate": 6.864814130091879e-08,
      "loss": 0.1211,
      "step": 31471
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4997037477748012,
      "learning_rate": 6.853216732855883e-08,
      "loss": 0.2892,
      "step": 31472
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4633829422879332,
      "learning_rate": 6.841629106524638e-08,
      "loss": 0.2222,
      "step": 31473
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.734538723717876,
      "learning_rate": 6.830051251212278e-08,
      "loss": 0.3054,
      "step": 31474
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.35630993288904017,
      "learning_rate": 6.818483167032597e-08,
      "loss": 0.1907,
      "step": 31475
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.4041728719433106,
      "learning_rate": 6.806924854099617e-08,
      "loss": 0.6885,
      "step": 31476
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5117206210356263,
      "learning_rate": 6.795376312526802e-08,
      "loss": 0.0165,
      "step": 31477
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3644581777088148,
      "learning_rate": 6.78383754242784e-08,
      "loss": 0.2549,
      "step": 31478
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3789140221562558,
      "learning_rate": 6.772308543916417e-08,
      "loss": 0.1675,
      "step": 31479
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.252375329160529,
      "learning_rate": 6.760789317105776e-08,
      "loss": 0.1667,
      "step": 31480
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.5277949190064668,
      "learning_rate": 6.749279862109382e-08,
      "loss": 0.0517,
      "step": 31481
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7513586494508326,
      "learning_rate": 6.737780179040365e-08,
      "loss": 0.4069,
      "step": 31482
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.5235639987070935,
      "learning_rate": 6.726290268011748e-08,
      "loss": 0.1857,
      "step": 31483
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2683978548442075,
      "learning_rate": 6.714810129136883e-08,
      "loss": 0.2145,
      "step": 31484
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.530468020711504,
      "learning_rate": 6.703339762528572e-08,
      "loss": 0.3247,
      "step": 31485
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.8826648102323956,
      "learning_rate": 6.691879168299497e-08,
      "loss": 0.179,
      "step": 31486
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4284717097250049,
      "learning_rate": 6.680428346562684e-08,
      "loss": 0.1889,
      "step": 31487
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3183886294899172,
      "learning_rate": 6.668987297430596e-08,
      "loss": 0.1485,
      "step": 31488
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3503231481729244,
      "learning_rate": 6.65755602101592e-08,
      "loss": 0.2098,
      "step": 31489
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.44688575027505834,
      "learning_rate": 6.646134517430903e-08,
      "loss": 0.2227,
      "step": 31490
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.46142105947512735,
      "learning_rate": 6.634722786788227e-08,
      "loss": 0.3068,
      "step": 31491
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7326854741229677,
      "learning_rate": 6.623320829200031e-08,
      "loss": 0.2133,
      "step": 31492
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.48314583548101947,
      "learning_rate": 6.611928644778331e-08,
      "loss": 0.2617,
      "step": 31493
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.4671079204187386,
      "learning_rate": 6.600546233635485e-08,
      "loss": 0.2553,
      "step": 31494
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.7171049063051826,
      "learning_rate": 6.58917359588318e-08,
      "loss": 0.836,
      "step": 31495
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.34270214410634503,
      "learning_rate": 6.57781073163366e-08,
      "loss": 0.2191,
      "step": 31496
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.40642204784648056,
      "learning_rate": 6.566457640998391e-08,
      "loss": 0.1593,
      "step": 31497
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3708444291016932,
      "learning_rate": 6.555114324089173e-08,
      "loss": 0.2563,
      "step": 31498
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.43602402119867306,
      "learning_rate": 6.543780781017695e-08,
      "loss": 0.1833,
      "step": 31499
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.9239684162146904,
      "learning_rate": 6.53245701189531e-08,
      "loss": 0.4261,
      "step": 31500
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.49151858336352056,
      "learning_rate": 6.521143016833598e-08,
      "loss": 0.1976,
      "step": 31501
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3533688286654862,
      "learning_rate": 6.50983879594369e-08,
      "loss": 0.2517,
      "step": 31502
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.3134051918055765,
      "learning_rate": 6.498544349336944e-08,
      "loss": 0.2299,
      "step": 31503
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.3101201938165532,
      "learning_rate": 6.48725967712438e-08,
      "loss": 0.5764,
      "step": 31504
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.1106499963797425,
      "learning_rate": 6.475984779417132e-08,
      "loss": 0.1067,
      "step": 31505
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.7587760471372629,
      "learning_rate": 6.464719656326001e-08,
      "loss": 0.3586,
      "step": 31506
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.2990189289964923,
      "learning_rate": 6.453464307961898e-08,
      "loss": 0.1762,
      "step": 31507
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.0525448736062075,
      "learning_rate": 6.442218734435624e-08,
      "loss": 0.409,
      "step": 31508
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.31222191672919597,
      "learning_rate": 6.430982935857533e-08,
      "loss": 0.2207,
      "step": 31509
    },
    {
      "epoch": 0.96,
      "grad_norm": 0.19758053186495855,
      "learning_rate": 6.419756912338537e-08,
      "loss": 0.1063,
      "step": 31510
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.387156649961728,
      "learning_rate": 6.408540663988772e-08,
      "loss": 0.2489,
      "step": 31511
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.311931718702428,
      "learning_rate": 6.397334190918814e-08,
      "loss": 0.4611,
      "step": 31512
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8567814418614956,
      "learning_rate": 6.386137493238798e-08,
      "loss": 0.6114,
      "step": 31513
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3061493129722838,
      "learning_rate": 6.374950571058858e-08,
      "loss": 0.2015,
      "step": 31514
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4018008184000367,
      "learning_rate": 6.363773424489129e-08,
      "loss": 0.2518,
      "step": 31515
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.46334230033856777,
      "learning_rate": 6.352606053639632e-08,
      "loss": 0.1845,
      "step": 31516
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7033771496998006,
      "learning_rate": 6.341448458620058e-08,
      "loss": 0.3515,
      "step": 31517
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.30394694080557577,
      "learning_rate": 6.33030063954021e-08,
      "loss": 0.1144,
      "step": 31518
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41747770977677773,
      "learning_rate": 6.319162596509886e-08,
      "loss": 0.3155,
      "step": 31519
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44097610266146964,
      "learning_rate": 6.308034329638668e-08,
      "loss": 0.1621,
      "step": 31520
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3703392901656506,
      "learning_rate": 6.296915839035911e-08,
      "loss": 0.2638,
      "step": 31521
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3187459324530892,
      "learning_rate": 6.285807124811083e-08,
      "loss": 0.1329,
      "step": 31522
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0201734028359584,
      "learning_rate": 6.274708187073431e-08,
      "loss": 0.2064,
      "step": 31523
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8015853093657723,
      "learning_rate": 6.263619025932199e-08,
      "loss": 0.328,
      "step": 31524
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3519872605060393,
      "learning_rate": 6.252539641496525e-08,
      "loss": 0.2032,
      "step": 31525
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.349308328730767,
      "learning_rate": 6.241470033875208e-08,
      "loss": 0.2763,
      "step": 31526
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2740265287585668,
      "learning_rate": 6.230410203177494e-08,
      "loss": 0.1005,
      "step": 31527
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4779146782200898,
      "learning_rate": 6.219360149511855e-08,
      "loss": 0.283,
      "step": 31528
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.28906382398056635,
      "learning_rate": 6.2083198729872e-08,
      "loss": 0.1707,
      "step": 31529
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4314443606769065,
      "learning_rate": 6.197289373712112e-08,
      "loss": 0.2483,
      "step": 31530
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8023082847888996,
      "learning_rate": 6.186268651795058e-08,
      "loss": 0.1584,
      "step": 31531
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3632362859124469,
      "learning_rate": 6.175257707344506e-08,
      "loss": 0.2557,
      "step": 31532
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6022842442957185,
      "learning_rate": 6.164256540468705e-08,
      "loss": 0.1401,
      "step": 31533
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36408127148358843,
      "learning_rate": 6.153265151276122e-08,
      "loss": 0.2484,
      "step": 31534
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9157368791386241,
      "learning_rate": 6.142283539874561e-08,
      "loss": 0.2498,
      "step": 31535
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5108653013902003,
      "learning_rate": 6.131311706372379e-08,
      "loss": 0.2444,
      "step": 31536
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3257228437429259,
      "learning_rate": 6.120349650877266e-08,
      "loss": 0.2423,
      "step": 31537
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27698017830474003,
      "learning_rate": 6.10939737349725e-08,
      "loss": 0.1681,
      "step": 31538
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9808716512283548,
      "learning_rate": 6.09845487434002e-08,
      "loss": 0.7433,
      "step": 31539
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32416351977683494,
      "learning_rate": 6.087522153513271e-08,
      "loss": 0.0814,
      "step": 31540
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9407147259583867,
      "learning_rate": 6.076599211124468e-08,
      "loss": 0.4031,
      "step": 31541
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5467774613290496,
      "learning_rate": 6.065686047281083e-08,
      "loss": 0.1941,
      "step": 31542
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5596494846370893,
      "learning_rate": 6.054782662090585e-08,
      "loss": 0.3092,
      "step": 31543
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2703722997528708,
      "learning_rate": 6.043889055660224e-08,
      "loss": 0.212,
      "step": 31544
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.1016708190323365,
      "learning_rate": 6.033005228097021e-08,
      "loss": 0.6799,
      "step": 31545
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.551785288329442,
      "learning_rate": 6.022131179508228e-08,
      "loss": 0.251,
      "step": 31546
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9182096811087007,
      "learning_rate": 6.011266910000757e-08,
      "loss": 0.4266,
      "step": 31547
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.26406137284601133,
      "learning_rate": 6.000412419681522e-08,
      "loss": 0.1669,
      "step": 31548
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.46586964152991583,
      "learning_rate": 5.98956770865733e-08,
      "loss": 0.2302,
      "step": 31549
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32624559800752534,
      "learning_rate": 5.978732777034757e-08,
      "loss": 0.2305,
      "step": 31550
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8219244744501841,
      "learning_rate": 5.96790762492061e-08,
      "loss": 0.1804,
      "step": 31551
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34892717669173257,
      "learning_rate": 5.957092252421137e-08,
      "loss": 0.2537,
      "step": 31552
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9279026905659827,
      "learning_rate": 5.94628665964303e-08,
      "loss": 0.0855,
      "step": 31553
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.178158541324497,
      "learning_rate": 5.9354908466923155e-08,
      "loss": 0.5482,
      "step": 31554
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4558566269124341,
      "learning_rate": 5.9247048136753525e-08,
      "loss": 0.2461,
      "step": 31555
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3586922774749623,
      "learning_rate": 5.913928560698279e-08,
      "loss": 0.2338,
      "step": 31556
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.33360009353475195,
      "learning_rate": 5.903162087867009e-08,
      "loss": 0.2025,
      "step": 31557
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23103281715347687,
      "learning_rate": 5.89240539528757e-08,
      "loss": 0.125,
      "step": 31558
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7701955747113304,
      "learning_rate": 5.8816584830657666e-08,
      "loss": 0.2617,
      "step": 31559
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.932957755649561,
      "learning_rate": 5.87092135130729e-08,
      "loss": 0.2705,
      "step": 31560
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3326721974516274,
      "learning_rate": 5.860194000117836e-08,
      "loss": 0.1793,
      "step": 31561
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.348112825222988,
      "learning_rate": 5.849476429602874e-08,
      "loss": 0.2674,
      "step": 31562
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0242828234499057,
      "learning_rate": 5.8387686398679865e-08,
      "loss": 0.0541,
      "step": 31563
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0527561059932178,
      "learning_rate": 5.828070631018201e-08,
      "loss": 0.5529,
      "step": 31564
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4374646829797385,
      "learning_rate": 5.8173824031591e-08,
      "loss": 0.2489,
      "step": 31565
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.45781176747594265,
      "learning_rate": 5.8067039563958205e-08,
      "loss": 0.1866,
      "step": 31566
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4863527297916172,
      "learning_rate": 5.7960352908331687e-08,
      "loss": 0.2383,
      "step": 31567
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32632652106524884,
      "learning_rate": 5.785376406576282e-08,
      "loss": 0.2303,
      "step": 31568
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3726590273718963,
      "learning_rate": 5.774727303730077e-08,
      "loss": 0.1997,
      "step": 31569
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4095172674670462,
      "learning_rate": 5.764087982399136e-08,
      "loss": 0.1775,
      "step": 31570
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.58914636140135,
      "learning_rate": 5.753458442688375e-08,
      "loss": 0.3142,
      "step": 31571
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0028012792207932,
      "learning_rate": 5.7428386847021566e-08,
      "loss": 0.1984,
      "step": 31572
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3529528316140492,
      "learning_rate": 5.732228708545062e-08,
      "loss": 0.2938,
      "step": 31573
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5124901455248364,
      "learning_rate": 5.721628514321453e-08,
      "loss": 0.1422,
      "step": 31574
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34079054090792754,
      "learning_rate": 5.711038102135691e-08,
      "loss": 0.2249,
      "step": 31575
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6059191683171107,
      "learning_rate": 5.700457472091803e-08,
      "loss": 0.2409,
      "step": 31576
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.47511543881361895,
      "learning_rate": 5.689886624294039e-08,
      "loss": 0.1385,
      "step": 31577
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40273034146870634,
      "learning_rate": 5.679325558846316e-08,
      "loss": 0.1753,
      "step": 31578
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2971932852420008,
      "learning_rate": 5.668774275852551e-08,
      "loss": 0.1691,
      "step": 31579
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3413278658356365,
      "learning_rate": 5.65823277541655e-08,
      "loss": 0.2747,
      "step": 31580
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0021451931146297,
      "learning_rate": 5.647701057642119e-08,
      "loss": 0.2772,
      "step": 31581
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6906702940723013,
      "learning_rate": 5.63717912263273e-08,
      "loss": 0.7322,
      "step": 31582
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5246970556525026,
      "learning_rate": 5.626666970491967e-08,
      "loss": 0.1307,
      "step": 31583
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3858194894667225,
      "learning_rate": 5.616164601323193e-08,
      "loss": 0.2758,
      "step": 31584
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44554618913437644,
      "learning_rate": 5.6056720152298794e-08,
      "loss": 0.2393,
      "step": 31585
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.46815082682626086,
      "learning_rate": 5.5951892123151664e-08,
      "loss": 0.3295,
      "step": 31586
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3146059000422326,
      "learning_rate": 5.584716192682082e-08,
      "loss": 0.0736,
      "step": 31587
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4270675494251589,
      "learning_rate": 5.574252956433879e-08,
      "loss": 0.2515,
      "step": 31588
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48238662485222744,
      "learning_rate": 5.5637995036733615e-08,
      "loss": 0.1762,
      "step": 31589
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9462817240317003,
      "learning_rate": 5.5533558345033376e-08,
      "loss": 0.1924,
      "step": 31590
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3206094787305323,
      "learning_rate": 5.542921949026614e-08,
      "loss": 0.2724,
      "step": 31591
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35491552892326167,
      "learning_rate": 5.532497847345886e-08,
      "loss": 0.1637,
      "step": 31592
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5770055975757237,
      "learning_rate": 5.522083529563627e-08,
      "loss": 0.3442,
      "step": 31593
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6065238180970084,
      "learning_rate": 5.5116789957824214e-08,
      "loss": 0.2542,
      "step": 31594
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.705128945550468,
      "learning_rate": 5.5012842461045215e-08,
      "loss": 0.6991,
      "step": 31595
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1532123443475318,
      "learning_rate": 5.490899280632178e-08,
      "loss": 0.0691,
      "step": 31596
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0143902070575577,
      "learning_rate": 5.480524099467643e-08,
      "loss": 0.3765,
      "step": 31597
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.24426150397257418,
      "learning_rate": 5.4701587027129466e-08,
      "loss": 0.2052,
      "step": 31598
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6664540943375465,
      "learning_rate": 5.459803090470006e-08,
      "loss": 0.8124,
      "step": 31599
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7381345875641581,
      "learning_rate": 5.449457262840852e-08,
      "loss": 0.3023,
      "step": 31600
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.01150889287064,
      "learning_rate": 5.4391212199271795e-08,
      "loss": 0.4324,
      "step": 31601
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2605205977231576,
      "learning_rate": 5.4287949618305745e-08,
      "loss": 0.1566,
      "step": 31602
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32684627851012094,
      "learning_rate": 5.418478488652956e-08,
      "loss": 0.2239,
      "step": 31603
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5018902729049148,
      "learning_rate": 5.4081718004953545e-08,
      "loss": 0.448,
      "step": 31604
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.1733562822544244,
      "learning_rate": 5.397874897459576e-08,
      "loss": 0.0695,
      "step": 31605
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6990692441737726,
      "learning_rate": 5.387587779646764e-08,
      "loss": 0.3491,
      "step": 31606
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34698027323833663,
      "learning_rate": 5.377310447158057e-08,
      "loss": 0.176,
      "step": 31607
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.375576318023918,
      "learning_rate": 5.367042900094821e-08,
      "loss": 0.7152,
      "step": 31608
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3084079196996659,
      "learning_rate": 5.356785138557752e-08,
      "loss": 0.2219,
      "step": 31609
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6273330174141774,
      "learning_rate": 5.34653716264788e-08,
      "loss": 0.3421,
      "step": 31610
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3481530226165243,
      "learning_rate": 5.336298972466125e-08,
      "loss": 0.1991,
      "step": 31611
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.032528857012565,
      "learning_rate": 5.326070568113184e-08,
      "loss": 0.448,
      "step": 31612
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5349761490344773,
      "learning_rate": 5.3158519496895324e-08,
      "loss": 0.048,
      "step": 31613
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3184537622723038,
      "learning_rate": 5.305643117295978e-08,
      "loss": 0.1398,
      "step": 31614
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.29093031734360064,
      "learning_rate": 5.295444071032663e-08,
      "loss": 0.1866,
      "step": 31615
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3370745496554552,
      "learning_rate": 5.285254811000173e-08,
      "loss": 0.1989,
      "step": 31616
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.634301503071311,
      "learning_rate": 5.275075337298541e-08,
      "loss": 0.7437,
      "step": 31617
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8475889836906523,
      "learning_rate": 5.264905650028129e-08,
      "loss": 0.2634,
      "step": 31618
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7552125031189652,
      "learning_rate": 5.254745749288859e-08,
      "loss": 0.3723,
      "step": 31619
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37395212659552074,
      "learning_rate": 5.24459563518076e-08,
      "loss": 0.1683,
      "step": 31620
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34691243573156716,
      "learning_rate": 5.234455307803532e-08,
      "loss": 0.264,
      "step": 31621
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44745059350609434,
      "learning_rate": 5.2243247672570944e-08,
      "loss": 0.1715,
      "step": 31622
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5342585922888388,
      "learning_rate": 5.2142040136411444e-08,
      "loss": 0.2193,
      "step": 31623
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.33378194476127604,
      "learning_rate": 5.204093047055048e-08,
      "loss": 0.0716,
      "step": 31624
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2823403946874082,
      "learning_rate": 5.193991867598613e-08,
      "loss": 0.2083,
      "step": 31625
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8177444339410974,
      "learning_rate": 5.183900475370873e-08,
      "loss": 0.276,
      "step": 31626
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.358843006322742,
      "learning_rate": 5.173818870471303e-08,
      "loss": 0.2847,
      "step": 31627
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0535867682678561,
      "learning_rate": 5.1637470529989354e-08,
      "loss": 0.2737,
      "step": 31628
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.28772466350993364,
      "learning_rate": 5.1536850230531344e-08,
      "loss": 0.1697,
      "step": 31629
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5906591096623905,
      "learning_rate": 5.14363278073271e-08,
      "loss": 0.7418,
      "step": 31630
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8965956164910167,
      "learning_rate": 5.133590326136473e-08,
      "loss": 0.1303,
      "step": 31631
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4579358373840945,
      "learning_rate": 5.123557659363454e-08,
      "loss": 0.212,
      "step": 31632
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2618966998327288,
      "learning_rate": 5.11353478051213e-08,
      "loss": 0.19,
      "step": 31633
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2933926561220799,
      "learning_rate": 5.103521689681201e-08,
      "loss": 0.2048,
      "step": 31634
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7455544141753631,
      "learning_rate": 5.093518386969254e-08,
      "loss": 0.2375,
      "step": 31635
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0599101525303296,
      "learning_rate": 5.0835248724745435e-08,
      "loss": 0.4537,
      "step": 31636
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6256885408308825,
      "learning_rate": 5.073541146295658e-08,
      "loss": 0.2479,
      "step": 31637
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36125806408685807,
      "learning_rate": 5.063567208530518e-08,
      "loss": 0.27,
      "step": 31638
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3115454786625712,
      "learning_rate": 5.0536030592773786e-08,
      "loss": 0.238,
      "step": 31639
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2322139191299037,
      "learning_rate": 5.0436486986343846e-08,
      "loss": 0.4716,
      "step": 31640
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.969174055645542,
      "learning_rate": 5.033704126699235e-08,
      "loss": 0.0427,
      "step": 31641
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2227047497053491,
      "learning_rate": 5.023769343569962e-08,
      "loss": 0.118,
      "step": 31642
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.38042121338853785,
      "learning_rate": 5.013844349344266e-08,
      "loss": 0.2437,
      "step": 31643
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6482612320930374,
      "learning_rate": 5.0039291441196234e-08,
      "loss": 0.2549,
      "step": 31644
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3417753226330861,
      "learning_rate": 4.994023727993846e-08,
      "loss": 0.2731,
      "step": 31645
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5272761926357454,
      "learning_rate": 4.9841281010641893e-08,
      "loss": 0.2036,
      "step": 31646
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48212372206054116,
      "learning_rate": 4.9742422634281305e-08,
      "loss": 0.2893,
      "step": 31647
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.49835152734651533,
      "learning_rate": 4.9643662151829255e-08,
      "loss": 0.236,
      "step": 31648
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9951809439862567,
      "learning_rate": 4.954499956425607e-08,
      "loss": 0.5351,
      "step": 31649
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35839864940249233,
      "learning_rate": 4.94464348725332e-08,
      "loss": 0.2294,
      "step": 31650
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40947474326416433,
      "learning_rate": 4.9347968077632094e-08,
      "loss": 0.1901,
      "step": 31651
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3108663351367524,
      "learning_rate": 4.924959918051864e-08,
      "loss": 0.1807,
      "step": 31652
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6802985177428328,
      "learning_rate": 4.915132818216206e-08,
      "loss": 0.3835,
      "step": 31653
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4140488884917433,
      "learning_rate": 4.905315508352826e-08,
      "loss": 0.1481,
      "step": 31654
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2628146853802893,
      "learning_rate": 4.8955079885584235e-08,
      "loss": 0.4376,
      "step": 31655
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4174062700889053,
      "learning_rate": 4.885710258929477e-08,
      "loss": 0.2506,
      "step": 31656
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27404300117130787,
      "learning_rate": 4.875922319562354e-08,
      "loss": 0.2194,
      "step": 31657
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1019953262300493,
      "learning_rate": 4.866144170553422e-08,
      "loss": 0.34,
      "step": 31658
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4281864036377079,
      "learning_rate": 4.8563758119986035e-08,
      "loss": 0.0562,
      "step": 31659
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6913164386430515,
      "learning_rate": 4.8466172439943785e-08,
      "loss": 0.3315,
      "step": 31660
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35315526913247214,
      "learning_rate": 4.8368684666364466e-08,
      "loss": 0.2051,
      "step": 31661
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4438636714746492,
      "learning_rate": 4.827129480020953e-08,
      "loss": 0.3257,
      "step": 31662
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34826201891058345,
      "learning_rate": 4.8174002842436006e-08,
      "loss": 0.1557,
      "step": 31663
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42083759846818397,
      "learning_rate": 4.8076808794000895e-08,
      "loss": 0.2132,
      "step": 31664
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3325029302781388,
      "learning_rate": 4.797971265586121e-08,
      "loss": 0.1635,
      "step": 31665
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6112097252390181,
      "learning_rate": 4.7882714428971744e-08,
      "loss": 0.3592,
      "step": 31666
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.2424593945121645,
      "learning_rate": 4.778581411428618e-08,
      "loss": 0.1577,
      "step": 31667
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32556128230452946,
      "learning_rate": 4.7689011712759304e-08,
      "loss": 0.2176,
      "step": 31668
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5847257846500157,
      "learning_rate": 4.759230722534258e-08,
      "loss": 0.2267,
      "step": 31669
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3572354571516621,
      "learning_rate": 4.7495700652987474e-08,
      "loss": 0.2065,
      "step": 31670
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9338353673348456,
      "learning_rate": 4.739919199664322e-08,
      "loss": 0.4088,
      "step": 31671
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4058476723062718,
      "learning_rate": 4.73027812572624e-08,
      "loss": 0.0999,
      "step": 31672
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.676595038227454,
      "learning_rate": 4.72064684357898e-08,
      "loss": 0.8503,
      "step": 31673
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3492429844429113,
      "learning_rate": 4.7110253533175775e-08,
      "loss": 0.1519,
      "step": 31674
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3508729822895883,
      "learning_rate": 4.701413655036624e-08,
      "loss": 0.2727,
      "step": 31675
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3062045315477955,
      "learning_rate": 4.69181174883071e-08,
      "loss": 0.1562,
      "step": 31676
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7738606433738008,
      "learning_rate": 4.6822196347942054e-08,
      "loss": 0.39,
      "step": 31677
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8487363593019625,
      "learning_rate": 4.672637313021477e-08,
      "loss": 0.2559,
      "step": 31678
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37874759943993064,
      "learning_rate": 4.6630647836068965e-08,
      "loss": 0.2206,
      "step": 31679
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4738622345695187,
      "learning_rate": 4.653502046644498e-08,
      "loss": 0.159,
      "step": 31680
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.25170052255892533,
      "learning_rate": 4.6439491022286506e-08,
      "loss": 0.1847,
      "step": 31681
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.0593381897670913,
      "learning_rate": 4.634405950452947e-08,
      "loss": 0.6925,
      "step": 31682
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3414173206378047,
      "learning_rate": 4.624872591411644e-08,
      "loss": 0.1524,
      "step": 31683
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2979981015096072,
      "learning_rate": 4.615349025198335e-08,
      "loss": 0.2159,
      "step": 31684
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1520945654116481,
      "learning_rate": 4.60583525190661e-08,
      "loss": 0.2317,
      "step": 31685
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.49589605911986795,
      "learning_rate": 4.5963312716302835e-08,
      "loss": 0.2883,
      "step": 31686
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4107419569999301,
      "learning_rate": 4.586837084462836e-08,
      "loss": 0.2259,
      "step": 31687
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3641169012090908,
      "learning_rate": 4.577352690497527e-08,
      "loss": 0.2513,
      "step": 31688
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9467693545087971,
      "learning_rate": 4.567878089827837e-08,
      "loss": 0.2032,
      "step": 31689
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.224804481863662,
      "learning_rate": 4.558413282546914e-08,
      "loss": 0.6303,
      "step": 31690
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0311791712085125,
      "learning_rate": 4.548958268747794e-08,
      "loss": 0.5729,
      "step": 31691
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3672283322974823,
      "learning_rate": 4.539513048523625e-08,
      "loss": 0.2416,
      "step": 31692
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21821248337544943,
      "learning_rate": 4.5300776219672216e-08,
      "loss": 0.1764,
      "step": 31693
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22697512171389495,
      "learning_rate": 4.52065198917151e-08,
      "loss": 0.0705,
      "step": 31694
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8014956789906417,
      "learning_rate": 4.511236150229192e-08,
      "loss": 0.3382,
      "step": 31695
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6868372030889163,
      "learning_rate": 4.501830105232863e-08,
      "loss": 0.1563,
      "step": 31696
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3673318462675117,
      "learning_rate": 4.492433854275113e-08,
      "loss": 0.2735,
      "step": 31697
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7756573315340762,
      "learning_rate": 4.483047397448314e-08,
      "loss": 0.0823,
      "step": 31698
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3125337750833124,
      "learning_rate": 4.4736707348449484e-08,
      "loss": 0.2431,
      "step": 31699
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.022097076536052,
      "learning_rate": 4.4643038665570515e-08,
      "loss": 0.5365,
      "step": 31700
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48397591435123116,
      "learning_rate": 4.454946792676995e-08,
      "loss": 0.2566,
      "step": 31701
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2821673092146756,
      "learning_rate": 4.445599513296705e-08,
      "loss": 0.1293,
      "step": 31702
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5298458718689817,
      "learning_rate": 4.4362620285083314e-08,
      "loss": 0.2324,
      "step": 31703
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.359433983510602,
      "learning_rate": 4.426934338403466e-08,
      "loss": 0.2166,
      "step": 31704
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.9392060170529348,
      "learning_rate": 4.417616443074035e-08,
      "loss": 0.8613,
      "step": 31705
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3583171432235572,
      "learning_rate": 4.408308342611634e-08,
      "loss": 0.1617,
      "step": 31706
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4909166559567204,
      "learning_rate": 4.3990100371079646e-08,
      "loss": 0.1614,
      "step": 31707
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.408265969530108,
      "learning_rate": 4.3897215266544e-08,
      "loss": 0.7388,
      "step": 31708
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3787811117554798,
      "learning_rate": 4.3804428113423115e-08,
      "loss": 0.4691,
      "step": 31709
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.32012111224413803,
      "learning_rate": 4.3711738912630696e-08,
      "loss": 0.2791,
      "step": 31710
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3118199094282044,
      "learning_rate": 4.3619147665078245e-08,
      "loss": 0.1784,
      "step": 31711
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4635153707518283,
      "learning_rate": 4.352665437167725e-08,
      "loss": 0.2616,
      "step": 31712
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7709915380633185,
      "learning_rate": 4.3434259033335865e-08,
      "loss": 0.2506,
      "step": 31713
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.49361564678143577,
      "learning_rate": 4.33419616509656e-08,
      "loss": 0.2754,
      "step": 31714
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3199318151997126,
      "learning_rate": 4.3249762225473504e-08,
      "loss": 0.1466,
      "step": 31715
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36847517451060247,
      "learning_rate": 4.315766075776551e-08,
      "loss": 0.2645,
      "step": 31716
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5290535857509074,
      "learning_rate": 4.306565724874867e-08,
      "loss": 0.2088,
      "step": 31717
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.5083432264914696,
      "learning_rate": 4.2973751699327827e-08,
      "loss": 0.7112,
      "step": 31718
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3235869849540818,
      "learning_rate": 4.2881944110408913e-08,
      "loss": 0.1081,
      "step": 31719
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37830974446830234,
      "learning_rate": 4.279023448289232e-08,
      "loss": 0.2555,
      "step": 31720
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8374605316062678,
      "learning_rate": 4.269862281768289e-08,
      "loss": 0.2421,
      "step": 31721
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.311110848155154,
      "learning_rate": 4.260710911567989e-08,
      "loss": 0.2245,
      "step": 31722
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2427769152028477,
      "learning_rate": 4.251569337778483e-08,
      "loss": 0.131,
      "step": 31723
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2895294814872836,
      "learning_rate": 4.2424375604896986e-08,
      "loss": 0.1679,
      "step": 31724
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7305949000896494,
      "learning_rate": 4.2333155797914525e-08,
      "loss": 0.7568,
      "step": 31725
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.0599042173785298,
      "learning_rate": 4.224203395773563e-08,
      "loss": 0.4107,
      "step": 31726
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.46117134769819623,
      "learning_rate": 4.215101008525513e-08,
      "loss": 0.2961,
      "step": 31727
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4041245455752509,
      "learning_rate": 4.2060084181370084e-08,
      "loss": 0.2205,
      "step": 31728
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5005835967190996,
      "learning_rate": 4.196925624697423e-08,
      "loss": 0.3141,
      "step": 31729
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.42883767117931937,
      "learning_rate": 4.18785262829613e-08,
      "loss": 0.2313,
      "step": 31730
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4784346392001905,
      "learning_rate": 4.178789429022501e-08,
      "loss": 0.2084,
      "step": 31731
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3394187544593695,
      "learning_rate": 4.169736026965576e-08,
      "loss": 0.0704,
      "step": 31732
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.646976410164687,
      "learning_rate": 4.1606924222143965e-08,
      "loss": 0.274,
      "step": 31733
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2798754980250208,
      "learning_rate": 4.151658614858001e-08,
      "loss": 0.2059,
      "step": 31734
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4232552804189938,
      "learning_rate": 4.1426346049853184e-08,
      "loss": 0.2348,
      "step": 31735
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4006742321659174,
      "learning_rate": 4.133620392685056e-08,
      "loss": 0.7457,
      "step": 31736
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6761581044901717,
      "learning_rate": 4.124615978045921e-08,
      "loss": 0.2264,
      "step": 31737
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.36317740586227143,
      "learning_rate": 4.1156213611565074e-08,
      "loss": 0.2495,
      "step": 31738
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7665729973863952,
      "learning_rate": 4.106636542105302e-08,
      "loss": 0.2248,
      "step": 31739
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.33694970072948754,
      "learning_rate": 4.097661520980678e-08,
      "loss": 0.251,
      "step": 31740
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.21094772199264405,
      "learning_rate": 4.088696297871009e-08,
      "loss": 0.0695,
      "step": 31741
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6306190672900291,
      "learning_rate": 4.079740872864446e-08,
      "loss": 0.2873,
      "step": 31742
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3614478814366908,
      "learning_rate": 4.070795246049031e-08,
      "loss": 0.196,
      "step": 31743
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.7123635240344937,
      "learning_rate": 4.0618594175128037e-08,
      "loss": 0.7026,
      "step": 31744
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6117416870850665,
      "learning_rate": 4.052933387343805e-08,
      "loss": 0.2389,
      "step": 31745
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.435904863904552,
      "learning_rate": 4.0440171556297426e-08,
      "loss": 0.3092,
      "step": 31746
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.29526213331261947,
      "learning_rate": 4.035110722458324e-08,
      "loss": 0.1771,
      "step": 31747
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.08900979282093,
      "learning_rate": 4.026214087917146e-08,
      "loss": 0.2682,
      "step": 31748
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3957603132061105,
      "learning_rate": 4.017327252093805e-08,
      "loss": 0.4021,
      "step": 31749
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27462339859378243,
      "learning_rate": 4.0084502150758985e-08,
      "loss": 0.0736,
      "step": 31750
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4320299678932226,
      "learning_rate": 3.9995829769503556e-08,
      "loss": 0.2339,
      "step": 31751
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3322682863154028,
      "learning_rate": 3.990725537804774e-08,
      "loss": 0.2017,
      "step": 31752
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4784471336354287,
      "learning_rate": 3.981877897726194e-08,
      "loss": 0.303,
      "step": 31753
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6544334229707344,
      "learning_rate": 3.973040056801436e-08,
      "loss": 0.2595,
      "step": 31754
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7051191038230443,
      "learning_rate": 3.964212015117874e-08,
      "loss": 0.3535,
      "step": 31755
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4161608270435194,
      "learning_rate": 3.955393772761995e-08,
      "loss": 0.1671,
      "step": 31756
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6086478825537105,
      "learning_rate": 3.946585329820729e-08,
      "loss": 0.3274,
      "step": 31757
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3455508593239007,
      "learning_rate": 3.9377866863807846e-08,
      "loss": 0.2309,
      "step": 31758
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27344039737135384,
      "learning_rate": 3.9289978425285366e-08,
      "loss": 0.1463,
      "step": 31759
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.39207169267312164,
      "learning_rate": 3.9202187983505836e-08,
      "loss": 0.0639,
      "step": 31760
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3592096904901618,
      "learning_rate": 3.911449553933189e-08,
      "loss": 0.1802,
      "step": 31761
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9700968107021506,
      "learning_rate": 3.902690109362839e-08,
      "loss": 0.484,
      "step": 31762
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48055724219140383,
      "learning_rate": 3.8939404647255765e-08,
      "loss": 0.2161,
      "step": 31763
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5195885689056725,
      "learning_rate": 3.885200620107443e-08,
      "loss": 0.3291,
      "step": 31764
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2850841143771091,
      "learning_rate": 3.876470575594482e-08,
      "loss": 0.1708,
      "step": 31765
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.3213846348542513,
      "learning_rate": 3.867750331272513e-08,
      "loss": 0.625,
      "step": 31766
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1964551557561527,
      "learning_rate": 3.859039887227467e-08,
      "loss": 0.0629,
      "step": 31767
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48909999697626894,
      "learning_rate": 3.8503392435449424e-08,
      "loss": 0.2204,
      "step": 31768
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.20761306317271258,
      "learning_rate": 3.841648400310538e-08,
      "loss": 0.126,
      "step": 31769
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2950382710251269,
      "learning_rate": 3.832967357609851e-08,
      "loss": 0.2425,
      "step": 31770
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9838824038356003,
      "learning_rate": 3.824296115528148e-08,
      "loss": 0.271,
      "step": 31771
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6782140952514574,
      "learning_rate": 3.815634674150803e-08,
      "loss": 0.3622,
      "step": 31772
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9140424971267621,
      "learning_rate": 3.806983033563083e-08,
      "loss": 0.3514,
      "step": 31773
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2926142308599525,
      "learning_rate": 3.79834119385003e-08,
      "loss": 0.17,
      "step": 31774
    },
    {
      "epoch": 0.97,
      "grad_norm": 2.234406092560628,
      "learning_rate": 3.789709155096577e-08,
      "loss": 0.9112,
      "step": 31775
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.31903656133286257,
      "learning_rate": 3.781086917387877e-08,
      "loss": 0.2158,
      "step": 31776
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43241979885627024,
      "learning_rate": 3.772474480808641e-08,
      "loss": 0.198,
      "step": 31777
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.22168838708609678,
      "learning_rate": 3.7638718454435783e-08,
      "loss": 0.0642,
      "step": 31778
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5581533527129746,
      "learning_rate": 3.755279011377399e-08,
      "loss": 0.3029,
      "step": 31779
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.44751694250760043,
      "learning_rate": 3.746695978694481e-08,
      "loss": 0.2327,
      "step": 31780
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4843752928378462,
      "learning_rate": 3.738122747479534e-08,
      "loss": 0.3013,
      "step": 31781
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3893116249786244,
      "learning_rate": 3.7295593178166e-08,
      "loss": 0.2156,
      "step": 31782
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9606638357127812,
      "learning_rate": 3.7210056897901695e-08,
      "loss": 0.362,
      "step": 31783
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34377475273255764,
      "learning_rate": 3.7124618634842843e-08,
      "loss": 0.1983,
      "step": 31784
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6896215676504984,
      "learning_rate": 3.70392783898299e-08,
      "loss": 0.5335,
      "step": 31785
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.30685724412163323,
      "learning_rate": 3.69540361637033e-08,
      "loss": 0.1583,
      "step": 31786
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3867590612834958,
      "learning_rate": 3.6868891957300145e-08,
      "loss": 0.1951,
      "step": 31787
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.37085581757228814,
      "learning_rate": 3.678384577146088e-08,
      "loss": 0.2535,
      "step": 31788
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4681223187149009,
      "learning_rate": 3.669889760701928e-08,
      "loss": 0.1908,
      "step": 31789
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9336439997444197,
      "learning_rate": 3.661404746481245e-08,
      "loss": 0.4819,
      "step": 31790
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.139478362094672,
      "learning_rate": 3.6529295345675284e-08,
      "loss": 0.4212,
      "step": 31791
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6452343729608401,
      "learning_rate": 3.6444641250440446e-08,
      "loss": 0.285,
      "step": 31792
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.2940148627434283,
      "learning_rate": 3.6360085179942826e-08,
      "loss": 0.2015,
      "step": 31793
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.48598421627386745,
      "learning_rate": 3.627562713501176e-08,
      "loss": 0.3047,
      "step": 31794
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.4225277703026986,
      "learning_rate": 3.619126711648102e-08,
      "loss": 0.0494,
      "step": 31795
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.41671477019851755,
      "learning_rate": 3.610700512517884e-08,
      "loss": 0.195,
      "step": 31796
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.33959397657514284,
      "learning_rate": 3.602284116193344e-08,
      "loss": 0.1974,
      "step": 31797
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.099866066446146,
      "learning_rate": 3.593877522757527e-08,
      "loss": 0.3704,
      "step": 31798
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.27420338040134384,
      "learning_rate": 3.585480732292923e-08,
      "loss": 0.1577,
      "step": 31799
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4649895269793998,
      "learning_rate": 3.577093744882243e-08,
      "loss": 0.281,
      "step": 31800
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43060361813962544,
      "learning_rate": 3.5687165606079765e-08,
      "loss": 0.2479,
      "step": 31801
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5327096672738962,
      "learning_rate": 3.5603491795525025e-08,
      "loss": 0.1859,
      "step": 31802
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.235715695357429,
      "learning_rate": 3.55199160179831e-08,
      "loss": 0.4518,
      "step": 31803
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9848378864711641,
      "learning_rate": 3.5436438274273346e-08,
      "loss": 0.2386,
      "step": 31804
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3262118065600015,
      "learning_rate": 3.535305856522064e-08,
      "loss": 0.2633,
      "step": 31805
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3868649825023691,
      "learning_rate": 3.5269776891641017e-08,
      "loss": 0.1665,
      "step": 31806
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.5121115500982957,
      "learning_rate": 3.5186593254358245e-08,
      "loss": 0.3197,
      "step": 31807
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.25308455593454526,
      "learning_rate": 3.510350765418724e-08,
      "loss": 0.1177,
      "step": 31808
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.45170662956373336,
      "learning_rate": 3.502052009194734e-08,
      "loss": 0.2653,
      "step": 31809
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.31194113954708685,
      "learning_rate": 3.4937630568454564e-08,
      "loss": 0.0679,
      "step": 31810
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.40501888646668643,
      "learning_rate": 3.4854839084524917e-08,
      "loss": 0.2974,
      "step": 31811
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.34524498432734624,
      "learning_rate": 3.477214564097109e-08,
      "loss": 0.221,
      "step": 31812
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6996793047236523,
      "learning_rate": 3.468955023860798e-08,
      "loss": 0.2284,
      "step": 31813
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.9771085302811502,
      "learning_rate": 3.460705287824939e-08,
      "loss": 0.4129,
      "step": 31814
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.26743079930080405,
      "learning_rate": 3.452465356070467e-08,
      "loss": 0.1669,
      "step": 31815
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.1792759084187008,
      "learning_rate": 3.444235228678538e-08,
      "loss": 0.3689,
      "step": 31816
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3085188989044728,
      "learning_rate": 3.4360149057302006e-08,
      "loss": 0.2247,
      "step": 31817
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.6103876063938911,
      "learning_rate": 3.427804387306166e-08,
      "loss": 0.6913,
      "step": 31818
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.18166137794310586,
      "learning_rate": 3.4196036734873704e-08,
      "loss": 0.0909,
      "step": 31819
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3043369845741185,
      "learning_rate": 3.411412764354416e-08,
      "loss": 0.2176,
      "step": 31820
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.8302679154944916,
      "learning_rate": 3.403231659987905e-08,
      "loss": 0.1923,
      "step": 31821
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7552572398307955,
      "learning_rate": 3.395060360468439e-08,
      "loss": 0.3638,
      "step": 31822
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.35092520371281355,
      "learning_rate": 3.3868988658761756e-08,
      "loss": 0.2246,
      "step": 31823
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4214905775163899,
      "learning_rate": 3.378747176291497e-08,
      "loss": 0.2414,
      "step": 31824
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4762027890035974,
      "learning_rate": 3.3706052917946705e-08,
      "loss": 0.1804,
      "step": 31825
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.2497855558037902,
      "learning_rate": 3.362473212465855e-08,
      "loss": 0.4437,
      "step": 31826
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.4442014447571811,
      "learning_rate": 3.354350938384876e-08,
      "loss": 0.2658,
      "step": 31827
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.23065883865440281,
      "learning_rate": 3.3462384696316685e-08,
      "loss": 0.117,
      "step": 31828
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3794600285006709,
      "learning_rate": 3.338135806286169e-08,
      "loss": 0.2978,
      "step": 31829
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.43094441773254694,
      "learning_rate": 3.330042948428092e-08,
      "loss": 0.1962,
      "step": 31830
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.7670815245773812,
      "learning_rate": 3.321959896136928e-08,
      "loss": 0.3664,
      "step": 31831
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.6671260175828556,
      "learning_rate": 3.3138866494922814e-08,
      "loss": 0.1591,
      "step": 31832
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.8952268707381728,
      "learning_rate": 3.305823208573644e-08,
      "loss": 0.4122,
      "step": 31833
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3920812548053785,
      "learning_rate": 3.297769573460174e-08,
      "loss": 0.1853,
      "step": 31834
    },
    {
      "epoch": 0.97,
      "grad_norm": 0.3539745365207464,
      "learning_rate": 3.289725744231254e-08,
      "loss": 0.2847,
      "step": 31835
    },
    {
      "epoch": 0.97,
      "grad_norm": 1.252873026232003,
      "learning_rate": 3.2816917209660406e-08,
      "loss": 0.5029,
      "step": 31836
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4528543876262393,
      "learning_rate": 3.2736675037433605e-08,
      "loss": 0.2014,
      "step": 31837
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.26614830135202805,
      "learning_rate": 3.265653092642373e-08,
      "loss": 0.1637,
      "step": 31838
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4513262958471561,
      "learning_rate": 3.2576484877419045e-08,
      "loss": 0.1604,
      "step": 31839
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45120877433661183,
      "learning_rate": 3.249653689120669e-08,
      "loss": 0.2925,
      "step": 31840
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4313165573515619,
      "learning_rate": 3.2416686968572697e-08,
      "loss": 0.1968,
      "step": 31841
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4363932120770726,
      "learning_rate": 3.2336935110303115e-08,
      "loss": 0.2884,
      "step": 31842
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5475939420281988,
      "learning_rate": 3.2257281317182865e-08,
      "loss": 0.1877,
      "step": 31843
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4659099328739058,
      "learning_rate": 3.217772558999466e-08,
      "loss": 0.8784,
      "step": 31844
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.253746238531329,
      "learning_rate": 3.2098267929522306e-08,
      "loss": 0.2795,
      "step": 31845
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3423559938884366,
      "learning_rate": 3.201890833654631e-08,
      "loss": 0.2141,
      "step": 31846
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2796236047078114,
      "learning_rate": 3.1939646811849356e-08,
      "loss": 0.1979,
      "step": 31847
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4673993410355499,
      "learning_rate": 3.186048335620973e-08,
      "loss": 0.2451,
      "step": 31848
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7956302889599596,
      "learning_rate": 3.178141797040568e-08,
      "loss": 0.242,
      "step": 31849
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9865534844144235,
      "learning_rate": 3.17024506552166e-08,
      "loss": 0.2949,
      "step": 31850
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3483559514205175,
      "learning_rate": 3.1623581411419633e-08,
      "loss": 0.1389,
      "step": 31851
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.47481349356786295,
      "learning_rate": 3.154481023978973e-08,
      "loss": 0.2183,
      "step": 31852
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3346084426300832,
      "learning_rate": 3.146613714110069e-08,
      "loss": 0.2553,
      "step": 31853
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0661175487568066,
      "learning_rate": 3.138756211612859e-08,
      "loss": 0.4984,
      "step": 31854
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7542418231162132,
      "learning_rate": 3.1309085165646124e-08,
      "loss": 0.3336,
      "step": 31855
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36780225800694877,
      "learning_rate": 3.123070629042491e-08,
      "loss": 0.1526,
      "step": 31856
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5052582614954438,
      "learning_rate": 3.115242549123654e-08,
      "loss": 0.308,
      "step": 31857
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.24321137537330112,
      "learning_rate": 3.1074242768850405e-08,
      "loss": 0.1515,
      "step": 31858
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5257893674881536,
      "learning_rate": 3.099615812403589e-08,
      "loss": 0.3415,
      "step": 31859
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.302675288006012,
      "learning_rate": 3.0918171557561275e-08,
      "loss": 0.0637,
      "step": 31860
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41703731669186755,
      "learning_rate": 3.084028307019371e-08,
      "loss": 0.2868,
      "step": 31861
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4080765991757154,
      "learning_rate": 3.076249266270037e-08,
      "loss": 0.3759,
      "step": 31862
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3361331943776982,
      "learning_rate": 3.06848003358462e-08,
      "loss": 0.5421,
      "step": 31863
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45662498662253653,
      "learning_rate": 3.060720609039502e-08,
      "loss": 0.2452,
      "step": 31864
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2913693016765895,
      "learning_rate": 3.052970992711069e-08,
      "loss": 0.1829,
      "step": 31865
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43712022531916367,
      "learning_rate": 3.0452311846754786e-08,
      "loss": 0.3058,
      "step": 31866
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4524871934341059,
      "learning_rate": 3.037501185009006e-08,
      "loss": 0.1057,
      "step": 31867
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.271675910238752,
      "learning_rate": 3.02978099378759e-08,
      "loss": 0.5051,
      "step": 31868
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5568291015042722,
      "learning_rate": 3.0220706110872797e-08,
      "loss": 0.0688,
      "step": 31869
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4045408912824432,
      "learning_rate": 3.0143700369837937e-08,
      "loss": 0.281,
      "step": 31870
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.32287503666637996,
      "learning_rate": 3.00667927155307e-08,
      "loss": 0.2401,
      "step": 31871
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.58694263190838,
      "learning_rate": 2.998998314870716e-08,
      "loss": 0.7736,
      "step": 31872
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6380021978246081,
      "learning_rate": 2.9913271670122256e-08,
      "loss": 0.2223,
      "step": 31873
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3587142339406872,
      "learning_rate": 2.983665828053206e-08,
      "loss": 0.268,
      "step": 31874
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7481950033378012,
      "learning_rate": 2.976014298068819e-08,
      "loss": 0.2453,
      "step": 31875
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.44218448319476794,
      "learning_rate": 2.9683725771345595e-08,
      "loss": 0.2079,
      "step": 31876
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23149547163285594,
      "learning_rate": 2.9607406653253678e-08,
      "loss": 0.164,
      "step": 31877
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3283914518168628,
      "learning_rate": 2.953118562716628e-08,
      "loss": 0.1488,
      "step": 31878
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6045415767244597,
      "learning_rate": 2.9455062693831694e-08,
      "loss": 0.371,
      "step": 31879
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5057017346026782,
      "learning_rate": 2.9379037853998205e-08,
      "loss": 0.4659,
      "step": 31880
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7887060658404448,
      "learning_rate": 2.9303111108415218e-08,
      "loss": 0.4309,
      "step": 31881
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3219648279147238,
      "learning_rate": 2.9227282457828797e-08,
      "loss": 0.2225,
      "step": 31882
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5834831525457288,
      "learning_rate": 2.915155190298502e-08,
      "loss": 0.3188,
      "step": 31883
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5314827149198162,
      "learning_rate": 2.9075919444628842e-08,
      "loss": 0.2306,
      "step": 31884
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2445952408086103,
      "learning_rate": 2.900038508350522e-08,
      "loss": 0.3374,
      "step": 31885
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20188162195633838,
      "learning_rate": 2.89249488203569e-08,
      "loss": 0.0687,
      "step": 31886
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9046237065167928,
      "learning_rate": 2.8849610655925508e-08,
      "loss": 0.4363,
      "step": 31887
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3189219049557477,
      "learning_rate": 2.877437059095267e-08,
      "loss": 0.1861,
      "step": 31888
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33410526645745964,
      "learning_rate": 2.86992286261778e-08,
      "loss": 0.2465,
      "step": 31889
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8792886043981617,
      "learning_rate": 2.8624184762341413e-08,
      "loss": 0.2755,
      "step": 31890
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2681345941243718,
      "learning_rate": 2.8549239000181804e-08,
      "loss": 0.2296,
      "step": 31891
    },
    {
      "epoch": 0.98,
      "grad_norm": 3.562476992766474,
      "learning_rate": 2.8474391340435058e-08,
      "loss": 0.2771,
      "step": 31892
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.48756912983083656,
      "learning_rate": 2.839964178383947e-08,
      "loss": 0.1977,
      "step": 31893
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.366258066672097,
      "learning_rate": 2.8324990331127788e-08,
      "loss": 0.2787,
      "step": 31894
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1446654778180895,
      "learning_rate": 2.82504369830372e-08,
      "loss": 0.0678,
      "step": 31895
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9721478078806587,
      "learning_rate": 2.8175981740299342e-08,
      "loss": 0.2828,
      "step": 31896
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3420671130494036,
      "learning_rate": 2.810162460364696e-08,
      "loss": 0.1748,
      "step": 31897
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9064998876813045,
      "learning_rate": 2.8027365573812803e-08,
      "loss": 0.4229,
      "step": 31898
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9742093250958148,
      "learning_rate": 2.795320465152629e-08,
      "loss": 0.2626,
      "step": 31899
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35752659114633345,
      "learning_rate": 2.787914183751794e-08,
      "loss": 0.2743,
      "step": 31900
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3478360500463694,
      "learning_rate": 2.780517713251496e-08,
      "loss": 0.1642,
      "step": 31901
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5775628342787108,
      "learning_rate": 2.773131053724676e-08,
      "loss": 0.3273,
      "step": 31902
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.48064716006474095,
      "learning_rate": 2.7657542052438312e-08,
      "loss": 0.01,
      "step": 31903
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2788996836808245,
      "learning_rate": 2.758387167881682e-08,
      "loss": 0.1585,
      "step": 31904
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2626836060340327,
      "learning_rate": 2.7510299417106147e-08,
      "loss": 0.0675,
      "step": 31905
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3563205780488935,
      "learning_rate": 2.7436825268031264e-08,
      "loss": 0.2078,
      "step": 31906
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38600817866404663,
      "learning_rate": 2.736344923231382e-08,
      "loss": 0.284,
      "step": 31907
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6833732126316525,
      "learning_rate": 2.729017131067657e-08,
      "loss": 0.2711,
      "step": 31908
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4420424189464651,
      "learning_rate": 2.721699150384005e-08,
      "loss": 0.7657,
      "step": 31909
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34770779618522446,
      "learning_rate": 2.7143909812523682e-08,
      "loss": 0.1554,
      "step": 31910
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5553020249560616,
      "learning_rate": 2.7070926237446894e-08,
      "loss": 0.3506,
      "step": 31911
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3289218851096858,
      "learning_rate": 2.6998040779329105e-08,
      "loss": 0.2353,
      "step": 31912
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45921869478404154,
      "learning_rate": 2.692525343888419e-08,
      "loss": 0.2302,
      "step": 31913
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23386226590462691,
      "learning_rate": 2.685256421683158e-08,
      "loss": 0.0665,
      "step": 31914
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4634446721474383,
      "learning_rate": 2.677997311388514e-08,
      "loss": 0.2628,
      "step": 31915
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4759303528698813,
      "learning_rate": 2.6707480130758746e-08,
      "loss": 0.2486,
      "step": 31916
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7926677190444631,
      "learning_rate": 2.6635085268165162e-08,
      "loss": 0.3868,
      "step": 31917
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33937754729686703,
      "learning_rate": 2.6562788526818262e-08,
      "loss": 0.2419,
      "step": 31918
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.336811707161186,
      "learning_rate": 2.649058990742748e-08,
      "loss": 0.1134,
      "step": 31919
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39398965607951325,
      "learning_rate": 2.641848941070446e-08,
      "loss": 0.2659,
      "step": 31920
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1851800220069888,
      "learning_rate": 2.634648703735865e-08,
      "loss": 0.0701,
      "step": 31921
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40787033710351334,
      "learning_rate": 2.6274582788097246e-08,
      "loss": 0.1869,
      "step": 31922
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38032551046069757,
      "learning_rate": 2.6202776663627472e-08,
      "loss": 0.1252,
      "step": 31923
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3618357288225933,
      "learning_rate": 2.6131068664657644e-08,
      "loss": 0.2535,
      "step": 31924
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3304154914696061,
      "learning_rate": 2.6059458791891644e-08,
      "loss": 0.2382,
      "step": 31925
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.166369314574786,
      "learning_rate": 2.5987947046035576e-08,
      "loss": 0.4529,
      "step": 31926
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9735742398021544,
      "learning_rate": 2.59165334277911e-08,
      "loss": 0.553,
      "step": 31927
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5916606638150286,
      "learning_rate": 2.58452179378621e-08,
      "loss": 0.2056,
      "step": 31928
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35578840037408355,
      "learning_rate": 2.577400057694912e-08,
      "loss": 0.1907,
      "step": 31929
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5069481668904634,
      "learning_rate": 2.5702881345753827e-08,
      "loss": 0.3146,
      "step": 31930
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4842214133096374,
      "learning_rate": 2.563186024497566e-08,
      "loss": 0.2002,
      "step": 31931
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4005143833602348,
      "learning_rate": 2.5560937275312948e-08,
      "loss": 0.154,
      "step": 31932
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3650424085080018,
      "learning_rate": 2.549011243746402e-08,
      "loss": 0.2602,
      "step": 31933
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8287328261298236,
      "learning_rate": 2.54193857321261e-08,
      "loss": 0.2488,
      "step": 31934
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.484849867478517,
      "learning_rate": 2.5348757159993075e-08,
      "loss": 0.3071,
      "step": 31935
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3167736197501422,
      "learning_rate": 2.5278226721762167e-08,
      "loss": 0.2373,
      "step": 31936
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4414223931160822,
      "learning_rate": 2.5207794418125043e-08,
      "loss": 0.2403,
      "step": 31937
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5319891683408021,
      "learning_rate": 2.51374602497767e-08,
      "loss": 0.1699,
      "step": 31938
    },
    {
      "epoch": 0.98,
      "grad_norm": 2.7632716075498402,
      "learning_rate": 2.5067224217407706e-08,
      "loss": 0.6773,
      "step": 31939
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.39846433576930645,
      "learning_rate": 2.4997086321709717e-08,
      "loss": 0.1104,
      "step": 31940
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4176290798641794,
      "learning_rate": 2.4927046563373303e-08,
      "loss": 0.2862,
      "step": 31941
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36994596754855813,
      "learning_rate": 2.4857104943085685e-08,
      "loss": 0.1616,
      "step": 31942
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4060281294037785,
      "learning_rate": 2.4787261461536314e-08,
      "loss": 0.2663,
      "step": 31943
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4115126183512834,
      "learning_rate": 2.471751611941242e-08,
      "loss": 0.1587,
      "step": 31944
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.257910079441486,
      "learning_rate": 2.464786891740012e-08,
      "loss": 0.4322,
      "step": 31945
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1538048047428,
      "learning_rate": 2.4578319856183308e-08,
      "loss": 0.3752,
      "step": 31946
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3339097523442077,
      "learning_rate": 2.4508868936449215e-08,
      "loss": 0.1799,
      "step": 31947
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34525332965658645,
      "learning_rate": 2.4439516158877295e-08,
      "loss": 0.2773,
      "step": 31948
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2478086938357402,
      "learning_rate": 2.4370261524153672e-08,
      "loss": 0.0386,
      "step": 31949
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6344287526305425,
      "learning_rate": 2.4301105032956683e-08,
      "loss": 0.3467,
      "step": 31950
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2661177385614464,
      "learning_rate": 2.4232046685968014e-08,
      "loss": 0.1712,
      "step": 31951
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.442005267556059,
      "learning_rate": 2.4163086483867115e-08,
      "loss": 0.2049,
      "step": 31952
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2847820947291084,
      "learning_rate": 2.4094224427331225e-08,
      "loss": 0.4626,
      "step": 31953
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3535030491423104,
      "learning_rate": 2.40254605170398e-08,
      "loss": 0.2802,
      "step": 31954
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22625658703927684,
      "learning_rate": 2.3956794753668967e-08,
      "loss": 0.0881,
      "step": 31955
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4339449710114217,
      "learning_rate": 2.3888227137892628e-08,
      "loss": 0.259,
      "step": 31956
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4223511465674399,
      "learning_rate": 2.381975767038802e-08,
      "loss": 0.0659,
      "step": 31957
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6936412909374426,
      "learning_rate": 2.3751386351825722e-08,
      "loss": 0.24,
      "step": 31958
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37017715055656497,
      "learning_rate": 2.3683113182880747e-08,
      "loss": 0.3,
      "step": 31959
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29640571995502163,
      "learning_rate": 2.3614938164223668e-08,
      "loss": 0.1697,
      "step": 31960
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8468654740931387,
      "learning_rate": 2.354686129652506e-08,
      "loss": 0.6839,
      "step": 31961
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1226216427654003,
      "learning_rate": 2.34788825804555e-08,
      "loss": 0.414,
      "step": 31962
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6790095866095608,
      "learning_rate": 2.3411002016683337e-08,
      "loss": 0.2832,
      "step": 31963
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.188105932833476,
      "learning_rate": 2.3343219605876933e-08,
      "loss": 0.0676,
      "step": 31964
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.573671174838506,
      "learning_rate": 2.3275535348702414e-08,
      "loss": 0.3066,
      "step": 31965
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29521137456449,
      "learning_rate": 2.3207949245824812e-08,
      "loss": 0.2218,
      "step": 31966
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8226797160940991,
      "learning_rate": 2.3140461297912475e-08,
      "loss": 0.3488,
      "step": 31967
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6242843473771448,
      "learning_rate": 2.3073071505624877e-08,
      "loss": 0.1958,
      "step": 31968
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.941079531327988,
      "learning_rate": 2.3005779869628153e-08,
      "loss": 0.428,
      "step": 31969
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35522260108612386,
      "learning_rate": 2.2938586390583993e-08,
      "loss": 0.205,
      "step": 31970
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29960623248724927,
      "learning_rate": 2.2871491069151874e-08,
      "loss": 0.2181,
      "step": 31971
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5228491812763727,
      "learning_rate": 2.2804493905993487e-08,
      "loss": 0.7593,
      "step": 31972
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1661774137955021,
      "learning_rate": 2.27375949017683e-08,
      "loss": 0.0732,
      "step": 31973
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34880694217843394,
      "learning_rate": 2.267079405713246e-08,
      "loss": 0.2304,
      "step": 31974
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8155401093242031,
      "learning_rate": 2.260409137274433e-08,
      "loss": 0.2637,
      "step": 31975
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8096147889489684,
      "learning_rate": 2.253748684926005e-08,
      "loss": 0.4008,
      "step": 31976
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3468589958588413,
      "learning_rate": 2.2470980487335758e-08,
      "loss": 0.2202,
      "step": 31977
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8270818585194532,
      "learning_rate": 2.2404572287624272e-08,
      "loss": 0.2521,
      "step": 31978
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4074319858038386,
      "learning_rate": 2.2338262250779508e-08,
      "loss": 0.1999,
      "step": 31979
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4513136918399254,
      "learning_rate": 2.227205037745539e-08,
      "loss": 0.7144,
      "step": 31980
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2609235378012746,
      "learning_rate": 2.2205936668300287e-08,
      "loss": 0.4495,
      "step": 31981
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2094630746962939,
      "learning_rate": 2.213992112396701e-08,
      "loss": 0.1215,
      "step": 31982
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.31626328132295406,
      "learning_rate": 2.2074003745105043e-08,
      "loss": 0.2033,
      "step": 31983
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3192253795426574,
      "learning_rate": 2.2008184532361643e-08,
      "loss": 0.2269,
      "step": 31984
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.634821946794049,
      "learning_rate": 2.194246348638407e-08,
      "loss": 0.3473,
      "step": 31985
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2411095520532682,
      "learning_rate": 2.1876840607820692e-08,
      "loss": 0.0476,
      "step": 31986
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4241434822318481,
      "learning_rate": 2.1811315897316552e-08,
      "loss": 0.2623,
      "step": 31987
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4859432815416257,
      "learning_rate": 2.174588935551447e-08,
      "loss": 0.1895,
      "step": 31988
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3540824613269402,
      "learning_rate": 2.1680560983060595e-08,
      "loss": 0.691,
      "step": 31989
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34008470142905606,
      "learning_rate": 2.1615330780596634e-08,
      "loss": 0.2242,
      "step": 31990
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.27144788091766975,
      "learning_rate": 2.1550198748764295e-08,
      "loss": 0.1485,
      "step": 31991
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3825456349649073,
      "learning_rate": 2.148516488820418e-08,
      "loss": 0.1685,
      "step": 31992
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.495053486070988,
      "learning_rate": 2.1420229199556885e-08,
      "loss": 0.2774,
      "step": 31993
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7331354727215961,
      "learning_rate": 2.135539168346079e-08,
      "loss": 0.2051,
      "step": 31994
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33699997697661205,
      "learning_rate": 2.1290652340553163e-08,
      "loss": 0.2771,
      "step": 31995
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3708700282545771,
      "learning_rate": 2.122601117147238e-08,
      "loss": 0.0644,
      "step": 31996
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35514946930034164,
      "learning_rate": 2.1161468176852383e-08,
      "loss": 0.182,
      "step": 31997
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4050810529354156,
      "learning_rate": 2.109702335733044e-08,
      "loss": 0.7137,
      "step": 31998
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2669044663242457,
      "learning_rate": 2.103267671353937e-08,
      "loss": 0.1813,
      "step": 31999
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7429575290479663,
      "learning_rate": 2.0968428246112004e-08,
      "loss": 0.3599,
      "step": 32000
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.28061500170175374,
      "learning_rate": 2.0904277955680062e-08,
      "loss": 0.1748,
      "step": 32001
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.381390836737068,
      "learning_rate": 2.084022584287637e-08,
      "loss": 0.2708,
      "step": 32002
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2748006551826548,
      "learning_rate": 2.0776271908330426e-08,
      "loss": 0.1083,
      "step": 32003
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.550307924980024,
      "learning_rate": 2.0712416152670612e-08,
      "loss": 0.5674,
      "step": 32004
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.635819910828943,
      "learning_rate": 2.0648658576525315e-08,
      "loss": 0.0803,
      "step": 32005
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40609094779979793,
      "learning_rate": 2.0584999180521814e-08,
      "loss": 0.2957,
      "step": 32006
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.139916677042059,
      "learning_rate": 2.0521437965288493e-08,
      "loss": 0.4322,
      "step": 32007
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3534156062489102,
      "learning_rate": 2.0457974931447078e-08,
      "loss": 0.2786,
      "step": 32008
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7242179092919768,
      "learning_rate": 2.0394610079624842e-08,
      "loss": 0.2563,
      "step": 32009
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29028609946379014,
      "learning_rate": 2.0331343410443515e-08,
      "loss": 0.1682,
      "step": 32010
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9631037519145113,
      "learning_rate": 2.026817492452704e-08,
      "loss": 0.4855,
      "step": 32011
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.21652969369440464,
      "learning_rate": 2.020510462249492e-08,
      "loss": 0.0665,
      "step": 32012
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35405319089897946,
      "learning_rate": 2.0142132504969992e-08,
      "loss": 0.258,
      "step": 32013
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3454031726716577,
      "learning_rate": 2.0079258572569537e-08,
      "loss": 0.1587,
      "step": 32014
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5674597071306132,
      "learning_rate": 2.0016482825913064e-08,
      "loss": 0.3301,
      "step": 32015
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2868900519513167,
      "learning_rate": 1.9953805265618963e-08,
      "loss": 0.3691,
      "step": 32016
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.891604213250486,
      "learning_rate": 1.9891225892303412e-08,
      "loss": 0.3833,
      "step": 32017
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43608620572181955,
      "learning_rate": 1.982874470658147e-08,
      "loss": 0.2274,
      "step": 32018
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9694350221515077,
      "learning_rate": 1.9766361709068203e-08,
      "loss": 0.4193,
      "step": 32019
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.30406933097882666,
      "learning_rate": 1.9704076900377567e-08,
      "loss": 0.2216,
      "step": 32020
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45671661664386176,
      "learning_rate": 1.96418902811224e-08,
      "loss": 0.2663,
      "step": 32021
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4080585636753636,
      "learning_rate": 1.9579801851913326e-08,
      "loss": 0.0707,
      "step": 32022
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46546620946093986,
      "learning_rate": 1.951781161336319e-08,
      "loss": 0.1373,
      "step": 32023
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33958312435131927,
      "learning_rate": 1.9455919566080393e-08,
      "loss": 0.2355,
      "step": 32024
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3251978676448688,
      "learning_rate": 1.939412571067445e-08,
      "loss": 0.2325,
      "step": 32025
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8886985633524588,
      "learning_rate": 1.933243004775265e-08,
      "loss": 0.3543,
      "step": 32026
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7806987359672422,
      "learning_rate": 1.9270832577923394e-08,
      "loss": 0.2412,
      "step": 32027
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6206625787402087,
      "learning_rate": 1.9209333301790643e-08,
      "loss": 0.2449,
      "step": 32028
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3852607350958285,
      "learning_rate": 1.9147932219961697e-08,
      "loss": 0.1856,
      "step": 32029
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2589435341470849,
      "learning_rate": 1.9086629333038285e-08,
      "loss": 0.1448,
      "step": 32030
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34730232000175637,
      "learning_rate": 1.9025424641625488e-08,
      "loss": 0.2183,
      "step": 32031
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8962313377273265,
      "learning_rate": 1.8964318146325043e-08,
      "loss": 0.3698,
      "step": 32032
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.32907723805856787,
      "learning_rate": 1.8903309847737583e-08,
      "loss": 0.1894,
      "step": 32033
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.674539856559061,
      "learning_rate": 1.8842399746463735e-08,
      "loss": 0.7481,
      "step": 32034
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6598790563054706,
      "learning_rate": 1.878158784310191e-08,
      "loss": 0.2598,
      "step": 32035
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3448721929454724,
      "learning_rate": 1.872087413825163e-08,
      "loss": 0.2445,
      "step": 32036
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42975430755682287,
      "learning_rate": 1.8660258632510196e-08,
      "loss": 0.2575,
      "step": 32037
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5386795940988478,
      "learning_rate": 1.8599741326472688e-08,
      "loss": 0.1959,
      "step": 32038
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8327594003227654,
      "learning_rate": 1.853932222073529e-08,
      "loss": 0.5201,
      "step": 32039
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22720723923793182,
      "learning_rate": 1.8479001315891974e-08,
      "loss": 0.0664,
      "step": 32040
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8971447658710534,
      "learning_rate": 1.8418778612537825e-08,
      "loss": 0.4488,
      "step": 32041
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3852973711149659,
      "learning_rate": 1.8358654111263473e-08,
      "loss": 0.1662,
      "step": 32042
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35477781626187394,
      "learning_rate": 1.8298627812660675e-08,
      "loss": 0.2813,
      "step": 32043
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4261364474647689,
      "learning_rate": 1.8238699717320064e-08,
      "loss": 0.2262,
      "step": 32044
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9296477220919793,
      "learning_rate": 1.817886982583228e-08,
      "loss": 0.439,
      "step": 32045
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.27465242859581757,
      "learning_rate": 1.8119138138785742e-08,
      "loss": 0.0658,
      "step": 32046
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3915247605575569,
      "learning_rate": 1.8059504656766648e-08,
      "loss": 0.2821,
      "step": 32047
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4602891525358311,
      "learning_rate": 1.7999969380363413e-08,
      "loss": 0.0835,
      "step": 32048
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2614750556243743,
      "learning_rate": 1.7940532310160018e-08,
      "loss": 0.1906,
      "step": 32049
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.620793071611421,
      "learning_rate": 1.788119344674266e-08,
      "loss": 0.7173,
      "step": 32050
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.29367238381239313,
      "learning_rate": 1.7821952790695317e-08,
      "loss": 0.1727,
      "step": 32051
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7746955450326342,
      "learning_rate": 1.7762810342599744e-08,
      "loss": 0.3966,
      "step": 32052
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7263919442620175,
      "learning_rate": 1.77037661030377e-08,
      "loss": 0.285,
      "step": 32053
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5316325512583223,
      "learning_rate": 1.764482007259094e-08,
      "loss": 0.33,
      "step": 32054
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2930271751881219,
      "learning_rate": 1.7585972251839e-08,
      "loss": 0.1726,
      "step": 32055
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40422136914356,
      "learning_rate": 1.752722264136142e-08,
      "loss": 0.2822,
      "step": 32056
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3167516795607233,
      "learning_rate": 1.7468571241735512e-08,
      "loss": 0.0596,
      "step": 32057
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4671135914672277,
      "learning_rate": 1.741001805353748e-08,
      "loss": 0.2532,
      "step": 32058
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4474625288667471,
      "learning_rate": 1.7351563077345757e-08,
      "loss": 0.1881,
      "step": 32059
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.389109657131234,
      "learning_rate": 1.7293206313733213e-08,
      "loss": 0.2164,
      "step": 32060
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.26643505447446003,
      "learning_rate": 1.7234947763274946e-08,
      "loss": 0.2174,
      "step": 32061
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6908794432652706,
      "learning_rate": 1.7176787426544938e-08,
      "loss": 0.248,
      "step": 32062
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2967078132418952,
      "learning_rate": 1.7118725304113847e-08,
      "loss": 0.5536,
      "step": 32063
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.32345431846005684,
      "learning_rate": 1.706076139655344e-08,
      "loss": 0.137,
      "step": 32064
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6113633166124999,
      "learning_rate": 1.7002895704433253e-08,
      "loss": 0.34,
      "step": 32065
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.43817696413855706,
      "learning_rate": 1.6945128228325058e-08,
      "loss": 0.0086,
      "step": 32066
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2536184901805482,
      "learning_rate": 1.6887458968795066e-08,
      "loss": 0.2071,
      "step": 32067
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38113796167437275,
      "learning_rate": 1.6829887926410605e-08,
      "loss": 0.1375,
      "step": 32068
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4999846800118882,
      "learning_rate": 1.6772415101738993e-08,
      "loss": 0.2838,
      "step": 32069
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4619780825911824,
      "learning_rate": 1.6715040495344226e-08,
      "loss": 0.2335,
      "step": 32070
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9346320432652462,
      "learning_rate": 1.6657764107793628e-08,
      "loss": 0.4317,
      "step": 32071
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3245505634366695,
      "learning_rate": 1.660058593964786e-08,
      "loss": 0.225,
      "step": 32072
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.8210582602389525,
      "learning_rate": 1.654350599147092e-08,
      "loss": 0.7671,
      "step": 32073
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.28369990346209467,
      "learning_rate": 1.6486524263823467e-08,
      "loss": 0.1714,
      "step": 32074
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9046601302466655,
      "learning_rate": 1.6429640757266162e-08,
      "loss": 0.0622,
      "step": 32075
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5302211709929558,
      "learning_rate": 1.6372855472359673e-08,
      "loss": 0.2589,
      "step": 32076
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7828833683428044,
      "learning_rate": 1.6316168409662435e-08,
      "loss": 0.2574,
      "step": 32077
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.40182944942559434,
      "learning_rate": 1.6259579569731787e-08,
      "loss": 0.208,
      "step": 32078
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2760330950288958,
      "learning_rate": 1.6203088953123947e-08,
      "loss": 0.2266,
      "step": 32079
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5518941212850625,
      "learning_rate": 1.6146696560395136e-08,
      "loss": 0.8695,
      "step": 32080
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2879825907162308,
      "learning_rate": 1.609040239210158e-08,
      "loss": 0.1003,
      "step": 32081
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.908947598078393,
      "learning_rate": 1.6034206448793945e-08,
      "loss": 0.3475,
      "step": 32082
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37935250754946204,
      "learning_rate": 1.5978108731028454e-08,
      "loss": 0.1838,
      "step": 32083
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5268698672914838,
      "learning_rate": 1.5922109239354666e-08,
      "loss": 0.3347,
      "step": 32084
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.42448814415858865,
      "learning_rate": 1.5866207974324365e-08,
      "loss": 0.2084,
      "step": 32085
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7859845303131227,
      "learning_rate": 1.581040493648711e-08,
      "loss": 0.3454,
      "step": 32086
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.37279463012257874,
      "learning_rate": 1.5754700126393573e-08,
      "loss": 0.179,
      "step": 32087
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2758728520307805,
      "learning_rate": 1.5699093544589984e-08,
      "loss": 0.1444,
      "step": 32088
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5763113018389796,
      "learning_rate": 1.564358519162368e-08,
      "loss": 0.793,
      "step": 32089
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22058492993977366,
      "learning_rate": 1.558817506804089e-08,
      "loss": 0.1827,
      "step": 32090
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7936003858581091,
      "learning_rate": 1.5532863174387848e-08,
      "loss": 0.2993,
      "step": 32091
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.35833424793526086,
      "learning_rate": 1.5477649511207447e-08,
      "loss": 0.1973,
      "step": 32092
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4121617894234533,
      "learning_rate": 1.5422534079042596e-08,
      "loss": 0.543,
      "step": 32093
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7451477199212901,
      "learning_rate": 1.5367516878437293e-08,
      "loss": 0.2038,
      "step": 32094
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.48797902625059925,
      "learning_rate": 1.5312597909931116e-08,
      "loss": 0.3098,
      "step": 32095
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2529533768311861,
      "learning_rate": 1.5257777174064736e-08,
      "loss": 0.0676,
      "step": 32096
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.30430339960570213,
      "learning_rate": 1.5203054671378836e-08,
      "loss": 0.261,
      "step": 32097
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1873676379029412,
      "learning_rate": 1.5148430402409653e-08,
      "loss": 0.5155,
      "step": 32098
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.23970911229273711,
      "learning_rate": 1.5093904367695644e-08,
      "loss": 0.1383,
      "step": 32099
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.119211462423847,
      "learning_rate": 1.5039476567774158e-08,
      "loss": 0.071,
      "step": 32100
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4223126220914341,
      "learning_rate": 1.4985147003178103e-08,
      "loss": 0.2576,
      "step": 32101
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.32327978161016285,
      "learning_rate": 1.4930915674445934e-08,
      "loss": 0.1969,
      "step": 32102
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46529314620537887,
      "learning_rate": 1.487678258210723e-08,
      "loss": 0.2463,
      "step": 32103
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.0871663867752615,
      "learning_rate": 1.4822747726697117e-08,
      "loss": 0.4205,
      "step": 32104
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6850965882553286,
      "learning_rate": 1.4768811108746284e-08,
      "loss": 0.0986,
      "step": 32105
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38481972166767553,
      "learning_rate": 1.4714972728785415e-08,
      "loss": 0.2695,
      "step": 32106
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.1377935735417515,
      "learning_rate": 1.4661232587344088e-08,
      "loss": 0.3165,
      "step": 32107
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.47573663803601524,
      "learning_rate": 1.4607590684950768e-08,
      "loss": 0.2881,
      "step": 32108
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.1836001297749966,
      "learning_rate": 1.4554047022133921e-08,
      "loss": 0.1197,
      "step": 32109
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36269324974839195,
      "learning_rate": 1.4500601599419795e-08,
      "loss": 0.2383,
      "step": 32110
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.693724555443804,
      "learning_rate": 1.4447254417334634e-08,
      "loss": 0.2073,
      "step": 32111
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9805110336858619,
      "learning_rate": 1.4394005476403573e-08,
      "loss": 0.4007,
      "step": 32112
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.025153262355897,
      "learning_rate": 1.4340854777149527e-08,
      "loss": 0.1922,
      "step": 32113
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.41444869868825857,
      "learning_rate": 1.4287802320096522e-08,
      "loss": 0.2731,
      "step": 32114
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.282224078865603,
      "learning_rate": 1.4234848105764143e-08,
      "loss": 0.2091,
      "step": 32115
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2051730667193699,
      "learning_rate": 1.4181992134676414e-08,
      "loss": 0.4367,
      "step": 32116
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4933524676766534,
      "learning_rate": 1.412923440735181e-08,
      "loss": 0.2563,
      "step": 32117
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.20323415483499374,
      "learning_rate": 1.4076574924309916e-08,
      "loss": 0.0716,
      "step": 32118
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.34938938040574946,
      "learning_rate": 1.4024013686068095e-08,
      "loss": 0.2379,
      "step": 32119
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45457179055947106,
      "learning_rate": 1.397155069314371e-08,
      "loss": 0.2489,
      "step": 32120
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45824910010737174,
      "learning_rate": 1.3919185946053016e-08,
      "loss": 0.309,
      "step": 32121
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.2329575079560737,
      "learning_rate": 1.3866919445312266e-08,
      "loss": 0.2193,
      "step": 32122
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6316631315012852,
      "learning_rate": 1.3814751191433273e-08,
      "loss": 0.2961,
      "step": 32123
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3613702896206545,
      "learning_rate": 1.3762681184931181e-08,
      "loss": 0.2051,
      "step": 32124
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.6234200765748081,
      "learning_rate": 1.3710709426317803e-08,
      "loss": 0.7007,
      "step": 32125
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.317399636358798,
      "learning_rate": 1.365883591610495e-08,
      "loss": 0.2206,
      "step": 32126
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.2664711539614568,
      "learning_rate": 1.3607060654801108e-08,
      "loss": 0.1481,
      "step": 32127
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33616814681324936,
      "learning_rate": 1.3555383642918085e-08,
      "loss": 0.1493,
      "step": 32128
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.4517043453461179,
      "learning_rate": 1.3503804880962146e-08,
      "loss": 0.2211,
      "step": 32129
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8191177325995321,
      "learning_rate": 1.3452324369441772e-08,
      "loss": 0.3412,
      "step": 32130
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3518933184656101,
      "learning_rate": 1.3400942108864334e-08,
      "loss": 0.0941,
      "step": 32131
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.31170290381200605,
      "learning_rate": 1.3349658099733875e-08,
      "loss": 0.2524,
      "step": 32132
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.30110660693847957,
      "learning_rate": 1.3298472342556657e-08,
      "loss": 0.1881,
      "step": 32133
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.3204119944266979,
      "learning_rate": 1.3247384837833388e-08,
      "loss": 0.655,
      "step": 32134
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.8991291453121033,
      "learning_rate": 1.3196395586069221e-08,
      "loss": 0.0538,
      "step": 32135
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46987432519672356,
      "learning_rate": 1.3145504587765979e-08,
      "loss": 0.2377,
      "step": 32136
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3950811146207323,
      "learning_rate": 1.3094711843422148e-08,
      "loss": 0.1866,
      "step": 32137
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.28421338684362346,
      "learning_rate": 1.3044017353538441e-08,
      "loss": 0.2637,
      "step": 32138
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.45332072373924515,
      "learning_rate": 1.2993421118614458e-08,
      "loss": 0.1339,
      "step": 32139
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.7379363204874887,
      "learning_rate": 1.2942923139147579e-08,
      "loss": 0.6011,
      "step": 32140
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3237870085573402,
      "learning_rate": 1.2892523415634073e-08,
      "loss": 0.1127,
      "step": 32141
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3469362960567667,
      "learning_rate": 1.284222194857021e-08,
      "loss": 0.2196,
      "step": 32142
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.5141754306073936,
      "learning_rate": 1.279201873845115e-08,
      "loss": 0.6412,
      "step": 32143
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33642996335014663,
      "learning_rate": 1.2741913785768723e-08,
      "loss": 0.1994,
      "step": 32144
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7634589859909686,
      "learning_rate": 1.2691907091019195e-08,
      "loss": 0.3542,
      "step": 32145
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.27329804307587857,
      "learning_rate": 1.2641998654693288e-08,
      "loss": 0.1692,
      "step": 32146
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.087428908803341,
      "learning_rate": 1.259218847728061e-08,
      "loss": 0.4389,
      "step": 32147
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.36187771592109913,
      "learning_rate": 1.2542476559272987e-08,
      "loss": 0.1108,
      "step": 32148
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.49281133398587645,
      "learning_rate": 1.2492862901157809e-08,
      "loss": 0.2827,
      "step": 32149
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.38194184059766645,
      "learning_rate": 1.2443347503424686e-08,
      "loss": 0.1944,
      "step": 32150
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.463343748014058,
      "learning_rate": 1.2393930366561003e-08,
      "loss": 0.2462,
      "step": 32151
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46946544055443484,
      "learning_rate": 1.2344611491050818e-08,
      "loss": 0.254,
      "step": 32152
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.9242669869532438,
      "learning_rate": 1.229539087738152e-08,
      "loss": 0.4538,
      "step": 32153
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.7065543526533075,
      "learning_rate": 1.2246268526037165e-08,
      "loss": 0.2695,
      "step": 32154
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.33648317223900404,
      "learning_rate": 1.219724443749959e-08,
      "loss": 0.1494,
      "step": 32155
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3672115775636346,
      "learning_rate": 1.2148318612251742e-08,
      "loss": 0.2834,
      "step": 32156
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.22329592762693806,
      "learning_rate": 1.2099491050775459e-08,
      "loss": 0.1176,
      "step": 32157
    },
    {
      "epoch": 0.98,
      "grad_norm": 1.4458197008689242,
      "learning_rate": 1.2050761753551466e-08,
      "loss": 0.4069,
      "step": 32158
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5823881069471489,
      "learning_rate": 1.2002130721057159e-08,
      "loss": 0.1237,
      "step": 32159
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.3316784353718569,
      "learning_rate": 1.1953597953773266e-08,
      "loss": 0.2243,
      "step": 32160
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.46720877380209413,
      "learning_rate": 1.1905163452176072e-08,
      "loss": 0.2826,
      "step": 32161
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.5121349080421672,
      "learning_rate": 1.1856827216742972e-08,
      "loss": 0.328,
      "step": 32162
    },
    {
      "epoch": 0.98,
      "grad_norm": 0.6055473471949322,
      "learning_rate": 1.1808589247949142e-08,
      "loss": 0.2466,
      "step": 32163
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6315023760643984,
      "learning_rate": 1.1760449546268649e-08,
      "loss": 0.2835,
      "step": 32164
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35762092488773467,
      "learning_rate": 1.1712408112174445e-08,
      "loss": 0.1817,
      "step": 32165
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.48470970586365636,
      "learning_rate": 1.1664464946141707e-08,
      "loss": 0.2299,
      "step": 32166
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3262426291314105,
      "learning_rate": 1.1616620048638948e-08,
      "loss": 0.1548,
      "step": 32167
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3803579368237341,
      "learning_rate": 1.1568873420139127e-08,
      "loss": 0.2115,
      "step": 32168
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.355767667849509,
      "learning_rate": 1.1521225061111862e-08,
      "loss": 0.2452,
      "step": 32169
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9224056751277666,
      "learning_rate": 1.1473674972025673e-08,
      "loss": 0.3949,
      "step": 32170
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.808075790505799,
      "learning_rate": 1.1426223153346849e-08,
      "loss": 0.4116,
      "step": 32171
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8693747384588844,
      "learning_rate": 1.1378869605545017e-08,
      "loss": 0.2635,
      "step": 32172
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4030161985322677,
      "learning_rate": 1.133161432908314e-08,
      "loss": 0.2476,
      "step": 32173
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32501867108299687,
      "learning_rate": 1.1284457324428621e-08,
      "loss": 0.1995,
      "step": 32174
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5043324016507265,
      "learning_rate": 1.1237398592043313e-08,
      "loss": 0.3299,
      "step": 32175
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29139657924669515,
      "learning_rate": 1.11904381323924e-08,
      "loss": 0.0763,
      "step": 32176
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0026724642837663,
      "learning_rate": 1.1143575945935514e-08,
      "loss": 0.5153,
      "step": 32177
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38853473554518675,
      "learning_rate": 1.1096812033135618e-08,
      "loss": 0.1748,
      "step": 32178
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5453878375499224,
      "learning_rate": 1.1050146394452343e-08,
      "loss": 0.3121,
      "step": 32179
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3440066382486867,
      "learning_rate": 1.1003579030344213e-08,
      "loss": 0.2307,
      "step": 32180
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9742930313693701,
      "learning_rate": 1.095710994126975e-08,
      "loss": 0.2461,
      "step": 32181
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8032304061553407,
      "learning_rate": 1.0910739127685254e-08,
      "loss": 0.2958,
      "step": 32182
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3102131906723526,
      "learning_rate": 1.0864466590048139e-08,
      "loss": 0.1703,
      "step": 32183
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24359551514762842,
      "learning_rate": 1.0818292328813596e-08,
      "loss": 0.125,
      "step": 32184
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41170459280618266,
      "learning_rate": 1.0772216344435704e-08,
      "loss": 0.1867,
      "step": 32185
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5157290578809136,
      "learning_rate": 1.0726238637367437e-08,
      "loss": 0.3194,
      "step": 32186
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29717006577141847,
      "learning_rate": 1.0680359208061764e-08,
      "loss": 0.165,
      "step": 32187
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7610419825494835,
      "learning_rate": 1.0634578056968326e-08,
      "loss": 0.427,
      "step": 32188
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9462592725301149,
      "learning_rate": 1.0588895184540093e-08,
      "loss": 0.2568,
      "step": 32189
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3851599464892579,
      "learning_rate": 1.0543310591224487e-08,
      "loss": 0.6942,
      "step": 32190
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3727778430142346,
      "learning_rate": 1.0497824277471146e-08,
      "loss": 0.1681,
      "step": 32191
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28693923946410005,
      "learning_rate": 1.0452436243727494e-08,
      "loss": 0.236,
      "step": 32192
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.871057277666778,
      "learning_rate": 1.0407146490439834e-08,
      "loss": 0.0855,
      "step": 32193
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.24848587625721077,
      "learning_rate": 1.0361955018053372e-08,
      "loss": 0.0695,
      "step": 32194
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.967343327032107,
      "learning_rate": 1.0316861827013303e-08,
      "loss": 0.4839,
      "step": 32195
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29296228968043525,
      "learning_rate": 1.0271866917761497e-08,
      "loss": 0.1733,
      "step": 32196
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9346968884482095,
      "learning_rate": 1.0226970290744264e-08,
      "loss": 0.387,
      "step": 32197
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3131839425480338,
      "learning_rate": 1.0182171946400144e-08,
      "loss": 0.2193,
      "step": 32198
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.670553685786435,
      "learning_rate": 1.0137471885171002e-08,
      "loss": 0.7354,
      "step": 32199
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5126716525573422,
      "learning_rate": 1.0092870107496488e-08,
      "loss": 0.0718,
      "step": 32200
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5542748067562958,
      "learning_rate": 1.0048366613815142e-08,
      "loss": 0.2871,
      "step": 32201
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2762852923406167,
      "learning_rate": 1.0003961404565498e-08,
      "loss": 0.1345,
      "step": 32202
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.25122559486338397,
      "learning_rate": 9.959654480183878e-09,
      "loss": 0.2062,
      "step": 32203
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6523727296683279,
      "learning_rate": 9.915445841106596e-09,
      "loss": 0.222,
      "step": 32204
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0042913671579565,
      "learning_rate": 9.871335487767753e-09,
      "loss": 0.3776,
      "step": 32205
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3597692261769846,
      "learning_rate": 9.827323420601442e-09,
      "loss": 0.2205,
      "step": 32206
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1063895659071699,
      "learning_rate": 9.783409640042874e-09,
      "loss": 0.3238,
      "step": 32207
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5385643459741185,
      "learning_rate": 9.739594146520592e-09,
      "loss": 0.2919,
      "step": 32208
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38136143736287037,
      "learning_rate": 9.695876940468696e-09,
      "loss": 0.1911,
      "step": 32209
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3535770784299864,
      "learning_rate": 9.65225802231573e-09,
      "loss": 0.234,
      "step": 32210
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2137348258680491,
      "learning_rate": 9.608737392490242e-09,
      "loss": 0.0586,
      "step": 32211
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.26262558519177975,
      "learning_rate": 9.565315051421886e-09,
      "loss": 0.1567,
      "step": 32212
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6415978278260759,
      "learning_rate": 9.52199099953699e-09,
      "loss": 0.2561,
      "step": 32213
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6710415278983436,
      "learning_rate": 9.478765237262987e-09,
      "loss": 0.2651,
      "step": 32214
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2861102251127599,
      "learning_rate": 9.435637765022876e-09,
      "loss": 0.2152,
      "step": 32215
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4786796816100174,
      "learning_rate": 9.39260858324298e-09,
      "loss": 0.336,
      "step": 32216
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1368886980236081,
      "learning_rate": 9.349677692346292e-09,
      "loss": 0.255,
      "step": 32217
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8088941598919508,
      "learning_rate": 9.306845092754702e-09,
      "loss": 0.3147,
      "step": 32218
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32051848104175545,
      "learning_rate": 9.26411078488898e-09,
      "loss": 0.1905,
      "step": 32219
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6437206063176869,
      "learning_rate": 9.221474769171013e-09,
      "loss": 0.0185,
      "step": 32220
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33379753771360365,
      "learning_rate": 9.178937046019353e-09,
      "loss": 0.2553,
      "step": 32221
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4164199601042973,
      "learning_rate": 9.136497615852557e-09,
      "loss": 0.1799,
      "step": 32222
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40592960204596057,
      "learning_rate": 9.094156479089177e-09,
      "loss": 0.2374,
      "step": 32223
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5519569300128464,
      "learning_rate": 9.051913636144439e-09,
      "loss": 0.2113,
      "step": 32224
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.402693865757136,
      "learning_rate": 9.009769087433562e-09,
      "loss": 0.7776,
      "step": 32225
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.26572130953530243,
      "learning_rate": 8.967722833372882e-09,
      "loss": 0.166,
      "step": 32226
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.47424874900186126,
      "learning_rate": 8.925774874375403e-09,
      "loss": 0.2955,
      "step": 32227
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3875131155257168,
      "learning_rate": 8.883925210853018e-09,
      "loss": 0.163,
      "step": 32228
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5597089800077221,
      "learning_rate": 8.842173843218726e-09,
      "loss": 0.2573,
      "step": 32229
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7163047989303502,
      "learning_rate": 8.800520771882204e-09,
      "loss": 0.2305,
      "step": 32230
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.49493971710795304,
      "learning_rate": 8.758965997254232e-09,
      "loss": 0.2451,
      "step": 32231
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3169930216031328,
      "learning_rate": 8.717509519742262e-09,
      "loss": 0.071,
      "step": 32232
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27266069796172937,
      "learning_rate": 8.676151339755967e-09,
      "loss": 0.1937,
      "step": 32233
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5112538538168315,
      "learning_rate": 8.634891457700579e-09,
      "loss": 0.3534,
      "step": 32234
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.252510429022797,
      "learning_rate": 8.593729873982437e-09,
      "loss": 0.0993,
      "step": 32235
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8375125955236133,
      "learning_rate": 8.552666589007885e-09,
      "loss": 0.6804,
      "step": 32236
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30412540308911873,
      "learning_rate": 8.511701603178823e-09,
      "loss": 0.1701,
      "step": 32237
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9332782135939972,
      "learning_rate": 8.470834916900483e-09,
      "loss": 0.373,
      "step": 32238
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33798336799513545,
      "learning_rate": 8.430066530572546e-09,
      "loss": 0.2321,
      "step": 32239
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7707345849476555,
      "learning_rate": 8.389396444596909e-09,
      "loss": 0.3337,
      "step": 32240
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2925396953493127,
      "learning_rate": 8.348824659375476e-09,
      "loss": 0.0675,
      "step": 32241
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3912267839185163,
      "learning_rate": 8.308351175304596e-09,
      "loss": 0.2792,
      "step": 32242
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.027800609448312,
      "learning_rate": 8.26797599278506e-09,
      "loss": 0.5139,
      "step": 32243
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2758114737097792,
      "learning_rate": 8.227699112212107e-09,
      "loss": 0.1633,
      "step": 32244
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31268963605083044,
      "learning_rate": 8.187520533983195e-09,
      "loss": 0.1968,
      "step": 32245
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2714984576196906,
      "learning_rate": 8.147440258493566e-09,
      "loss": 0.1658,
      "step": 32246
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.094266508755076,
      "learning_rate": 8.107458286136238e-09,
      "loss": 0.3931,
      "step": 32247
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6626508000516673,
      "learning_rate": 8.06757461730534e-09,
      "loss": 0.2524,
      "step": 32248
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2033565773485388,
      "learning_rate": 8.027789252395001e-09,
      "loss": 0.3741,
      "step": 32249
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3458971570513096,
      "learning_rate": 7.9881021917938e-09,
      "loss": 0.1674,
      "step": 32250
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5803431761550067,
      "learning_rate": 7.948513435893646e-09,
      "loss": 0.292,
      "step": 32251
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3160253021809296,
      "learning_rate": 7.909022985084225e-09,
      "loss": 0.2298,
      "step": 32252
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3159148042567457,
      "learning_rate": 7.869630839753007e-09,
      "loss": 0.2219,
      "step": 32253
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8698033765010689,
      "learning_rate": 7.830337000288568e-09,
      "loss": 0.0609,
      "step": 32254
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8908221791964879,
      "learning_rate": 7.791141467077268e-09,
      "loss": 0.3401,
      "step": 32255
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38330449882640966,
      "learning_rate": 7.752044240504353e-09,
      "loss": 0.2143,
      "step": 32256
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.34361933765419794,
      "learning_rate": 7.713045320955071e-09,
      "loss": 0.2649,
      "step": 32257
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9313618551543857,
      "learning_rate": 7.674144708812448e-09,
      "loss": 0.2082,
      "step": 32258
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4710749637869374,
      "learning_rate": 7.635342404459511e-09,
      "loss": 0.1413,
      "step": 32259
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3710135228221222,
      "learning_rate": 7.596638408278178e-09,
      "loss": 0.255,
      "step": 32260
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1259513765317317,
      "learning_rate": 7.558032720648145e-09,
      "loss": 0.5056,
      "step": 32261
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2987715118623093,
      "learning_rate": 7.519525341951328e-09,
      "loss": 0.233,
      "step": 32262
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28714557581993283,
      "learning_rate": 7.481116272565203e-09,
      "loss": 0.0649,
      "step": 32263
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4204742453039772,
      "learning_rate": 7.442805512867246e-09,
      "loss": 0.2687,
      "step": 32264
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43786411010819476,
      "learning_rate": 7.404593063234933e-09,
      "loss": 0.2383,
      "step": 32265
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8742195620383388,
      "learning_rate": 7.366478924044629e-09,
      "loss": 0.3689,
      "step": 32266
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9869718790049078,
      "learning_rate": 7.3284630956715896e-09,
      "loss": 0.055,
      "step": 32267
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.44246045995432726,
      "learning_rate": 7.29054557848774e-09,
      "loss": 0.279,
      "step": 32268
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2718245065441951,
      "learning_rate": 7.2527263728683354e-09,
      "loss": 0.2054,
      "step": 32269
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7048962133017613,
      "learning_rate": 7.21500547918419e-09,
      "loss": 0.7081,
      "step": 32270
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6444149848849412,
      "learning_rate": 7.177382897807228e-09,
      "loss": 0.069,
      "step": 32271
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2756369593039999,
      "learning_rate": 7.139858629107155e-09,
      "loss": 0.1278,
      "step": 32272
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4204418008822093,
      "learning_rate": 7.102432673453674e-09,
      "loss": 0.2248,
      "step": 32273
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5333198031198646,
      "learning_rate": 7.06510503121427e-09,
      "loss": 0.2179,
      "step": 32274
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.369107568046873,
      "learning_rate": 7.027875702756426e-09,
      "loss": 0.2778,
      "step": 32275
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1891209018600108,
      "learning_rate": 6.990744688446516e-09,
      "loss": 0.0975,
      "step": 32276
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.187207225452877,
      "learning_rate": 6.953711988649803e-09,
      "loss": 0.4395,
      "step": 32277
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3812726636794636,
      "learning_rate": 6.91677760373044e-09,
      "loss": 0.1609,
      "step": 32278
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5482957061688172,
      "learning_rate": 6.879941534052581e-09,
      "loss": 0.3647,
      "step": 32279
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.25897770174436896,
      "learning_rate": 6.843203779977048e-09,
      "loss": 0.1319,
      "step": 32280
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4370781219258231,
      "learning_rate": 6.806564341866883e-09,
      "loss": 0.2767,
      "step": 32281
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31370338323287467,
      "learning_rate": 6.770023220081801e-09,
      "loss": 0.065,
      "step": 32282
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.41049171121840133,
      "learning_rate": 6.733580414981511e-09,
      "loss": 0.2797,
      "step": 32283
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42553645944993657,
      "learning_rate": 6.697235926924617e-09,
      "loss": 0.1368,
      "step": 32284
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2564159492211806,
      "learning_rate": 6.6609897562675e-09,
      "loss": 0.7318,
      "step": 32285
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3160354987824929,
      "learning_rate": 6.6248419033687615e-09,
      "loss": 0.2233,
      "step": 32286
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2739629108943658,
      "learning_rate": 6.588792368583674e-09,
      "loss": 0.1676,
      "step": 32287
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5289800141324135,
      "learning_rate": 6.552841152265288e-09,
      "loss": 0.8278,
      "step": 32288
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.755053604199937,
      "learning_rate": 6.516988254768874e-09,
      "loss": 0.2295,
      "step": 32289
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0485612651149292,
      "learning_rate": 6.481233676446375e-09,
      "loss": 0.4282,
      "step": 32290
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6106572577998921,
      "learning_rate": 6.4455774176497285e-09,
      "loss": 0.1494,
      "step": 32291
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29407320859645303,
      "learning_rate": 6.410019478729767e-09,
      "loss": 0.2267,
      "step": 32292
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2476616257750523,
      "learning_rate": 6.374559860036212e-09,
      "loss": 0.1722,
      "step": 32293
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.759526567034114,
      "learning_rate": 6.33919856191878e-09,
      "loss": 0.6845,
      "step": 32294
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7502590580928352,
      "learning_rate": 6.303935584723864e-09,
      "loss": 0.2034,
      "step": 32295
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38078040632001126,
      "learning_rate": 6.268770928800072e-09,
      "loss": 0.238,
      "step": 32296
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4169560505549275,
      "learning_rate": 6.233704594491574e-09,
      "loss": 0.2156,
      "step": 32297
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0090304557216685,
      "learning_rate": 6.19873658214476e-09,
      "loss": 0.3971,
      "step": 32298
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32433114740614016,
      "learning_rate": 6.163866892102688e-09,
      "loss": 0.223,
      "step": 32299
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6002316538953385,
      "learning_rate": 6.129095524709527e-09,
      "loss": 0.1899,
      "step": 32300
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5428272713385012,
      "learning_rate": 6.094422480307227e-09,
      "loss": 0.3095,
      "step": 32301
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.14318118251150289,
      "learning_rate": 6.059847759235515e-09,
      "loss": 0.0694,
      "step": 32302
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3167316707849783,
      "learning_rate": 6.02537136183523e-09,
      "loss": 0.4362,
      "step": 32303
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2913852430395657,
      "learning_rate": 5.990993288446101e-09,
      "loss": 0.1992,
      "step": 32304
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.484036266368118,
      "learning_rate": 5.956713539406744e-09,
      "loss": 0.289,
      "step": 32305
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7407180728489056,
      "learning_rate": 5.922532115052448e-09,
      "loss": 0.2216,
      "step": 32306
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7843715311711873,
      "learning_rate": 5.888449015721831e-09,
      "loss": 0.4074,
      "step": 32307
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7228940493888342,
      "learning_rate": 5.854464241749069e-09,
      "loss": 0.2092,
      "step": 32308
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0658397417892778,
      "learning_rate": 5.82057779346834e-09,
      "loss": 0.3199,
      "step": 32309
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3044751796099463,
      "learning_rate": 5.786789671212712e-09,
      "loss": 0.1833,
      "step": 32310
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.26210432975780285,
      "learning_rate": 5.753099875316359e-09,
      "loss": 0.2196,
      "step": 32311
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.464634567978091,
      "learning_rate": 5.71950840610902e-09,
      "loss": 0.1088,
      "step": 32312
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.2235275749538956,
      "learning_rate": 5.68601526392154e-09,
      "loss": 0.4467,
      "step": 32313
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4382509912716005,
      "learning_rate": 5.652620449083657e-09,
      "loss": 0.2393,
      "step": 32314
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4591202475411601,
      "learning_rate": 5.619323961922884e-09,
      "loss": 0.2432,
      "step": 32315
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3444480184772641,
      "learning_rate": 5.5861258027689606e-09,
      "loss": 0.2599,
      "step": 32316
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4075451491555957,
      "learning_rate": 5.553025971947179e-09,
      "loss": 0.2623,
      "step": 32317
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9545635384846848,
      "learning_rate": 5.520024469783947e-09,
      "loss": 0.3916,
      "step": 32318
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29674980333110146,
      "learning_rate": 5.487121296602338e-09,
      "loss": 0.1701,
      "step": 32319
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31026068365516873,
      "learning_rate": 5.454316452727648e-09,
      "loss": 0.1636,
      "step": 32320
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.365231139941028,
      "learning_rate": 5.421609938481842e-09,
      "loss": 0.1568,
      "step": 32321
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3449264507812352,
      "learning_rate": 5.389001754186885e-09,
      "loss": 0.2682,
      "step": 32322
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3544078316345922,
      "learning_rate": 5.356491900163629e-09,
      "loss": 0.1729,
      "step": 32323
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5387080826711863,
      "learning_rate": 5.3240803767307114e-09,
      "loss": 0.3332,
      "step": 32324
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8432605301659948,
      "learning_rate": 5.291767184210095e-09,
      "loss": 0.2518,
      "step": 32325
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0604430134608487,
      "learning_rate": 5.259552322917083e-09,
      "loss": 0.1944,
      "step": 32326
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7092673696342576,
      "learning_rate": 5.2274357931692e-09,
      "loss": 0.2465,
      "step": 32327
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28570182494808505,
      "learning_rate": 5.195417595281749e-09,
      "loss": 0.2102,
      "step": 32328
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5475817412849582,
      "learning_rate": 5.163497729571143e-09,
      "loss": 0.3317,
      "step": 32329
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28741210251134713,
      "learning_rate": 5.131676196351576e-09,
      "loss": 0.0906,
      "step": 32330
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.1411331651722305,
      "learning_rate": 5.09995299593391e-09,
      "loss": 0.4483,
      "step": 32331
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3524589563566611,
      "learning_rate": 5.068328128632338e-09,
      "loss": 0.1566,
      "step": 32332
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5251792007814033,
      "learning_rate": 5.036801594757723e-09,
      "loss": 0.3062,
      "step": 32333
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33348672038978616,
      "learning_rate": 5.005373394619817e-09,
      "loss": 0.228,
      "step": 32334
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.525197839810897,
      "learning_rate": 4.974043528527262e-09,
      "loss": 0.521,
      "step": 32335
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6825919538921297,
      "learning_rate": 4.942811996788699e-09,
      "loss": 0.0705,
      "step": 32336
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.49300957244244836,
      "learning_rate": 4.911678799711661e-09,
      "loss": 0.2926,
      "step": 32337
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32424761876922176,
      "learning_rate": 4.880643937602569e-09,
      "loss": 0.1397,
      "step": 32338
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27142241976179193,
      "learning_rate": 4.849707410766735e-09,
      "loss": 0.177,
      "step": 32339
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5145598646209952,
      "learning_rate": 4.8188692195083596e-09,
      "loss": 0.3338,
      "step": 32340
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5345942169379576,
      "learning_rate": 4.788129364130534e-09,
      "loss": 0.1998,
      "step": 32341
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39282351234871765,
      "learning_rate": 4.7574878449363485e-09,
      "loss": 0.2603,
      "step": 32342
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8665916057648697,
      "learning_rate": 4.726944662226674e-09,
      "loss": 0.3404,
      "step": 32343
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.719588526420925,
      "learning_rate": 4.69649981630127e-09,
      "loss": 0.5704,
      "step": 32344
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4526927440241424,
      "learning_rate": 4.666153307462118e-09,
      "loss": 0.2085,
      "step": 32345
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.34893132635585367,
      "learning_rate": 4.635905136005647e-09,
      "loss": 0.2428,
      "step": 32346
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2671649005415675,
      "learning_rate": 4.605755302229398e-09,
      "loss": 0.1466,
      "step": 32347
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4448758438093454,
      "learning_rate": 4.575703806430909e-09,
      "loss": 0.2011,
      "step": 32348
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8238989792031394,
      "learning_rate": 4.545750648906611e-09,
      "loss": 0.3174,
      "step": 32349
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6227971681535281,
      "learning_rate": 4.515895829948491e-09,
      "loss": 0.3002,
      "step": 32350
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.36575628988793407,
      "learning_rate": 4.48613934985298e-09,
      "loss": 0.2159,
      "step": 32351
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31140522585127817,
      "learning_rate": 4.456481208912067e-09,
      "loss": 0.212,
      "step": 32352
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5843984125773127,
      "learning_rate": 4.426921407416629e-09,
      "loss": 0.7074,
      "step": 32353
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.803561600960507,
      "learning_rate": 4.397459945658656e-09,
      "loss": 0.0761,
      "step": 32354
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33417966202261923,
      "learning_rate": 4.368096823926804e-09,
      "loss": 0.2412,
      "step": 32355
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5557216196009414,
      "learning_rate": 4.3388320425108435e-09,
      "loss": 0.0858,
      "step": 32356
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4443117896922073,
      "learning_rate": 4.30966560169832e-09,
      "loss": 0.246,
      "step": 32357
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31427600895739455,
      "learning_rate": 4.280597501776784e-09,
      "loss": 0.2134,
      "step": 32358
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3576074891796748,
      "learning_rate": 4.251627743031561e-09,
      "loss": 0.1997,
      "step": 32359
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4947417294411885,
      "learning_rate": 4.2227563257479786e-09,
      "loss": 0.2171,
      "step": 32360
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4419800045497937,
      "learning_rate": 4.193983250210254e-09,
      "loss": 0.7813,
      "step": 32361
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.472397195612602,
      "learning_rate": 4.165308516700384e-09,
      "loss": 0.084,
      "step": 32362
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5335521835519687,
      "learning_rate": 4.1367321255014744e-09,
      "loss": 0.3406,
      "step": 32363
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3313773036230896,
      "learning_rate": 4.108254076894414e-09,
      "loss": 0.1527,
      "step": 32364
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3342152871778582,
      "learning_rate": 4.079874371160086e-09,
      "loss": 0.2107,
      "step": 32365
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7176160628546419,
      "learning_rate": 4.051593008576049e-09,
      "loss": 0.3679,
      "step": 32366
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7722733578048018,
      "learning_rate": 4.023409989422078e-09,
      "loss": 0.2588,
      "step": 32367
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9059814727456331,
      "learning_rate": 3.995325313975728e-09,
      "loss": 0.5034,
      "step": 32368
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.30766901477016523,
      "learning_rate": 3.967338982511226e-09,
      "loss": 0.1821,
      "step": 32369
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3354957670731993,
      "learning_rate": 3.939450995306127e-09,
      "loss": 0.2673,
      "step": 32370
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.16262856297790593,
      "learning_rate": 3.911661352632435e-09,
      "loss": 0.0673,
      "step": 32371
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0835337080406733,
      "learning_rate": 3.883970054766595e-09,
      "loss": 0.3815,
      "step": 32372
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2757334280520292,
      "learning_rate": 3.856377101978392e-09,
      "loss": 0.1693,
      "step": 32373
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5394999494131407,
      "learning_rate": 3.828882494540942e-09,
      "loss": 0.4618,
      "step": 32374
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7830761198079915,
      "learning_rate": 3.8014862327229175e-09,
      "loss": 0.241,
      "step": 32375
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.34950158754738214,
      "learning_rate": 3.7741883167963234e-09,
      "loss": 0.2579,
      "step": 32376
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5177933432137064,
      "learning_rate": 3.7469887470287235e-09,
      "loss": 0.1749,
      "step": 32377
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3776975803598984,
      "learning_rate": 3.7198875236865717e-09,
      "loss": 0.2242,
      "step": 32378
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.7138331169023566,
      "learning_rate": 3.692884647037431e-09,
      "loss": 0.7249,
      "step": 32379
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21212111238841325,
      "learning_rate": 3.6659801173466456e-09,
      "loss": 0.0725,
      "step": 32380
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.34118850727541084,
      "learning_rate": 3.6391739348795586e-09,
      "loss": 0.2675,
      "step": 32381
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2922536835375063,
      "learning_rate": 3.6124660999004025e-09,
      "loss": 0.1715,
      "step": 32382
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.8224211880364833,
      "learning_rate": 3.58585661267008e-09,
      "loss": 0.6709,
      "step": 32383
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7172065075616229,
      "learning_rate": 3.5593454734506038e-09,
      "loss": 0.2475,
      "step": 32384
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9237836952326477,
      "learning_rate": 3.532932682505097e-09,
      "loss": 0.4372,
      "step": 32385
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3677362807912506,
      "learning_rate": 3.506618240091131e-09,
      "loss": 0.1472,
      "step": 32386
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7416538246321323,
      "learning_rate": 3.480402146468498e-09,
      "loss": 0.3417,
      "step": 32387
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.27959178915190513,
      "learning_rate": 3.4542844018947696e-09,
      "loss": 0.2142,
      "step": 32388
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2816386785275723,
      "learning_rate": 3.428265006627518e-09,
      "loss": 0.1643,
      "step": 32389
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0535484051576347,
      "learning_rate": 3.402343960922094e-09,
      "loss": 0.0656,
      "step": 32390
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3416624894342098,
      "learning_rate": 3.3765212650327394e-09,
      "loss": 0.1514,
      "step": 32391
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5287297996745114,
      "learning_rate": 3.350796919215915e-09,
      "loss": 0.3163,
      "step": 32392
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3450771973568955,
      "learning_rate": 3.3251709237225315e-09,
      "loss": 0.2438,
      "step": 32393
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6637799801314317,
      "learning_rate": 3.2996432788057197e-09,
      "loss": 0.5736,
      "step": 32394
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4988205693903391,
      "learning_rate": 3.2742139847163902e-09,
      "loss": 0.1967,
      "step": 32395
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3535777704081858,
      "learning_rate": 3.2488830417043427e-09,
      "loss": 0.2404,
      "step": 32396
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2632936781837482,
      "learning_rate": 3.223650450019378e-09,
      "loss": 0.1261,
      "step": 32397
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4675611203358693,
      "learning_rate": 3.1985162099101853e-09,
      "loss": 0.2182,
      "step": 32398
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31983373355655226,
      "learning_rate": 3.173480321622124e-09,
      "loss": 0.225,
      "step": 32399
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6614303283681086,
      "learning_rate": 3.1485427854038853e-09,
      "loss": 0.3327,
      "step": 32400
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35707321322552676,
      "learning_rate": 3.1237036014986066e-09,
      "loss": 0.2121,
      "step": 32401
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.124787389684213,
      "learning_rate": 3.098962770152758e-09,
      "loss": 0.5201,
      "step": 32402
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7860297464488973,
      "learning_rate": 3.074320291608368e-09,
      "loss": 0.1651,
      "step": 32403
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3298006299823672,
      "learning_rate": 3.0497761661074654e-09,
      "loss": 0.14,
      "step": 32404
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.39639539520515776,
      "learning_rate": 3.0253303938931886e-09,
      "loss": 0.2857,
      "step": 32405
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28905315185607505,
      "learning_rate": 3.0009829752053466e-09,
      "loss": 0.2099,
      "step": 32406
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6975267548547136,
      "learning_rate": 2.9767339102826366e-09,
      "loss": 0.199,
      "step": 32407
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5073328746349692,
      "learning_rate": 2.9525831993648667e-09,
      "loss": 0.1671,
      "step": 32408
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4612662573476004,
      "learning_rate": 2.928530842688515e-09,
      "loss": 0.2942,
      "step": 32409
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.49287009503637425,
      "learning_rate": 2.904576840491169e-09,
      "loss": 0.212,
      "step": 32410
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5226528350520109,
      "learning_rate": 2.880721193008196e-09,
      "loss": 0.3507,
      "step": 32411
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.42937818499634467,
      "learning_rate": 2.8569639004727424e-09,
      "loss": 0.1767,
      "step": 32412
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8589560452634404,
      "learning_rate": 2.833304963121286e-09,
      "loss": 0.4255,
      "step": 32413
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3928223196954599,
      "learning_rate": 2.809744381185864e-09,
      "loss": 0.1685,
      "step": 32414
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5889701818597224,
      "learning_rate": 2.7862821548962916e-09,
      "loss": 0.3642,
      "step": 32415
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.48349036608458656,
      "learning_rate": 2.762918284485716e-09,
      "loss": 0.1345,
      "step": 32416
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3459718672826302,
      "learning_rate": 2.7396527701828436e-09,
      "loss": 0.2315,
      "step": 32417
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7985337057361624,
      "learning_rate": 2.7164856122163797e-09,
      "loss": 0.405,
      "step": 32418
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32460711096198275,
      "learning_rate": 2.6934168108150306e-09,
      "loss": 0.1961,
      "step": 32419
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5150153353835948,
      "learning_rate": 2.6704463662052814e-09,
      "loss": 0.258,
      "step": 32420
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6296416779510616,
      "learning_rate": 2.6475742786136182e-09,
      "loss": 0.1882,
      "step": 32421
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4645469673449176,
      "learning_rate": 2.624800548264306e-09,
      "loss": 0.7967,
      "step": 32422
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.31307927777501326,
      "learning_rate": 2.602125175381609e-09,
      "loss": 0.1671,
      "step": 32423
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3480097962138839,
      "learning_rate": 2.579548160188683e-09,
      "loss": 0.2872,
      "step": 32424
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2963757600403414,
      "learning_rate": 2.557069502908682e-09,
      "loss": 0.0651,
      "step": 32425
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8073478110592133,
      "learning_rate": 2.534689203760321e-09,
      "loss": 0.3667,
      "step": 32426
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5437585247128915,
      "learning_rate": 2.5124072629667538e-09,
      "loss": 0.1931,
      "step": 32427
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2766154262172796,
      "learning_rate": 2.4902236807444745e-09,
      "loss": 0.2139,
      "step": 32428
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4845918611014987,
      "learning_rate": 2.4681384573144174e-09,
      "loss": 0.2102,
      "step": 32429
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.436568456969257,
      "learning_rate": 2.4461515928908553e-09,
      "loss": 0.2222,
      "step": 32430
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4536646114230383,
      "learning_rate": 2.4242630876925023e-09,
      "loss": 0.7184,
      "step": 32431
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2807798412530708,
      "learning_rate": 2.4024729419347413e-09,
      "loss": 0.1672,
      "step": 32432
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.6630715830434752,
      "learning_rate": 2.3807811558296258e-09,
      "loss": 0.6876,
      "step": 32433
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8091206451479053,
      "learning_rate": 2.359187729592538e-09,
      "loss": 0.2339,
      "step": 32434
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.33380695488147244,
      "learning_rate": 2.3376926634366417e-09,
      "loss": 0.2744,
      "step": 32435
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.37481545838723945,
      "learning_rate": 2.316295957570658e-09,
      "loss": 0.1806,
      "step": 32436
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5755699944060393,
      "learning_rate": 2.29499761220664e-09,
      "loss": 0.3192,
      "step": 32437
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.251490686111125,
      "learning_rate": 2.273797627554419e-09,
      "loss": 0.0793,
      "step": 32438
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0767063265173267,
      "learning_rate": 2.2526960038227187e-09,
      "loss": 0.4238,
      "step": 32439
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3597849333114178,
      "learning_rate": 2.2316927412180387e-09,
      "loss": 0.2007,
      "step": 32440
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9932626145661696,
      "learning_rate": 2.210787839947992e-09,
      "loss": 0.4499,
      "step": 32441
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28897341688984346,
      "learning_rate": 2.1899813002179695e-09,
      "loss": 0.2189,
      "step": 32442
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6711678306952332,
      "learning_rate": 2.1692731222333617e-09,
      "loss": 0.2333,
      "step": 32443
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8784861262234593,
      "learning_rate": 2.14866330619623e-09,
      "loss": 0.3752,
      "step": 32444
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.21332239585883048,
      "learning_rate": 2.1281518523097456e-09,
      "loss": 0.0674,
      "step": 32445
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3663613777298841,
      "learning_rate": 2.107738760777078e-09,
      "loss": 0.2571,
      "step": 32446
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.22041987494267215,
      "learning_rate": 2.0874240317980687e-09,
      "loss": 0.1669,
      "step": 32447
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.5555481901078083,
      "learning_rate": 2.0672076655725572e-09,
      "loss": 0.631,
      "step": 32448
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.0152546896784964,
      "learning_rate": 2.0470896623003834e-09,
      "loss": 0.418,
      "step": 32449
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5723966290611089,
      "learning_rate": 2.027070022178057e-09,
      "loss": 0.264,
      "step": 32450
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.38763650376473563,
      "learning_rate": 2.0071487454031977e-09,
      "loss": 0.2116,
      "step": 32451
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7651811798233589,
      "learning_rate": 1.9873258321712054e-09,
      "loss": 0.3367,
      "step": 32452
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.349407045486783,
      "learning_rate": 1.9676012826785883e-09,
      "loss": 0.2325,
      "step": 32453
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9882868046572288,
      "learning_rate": 1.9479750971185265e-09,
      "loss": 0.4279,
      "step": 32454
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32809603309037577,
      "learning_rate": 1.9284472756830873e-09,
      "loss": 0.1876,
      "step": 32455
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.17608979738974379,
      "learning_rate": 1.909017818566561e-09,
      "loss": 0.0674,
      "step": 32456
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.3735423929616557,
      "learning_rate": 1.8896867259576846e-09,
      "loss": 0.5496,
      "step": 32457
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3139479740694737,
      "learning_rate": 1.870453998049637e-09,
      "loss": 0.2356,
      "step": 32458
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.43924678568432823,
      "learning_rate": 1.8513196350278263e-09,
      "loss": 0.27,
      "step": 32459
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4719223617590386,
      "learning_rate": 1.83228363708432e-09,
      "loss": 0.218,
      "step": 32460
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.6656393405585633,
      "learning_rate": 1.8133460044034157e-09,
      "loss": 0.3525,
      "step": 32461
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8546718801834898,
      "learning_rate": 1.7945067371738512e-09,
      "loss": 0.1609,
      "step": 32462
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.8316650369649525,
      "learning_rate": 1.7757658355788131e-09,
      "loss": 0.3394,
      "step": 32463
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3832596799095679,
      "learning_rate": 1.757123299803709e-09,
      "loss": 0.1624,
      "step": 32464
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28959266879755297,
      "learning_rate": 1.7385791300317255e-09,
      "loss": 0.2145,
      "step": 32465
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.2950756781812049,
      "learning_rate": 1.7201333264460496e-09,
      "loss": 0.1617,
      "step": 32466
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.9741516695404733,
      "learning_rate": 1.7017858892276473e-09,
      "loss": 0.454,
      "step": 32467
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.29740394768809036,
      "learning_rate": 1.6835368185563749e-09,
      "loss": 0.0681,
      "step": 32468
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.381160187030201,
      "learning_rate": 1.6653861146120886e-09,
      "loss": 0.2393,
      "step": 32469
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.4497116607535383,
      "learning_rate": 1.6473337775746444e-09,
      "loss": 0.2508,
      "step": 32470
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.51721845329503,
      "learning_rate": 1.629379807620568e-09,
      "loss": 0.2173,
      "step": 32471
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4583985278184883,
      "learning_rate": 1.6115242049252744e-09,
      "loss": 0.6157,
      "step": 32472
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28363755692138043,
      "learning_rate": 1.5937669696663994e-09,
      "loss": 0.1724,
      "step": 32473
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.32420804237772954,
      "learning_rate": 1.5761081020182478e-09,
      "loss": 0.1683,
      "step": 32474
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.932013447788425,
      "learning_rate": 1.5585476021540147e-09,
      "loss": 0.2835,
      "step": 32475
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.49923983260456756,
      "learning_rate": 1.5410854702468948e-09,
      "loss": 0.3367,
      "step": 32476
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35811513290971436,
      "learning_rate": 1.5237217064689725e-09,
      "loss": 0.1858,
      "step": 32477
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.40584757779930036,
      "learning_rate": 1.5064563109901121e-09,
      "loss": 0.2685,
      "step": 32478
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7315966371106077,
      "learning_rate": 1.4892892839801777e-09,
      "loss": 0.2765,
      "step": 32479
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.4807531231164601,
      "learning_rate": 1.4722206256090331e-09,
      "loss": 0.3639,
      "step": 32480
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5697527710272818,
      "learning_rate": 1.4552503360443226e-09,
      "loss": 0.0383,
      "step": 32481
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3895528038760407,
      "learning_rate": 1.4383784154525793e-09,
      "loss": 0.2173,
      "step": 32482
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.28527780896543936,
      "learning_rate": 1.4216048639992264e-09,
      "loss": 0.2278,
      "step": 32483
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.1687096847393558,
      "learning_rate": 1.4049296818519075e-09,
      "loss": 0.0845,
      "step": 32484
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.7686463353101939,
      "learning_rate": 1.3883528691716053e-09,
      "loss": 0.3918,
      "step": 32485
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.35459577741855436,
      "learning_rate": 1.3718744261226325e-09,
      "loss": 0.1638,
      "step": 32486
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.5515751464549336,
      "learning_rate": 1.3554943528670816e-09,
      "loss": 0.3222,
      "step": 32487
    },
    {
      "epoch": 0.99,
      "grad_norm": 1.257327047994915,
      "learning_rate": 1.3392126495670454e-09,
      "loss": 0.4234,
      "step": 32488
    },
    {
      "epoch": 0.99,
      "grad_norm": 0.3686406414930253,
      "learning_rate": 1.3230293163812858e-09,
      "loss": 0.2684,
      "step": 32489
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8155935089985257,
      "learning_rate": 1.3069443534696746e-09,
      "loss": 0.0753,
      "step": 32490
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0742014232212196,
      "learning_rate": 1.2909577609898639e-09,
      "loss": 0.4775,
      "step": 32491
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.261917362934881,
      "learning_rate": 1.2750695391006152e-09,
      "loss": 0.1561,
      "step": 32492
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5330250447529064,
      "learning_rate": 1.2592796879562496e-09,
      "loss": 0.3214,
      "step": 32493
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33661317484339304,
      "learning_rate": 1.2435882077133088e-09,
      "loss": 0.2358,
      "step": 32494
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0357074651703249,
      "learning_rate": 1.2279950985261134e-09,
      "loss": 0.4566,
      "step": 32495
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.29615252711987716,
      "learning_rate": 1.2125003605478747e-09,
      "loss": 0.1722,
      "step": 32496
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9615313034726116,
      "learning_rate": 1.197103993931803e-09,
      "loss": 0.4751,
      "step": 32497
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5227630181719045,
      "learning_rate": 1.1818059988277786e-09,
      "loss": 0.4981,
      "step": 32498
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.306876262983503,
      "learning_rate": 1.1666063753867918e-09,
      "loss": 0.1924,
      "step": 32499
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3954874550858749,
      "learning_rate": 1.151505123759833e-09,
      "loss": 0.2484,
      "step": 32500
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.25556081679628584,
      "learning_rate": 1.1365022440945616e-09,
      "loss": 0.2008,
      "step": 32501
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4344682662635186,
      "learning_rate": 1.1215977365375274e-09,
      "loss": 0.212,
      "step": 32502
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7813041112476216,
      "learning_rate": 1.10679160123639e-09,
      "loss": 0.2542,
      "step": 32503
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7275083993575128,
      "learning_rate": 1.0920838383376986e-09,
      "loss": 0.3685,
      "step": 32504
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30875883162992773,
      "learning_rate": 1.077474447983562e-09,
      "loss": 0.1863,
      "step": 32505
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3380445304722824,
      "learning_rate": 1.0629634303205294e-09,
      "loss": 0.2286,
      "step": 32506
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3901424326888251,
      "learning_rate": 1.0485507854895993e-09,
      "loss": 0.1861,
      "step": 32507
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1869692132856433,
      "learning_rate": 1.0342365136339904e-09,
      "loss": 0.4059,
      "step": 32508
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3416000440012232,
      "learning_rate": 1.0200206148924807e-09,
      "loss": 0.1443,
      "step": 32509
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6072385832492174,
      "learning_rate": 1.0059030894071786e-09,
      "loss": 0.184,
      "step": 32510
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8590303273187545,
      "learning_rate": 9.918839373146415e-10,
      "loss": 0.3479,
      "step": 32511
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33751977744058415,
      "learning_rate": 9.779631587547578e-10,
      "loss": 0.2347,
      "step": 32512
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6356215938656091,
      "learning_rate": 9.641407538640847e-10,
      "loss": 0.3395,
      "step": 32513
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.36016328001709624,
      "learning_rate": 9.504167227780691e-10,
      "loss": 0.21,
      "step": 32514
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.3217458514087639,
      "learning_rate": 9.367910656321588e-10,
      "loss": 0.8075,
      "step": 32515
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.16092449698199673,
      "learning_rate": 9.232637825595803e-10,
      "loss": 0.0677,
      "step": 32516
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.539549914607191,
      "learning_rate": 9.098348736946705e-10,
      "loss": 0.2993,
      "step": 32517
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.34948974973710517,
      "learning_rate": 8.965043391684358e-10,
      "loss": 0.0653,
      "step": 32518
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30885444433686743,
      "learning_rate": 8.832721791129928e-10,
      "loss": 0.2578,
      "step": 32519
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7028893436769172,
      "learning_rate": 8.701383936582375e-10,
      "loss": 0.2522,
      "step": 32520
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4476403565256466,
      "learning_rate": 8.571029829329558e-10,
      "loss": 0.4208,
      "step": 32521
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2026712479497383,
      "learning_rate": 8.441659470659335e-10,
      "loss": 0.4428,
      "step": 32522
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2807741276345384,
      "learning_rate": 8.313272861837363e-10,
      "loss": 0.164,
      "step": 32523
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.32666026978899065,
      "learning_rate": 8.185870004129292e-10,
      "loss": 0.22,
      "step": 32524
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3586640912039219,
      "learning_rate": 8.059450898789678e-10,
      "loss": 0.1503,
      "step": 32525
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4842895958422297,
      "learning_rate": 7.934015547073071e-10,
      "loss": 0.5009,
      "step": 32526
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2832164762822568,
      "learning_rate": 7.809563950189614e-10,
      "loss": 0.0647,
      "step": 32527
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3802744573452773,
      "learning_rate": 7.686096109382756e-10,
      "loss": 0.2722,
      "step": 32528
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0785554707498826,
      "learning_rate": 7.563612025873746e-10,
      "loss": 0.2509,
      "step": 32529
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5166028555978939,
      "learning_rate": 7.442111700839416e-10,
      "loss": 0.2953,
      "step": 32530
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3936036769721855,
      "learning_rate": 7.321595135501014e-10,
      "loss": 0.2301,
      "step": 32531
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3578574329958259,
      "learning_rate": 7.202062331024273e-10,
      "loss": 0.2567,
      "step": 32532
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.27917149770920135,
      "learning_rate": 7.083513288608235e-10,
      "loss": 0.126,
      "step": 32533
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4677515446808362,
      "learning_rate": 6.965948009396428e-10,
      "loss": 0.2404,
      "step": 32534
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5120586603710914,
      "learning_rate": 6.84936649456569e-10,
      "loss": 0.191,
      "step": 32535
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.25385822060407226,
      "learning_rate": 6.733768745248448e-10,
      "loss": 0.1689,
      "step": 32536
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.49766537228054375,
      "learning_rate": 6.61915476257713e-10,
      "loss": 0.2776,
      "step": 32537
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9194826202753326,
      "learning_rate": 6.505524547706365e-10,
      "loss": 0.2546,
      "step": 32538
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9847344998025415,
      "learning_rate": 6.392878101724176e-10,
      "loss": 0.4019,
      "step": 32539
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5492298838322837,
      "learning_rate": 6.281215425751885e-10,
      "loss": 0.2597,
      "step": 32540
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5072141020197924,
      "learning_rate": 6.170536520888615e-10,
      "loss": 0.263,
      "step": 32541
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.34450402647693884,
      "learning_rate": 6.060841388222383e-10,
      "loss": 0.2218,
      "step": 32542
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2592126599486304,
      "learning_rate": 5.952130028830105e-10,
      "loss": 0.1684,
      "step": 32543
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7743908338546613,
      "learning_rate": 5.844402443788699e-10,
      "loss": 0.2341,
      "step": 32544
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5759228664506537,
      "learning_rate": 5.737658634152877e-10,
      "loss": 0.2345,
      "step": 32545
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38609785720223977,
      "learning_rate": 5.631898600966246e-10,
      "loss": 0.21,
      "step": 32546
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9748190606353706,
      "learning_rate": 5.52712234528352e-10,
      "loss": 0.4833,
      "step": 32547
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3058905471994023,
      "learning_rate": 5.423329868115002e-10,
      "loss": 0.215,
      "step": 32548
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2155749812029286,
      "learning_rate": 5.3205211705043e-10,
      "loss": 0.4447,
      "step": 32549
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4035899988142494,
      "learning_rate": 5.218696253450617e-10,
      "loss": 0.2427,
      "step": 32550
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.46192299521182056,
      "learning_rate": 5.117855117953152e-10,
      "loss": 0.1912,
      "step": 32551
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30430956865457254,
      "learning_rate": 5.017997765011107e-10,
      "loss": 0.1664,
      "step": 32552
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4467718326190671,
      "learning_rate": 4.919124195612579e-10,
      "loss": 0.2211,
      "step": 32553
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.43566367725231814,
      "learning_rate": 4.821234410712361e-10,
      "loss": 0.2761,
      "step": 32554
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33950056352166513,
      "learning_rate": 4.724328411287448e-10,
      "loss": 0.1997,
      "step": 32555
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0851953641373013,
      "learning_rate": 4.628406198292634e-10,
      "loss": 0.4682,
      "step": 32556
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2056602130569458,
      "learning_rate": 4.533467772660505e-10,
      "loss": 0.3208,
      "step": 32557
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.539871780758275,
      "learning_rate": 4.4395131353347497e-10,
      "loss": 0.7764,
      "step": 32558
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.44060170750009847,
      "learning_rate": 4.3465422872257523e-10,
      "loss": 0.1439,
      "step": 32559
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.28963133033633004,
      "learning_rate": 4.2545552292661e-10,
      "loss": 0.2535,
      "step": 32560
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.7116766802179786,
      "learning_rate": 4.1635519623550725e-10,
      "loss": 0.0625,
      "step": 32561
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.43655349679631333,
      "learning_rate": 4.073532487380849e-10,
      "loss": 0.1955,
      "step": 32562
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6395395953724666,
      "learning_rate": 3.9844968052427103e-10,
      "loss": 0.3051,
      "step": 32563
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.24050026201632643,
      "learning_rate": 3.8964449167955276e-10,
      "loss": 0.1516,
      "step": 32564
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.8404432103684627,
      "learning_rate": 3.80937682292748e-10,
      "loss": 0.796,
      "step": 32565
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3280884534604586,
      "learning_rate": 3.723292524482336e-10,
      "loss": 0.2372,
      "step": 32566
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.387364478400599,
      "learning_rate": 3.638192022314968e-10,
      "loss": 0.6315,
      "step": 32567
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33636940156306433,
      "learning_rate": 3.554075317258043e-10,
      "loss": 0.1459,
      "step": 32568
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.523231539923004,
      "learning_rate": 3.470942410133127e-10,
      "loss": 0.3024,
      "step": 32569
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8713108869591177,
      "learning_rate": 3.3887933017728857e-10,
      "loss": 0.2307,
      "step": 32570
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.48355596017784297,
      "learning_rate": 3.3076279929655787e-10,
      "loss": 0.3059,
      "step": 32571
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2632091048680283,
      "learning_rate": 3.2274464845327705e-10,
      "loss": 0.1559,
      "step": 32572
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5138819369363666,
      "learning_rate": 3.1482487772405145e-10,
      "loss": 0.2626,
      "step": 32573
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33744045858434896,
      "learning_rate": 3.0700348718881724e-10,
      "loss": 0.1318,
      "step": 32574
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1525588381916752,
      "learning_rate": 2.9928047692306947e-10,
      "loss": 0.3116,
      "step": 32575
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.6907359912618645,
      "learning_rate": 2.9165584700341366e-10,
      "loss": 0.7959,
      "step": 32576
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.35144347765061446,
      "learning_rate": 2.841295975053449e-10,
      "loss": 0.1534,
      "step": 32577
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.29116987837845254,
      "learning_rate": 2.767017285010276e-10,
      "loss": 0.2572,
      "step": 32578
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8323934969148267,
      "learning_rate": 2.693722400659571e-10,
      "loss": 0.2469,
      "step": 32579
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.0259891940606196,
      "learning_rate": 2.6214113227118754e-10,
      "loss": 0.3936,
      "step": 32580
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4200163152258231,
      "learning_rate": 2.5500840518777327e-10,
      "loss": 0.1232,
      "step": 32581
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3770442842933597,
      "learning_rate": 2.479740588856583e-10,
      "loss": 0.2686,
      "step": 32582
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.299754759348073,
      "learning_rate": 2.4103809343367646e-10,
      "loss": 0.1484,
      "step": 32583
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5674093775261768,
      "learning_rate": 2.342005089017718e-10,
      "loss": 0.36,
      "step": 32584
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.098564501479847,
      "learning_rate": 2.2746130535655776e-10,
      "loss": 0.5557,
      "step": 32585
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6955591327777603,
      "learning_rate": 2.208204828624272e-10,
      "loss": 0.3816,
      "step": 32586
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3244036068504695,
      "learning_rate": 2.1427804148710374e-10,
      "loss": 0.1991,
      "step": 32587
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.895957471850313,
      "learning_rate": 2.0783398129387012e-10,
      "loss": 0.275,
      "step": 32588
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4588343614580395,
      "learning_rate": 2.0148830234711924e-10,
      "loss": 0.3019,
      "step": 32589
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.23455731125756057,
      "learning_rate": 1.952410047068032e-10,
      "loss": 0.143,
      "step": 32590
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3399867728706195,
      "learning_rate": 1.8909208843731486e-10,
      "loss": 0.2213,
      "step": 32591
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.43677651476798185,
      "learning_rate": 1.8304155359749609e-10,
      "loss": 0.0851,
      "step": 32592
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1981705758701027,
      "learning_rate": 1.7708940024840916e-10,
      "loss": 0.4976,
      "step": 32593
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4155803733759097,
      "learning_rate": 1.7123562844556518e-10,
      "loss": 0.2231,
      "step": 32594
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.40600781336573144,
      "learning_rate": 1.6548023825002646e-10,
      "loss": 0.2637,
      "step": 32595
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3387153590116756,
      "learning_rate": 1.598232297161939e-10,
      "loss": 0.2262,
      "step": 32596
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6582833387744164,
      "learning_rate": 1.542646029006889e-10,
      "loss": 0.3571,
      "step": 32597
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9461019616077997,
      "learning_rate": 1.4880435785791236e-10,
      "loss": 0.1972,
      "step": 32598
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38469307747032394,
      "learning_rate": 1.43442494641155e-10,
      "loss": 0.2123,
      "step": 32599
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3965863422592206,
      "learning_rate": 1.3817901330370753e-10,
      "loss": 0.1599,
      "step": 32600
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.23576072360392641,
      "learning_rate": 1.3301391389775043e-10,
      "loss": 0.1652,
      "step": 32601
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5216911969425083,
      "learning_rate": 1.2794719647324372e-10,
      "loss": 0.3688,
      "step": 32602
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2989702429484906,
      "learning_rate": 1.2297886108014745e-10,
      "loss": 0.4611,
      "step": 32603
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7163433160811831,
      "learning_rate": 1.1810890776842165e-10,
      "loss": 0.3016,
      "step": 32604
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.33435525298441104,
      "learning_rate": 1.1333733658358548e-10,
      "loss": 0.1975,
      "step": 32605
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.801021062916197,
      "learning_rate": 1.0866414757559895e-10,
      "loss": 0.3722,
      "step": 32606
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.442130187037269,
      "learning_rate": 1.04089340788871e-10,
      "loss": 0.2005,
      "step": 32607
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.49544062917284964,
      "learning_rate": 9.961291626781055e-11,
      "loss": 0.2662,
      "step": 32608
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.28350911088750097,
      "learning_rate": 9.523487405682652e-11,
      "loss": 0.1711,
      "step": 32609
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.306211594843124,
      "learning_rate": 9.095521420032782e-11,
      "loss": 0.1602,
      "step": 32610
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4442731098729689,
      "learning_rate": 8.677393673939272e-11,
      "loss": 0.4639,
      "step": 32611
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9443089497308303,
      "learning_rate": 8.269104171509945e-11,
      "loss": 0.4794,
      "step": 32612
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3709732364136562,
      "learning_rate": 7.870652916741606e-11,
      "loss": 0.1981,
      "step": 32613
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.30017604354045524,
      "learning_rate": 7.482039913520034e-11,
      "loss": 0.2177,
      "step": 32614
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9920744734449357,
      "learning_rate": 7.103265165842032e-11,
      "loss": 0.4379,
      "step": 32615
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2040927165573572,
      "learning_rate": 6.734328677260315e-11,
      "loss": 0.0413,
      "step": 32616
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5072575210202142,
      "learning_rate": 6.375230451438619e-11,
      "loss": 0.2085,
      "step": 32617
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3021424154292928,
      "learning_rate": 6.025970492040678e-11,
      "loss": 0.1739,
      "step": 32618
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.41934344890972913,
      "learning_rate": 5.6865488022861404e-11,
      "loss": 0.1988,
      "step": 32619
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3244768994394752,
      "learning_rate": 5.3569653857277194e-11,
      "loss": 0.2388,
      "step": 32620
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8102014909517524,
      "learning_rate": 5.037220245474039e-11,
      "loss": 0.3802,
      "step": 32621
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5526026657911255,
      "learning_rate": 4.727313384633725e-11,
      "loss": 0.1885,
      "step": 32622
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5543740228544981,
      "learning_rate": 4.427244806426423e-11,
      "loss": 0.3331,
      "step": 32623
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6038194347006814,
      "learning_rate": 4.137014513738713e-11,
      "loss": 0.2146,
      "step": 32624
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3498984651429842,
      "learning_rate": 3.8566225092351304e-11,
      "loss": 0.2733,
      "step": 32625
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.6647386477341205,
      "learning_rate": 3.5860687960243e-11,
      "loss": 0.0663,
      "step": 32626
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38946401206509484,
      "learning_rate": 3.32535337643769e-11,
      "loss": 0.1736,
      "step": 32627
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.36462645954281175,
      "learning_rate": 3.0744762532508574e-11,
      "loss": 0.2354,
      "step": 32628
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.9261053982789746,
      "learning_rate": 2.8334374287952714e-11,
      "loss": 0.3345,
      "step": 32629
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.8365086701482908,
      "learning_rate": 2.6022369055134223e-11,
      "loss": 0.3529,
      "step": 32630
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2889720072143006,
      "learning_rate": 2.380874685736778e-11,
      "loss": 0.1961,
      "step": 32631
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3649620709420431,
      "learning_rate": 2.169350771574763e-11,
      "loss": 0.2739,
      "step": 32632
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.41999584680286667,
      "learning_rate": 1.9676651650257784e-11,
      "loss": 0.1548,
      "step": 32633
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.2044888493478367,
      "learning_rate": 1.7758178681992476e-11,
      "loss": 0.4972,
      "step": 32634
    },
    {
      "epoch": 1.0,
      "grad_norm": 2.0513147067820583,
      "learning_rate": 1.5938088829825503e-11,
      "loss": 0.082,
      "step": 32635
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3243647477338183,
      "learning_rate": 1.4216382111520433e-11,
      "loss": 0.1814,
      "step": 32636
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.2411436491749102,
      "learning_rate": 1.2593058542620384e-11,
      "loss": 0.21,
      "step": 32637
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.5629492188636507,
      "learning_rate": 1.1068118140888928e-11,
      "loss": 0.8379,
      "step": 32638
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7189856495674793,
      "learning_rate": 9.641560920758963e-12,
      "loss": 0.2505,
      "step": 32639
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.5667492938156989,
      "learning_rate": 8.31338689666339e-12,
      "loss": 0.1776,
      "step": 32640
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.3862802475224262,
      "learning_rate": 7.0835960808146584e-12,
      "loss": 0.2865,
      "step": 32641
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1936044553884259,
      "learning_rate": 5.952188485425226e-12,
      "loss": 0.0662,
      "step": 32642
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.36694003770569533,
      "learning_rate": 4.919164122707543e-12,
      "loss": 0.2765,
      "step": 32643
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1111519615038195,
      "learning_rate": 3.9845230015433944e-12,
      "loss": 0.0927,
      "step": 32644
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.38579067535888956,
      "learning_rate": 3.148265130814565e-12,
      "loss": 0.228,
      "step": 32645
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.48999933692353365,
      "learning_rate": 2.410390520513062e-12,
      "loss": 0.1937,
      "step": 32646
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.7918168446690911,
      "learning_rate": 1.7708991773002227e-12,
      "loss": 0.3521,
      "step": 32647
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.49458368846475403,
      "learning_rate": 1.2297911067271628e-12,
      "loss": 0.2537,
      "step": 32648
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4305579529348849,
      "learning_rate": 7.870663143449975e-13,
      "loss": 0.3207,
      "step": 32649
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.35945471670199075,
      "learning_rate": 4.4272480459461864e-13,
      "loss": 0.1553,
      "step": 32650
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.31863065513401234,
      "learning_rate": 1.967665808066954e-13,
      "loss": 0.1978,
      "step": 32651
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.4043516614775359,
      "learning_rate": 4.919164520167385e-14,
      "loss": 0.0633,
      "step": 32652
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.4422801142352744,
      "learning_rate": 0.0,
      "loss": 0.0439,
      "step": 32653
    },
    {
      "epoch": 1.0,
      "step": 32653,
      "total_flos": 0.0,
      "train_loss": 0.14857381297872044,
      "train_runtime": 208564.7172,
      "train_samples_per_second": 40.097,
      "train_steps_per_second": 0.157
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 32653,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 1000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}