{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 534,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0037488284910965324,
      "grad_norm": 3.1443588733673096,
      "learning_rate": 0.0,
      "loss": 0.7334,
      "step": 1
    },
    {
      "epoch": 0.007497656982193065,
      "grad_norm": 2.8891870975494385,
      "learning_rate": 3.7037037037037036e-07,
      "loss": 0.8449,
      "step": 2
    },
    {
      "epoch": 0.011246485473289597,
      "grad_norm": 3.069650888442993,
      "learning_rate": 7.407407407407407e-07,
      "loss": 0.8849,
      "step": 3
    },
    {
      "epoch": 0.01499531396438613,
      "grad_norm": 3.2767655849456787,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.918,
      "step": 4
    },
    {
      "epoch": 0.01874414245548266,
      "grad_norm": 3.732121467590332,
      "learning_rate": 1.4814814814814815e-06,
      "loss": 0.9034,
      "step": 5
    },
    {
      "epoch": 0.022492970946579195,
      "grad_norm": 3.3119568824768066,
      "learning_rate": 1.8518518518518519e-06,
      "loss": 0.8088,
      "step": 6
    },
    {
      "epoch": 0.026241799437675725,
      "grad_norm": 2.831653118133545,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.7799,
      "step": 7
    },
    {
      "epoch": 0.02999062792877226,
      "grad_norm": 2.624796152114868,
      "learning_rate": 2.5925925925925925e-06,
      "loss": 0.7546,
      "step": 8
    },
    {
      "epoch": 0.033739456419868794,
      "grad_norm": 2.72420597076416,
      "learning_rate": 2.962962962962963e-06,
      "loss": 0.9903,
      "step": 9
    },
    {
      "epoch": 0.03748828491096532,
      "grad_norm": 2.9389147758483887,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.8715,
      "step": 10
    },
    {
      "epoch": 0.041237113402061855,
      "grad_norm": 3.242018461227417,
      "learning_rate": 3.7037037037037037e-06,
      "loss": 1.1089,
      "step": 11
    },
    {
      "epoch": 0.04498594189315839,
      "grad_norm": 3.1691386699676514,
      "learning_rate": 4.074074074074074e-06,
      "loss": 0.9839,
      "step": 12
    },
    {
      "epoch": 0.04873477038425492,
      "grad_norm": 3.211369037628174,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.6889,
      "step": 13
    },
    {
      "epoch": 0.05248359887535145,
      "grad_norm": 2.753861665725708,
      "learning_rate": 4.814814814814815e-06,
      "loss": 0.9228,
      "step": 14
    },
    {
      "epoch": 0.056232427366447985,
      "grad_norm": 2.5858447551727295,
      "learning_rate": 5.185185185185185e-06,
      "loss": 0.6551,
      "step": 15
    },
    {
      "epoch": 0.05998125585754452,
      "grad_norm": 2.531930446624756,
      "learning_rate": 5.555555555555557e-06,
      "loss": 0.7092,
      "step": 16
    },
    {
      "epoch": 0.06373008434864105,
      "grad_norm": 3.0797035694122314,
      "learning_rate": 5.925925925925926e-06,
      "loss": 0.6336,
      "step": 17
    },
    {
      "epoch": 0.06747891283973759,
      "grad_norm": 2.6957359313964844,
      "learning_rate": 6.296296296296297e-06,
      "loss": 0.7413,
      "step": 18
    },
    {
      "epoch": 0.07122774133083412,
      "grad_norm": 2.9657933712005615,
      "learning_rate": 6.666666666666667e-06,
      "loss": 0.9917,
      "step": 19
    },
    {
      "epoch": 0.07497656982193064,
      "grad_norm": 2.8018271923065186,
      "learning_rate": 7.0370370370370375e-06,
      "loss": 0.7026,
      "step": 20
    },
    {
      "epoch": 0.07872539831302718,
      "grad_norm": 3.1438984870910645,
      "learning_rate": 7.4074074074074075e-06,
      "loss": 1.0012,
      "step": 21
    },
    {
      "epoch": 0.08247422680412371,
      "grad_norm": 2.5763065814971924,
      "learning_rate": 7.77777777777778e-06,
      "loss": 0.7979,
      "step": 22
    },
    {
      "epoch": 0.08622305529522024,
      "grad_norm": 2.6274800300598145,
      "learning_rate": 8.148148148148148e-06,
      "loss": 0.6811,
      "step": 23
    },
    {
      "epoch": 0.08997188378631678,
      "grad_norm": 2.6588563919067383,
      "learning_rate": 8.518518518518519e-06,
      "loss": 0.7003,
      "step": 24
    },
    {
      "epoch": 0.09372071227741331,
      "grad_norm": 2.8751745223999023,
      "learning_rate": 8.888888888888888e-06,
      "loss": 1.1499,
      "step": 25
    },
    {
      "epoch": 0.09746954076850985,
      "grad_norm": 2.9690346717834473,
      "learning_rate": 9.25925925925926e-06,
      "loss": 0.8019,
      "step": 26
    },
    {
      "epoch": 0.10121836925960637,
      "grad_norm": 2.8583548069000244,
      "learning_rate": 9.62962962962963e-06,
      "loss": 0.926,
      "step": 27
    },
    {
      "epoch": 0.1049671977507029,
      "grad_norm": 2.719120502471924,
      "learning_rate": 1e-05,
      "loss": 0.8262,
      "step": 28
    },
    {
      "epoch": 0.10871602624179943,
      "grad_norm": 2.5264711380004883,
      "learning_rate": 9.999904010783725e-06,
      "loss": 0.7492,
      "step": 29
    },
    {
      "epoch": 0.11246485473289597,
      "grad_norm": 2.5457921028137207,
      "learning_rate": 9.999616046820467e-06,
      "loss": 0.8453,
      "step": 30
    },
    {
      "epoch": 0.1162136832239925,
      "grad_norm": 2.66349720954895,
      "learning_rate": 9.999136119166803e-06,
      "loss": 0.6645,
      "step": 31
    },
    {
      "epoch": 0.11996251171508904,
      "grad_norm": 2.695261240005493,
      "learning_rate": 9.998464246249885e-06,
      "loss": 1.0498,
      "step": 32
    },
    {
      "epoch": 0.12371134020618557,
      "grad_norm": 2.371567487716675,
      "learning_rate": 9.997600453866734e-06,
      "loss": 0.9992,
      "step": 33
    },
    {
      "epoch": 0.1274601686972821,
      "grad_norm": 2.551347255706787,
      "learning_rate": 9.99654477518325e-06,
      "loss": 0.786,
      "step": 34
    },
    {
      "epoch": 0.13120899718837864,
      "grad_norm": 2.8569281101226807,
      "learning_rate": 9.995297250732942e-06,
      "loss": 0.7571,
      "step": 35
    },
    {
      "epoch": 0.13495782567947517,
      "grad_norm": 2.9101366996765137,
      "learning_rate": 9.99385792841537e-06,
      "loss": 0.8682,
      "step": 36
    },
    {
      "epoch": 0.1387066541705717,
      "grad_norm": 2.8088221549987793,
      "learning_rate": 9.9922268634943e-06,
      "loss": 0.9882,
      "step": 37
    },
    {
      "epoch": 0.14245548266166824,
      "grad_norm": 2.6386826038360596,
      "learning_rate": 9.99040411859559e-06,
      "loss": 0.8706,
      "step": 38
    },
    {
      "epoch": 0.14620431115276475,
      "grad_norm": 2.306828022003174,
      "learning_rate": 9.98838976370478e-06,
      "loss": 0.7294,
      "step": 39
    },
    {
      "epoch": 0.14995313964386128,
      "grad_norm": 3.0453381538391113,
      "learning_rate": 9.986183876164412e-06,
      "loss": 0.7128,
      "step": 40
    },
    {
      "epoch": 0.15370196813495782,
      "grad_norm": 2.72247314453125,
      "learning_rate": 9.983786540671052e-06,
      "loss": 0.8392,
      "step": 41
    },
    {
      "epoch": 0.15745079662605435,
      "grad_norm": 2.6883301734924316,
      "learning_rate": 9.981197849272039e-06,
      "loss": 0.6254,
      "step": 42
    },
    {
      "epoch": 0.16119962511715089,
      "grad_norm": 2.640571355819702,
      "learning_rate": 9.978417901361958e-06,
      "loss": 0.7416,
      "step": 43
    },
    {
      "epoch": 0.16494845360824742,
      "grad_norm": 2.496026039123535,
      "learning_rate": 9.975446803678818e-06,
      "loss": 0.9206,
      "step": 44
    },
    {
      "epoch": 0.16869728209934395,
      "grad_norm": 2.7780568599700928,
      "learning_rate": 9.972284670299957e-06,
      "loss": 0.8848,
      "step": 45
    },
    {
      "epoch": 0.1724461105904405,
      "grad_norm": 2.521110773086548,
      "learning_rate": 9.968931622637652e-06,
      "loss": 0.6183,
      "step": 46
    },
    {
      "epoch": 0.17619493908153702,
      "grad_norm": 2.378988742828369,
      "learning_rate": 9.965387789434474e-06,
      "loss": 0.7197,
      "step": 47
    },
    {
      "epoch": 0.17994376757263356,
      "grad_norm": 2.662712574005127,
      "learning_rate": 9.961653306758326e-06,
      "loss": 0.8191,
      "step": 48
    },
    {
      "epoch": 0.1836925960637301,
      "grad_norm": 2.818058729171753,
      "learning_rate": 9.95772831799724e-06,
      "loss": 0.945,
      "step": 49
    },
    {
      "epoch": 0.18744142455482662,
      "grad_norm": 2.4332478046417236,
      "learning_rate": 9.953612973853853e-06,
      "loss": 0.743,
      "step": 50
    },
    {
      "epoch": 0.19119025304592316,
      "grad_norm": 2.722165584564209,
      "learning_rate": 9.949307432339625e-06,
      "loss": 0.7276,
      "step": 51
    },
    {
      "epoch": 0.1949390815370197,
      "grad_norm": 3.4195361137390137,
      "learning_rate": 9.944811858768782e-06,
      "loss": 0.9687,
      "step": 52
    },
    {
      "epoch": 0.19868791002811623,
      "grad_norm": 3.273954391479492,
      "learning_rate": 9.940126425751957e-06,
      "loss": 0.8701,
      "step": 53
    },
    {
      "epoch": 0.20243673851921273,
      "grad_norm": 2.8895037174224854,
      "learning_rate": 9.935251313189564e-06,
      "loss": 1.1019,
      "step": 54
    },
    {
      "epoch": 0.20618556701030927,
      "grad_norm": 2.51041316986084,
      "learning_rate": 9.930186708264902e-06,
      "loss": 0.6458,
      "step": 55
    },
    {
      "epoch": 0.2099343955014058,
      "grad_norm": 2.807650566101074,
      "learning_rate": 9.92493280543695e-06,
      "loss": 0.7822,
      "step": 56
    },
    {
      "epoch": 0.21368322399250234,
      "grad_norm": 2.7724640369415283,
      "learning_rate": 9.919489806432915e-06,
      "loss": 0.9093,
      "step": 57
    },
    {
      "epoch": 0.21743205248359887,
      "grad_norm": 2.830974578857422,
      "learning_rate": 9.913857920240481e-06,
      "loss": 1.0478,
      "step": 58
    },
    {
      "epoch": 0.2211808809746954,
      "grad_norm": 2.7909905910491943,
      "learning_rate": 9.908037363099782e-06,
      "loss": 0.6055,
      "step": 59
    },
    {
      "epoch": 0.22492970946579194,
      "grad_norm": 2.5817978382110596,
      "learning_rate": 9.90202835849511e-06,
      "loss": 0.5351,
      "step": 60
    },
    {
      "epoch": 0.22867853795688847,
      "grad_norm": 2.8909380435943604,
      "learning_rate": 9.895831137146319e-06,
      "loss": 0.7823,
      "step": 61
    },
    {
      "epoch": 0.232427366447985,
      "grad_norm": 2.5689468383789062,
      "learning_rate": 9.889445936999978e-06,
      "loss": 0.7244,
      "step": 62
    },
    {
      "epoch": 0.23617619493908154,
      "grad_norm": 2.3992927074432373,
      "learning_rate": 9.882873003220229e-06,
      "loss": 0.6762,
      "step": 63
    },
    {
      "epoch": 0.23992502343017807,
      "grad_norm": 2.6709885597229004,
      "learning_rate": 9.876112588179378e-06,
      "loss": 1.0162,
      "step": 64
    },
    {
      "epoch": 0.2436738519212746,
      "grad_norm": 2.898042917251587,
      "learning_rate": 9.869164951448201e-06,
      "loss": 0.8213,
      "step": 65
    },
    {
      "epoch": 0.24742268041237114,
      "grad_norm": 2.4519455432891846,
      "learning_rate": 9.86203035978598e-06,
      "loss": 0.6992,
      "step": 66
    },
    {
      "epoch": 0.2511715089034677,
      "grad_norm": 2.6127779483795166,
      "learning_rate": 9.854709087130261e-06,
      "loss": 0.7047,
      "step": 67
    },
    {
      "epoch": 0.2549203373945642,
      "grad_norm": 2.591939926147461,
      "learning_rate": 9.847201414586331e-06,
      "loss": 0.8114,
      "step": 68
    },
    {
      "epoch": 0.25866916588566075,
      "grad_norm": 2.322962522506714,
      "learning_rate": 9.839507630416436e-06,
      "loss": 0.567,
      "step": 69
    },
    {
      "epoch": 0.2624179943767573,
      "grad_norm": 2.7526068687438965,
      "learning_rate": 9.831628030028698e-06,
      "loss": 0.8195,
      "step": 70
    },
    {
      "epoch": 0.2661668228678538,
      "grad_norm": 3.0277445316314697,
      "learning_rate": 9.82356291596578e-06,
      "loss": 1.1821,
      "step": 71
    },
    {
      "epoch": 0.26991565135895035,
      "grad_norm": 2.836676836013794,
      "learning_rate": 9.81531259789328e-06,
      "loss": 1.1377,
      "step": 72
    },
    {
      "epoch": 0.2736644798500469,
      "grad_norm": 2.7574238777160645,
      "learning_rate": 9.80687739258782e-06,
      "loss": 0.8165,
      "step": 73
    },
    {
      "epoch": 0.2774133083411434,
      "grad_norm": 2.917651414871216,
      "learning_rate": 9.7982576239249e-06,
      "loss": 0.8276,
      "step": 74
    },
    {
      "epoch": 0.28116213683223995,
      "grad_norm": 2.5589005947113037,
      "learning_rate": 9.789453622866455e-06,
      "loss": 0.7906,
      "step": 75
    },
    {
      "epoch": 0.2849109653233365,
      "grad_norm": 2.742344379425049,
      "learning_rate": 9.78046572744815e-06,
      "loss": 0.9195,
      "step": 76
    },
    {
      "epoch": 0.28865979381443296,
      "grad_norm": 2.717301845550537,
      "learning_rate": 9.771294282766399e-06,
      "loss": 0.9222,
      "step": 77
    },
    {
      "epoch": 0.2924086223055295,
      "grad_norm": 2.7094359397888184,
      "learning_rate": 9.761939640965117e-06,
      "loss": 0.8209,
      "step": 78
    },
    {
      "epoch": 0.29615745079662603,
      "grad_norm": 2.3744776248931885,
      "learning_rate": 9.7524021612222e-06,
      "loss": 0.8834,
      "step": 79
    },
    {
      "epoch": 0.29990627928772257,
      "grad_norm": 2.6806893348693848,
      "learning_rate": 9.742682209735727e-06,
      "loss": 0.6742,
      "step": 80
    },
    {
      "epoch": 0.3036551077788191,
      "grad_norm": 2.5327813625335693,
      "learning_rate": 9.732780159709912e-06,
      "loss": 0.8571,
      "step": 81
    },
    {
      "epoch": 0.30740393626991563,
      "grad_norm": 2.423889636993408,
      "learning_rate": 9.722696391340762e-06,
      "loss": 0.6847,
      "step": 82
    },
    {
      "epoch": 0.31115276476101217,
      "grad_norm": 3.3473219871520996,
      "learning_rate": 9.712431291801483e-06,
      "loss": 0.997,
      "step": 83
    },
    {
      "epoch": 0.3149015932521087,
      "grad_norm": 2.3630363941192627,
      "learning_rate": 9.701985255227624e-06,
      "loss": 0.7263,
      "step": 84
    },
    {
      "epoch": 0.31865042174320524,
      "grad_norm": 2.9431068897247314,
      "learning_rate": 9.691358682701927e-06,
      "loss": 0.9122,
      "step": 85
    },
    {
      "epoch": 0.32239925023430177,
      "grad_norm": 2.4724693298339844,
      "learning_rate": 9.680551982238941e-06,
      "loss": 0.85,
      "step": 86
    },
    {
      "epoch": 0.3261480787253983,
      "grad_norm": 2.641633987426758,
      "learning_rate": 9.669565568769348e-06,
      "loss": 0.7426,
      "step": 87
    },
    {
      "epoch": 0.32989690721649484,
      "grad_norm": 2.5549933910369873,
      "learning_rate": 9.658399864124037e-06,
      "loss": 0.7846,
      "step": 88
    },
    {
      "epoch": 0.3336457357075914,
      "grad_norm": 2.469627618789673,
      "learning_rate": 9.647055297017901e-06,
      "loss": 0.745,
      "step": 89
    },
    {
      "epoch": 0.3373945641986879,
      "grad_norm": 2.484022378921509,
      "learning_rate": 9.635532303033386e-06,
      "loss": 0.7989,
      "step": 90
    },
    {
      "epoch": 0.34114339268978444,
      "grad_norm": 2.852081060409546,
      "learning_rate": 9.623831324603755e-06,
      "loss": 0.6607,
      "step": 91
    },
    {
      "epoch": 0.344892221180881,
      "grad_norm": 2.902130126953125,
      "learning_rate": 9.611952810996104e-06,
      "loss": 0.8182,
      "step": 92
    },
    {
      "epoch": 0.3486410496719775,
      "grad_norm": 2.2227494716644287,
      "learning_rate": 9.599897218294122e-06,
      "loss": 0.6317,
      "step": 93
    },
    {
      "epoch": 0.35238987816307404,
      "grad_norm": 2.668668508529663,
      "learning_rate": 9.587665009380565e-06,
      "loss": 0.8303,
      "step": 94
    },
    {
      "epoch": 0.3561387066541706,
      "grad_norm": 2.8259904384613037,
      "learning_rate": 9.575256653919494e-06,
      "loss": 0.8235,
      "step": 95
    },
    {
      "epoch": 0.3598875351452671,
      "grad_norm": 2.3626370429992676,
      "learning_rate": 9.562672628338233e-06,
      "loss": 0.8566,
      "step": 96
    },
    {
      "epoch": 0.36363636363636365,
      "grad_norm": 2.487170934677124,
      "learning_rate": 9.549913415809084e-06,
      "loss": 0.8119,
      "step": 97
    },
    {
      "epoch": 0.3673851921274602,
      "grad_norm": 2.4403631687164307,
      "learning_rate": 9.536979506230772e-06,
      "loss": 0.6435,
      "step": 98
    },
    {
      "epoch": 0.3711340206185567,
      "grad_norm": 2.5403356552124023,
      "learning_rate": 9.523871396209633e-06,
      "loss": 0.8494,
      "step": 99
    },
    {
      "epoch": 0.37488284910965325,
      "grad_norm": 2.5626585483551025,
      "learning_rate": 9.510589589040554e-06,
      "loss": 0.8538,
      "step": 100
    },
    {
      "epoch": 0.3786316776007498,
      "grad_norm": 2.8327720165252686,
      "learning_rate": 9.497134594687635e-06,
      "loss": 0.8465,
      "step": 101
    },
    {
      "epoch": 0.3823805060918463,
      "grad_norm": 2.6077094078063965,
      "learning_rate": 9.483506929764623e-06,
      "loss": 0.9154,
      "step": 102
    },
    {
      "epoch": 0.38612933458294285,
      "grad_norm": 2.6204569339752197,
      "learning_rate": 9.469707117515068e-06,
      "loss": 0.8607,
      "step": 103
    },
    {
      "epoch": 0.3898781630740394,
      "grad_norm": 2.281536340713501,
      "learning_rate": 9.455735687792233e-06,
      "loss": 0.6667,
      "step": 104
    },
    {
      "epoch": 0.3936269915651359,
      "grad_norm": 2.339271068572998,
      "learning_rate": 9.44159317703876e-06,
      "loss": 0.7999,
      "step": 105
    },
    {
      "epoch": 0.39737582005623245,
      "grad_norm": 2.6674795150756836,
      "learning_rate": 9.427280128266049e-06,
      "loss": 0.8966,
      "step": 106
    },
    {
      "epoch": 0.40112464854732893,
      "grad_norm": 2.6626410484313965,
      "learning_rate": 9.412797091033444e-06,
      "loss": 0.9888,
      "step": 107
    },
    {
      "epoch": 0.40487347703842547,
      "grad_norm": 2.451646327972412,
      "learning_rate": 9.398144621427095e-06,
      "loss": 0.7585,
      "step": 108
    },
    {
      "epoch": 0.408622305529522,
      "grad_norm": 2.4723331928253174,
      "learning_rate": 9.383323282038632e-06,
      "loss": 0.7715,
      "step": 109
    },
    {
      "epoch": 0.41237113402061853,
      "grad_norm": 3.4575955867767334,
      "learning_rate": 9.368333641943558e-06,
      "loss": 0.6064,
      "step": 110
    },
    {
      "epoch": 0.41611996251171507,
      "grad_norm": 2.347303867340088,
      "learning_rate": 9.353176276679397e-06,
      "loss": 0.8154,
      "step": 111
    },
    {
      "epoch": 0.4198687910028116,
      "grad_norm": 2.850472927093506,
      "learning_rate": 9.337851768223589e-06,
      "loss": 0.8903,
      "step": 112
    },
    {
      "epoch": 0.42361761949390814,
      "grad_norm": 2.8391191959381104,
      "learning_rate": 9.322360704971161e-06,
      "loss": 0.8927,
      "step": 113
    },
    {
      "epoch": 0.42736644798500467,
      "grad_norm": 2.2651166915893555,
      "learning_rate": 9.30670368171212e-06,
      "loss": 0.5307,
      "step": 114
    },
    {
      "epoch": 0.4311152764761012,
      "grad_norm": 2.3141026496887207,
      "learning_rate": 9.29088129960862e-06,
      "loss": 0.7067,
      "step": 115
    },
    {
      "epoch": 0.43486410496719774,
      "grad_norm": 2.7184128761291504,
      "learning_rate": 9.274894166171888e-06,
      "loss": 1.0238,
      "step": 116
    },
    {
      "epoch": 0.4386129334582943,
      "grad_norm": 2.4955923557281494,
      "learning_rate": 9.258742895238886e-06,
      "loss": 0.7917,
      "step": 117
    },
    {
      "epoch": 0.4423617619493908,
      "grad_norm": 2.485121726989746,
      "learning_rate": 9.242428106948748e-06,
      "loss": 0.9189,
      "step": 118
    },
    {
      "epoch": 0.44611059044048734,
      "grad_norm": 2.644423723220825,
      "learning_rate": 9.225950427718974e-06,
      "loss": 0.8239,
      "step": 119
    },
    {
      "epoch": 0.4498594189315839,
      "grad_norm": 2.267098903656006,
      "learning_rate": 9.209310490221368e-06,
      "loss": 0.5885,
      "step": 120
    },
    {
      "epoch": 0.4536082474226804,
      "grad_norm": 3.184373140335083,
      "learning_rate": 9.192508933357753e-06,
      "loss": 0.9993,
      "step": 121
    },
    {
      "epoch": 0.45735707591377694,
      "grad_norm": 2.420222282409668,
      "learning_rate": 9.175546402235443e-06,
      "loss": 0.6963,
      "step": 122
    },
    {
      "epoch": 0.4611059044048735,
      "grad_norm": 2.1986594200134277,
      "learning_rate": 9.158423548142459e-06,
      "loss": 0.6484,
      "step": 123
    },
    {
      "epoch": 0.46485473289597,
      "grad_norm": 2.687732219696045,
      "learning_rate": 9.141141028522544e-06,
      "loss": 0.8614,
      "step": 124
    },
    {
      "epoch": 0.46860356138706655,
      "grad_norm": 2.5353922843933105,
      "learning_rate": 9.123699506949903e-06,
      "loss": 0.901,
      "step": 125
    },
    {
      "epoch": 0.4723523898781631,
      "grad_norm": 2.416156768798828,
      "learning_rate": 9.106099653103729e-06,
      "loss": 0.6771,
      "step": 126
    },
    {
      "epoch": 0.4761012183692596,
      "grad_norm": 2.7599902153015137,
      "learning_rate": 9.088342142742493e-06,
      "loss": 0.6258,
      "step": 127
    },
    {
      "epoch": 0.47985004686035615,
      "grad_norm": 2.6150290966033936,
      "learning_rate": 9.070427657677996e-06,
      "loss": 0.923,
      "step": 128
    },
    {
      "epoch": 0.4835988753514527,
      "grad_norm": 2.746690034866333,
      "learning_rate": 9.052356885749191e-06,
      "loss": 0.7873,
      "step": 129
    },
    {
      "epoch": 0.4873477038425492,
      "grad_norm": 2.5986545085906982,
      "learning_rate": 9.034130520795774e-06,
      "loss": 0.6669,
      "step": 130
    },
    {
      "epoch": 0.49109653233364575,
      "grad_norm": 2.674226760864258,
      "learning_rate": 9.015749262631537e-06,
      "loss": 0.8936,
      "step": 131
    },
    {
      "epoch": 0.4948453608247423,
      "grad_norm": 2.7541725635528564,
      "learning_rate": 8.997213817017508e-06,
      "loss": 0.9551,
      "step": 132
    },
    {
      "epoch": 0.4985941893158388,
      "grad_norm": 2.3283989429473877,
      "learning_rate": 8.978524895634842e-06,
      "loss": 0.8066,
      "step": 133
    },
    {
      "epoch": 0.5023430178069354,
      "grad_norm": 2.7233939170837402,
      "learning_rate": 8.959683216057512e-06,
      "loss": 0.8723,
      "step": 134
    },
    {
      "epoch": 0.5060918462980318,
      "grad_norm": 3.2135050296783447,
      "learning_rate": 8.940689501724737e-06,
      "loss": 0.9622,
      "step": 135
    },
    {
      "epoch": 0.5098406747891284,
      "grad_norm": 2.69746994972229,
      "learning_rate": 8.921544481913218e-06,
      "loss": 0.9274,
      "step": 136
    },
    {
      "epoch": 0.5135895032802249,
      "grad_norm": 2.718744993209839,
      "learning_rate": 8.902248891709133e-06,
      "loss": 0.7959,
      "step": 137
    },
    {
      "epoch": 0.5173383317713215,
      "grad_norm": 2.5741381645202637,
      "learning_rate": 8.882803471979917e-06,
      "loss": 0.8901,
      "step": 138
    },
    {
      "epoch": 0.521087160262418,
      "grad_norm": 2.678241491317749,
      "learning_rate": 8.86320896934581e-06,
      "loss": 0.8386,
      "step": 139
    },
    {
      "epoch": 0.5248359887535146,
      "grad_norm": 3.3864355087280273,
      "learning_rate": 8.843466136151191e-06,
      "loss": 0.8401,
      "step": 140
    },
    {
      "epoch": 0.528584817244611,
      "grad_norm": 2.4285075664520264,
      "learning_rate": 8.823575730435694e-06,
      "loss": 0.7686,
      "step": 141
    },
    {
      "epoch": 0.5323336457357076,
      "grad_norm": 2.551071882247925,
      "learning_rate": 8.803538515905102e-06,
      "loss": 0.7005,
      "step": 142
    },
    {
      "epoch": 0.5360824742268041,
      "grad_norm": 2.8349006175994873,
      "learning_rate": 8.783355261902023e-06,
      "loss": 0.7764,
      "step": 143
    },
    {
      "epoch": 0.5398313027179007,
      "grad_norm": 2.5966458320617676,
      "learning_rate": 8.763026743376349e-06,
      "loss": 0.6911,
      "step": 144
    },
    {
      "epoch": 0.5435801312089972,
      "grad_norm": 2.6098828315734863,
      "learning_rate": 8.742553740855507e-06,
      "loss": 0.8884,
      "step": 145
    },
    {
      "epoch": 0.5473289597000938,
      "grad_norm": 2.8855576515197754,
      "learning_rate": 8.721937040414481e-06,
      "loss": 0.7172,
      "step": 146
    },
    {
      "epoch": 0.5510777881911902,
      "grad_norm": 3.230717420578003,
      "learning_rate": 8.70117743364564e-06,
      "loss": 0.7861,
      "step": 147
    },
    {
      "epoch": 0.5548266166822868,
      "grad_norm": 2.7732186317443848,
      "learning_rate": 8.680275717628336e-06,
      "loss": 0.657,
      "step": 148
    },
    {
      "epoch": 0.5585754451733833,
      "grad_norm": 2.7117972373962402,
      "learning_rate": 8.659232694898307e-06,
      "loss": 0.8825,
      "step": 149
    },
    {
      "epoch": 0.5623242736644799,
      "grad_norm": 2.5794482231140137,
      "learning_rate": 8.638049173416855e-06,
      "loss": 0.8581,
      "step": 150
    },
    {
      "epoch": 0.5660731021555764,
      "grad_norm": 2.9475417137145996,
      "learning_rate": 8.616725966539831e-06,
      "loss": 0.908,
      "step": 151
    },
    {
      "epoch": 0.569821930646673,
      "grad_norm": 2.9310085773468018,
      "learning_rate": 8.595263892986403e-06,
      "loss": 0.986,
      "step": 152
    },
    {
      "epoch": 0.5735707591377694,
      "grad_norm": 2.635514974594116,
      "learning_rate": 8.573663776807615e-06,
      "loss": 0.6768,
      "step": 153
    },
    {
      "epoch": 0.5773195876288659,
      "grad_norm": 2.826488733291626,
      "learning_rate": 8.551926447354759e-06,
      "loss": 1.0033,
      "step": 154
    },
    {
      "epoch": 0.5810684161199625,
      "grad_norm": 2.487631320953369,
      "learning_rate": 8.530052739247522e-06,
      "loss": 0.7031,
      "step": 155
    },
    {
      "epoch": 0.584817244611059,
      "grad_norm": 2.5516085624694824,
      "learning_rate": 8.508043492341944e-06,
      "loss": 0.7245,
      "step": 156
    },
    {
      "epoch": 0.5885660731021556,
      "grad_norm": 2.3507580757141113,
      "learning_rate": 8.485899551698166e-06,
      "loss": 0.8073,
      "step": 157
    },
    {
      "epoch": 0.5923149015932521,
      "grad_norm": 2.7757110595703125,
      "learning_rate": 8.463621767547998e-06,
      "loss": 0.8046,
      "step": 158
    },
    {
      "epoch": 0.5960637300843487,
      "grad_norm": 2.639394521713257,
      "learning_rate": 8.44121099526225e-06,
      "loss": 0.7603,
      "step": 159
    },
    {
      "epoch": 0.5998125585754451,
      "grad_norm": 2.907921314239502,
      "learning_rate": 8.418668095317912e-06,
      "loss": 0.9165,
      "step": 160
    },
    {
      "epoch": 0.6035613870665417,
      "grad_norm": 2.7137627601623535,
      "learning_rate": 8.395993933265102e-06,
      "loss": 0.8256,
      "step": 161
    },
    {
      "epoch": 0.6073102155576382,
      "grad_norm": 2.4688782691955566,
      "learning_rate": 8.373189379693838e-06,
      "loss": 0.9909,
      "step": 162
    },
    {
      "epoch": 0.6110590440487348,
      "grad_norm": 2.68072772026062,
      "learning_rate": 8.350255310200611e-06,
      "loss": 0.9874,
      "step": 163
    },
    {
      "epoch": 0.6148078725398313,
      "grad_norm": 3.0509040355682373,
      "learning_rate": 8.327192605354766e-06,
      "loss": 0.9298,
      "step": 164
    },
    {
      "epoch": 0.6185567010309279,
      "grad_norm": 2.4764254093170166,
      "learning_rate": 8.304002150664684e-06,
      "loss": 0.7797,
      "step": 165
    },
    {
      "epoch": 0.6223055295220243,
      "grad_norm": 2.992629051208496,
      "learning_rate": 8.280684836543794e-06,
      "loss": 0.8159,
      "step": 166
    },
    {
      "epoch": 0.6260543580131209,
      "grad_norm": 2.652869462966919,
      "learning_rate": 8.257241558276381e-06,
      "loss": 0.597,
      "step": 167
    },
    {
      "epoch": 0.6298031865042174,
      "grad_norm": 2.40967059135437,
      "learning_rate": 8.233673215983207e-06,
      "loss": 0.7821,
      "step": 168
    },
    {
      "epoch": 0.633552014995314,
      "grad_norm": 2.9071333408355713,
      "learning_rate": 8.209980714586955e-06,
      "loss": 0.9107,
      "step": 169
    },
    {
      "epoch": 0.6373008434864105,
      "grad_norm": 2.4595487117767334,
      "learning_rate": 8.18616496377748e-06,
      "loss": 0.8695,
      "step": 170
    },
    {
      "epoch": 0.6410496719775071,
      "grad_norm": 2.700758934020996,
      "learning_rate": 8.162226877976886e-06,
      "loss": 0.8954,
      "step": 171
    },
    {
      "epoch": 0.6447985004686035,
      "grad_norm": 3.0076358318328857,
      "learning_rate": 8.138167376304411e-06,
      "loss": 0.7885,
      "step": 172
    },
    {
      "epoch": 0.6485473289597001,
      "grad_norm": 2.321608543395996,
      "learning_rate": 8.113987382541138e-06,
      "loss": 0.6624,
      "step": 173
    },
    {
      "epoch": 0.6522961574507966,
      "grad_norm": 2.736940383911133,
      "learning_rate": 8.089687825094524e-06,
      "loss": 0.9155,
      "step": 174
    },
    {
      "epoch": 0.6560449859418932,
      "grad_norm": 2.9796676635742188,
      "learning_rate": 8.065269636962765e-06,
      "loss": 0.8578,
      "step": 175
    },
    {
      "epoch": 0.6597938144329897,
      "grad_norm": 2.3743815422058105,
      "learning_rate": 8.040733755698954e-06,
      "loss": 0.831,
      "step": 176
    },
    {
      "epoch": 0.6635426429240863,
      "grad_norm": 2.536311388015747,
      "learning_rate": 8.016081123375098e-06,
      "loss": 0.6518,
      "step": 177
    },
    {
      "epoch": 0.6672914714151827,
      "grad_norm": 2.936393976211548,
      "learning_rate": 7.991312686545939e-06,
      "loss": 0.8743,
      "step": 178
    },
    {
      "epoch": 0.6710402999062793,
      "grad_norm": 2.9552223682403564,
      "learning_rate": 7.96642939621261e-06,
      "loss": 1.0273,
      "step": 179
    },
    {
      "epoch": 0.6747891283973758,
      "grad_norm": 2.538328170776367,
      "learning_rate": 7.94143220778613e-06,
      "loss": 0.786,
      "step": 180
    },
    {
      "epoch": 0.6785379568884724,
      "grad_norm": 2.406303882598877,
      "learning_rate": 7.916322081050708e-06,
      "loss": 0.8003,
      "step": 181
    },
    {
      "epoch": 0.6822867853795689,
      "grad_norm": 2.377192735671997,
      "learning_rate": 7.8910999801269e-06,
      "loss": 0.7355,
      "step": 182
    },
    {
      "epoch": 0.6860356138706654,
      "grad_norm": 3.0302469730377197,
      "learning_rate": 7.865766873434582e-06,
      "loss": 1.0152,
      "step": 183
    },
    {
      "epoch": 0.689784442361762,
      "grad_norm": 2.5823426246643066,
      "learning_rate": 7.84032373365578e-06,
      "loss": 0.8919,
      "step": 184
    },
    {
      "epoch": 0.6935332708528584,
      "grad_norm": 2.891167640686035,
      "learning_rate": 7.814771537697312e-06,
      "loss": 0.8386,
      "step": 185
    },
    {
      "epoch": 0.697282099343955,
      "grad_norm": 2.61259126663208,
      "learning_rate": 7.789111266653285e-06,
      "loss": 0.8342,
      "step": 186
    },
    {
      "epoch": 0.7010309278350515,
      "grad_norm": 3.1016290187835693,
      "learning_rate": 7.76334390576742e-06,
      "loss": 0.8578,
      "step": 187
    },
    {
      "epoch": 0.7047797563261481,
      "grad_norm": 2.2901554107666016,
      "learning_rate": 7.737470444395227e-06,
      "loss": 0.6153,
      "step": 188
    },
    {
      "epoch": 0.7085285848172446,
      "grad_norm": 3.0746381282806396,
      "learning_rate": 7.71149187596602e-06,
      "loss": 0.7487,
      "step": 189
    },
    {
      "epoch": 0.7122774133083412,
      "grad_norm": 2.534573554992676,
      "learning_rate": 7.685409197944768e-06,
      "loss": 0.9354,
      "step": 190
    },
    {
      "epoch": 0.7160262417994376,
      "grad_norm": 2.7725582122802734,
      "learning_rate": 7.6592234117938e-06,
      "loss": 0.8641,
      "step": 191
    },
    {
      "epoch": 0.7197750702905342,
      "grad_norm": 2.284388780593872,
      "learning_rate": 7.63293552293435e-06,
      "loss": 0.8296,
      "step": 192
    },
    {
      "epoch": 0.7235238987816307,
      "grad_norm": 2.665536880493164,
      "learning_rate": 7.60654654070796e-06,
      "loss": 0.7976,
      "step": 193
    },
    {
      "epoch": 0.7272727272727273,
      "grad_norm": 2.496126651763916,
      "learning_rate": 7.580057478337717e-06,
      "loss": 0.8097,
      "step": 194
    },
    {
      "epoch": 0.7310215557638238,
      "grad_norm": 2.898500919342041,
      "learning_rate": 7.553469352889356e-06,
      "loss": 0.832,
      "step": 195
    },
    {
      "epoch": 0.7347703842549204,
      "grad_norm": 2.545947551727295,
      "learning_rate": 7.526783185232208e-06,
      "loss": 0.8465,
      "step": 196
    },
    {
      "epoch": 0.7385192127460168,
      "grad_norm": 2.320594549179077,
      "learning_rate": 7.500000000000001e-06,
      "loss": 0.6391,
      "step": 197
    },
    {
      "epoch": 0.7422680412371134,
      "grad_norm": 2.390089511871338,
      "learning_rate": 7.473120825551517e-06,
      "loss": 0.8494,
      "step": 198
    },
    {
      "epoch": 0.7460168697282099,
      "grad_norm": 3.2954750061035156,
      "learning_rate": 7.446146693931111e-06,
      "loss": 0.9194,
      "step": 199
    },
    {
      "epoch": 0.7497656982193065,
      "grad_norm": 3.3448970317840576,
      "learning_rate": 7.419078640829088e-06,
      "loss": 0.8001,
      "step": 200
    },
    {
      "epoch": 0.753514526710403,
      "grad_norm": 2.490509033203125,
      "learning_rate": 7.391917705541927e-06,
      "loss": 0.8049,
      "step": 201
    },
    {
      "epoch": 0.7572633552014996,
      "grad_norm": 3.141767740249634,
      "learning_rate": 7.364664930932385e-06,
      "loss": 0.6791,
      "step": 202
    },
    {
      "epoch": 0.761012183692596,
      "grad_norm": 2.646108388900757,
      "learning_rate": 7.337321363389453e-06,
      "loss": 0.6114,
      "step": 203
    },
    {
      "epoch": 0.7647610121836926,
      "grad_norm": 2.71254563331604,
      "learning_rate": 7.3098880527881755e-06,
      "loss": 0.8919,
      "step": 204
    },
    {
      "epoch": 0.7685098406747891,
      "grad_norm": 2.701137065887451,
      "learning_rate": 7.282366052449351e-06,
      "loss": 0.9196,
      "step": 205
    },
    {
      "epoch": 0.7722586691658857,
      "grad_norm": 2.5790905952453613,
      "learning_rate": 7.254756419099074e-06,
      "loss": 0.8328,
      "step": 206
    },
    {
      "epoch": 0.7760074976569822,
      "grad_norm": 2.885322332382202,
      "learning_rate": 7.227060212828171e-06,
      "loss": 0.7947,
      "step": 207
    },
    {
      "epoch": 0.7797563261480788,
      "grad_norm": 2.563446044921875,
      "learning_rate": 7.199278497051498e-06,
      "loss": 0.8801,
      "step": 208
    },
    {
      "epoch": 0.7835051546391752,
      "grad_norm": 2.7891945838928223,
      "learning_rate": 7.171412338467101e-06,
      "loss": 0.7966,
      "step": 209
    },
    {
      "epoch": 0.7872539831302718,
      "grad_norm": 2.888983726501465,
      "learning_rate": 7.143462807015271e-06,
      "loss": 0.7454,
      "step": 210
    },
    {
      "epoch": 0.7910028116213683,
      "grad_norm": 2.7022621631622314,
      "learning_rate": 7.115430975837457e-06,
      "loss": 0.9606,
      "step": 211
    },
    {
      "epoch": 0.7947516401124649,
      "grad_norm": 2.689760208129883,
      "learning_rate": 7.087317921235059e-06,
      "loss": 0.819,
      "step": 212
    },
    {
      "epoch": 0.7985004686035614,
      "grad_norm": 2.960847854614258,
      "learning_rate": 7.059124722628113e-06,
      "loss": 0.9319,
      "step": 213
    },
    {
      "epoch": 0.8022492970946579,
      "grad_norm": 3.116780996322632,
      "learning_rate": 7.030852462513827e-06,
      "loss": 1.0187,
      "step": 214
    },
    {
      "epoch": 0.8059981255857545,
      "grad_norm": 2.5313186645507812,
      "learning_rate": 7.002502226425042e-06,
      "loss": 0.9004,
      "step": 215
    },
    {
      "epoch": 0.8097469540768509,
      "grad_norm": 2.4676010608673096,
      "learning_rate": 6.974075102888535e-06,
      "loss": 0.805,
      "step": 216
    },
    {
      "epoch": 0.8134957825679475,
      "grad_norm": 2.8545048236846924,
      "learning_rate": 6.945572183383229e-06,
      "loss": 0.8092,
      "step": 217
    },
    {
      "epoch": 0.817244611059044,
      "grad_norm": 2.710291624069214,
      "learning_rate": 6.916994562298286e-06,
      "loss": 0.8639,
      "step": 218
    },
    {
      "epoch": 0.8209934395501406,
      "grad_norm": 2.850886344909668,
      "learning_rate": 6.888343336891088e-06,
      "loss": 0.8063,
      "step": 219
    },
    {
      "epoch": 0.8247422680412371,
      "grad_norm": 2.6347622871398926,
      "learning_rate": 6.859619607245102e-06,
      "loss": 0.9418,
      "step": 220
    },
    {
      "epoch": 0.8284910965323337,
      "grad_norm": 2.4911224842071533,
      "learning_rate": 6.830824476227646e-06,
      "loss": 0.633,
      "step": 221
    },
    {
      "epoch": 0.8322399250234301,
      "grad_norm": 2.8799281120300293,
      "learning_rate": 6.801959049447546e-06,
      "loss": 0.7747,
      "step": 222
    },
    {
      "epoch": 0.8359887535145267,
      "grad_norm": 2.7227516174316406,
      "learning_rate": 6.773024435212678e-06,
      "loss": 1.0301,
      "step": 223
    },
    {
      "epoch": 0.8397375820056232,
      "grad_norm": 2.352261781692505,
      "learning_rate": 6.744021744487422e-06,
      "loss": 0.7293,
      "step": 224
    },
    {
      "epoch": 0.8434864104967198,
      "grad_norm": 2.8626327514648438,
      "learning_rate": 6.714952090849996e-06,
      "loss": 1.022,
      "step": 225
    },
    {
      "epoch": 0.8472352389878163,
      "grad_norm": 2.3415591716766357,
      "learning_rate": 6.685816590449708e-06,
      "loss": 0.7533,
      "step": 226
    },
    {
      "epoch": 0.8509840674789129,
      "grad_norm": 2.774419069290161,
      "learning_rate": 6.6566163619641e-06,
      "loss": 0.7825,
      "step": 227
    },
    {
      "epoch": 0.8547328959700093,
      "grad_norm": 2.38519287109375,
      "learning_rate": 6.62735252655599e-06,
      "loss": 0.9026,
      "step": 228
    },
    {
      "epoch": 0.8584817244611059,
      "grad_norm": 2.562180280685425,
      "learning_rate": 6.598026207830428e-06,
      "loss": 0.8274,
      "step": 229
    },
    {
      "epoch": 0.8622305529522024,
      "grad_norm": 2.506265640258789,
      "learning_rate": 6.568638531791555e-06,
      "loss": 0.8533,
      "step": 230
    },
    {
      "epoch": 0.865979381443299,
      "grad_norm": 2.4109628200531006,
      "learning_rate": 6.539190626799366e-06,
      "loss": 0.718,
      "step": 231
    },
    {
      "epoch": 0.8697282099343955,
      "grad_norm": 2.856254816055298,
      "learning_rate": 6.5096836235263904e-06,
      "loss": 0.8667,
      "step": 232
    },
    {
      "epoch": 0.8734770384254921,
      "grad_norm": 2.8561127185821533,
      "learning_rate": 6.480118654914276e-06,
      "loss": 0.5914,
      "step": 233
    },
    {
      "epoch": 0.8772258669165885,
      "grad_norm": 2.726389169692993,
      "learning_rate": 6.4504968561302905e-06,
      "loss": 0.9917,
      "step": 234
    },
    {
      "epoch": 0.8809746954076851,
      "grad_norm": 2.7317206859588623,
      "learning_rate": 6.4208193645237314e-06,
      "loss": 0.9438,
      "step": 235
    },
    {
      "epoch": 0.8847235238987816,
      "grad_norm": 2.6856586933135986,
      "learning_rate": 6.391087319582264e-06,
      "loss": 0.8312,
      "step": 236
    },
    {
      "epoch": 0.8884723523898782,
      "grad_norm": 2.7664246559143066,
      "learning_rate": 6.3613018628881655e-06,
      "loss": 0.8756,
      "step": 237
    },
    {
      "epoch": 0.8922211808809747,
      "grad_norm": 2.266479730606079,
      "learning_rate": 6.331464138074493e-06,
      "loss": 0.7957,
      "step": 238
    },
    {
      "epoch": 0.8959700093720713,
      "grad_norm": 2.8850481510162354,
      "learning_rate": 6.301575290781174e-06,
      "loss": 0.9892,
      "step": 239
    },
    {
      "epoch": 0.8997188378631678,
      "grad_norm": 3.300827980041504,
      "learning_rate": 6.271636468611022e-06,
      "loss": 0.8914,
      "step": 240
    },
    {
      "epoch": 0.9034676663542643,
      "grad_norm": 2.5576953887939453,
      "learning_rate": 6.241648821085666e-06,
      "loss": 0.9061,
      "step": 241
    },
    {
      "epoch": 0.9072164948453608,
      "grad_norm": 2.373811721801758,
      "learning_rate": 6.211613499601419e-06,
      "loss": 0.8346,
      "step": 242
    },
    {
      "epoch": 0.9109653233364574,
      "grad_norm": 2.5030999183654785,
      "learning_rate": 6.181531657385068e-06,
      "loss": 0.8327,
      "step": 243
    },
    {
      "epoch": 0.9147141518275539,
      "grad_norm": 3.0258920192718506,
      "learning_rate": 6.1514044494496e-06,
      "loss": 0.7557,
      "step": 244
    },
    {
      "epoch": 0.9184629803186504,
      "grad_norm": 2.5062878131866455,
      "learning_rate": 6.1212330325498425e-06,
      "loss": 0.8887,
      "step": 245
    },
    {
      "epoch": 0.922211808809747,
      "grad_norm": 2.974062442779541,
      "learning_rate": 6.091018565138062e-06,
      "loss": 0.8752,
      "step": 246
    },
    {
      "epoch": 0.9259606373008434,
      "grad_norm": 2.3620338439941406,
      "learning_rate": 6.060762207319479e-06,
      "loss": 0.5664,
      "step": 247
    },
    {
      "epoch": 0.92970946579194,
      "grad_norm": 2.6449522972106934,
      "learning_rate": 6.03046512080772e-06,
      "loss": 0.7479,
      "step": 248
    },
    {
      "epoch": 0.9334582942830365,
      "grad_norm": 2.6556012630462646,
      "learning_rate": 6.000128468880223e-06,
      "loss": 0.9128,
      "step": 249
    },
    {
      "epoch": 0.9372071227741331,
      "grad_norm": 2.5893819332122803,
      "learning_rate": 5.9697534163335645e-06,
      "loss": 0.7899,
      "step": 250
    },
    {
      "epoch": 0.9409559512652296,
      "grad_norm": 2.8196489810943604,
      "learning_rate": 5.939341129438739e-06,
      "loss": 0.9707,
      "step": 251
    },
    {
      "epoch": 0.9447047797563262,
      "grad_norm": 2.6079866886138916,
      "learning_rate": 5.908892775896383e-06,
      "loss": 0.6921,
      "step": 252
    },
    {
      "epoch": 0.9484536082474226,
      "grad_norm": 2.749600648880005,
      "learning_rate": 5.878409524791931e-06,
      "loss": 0.8562,
      "step": 253
    },
    {
      "epoch": 0.9522024367385192,
      "grad_norm": 2.7583858966827393,
      "learning_rate": 5.847892546550738e-06,
      "loss": 0.7389,
      "step": 254
    },
    {
      "epoch": 0.9559512652296157,
      "grad_norm": 2.819322347640991,
      "learning_rate": 5.817343012893132e-06,
      "loss": 0.9122,
      "step": 255
    },
    {
      "epoch": 0.9597000937207123,
      "grad_norm": 2.7966537475585938,
      "learning_rate": 5.786762096789431e-06,
      "loss": 1.0095,
      "step": 256
    },
    {
      "epoch": 0.9634489222118088,
      "grad_norm": 2.982713460922241,
      "learning_rate": 5.756150972414904e-06,
      "loss": 1.0745,
      "step": 257
    },
    {
      "epoch": 0.9671977507029054,
      "grad_norm": 2.7742886543273926,
      "learning_rate": 5.725510815104685e-06,
      "loss": 0.7517,
      "step": 258
    },
    {
      "epoch": 0.9709465791940018,
      "grad_norm": 2.983306646347046,
      "learning_rate": 5.694842801308651e-06,
      "loss": 0.8668,
      "step": 259
    },
    {
      "epoch": 0.9746954076850984,
      "grad_norm": 2.550466299057007,
      "learning_rate": 5.664148108546242e-06,
      "loss": 0.7585,
      "step": 260
    },
    {
      "epoch": 0.9784442361761949,
      "grad_norm": 2.5021555423736572,
      "learning_rate": 5.633427915361261e-06,
      "loss": 0.7487,
      "step": 261
    },
    {
      "epoch": 0.9821930646672915,
      "grad_norm": 2.9954419136047363,
      "learning_rate": 5.6026834012766155e-06,
      "loss": 0.7231,
      "step": 262
    },
    {
      "epoch": 0.985941893158388,
      "grad_norm": 2.553372859954834,
      "learning_rate": 5.5719157467490305e-06,
      "loss": 0.7305,
      "step": 263
    },
    {
      "epoch": 0.9896907216494846,
      "grad_norm": 2.377293825149536,
      "learning_rate": 5.541126133123721e-06,
      "loss": 0.9153,
      "step": 264
    },
    {
      "epoch": 0.993439550140581,
      "grad_norm": 2.6041624546051025,
      "learning_rate": 5.510315742589042e-06,
      "loss": 0.7345,
      "step": 265
    },
    {
      "epoch": 0.9971883786316776,
      "grad_norm": 2.1862921714782715,
      "learning_rate": 5.479485758131089e-06,
      "loss": 0.8135,
      "step": 266
    },
    {
      "epoch": 1.0,
      "grad_norm": 3.3813092708587646,
      "learning_rate": 5.4486373634882805e-06,
      "loss": 0.7056,
      "step": 267
    },
    {
      "epoch": 1.0037488284910965,
      "grad_norm": 2.4991369247436523,
      "learning_rate": 5.417771743105908e-06,
      "loss": 0.5993,
      "step": 268
    },
    {
      "epoch": 1.007497656982193,
      "grad_norm": 2.8168036937713623,
      "learning_rate": 5.386890082090652e-06,
      "loss": 0.6456,
      "step": 269
    },
    {
      "epoch": 1.0112464854732897,
      "grad_norm": 2.7016103267669678,
      "learning_rate": 5.355993566165091e-06,
      "loss": 0.7714,
      "step": 270
    },
    {
      "epoch": 1.0149953139643861,
      "grad_norm": 2.449699878692627,
      "learning_rate": 5.325083381622165e-06,
      "loss": 0.7332,
      "step": 271
    },
    {
      "epoch": 1.0187441424554826,
      "grad_norm": 2.5395665168762207,
      "learning_rate": 5.294160715279626e-06,
      "loss": 0.7395,
      "step": 272
    },
    {
      "epoch": 1.022492970946579,
      "grad_norm": 2.3110768795013428,
      "learning_rate": 5.263226754434481e-06,
      "loss": 0.6565,
      "step": 273
    },
    {
      "epoch": 1.0262417994376758,
      "grad_norm": 2.3739070892333984,
      "learning_rate": 5.232282686817392e-06,
      "loss": 0.7151,
      "step": 274
    },
    {
      "epoch": 1.0299906279287723,
      "grad_norm": 2.500891923904419,
      "learning_rate": 5.201329700547077e-06,
      "loss": 0.6505,
      "step": 275
    },
    {
      "epoch": 1.0337394564198688,
      "grad_norm": 2.575612783432007,
      "learning_rate": 5.170368984084695e-06,
      "loss": 0.737,
      "step": 276
    },
    {
      "epoch": 1.0374882849109652,
      "grad_norm": 2.4420552253723145,
      "learning_rate": 5.139401726188208e-06,
      "loss": 0.5698,
      "step": 277
    },
    {
      "epoch": 1.041237113402062,
      "grad_norm": 2.4866912364959717,
      "learning_rate": 5.108429115866744e-06,
      "loss": 0.5371,
      "step": 278
    },
    {
      "epoch": 1.0449859418931584,
      "grad_norm": 2.7172107696533203,
      "learning_rate": 5.077452342334939e-06,
      "loss": 0.7068,
      "step": 279
    },
    {
      "epoch": 1.0487347703842549,
      "grad_norm": 2.302205801010132,
      "learning_rate": 5.046472594967279e-06,
      "loss": 0.6946,
      "step": 280
    },
    {
      "epoch": 1.0524835988753514,
      "grad_norm": 2.5750224590301514,
      "learning_rate": 5.01549106325243e-06,
      "loss": 0.7505,
      "step": 281
    },
    {
      "epoch": 1.056232427366448,
      "grad_norm": 2.5373618602752686,
      "learning_rate": 4.9845089367475715e-06,
      "loss": 0.7652,
      "step": 282
    },
    {
      "epoch": 1.0599812558575445,
      "grad_norm": 2.721522092819214,
      "learning_rate": 4.953527405032723e-06,
      "loss": 0.7254,
      "step": 283
    },
    {
      "epoch": 1.063730084348641,
      "grad_norm": 2.75787091255188,
      "learning_rate": 4.922547657665062e-06,
      "loss": 0.8159,
      "step": 284
    },
    {
      "epoch": 1.0674789128397375,
      "grad_norm": 2.2756996154785156,
      "learning_rate": 4.891570884133256e-06,
      "loss": 0.5339,
      "step": 285
    },
    {
      "epoch": 1.0712277413308342,
      "grad_norm": 2.541193962097168,
      "learning_rate": 4.860598273811793e-06,
      "loss": 0.6664,
      "step": 286
    },
    {
      "epoch": 1.0749765698219307,
      "grad_norm": 2.9506993293762207,
      "learning_rate": 4.829631015915306e-06,
      "loss": 0.6293,
      "step": 287
    },
    {
      "epoch": 1.0787253983130272,
      "grad_norm": 2.649301290512085,
      "learning_rate": 4.798670299452926e-06,
      "loss": 0.5073,
      "step": 288
    },
    {
      "epoch": 1.0824742268041236,
      "grad_norm": 2.1914241313934326,
      "learning_rate": 4.767717313182611e-06,
      "loss": 0.6079,
      "step": 289
    },
    {
      "epoch": 1.0862230552952203,
      "grad_norm": 2.44549298286438,
      "learning_rate": 4.736773245565521e-06,
      "loss": 0.5781,
      "step": 290
    },
    {
      "epoch": 1.0899718837863168,
      "grad_norm": 2.684919595718384,
      "learning_rate": 4.705839284720376e-06,
      "loss": 0.5453,
      "step": 291
    },
    {
      "epoch": 1.0937207122774133,
      "grad_norm": 2.83229660987854,
      "learning_rate": 4.6749166183778375e-06,
      "loss": 0.831,
      "step": 292
    },
    {
      "epoch": 1.0974695407685098,
      "grad_norm": 2.6568431854248047,
      "learning_rate": 4.64400643383491e-06,
      "loss": 0.587,
      "step": 293
    },
    {
      "epoch": 1.1012183692596063,
      "grad_norm": 2.890794277191162,
      "learning_rate": 4.613109917909349e-06,
      "loss": 0.7931,
      "step": 294
    },
    {
      "epoch": 1.104967197750703,
      "grad_norm": 2.916185140609741,
      "learning_rate": 4.582228256894093e-06,
      "loss": 0.7106,
      "step": 295
    },
    {
      "epoch": 1.1087160262417994,
      "grad_norm": 2.470686912536621,
      "learning_rate": 4.55136263651172e-06,
      "loss": 0.7789,
      "step": 296
    },
    {
      "epoch": 1.112464854732896,
      "grad_norm": 2.4361371994018555,
      "learning_rate": 4.520514241868912e-06,
      "loss": 0.6496,
      "step": 297
    },
    {
      "epoch": 1.1162136832239926,
      "grad_norm": 2.7543270587921143,
      "learning_rate": 4.489684257410959e-06,
      "loss": 0.6137,
      "step": 298
    },
    {
      "epoch": 1.119962511715089,
      "grad_norm": 2.6973302364349365,
      "learning_rate": 4.458873866876282e-06,
      "loss": 0.5914,
      "step": 299
    },
    {
      "epoch": 1.1237113402061856,
      "grad_norm": 2.2894020080566406,
      "learning_rate": 4.428084253250972e-06,
      "loss": 0.66,
      "step": 300
    },
    {
      "epoch": 1.127460168697282,
      "grad_norm": 2.703667402267456,
      "learning_rate": 4.397316598723385e-06,
      "loss": 0.6326,
      "step": 301
    },
    {
      "epoch": 1.1312089971883785,
      "grad_norm": 2.5640034675598145,
      "learning_rate": 4.3665720846387406e-06,
      "loss": 0.6076,
      "step": 302
    },
    {
      "epoch": 1.1349578256794752,
      "grad_norm": 2.6277530193328857,
      "learning_rate": 4.335851891453759e-06,
      "loss": 0.6344,
      "step": 303
    },
    {
      "epoch": 1.1387066541705717,
      "grad_norm": 2.1854584217071533,
      "learning_rate": 4.305157198691351e-06,
      "loss": 0.537,
      "step": 304
    },
    {
      "epoch": 1.1424554826616682,
      "grad_norm": 2.584956407546997,
      "learning_rate": 4.2744891848953156e-06,
      "loss": 0.6874,
      "step": 305
    },
    {
      "epoch": 1.1462043111527647,
      "grad_norm": 3.1633033752441406,
      "learning_rate": 4.2438490275850965e-06,
      "loss": 0.776,
      "step": 306
    },
    {
      "epoch": 1.1499531396438614,
      "grad_norm": 2.372377634048462,
      "learning_rate": 4.2132379032105695e-06,
      "loss": 0.6193,
      "step": 307
    },
    {
      "epoch": 1.1537019681349578,
      "grad_norm": 2.6179869174957275,
      "learning_rate": 4.182656987106869e-06,
      "loss": 0.8479,
      "step": 308
    },
    {
      "epoch": 1.1574507966260543,
      "grad_norm": 2.4828739166259766,
      "learning_rate": 4.152107453449263e-06,
      "loss": 0.5892,
      "step": 309
    },
    {
      "epoch": 1.1611996251171508,
      "grad_norm": 2.4343421459198,
      "learning_rate": 4.121590475208071e-06,
      "loss": 0.5757,
      "step": 310
    },
    {
      "epoch": 1.1649484536082475,
      "grad_norm": 2.660548210144043,
      "learning_rate": 4.091107224103619e-06,
      "loss": 0.7259,
      "step": 311
    },
    {
      "epoch": 1.168697282099344,
      "grad_norm": 2.5796167850494385,
      "learning_rate": 4.060658870561263e-06,
      "loss": 0.5966,
      "step": 312
    },
    {
      "epoch": 1.1724461105904405,
      "grad_norm": 2.7497189044952393,
      "learning_rate": 4.030246583666437e-06,
      "loss": 0.6979,
      "step": 313
    },
    {
      "epoch": 1.176194939081537,
      "grad_norm": 2.582590103149414,
      "learning_rate": 3.999871531119779e-06,
      "loss": 0.722,
      "step": 314
    },
    {
      "epoch": 1.1799437675726336,
      "grad_norm": 2.6821389198303223,
      "learning_rate": 3.969534879192281e-06,
      "loss": 0.7015,
      "step": 315
    },
    {
      "epoch": 1.1836925960637301,
      "grad_norm": 2.4163970947265625,
      "learning_rate": 3.9392377926805226e-06,
      "loss": 0.4959,
      "step": 316
    },
    {
      "epoch": 1.1874414245548266,
      "grad_norm": 2.1838912963867188,
      "learning_rate": 3.9089814348619386e-06,
      "loss": 0.6188,
      "step": 317
    },
    {
      "epoch": 1.191190253045923,
      "grad_norm": 2.767521619796753,
      "learning_rate": 3.878766967450158e-06,
      "loss": 0.8107,
      "step": 318
    },
    {
      "epoch": 1.1949390815370198,
      "grad_norm": 2.630614995956421,
      "learning_rate": 3.848595550550401e-06,
      "loss": 0.7719,
      "step": 319
    },
    {
      "epoch": 1.1986879100281163,
      "grad_norm": 2.7656266689300537,
      "learning_rate": 3.818468342614932e-06,
      "loss": 0.6614,
      "step": 320
    },
    {
      "epoch": 1.2024367385192127,
      "grad_norm": 2.6769886016845703,
      "learning_rate": 3.788386500398583e-06,
      "loss": 0.6055,
      "step": 321
    },
    {
      "epoch": 1.2061855670103092,
      "grad_norm": 2.4502344131469727,
      "learning_rate": 3.758351178914336e-06,
      "loss": 0.5417,
      "step": 322
    },
    {
      "epoch": 1.209934395501406,
      "grad_norm": 2.5828676223754883,
      "learning_rate": 3.728363531388979e-06,
      "loss": 0.6551,
      "step": 323
    },
    {
      "epoch": 1.2136832239925024,
      "grad_norm": 2.5377285480499268,
      "learning_rate": 3.6984247092188265e-06,
      "loss": 0.5738,
      "step": 324
    },
    {
      "epoch": 1.2174320524835989,
      "grad_norm": 3.0588667392730713,
      "learning_rate": 3.668535861925509e-06,
      "loss": 0.9216,
      "step": 325
    },
    {
      "epoch": 1.2211808809746953,
      "grad_norm": 2.466240167617798,
      "learning_rate": 3.6386981371118358e-06,
      "loss": 0.4853,
      "step": 326
    },
    {
      "epoch": 1.2249297094657918,
      "grad_norm": 2.7123985290527344,
      "learning_rate": 3.6089126804177373e-06,
      "loss": 0.5507,
      "step": 327
    },
    {
      "epoch": 1.2286785379568885,
      "grad_norm": 2.5103797912597656,
      "learning_rate": 3.5791806354762702e-06,
      "loss": 0.723,
      "step": 328
    },
    {
      "epoch": 1.232427366447985,
      "grad_norm": 2.6206247806549072,
      "learning_rate": 3.5495031438697103e-06,
      "loss": 0.6191,
      "step": 329
    },
    {
      "epoch": 1.2361761949390815,
      "grad_norm": 2.5603182315826416,
      "learning_rate": 3.519881345085724e-06,
      "loss": 0.6315,
      "step": 330
    },
    {
      "epoch": 1.2399250234301782,
      "grad_norm": 2.5217936038970947,
      "learning_rate": 3.4903163764736104e-06,
      "loss": 0.5989,
      "step": 331
    },
    {
      "epoch": 1.2436738519212747,
      "grad_norm": 2.4896228313446045,
      "learning_rate": 3.4608093732006367e-06,
      "loss": 0.6737,
      "step": 332
    },
    {
      "epoch": 1.2474226804123711,
      "grad_norm": 2.5744566917419434,
      "learning_rate": 3.4313614682084483e-06,
      "loss": 0.5537,
      "step": 333
    },
    {
      "epoch": 1.2511715089034676,
      "grad_norm": 2.5293214321136475,
      "learning_rate": 3.401973792169574e-06,
      "loss": 0.6486,
      "step": 334
    },
    {
      "epoch": 1.254920337394564,
      "grad_norm": 2.5666043758392334,
      "learning_rate": 3.372647473444012e-06,
      "loss": 0.5285,
      "step": 335
    },
    {
      "epoch": 1.2586691658856608,
      "grad_norm": 2.9129676818847656,
      "learning_rate": 3.343383638035902e-06,
      "loss": 0.8922,
      "step": 336
    },
    {
      "epoch": 1.2624179943767573,
      "grad_norm": 2.596280097961426,
      "learning_rate": 3.314183409550293e-06,
      "loss": 0.6352,
      "step": 337
    },
    {
      "epoch": 1.2661668228678538,
      "grad_norm": 2.8564982414245605,
      "learning_rate": 3.285047909150006e-06,
      "loss": 0.5762,
      "step": 338
    },
    {
      "epoch": 1.2699156513589505,
      "grad_norm": 2.6609466075897217,
      "learning_rate": 3.2559782555125793e-06,
      "loss": 0.6421,
      "step": 339
    },
    {
      "epoch": 1.273664479850047,
      "grad_norm": 2.636430501937866,
      "learning_rate": 3.226975564787322e-06,
      "loss": 0.8343,
      "step": 340
    },
    {
      "epoch": 1.2774133083411434,
      "grad_norm": 2.764880657196045,
      "learning_rate": 3.1980409505524546e-06,
      "loss": 0.8492,
      "step": 341
    },
    {
      "epoch": 1.28116213683224,
      "grad_norm": 2.4313178062438965,
      "learning_rate": 3.1691755237723538e-06,
      "loss": 0.5935,
      "step": 342
    },
    {
      "epoch": 1.2849109653233364,
      "grad_norm": 3.312020778656006,
      "learning_rate": 3.140380392754901e-06,
      "loss": 0.7306,
      "step": 343
    },
    {
      "epoch": 1.2886597938144329,
      "grad_norm": 2.7336678504943848,
      "learning_rate": 3.111656663108914e-06,
      "loss": 0.6683,
      "step": 344
    },
    {
      "epoch": 1.2924086223055296,
      "grad_norm": 2.6858274936676025,
      "learning_rate": 3.083005437701715e-06,
      "loss": 0.6287,
      "step": 345
    },
    {
      "epoch": 1.296157450796626,
      "grad_norm": 2.664910078048706,
      "learning_rate": 3.054427816616773e-06,
      "loss": 0.6392,
      "step": 346
    },
    {
      "epoch": 1.2999062792877225,
      "grad_norm": 2.3806943893432617,
      "learning_rate": 3.0259248971114663e-06,
      "loss": 0.6144,
      "step": 347
    },
    {
      "epoch": 1.3036551077788192,
      "grad_norm": 2.755476951599121,
      "learning_rate": 2.9974977735749596e-06,
      "loss": 0.6259,
      "step": 348
    },
    {
      "epoch": 1.3074039362699157,
      "grad_norm": 2.4375879764556885,
      "learning_rate": 2.969147537486175e-06,
      "loss": 0.6063,
      "step": 349
    },
    {
      "epoch": 1.3111527647610122,
      "grad_norm": 2.6043853759765625,
      "learning_rate": 2.9408752773718895e-06,
      "loss": 0.9196,
      "step": 350
    },
    {
      "epoch": 1.3149015932521086,
      "grad_norm": 2.462339401245117,
      "learning_rate": 2.9126820787649403e-06,
      "loss": 0.5322,
      "step": 351
    },
    {
      "epoch": 1.3186504217432051,
      "grad_norm": 2.3056864738464355,
      "learning_rate": 2.8845690241625437e-06,
      "loss": 0.6108,
      "step": 352
    },
    {
      "epoch": 1.3223992502343018,
      "grad_norm": 2.394547462463379,
      "learning_rate": 2.8565371929847286e-06,
      "loss": 0.6488,
      "step": 353
    },
    {
      "epoch": 1.3261480787253983,
      "grad_norm": 3.0996365547180176,
      "learning_rate": 2.828587661532901e-06,
      "loss": 0.6514,
      "step": 354
    },
    {
      "epoch": 1.3298969072164948,
      "grad_norm": 2.57309889793396,
      "learning_rate": 2.800721502948506e-06,
      "loss": 0.614,
      "step": 355
    },
    {
      "epoch": 1.3336457357075915,
      "grad_norm": 2.417073965072632,
      "learning_rate": 2.7729397871718306e-06,
      "loss": 0.5476,
      "step": 356
    },
    {
      "epoch": 1.337394564198688,
      "grad_norm": 2.636715888977051,
      "learning_rate": 2.7452435809009272e-06,
      "loss": 0.7593,
      "step": 357
    },
    {
      "epoch": 1.3411433926897844,
      "grad_norm": 3.096482992172241,
      "learning_rate": 2.7176339475506515e-06,
      "loss": 0.8159,
      "step": 358
    },
    {
      "epoch": 1.344892221180881,
      "grad_norm": 2.1638081073760986,
      "learning_rate": 2.6901119472118253e-06,
      "loss": 0.5549,
      "step": 359
    },
    {
      "epoch": 1.3486410496719774,
      "grad_norm": 2.6135005950927734,
      "learning_rate": 2.6626786366105493e-06,
      "loss": 0.6626,
      "step": 360
    },
    {
      "epoch": 1.352389878163074,
      "grad_norm": 2.5569889545440674,
      "learning_rate": 2.635335069067617e-06,
      "loss": 0.5365,
      "step": 361
    },
    {
      "epoch": 1.3561387066541706,
      "grad_norm": 2.9490528106689453,
      "learning_rate": 2.608082294458074e-06,
      "loss": 0.9714,
      "step": 362
    },
    {
      "epoch": 1.359887535145267,
      "grad_norm": 3.6054556369781494,
      "learning_rate": 2.5809213591709124e-06,
      "loss": 0.8305,
      "step": 363
    },
    {
      "epoch": 1.3636363636363638,
      "grad_norm": 2.648580312728882,
      "learning_rate": 2.553853306068888e-06,
      "loss": 0.7582,
      "step": 364
    },
    {
      "epoch": 1.3673851921274602,
      "grad_norm": 2.500655174255371,
      "learning_rate": 2.5268791744484865e-06,
      "loss": 0.5625,
      "step": 365
    },
    {
      "epoch": 1.3711340206185567,
      "grad_norm": 2.625460147857666,
      "learning_rate": 2.5000000000000015e-06,
      "loss": 0.8212,
      "step": 366
    },
    {
      "epoch": 1.3748828491096532,
      "grad_norm": 2.444305896759033,
      "learning_rate": 2.4732168147677927e-06,
      "loss": 0.582,
      "step": 367
    },
    {
      "epoch": 1.3786316776007497,
      "grad_norm": 2.8055059909820557,
      "learning_rate": 2.446530647110646e-06,
      "loss": 0.5747,
      "step": 368
    },
    {
      "epoch": 1.3823805060918464,
      "grad_norm": 2.7823076248168945,
      "learning_rate": 2.419942521662285e-06,
      "loss": 0.5465,
      "step": 369
    },
    {
      "epoch": 1.3861293345829429,
      "grad_norm": 3.1207704544067383,
      "learning_rate": 2.3934534592920416e-06,
      "loss": 0.8112,
      "step": 370
    },
    {
      "epoch": 1.3898781630740393,
      "grad_norm": 2.4543776512145996,
      "learning_rate": 2.367064477065652e-06,
      "loss": 0.7124,
      "step": 371
    },
    {
      "epoch": 1.393626991565136,
      "grad_norm": 2.6401026248931885,
      "learning_rate": 2.3407765882062024e-06,
      "loss": 0.6856,
      "step": 372
    },
    {
      "epoch": 1.3973758200562325,
      "grad_norm": 2.729567527770996,
      "learning_rate": 2.314590802055232e-06,
      "loss": 0.6516,
      "step": 373
    },
    {
      "epoch": 1.401124648547329,
      "grad_norm": 2.283984899520874,
      "learning_rate": 2.2885081240339813e-06,
      "loss": 0.4082,
      "step": 374
    },
    {
      "epoch": 1.4048734770384255,
      "grad_norm": 2.7015268802642822,
      "learning_rate": 2.262529555604774e-06,
      "loss": 0.7567,
      "step": 375
    },
    {
      "epoch": 1.408622305529522,
      "grad_norm": 2.3578450679779053,
      "learning_rate": 2.2366560942325833e-06,
      "loss": 0.6722,
      "step": 376
    },
    {
      "epoch": 1.4123711340206184,
      "grad_norm": 2.6497104167938232,
      "learning_rate": 2.2108887333467172e-06,
      "loss": 0.6284,
      "step": 377
    },
    {
      "epoch": 1.4161199625117151,
      "grad_norm": 2.5546791553497314,
      "learning_rate": 2.1852284623026906e-06,
      "loss": 0.5851,
      "step": 378
    },
    {
      "epoch": 1.4198687910028116,
      "grad_norm": 2.486978530883789,
      "learning_rate": 2.159676266344222e-06,
      "loss": 0.6988,
      "step": 379
    },
    {
      "epoch": 1.423617619493908,
      "grad_norm": 2.8536384105682373,
      "learning_rate": 2.1342331265654194e-06,
      "loss": 0.7193,
      "step": 380
    },
    {
      "epoch": 1.4273664479850048,
      "grad_norm": 2.9641923904418945,
      "learning_rate": 2.108900019873103e-06,
      "loss": 0.6805,
      "step": 381
    },
    {
      "epoch": 1.4311152764761013,
      "grad_norm": 2.463860034942627,
      "learning_rate": 2.0836779189492925e-06,
      "loss": 0.7775,
      "step": 382
    },
    {
      "epoch": 1.4348641049671977,
      "grad_norm": 2.7579569816589355,
      "learning_rate": 2.0585677922138696e-06,
      "loss": 0.6252,
      "step": 383
    },
    {
      "epoch": 1.4386129334582942,
      "grad_norm": 2.5609865188598633,
      "learning_rate": 2.033570603787391e-06,
      "loss": 0.6355,
      "step": 384
    },
    {
      "epoch": 1.4423617619493907,
      "grad_norm": 2.9254584312438965,
      "learning_rate": 2.0086873134540626e-06,
      "loss": 0.6698,
      "step": 385
    },
    {
      "epoch": 1.4461105904404874,
      "grad_norm": 3.046320915222168,
      "learning_rate": 1.9839188766249024e-06,
      "loss": 0.6857,
      "step": 386
    },
    {
      "epoch": 1.4498594189315839,
      "grad_norm": 2.7485857009887695,
      "learning_rate": 1.959266244301047e-06,
      "loss": 0.8151,
      "step": 387
    },
    {
      "epoch": 1.4536082474226804,
      "grad_norm": 2.6440517902374268,
      "learning_rate": 1.9347303630372373e-06,
      "loss": 0.63,
      "step": 388
    },
    {
      "epoch": 1.457357075913777,
      "grad_norm": 2.842188596725464,
      "learning_rate": 1.910312174905477e-06,
      "loss": 0.5113,
      "step": 389
    },
    {
      "epoch": 1.4611059044048735,
      "grad_norm": 2.662004232406616,
      "learning_rate": 1.886012617458864e-06,
      "loss": 0.7212,
      "step": 390
    },
    {
      "epoch": 1.46485473289597,
      "grad_norm": 3.0616824626922607,
      "learning_rate": 1.8618326236955908e-06,
      "loss": 0.4773,
      "step": 391
    },
    {
      "epoch": 1.4686035613870665,
      "grad_norm": 2.796537399291992,
      "learning_rate": 1.8377731220231144e-06,
      "loss": 0.7543,
      "step": 392
    },
    {
      "epoch": 1.472352389878163,
      "grad_norm": 2.2921605110168457,
      "learning_rate": 1.8138350362225193e-06,
      "loss": 0.5149,
      "step": 393
    },
    {
      "epoch": 1.4761012183692597,
      "grad_norm": 3.190627098083496,
      "learning_rate": 1.7900192854130465e-06,
      "loss": 0.7386,
      "step": 394
    },
    {
      "epoch": 1.4798500468603561,
      "grad_norm": 2.5983688831329346,
      "learning_rate": 1.7663267840167936e-06,
      "loss": 0.6061,
      "step": 395
    },
    {
      "epoch": 1.4835988753514526,
      "grad_norm": 2.47511887550354,
      "learning_rate": 1.7427584417236194e-06,
      "loss": 0.4628,
      "step": 396
    },
    {
      "epoch": 1.4873477038425493,
      "grad_norm": 2.6179940700531006,
      "learning_rate": 1.7193151634562071e-06,
      "loss": 0.7502,
      "step": 397
    },
    {
      "epoch": 1.4910965323336458,
      "grad_norm": 2.2079148292541504,
      "learning_rate": 1.695997849335319e-06,
      "loss": 0.7195,
      "step": 398
    },
    {
      "epoch": 1.4948453608247423,
      "grad_norm": 2.3705711364746094,
      "learning_rate": 1.672807394645236e-06,
      "loss": 0.679,
      "step": 399
    },
    {
      "epoch": 1.4985941893158388,
      "grad_norm": 2.5409562587738037,
      "learning_rate": 1.6497446897993885e-06,
      "loss": 0.6188,
      "step": 400
    },
    {
      "epoch": 1.5023430178069352,
      "grad_norm": 2.718851089477539,
      "learning_rate": 1.6268106203061628e-06,
      "loss": 0.4311,
      "step": 401
    },
    {
      "epoch": 1.5060918462980317,
      "grad_norm": 2.4857470989227295,
      "learning_rate": 1.6040060667348995e-06,
      "loss": 0.5452,
      "step": 402
    },
    {
      "epoch": 1.5098406747891284,
      "grad_norm": 2.272148609161377,
      "learning_rate": 1.581331904682089e-06,
      "loss": 0.6673,
      "step": 403
    },
    {
      "epoch": 1.513589503280225,
      "grad_norm": 2.118818759918213,
      "learning_rate": 1.5587890047377512e-06,
      "loss": 0.474,
      "step": 404
    },
    {
      "epoch": 1.5173383317713216,
      "grad_norm": 2.6302151679992676,
      "learning_rate": 1.5363782324520033e-06,
      "loss": 0.7036,
      "step": 405
    },
    {
      "epoch": 1.521087160262418,
      "grad_norm": 2.78167724609375,
      "learning_rate": 1.5141004483018323e-06,
      "loss": 0.5538,
      "step": 406
    },
    {
      "epoch": 1.5248359887535146,
      "grad_norm": 2.7051169872283936,
      "learning_rate": 1.4919565076580577e-06,
      "loss": 0.6842,
      "step": 407
    },
    {
      "epoch": 1.528584817244611,
      "grad_norm": 2.7234108448028564,
      "learning_rate": 1.4699472607524785e-06,
      "loss": 0.6473,
      "step": 408
    },
    {
      "epoch": 1.5323336457357075,
      "grad_norm": 2.808751106262207,
      "learning_rate": 1.4480735526452427e-06,
      "loss": 0.7719,
      "step": 409
    },
    {
      "epoch": 1.536082474226804,
      "grad_norm": 2.5814692974090576,
      "learning_rate": 1.426336223192386e-06,
      "loss": 0.6786,
      "step": 410
    },
    {
      "epoch": 1.5398313027179007,
      "grad_norm": 2.6848583221435547,
      "learning_rate": 1.4047361070135996e-06,
      "loss": 0.7245,
      "step": 411
    },
    {
      "epoch": 1.5435801312089972,
      "grad_norm": 2.525697708129883,
      "learning_rate": 1.3832740334601692e-06,
      "loss": 0.5269,
      "step": 412
    },
    {
      "epoch": 1.5473289597000939,
      "grad_norm": 2.4603147506713867,
      "learning_rate": 1.3619508265831445e-06,
      "loss": 0.7593,
      "step": 413
    },
    {
      "epoch": 1.5510777881911904,
      "grad_norm": 2.712268114089966,
      "learning_rate": 1.340767305101694e-06,
      "loss": 0.8538,
      "step": 414
    },
    {
      "epoch": 1.5548266166822868,
      "grad_norm": 2.391826629638672,
      "learning_rate": 1.319724282371664e-06,
      "loss": 0.7118,
      "step": 415
    },
    {
      "epoch": 1.5585754451733833,
      "grad_norm": 2.6861860752105713,
      "learning_rate": 1.2988225663543601e-06,
      "loss": 0.6899,
      "step": 416
    },
    {
      "epoch": 1.5623242736644798,
      "grad_norm": 3.0465526580810547,
      "learning_rate": 1.2780629595855203e-06,
      "loss": 0.6717,
      "step": 417
    },
    {
      "epoch": 1.5660731021555763,
      "grad_norm": 2.586912155151367,
      "learning_rate": 1.257446259144494e-06,
      "loss": 0.6787,
      "step": 418
    },
    {
      "epoch": 1.569821930646673,
      "grad_norm": 2.4131391048431396,
      "learning_rate": 1.2369732566236508e-06,
      "loss": 0.7233,
      "step": 419
    },
    {
      "epoch": 1.5735707591377694,
      "grad_norm": 2.3727686405181885,
      "learning_rate": 1.2166447380979801e-06,
      "loss": 0.7571,
      "step": 420
    },
    {
      "epoch": 1.577319587628866,
      "grad_norm": 2.27282452583313,
      "learning_rate": 1.1964614840949002e-06,
      "loss": 0.5189,
      "step": 421
    },
    {
      "epoch": 1.5810684161199626,
      "grad_norm": 2.7784337997436523,
      "learning_rate": 1.1764242695643075e-06,
      "loss": 0.4099,
      "step": 422
    },
    {
      "epoch": 1.584817244611059,
      "grad_norm": 2.5282742977142334,
      "learning_rate": 1.1565338638488117e-06,
      "loss": 0.5868,
      "step": 423
    },
    {
      "epoch": 1.5885660731021556,
      "grad_norm": 2.49729061126709,
      "learning_rate": 1.1367910306541918e-06,
      "loss": 0.6738,
      "step": 424
    },
    {
      "epoch": 1.592314901593252,
      "grad_norm": 2.5107173919677734,
      "learning_rate": 1.1171965280200831e-06,
      "loss": 0.5341,
      "step": 425
    },
    {
      "epoch": 1.5960637300843485,
      "grad_norm": 2.601574659347534,
      "learning_rate": 1.097751108290867e-06,
      "loss": 0.692,
      "step": 426
    },
    {
      "epoch": 1.599812558575445,
      "grad_norm": 2.5476865768432617,
      "learning_rate": 1.078455518086784e-06,
      "loss": 0.7559,
      "step": 427
    },
    {
      "epoch": 1.6035613870665417,
      "grad_norm": 2.2659873962402344,
      "learning_rate": 1.0593104982752645e-06,
      "loss": 0.5147,
      "step": 428
    },
    {
      "epoch": 1.6073102155576382,
      "grad_norm": 2.6162917613983154,
      "learning_rate": 1.0403167839424883e-06,
      "loss": 0.7862,
      "step": 429
    },
    {
      "epoch": 1.611059044048735,
      "grad_norm": 3.0038373470306396,
      "learning_rate": 1.0214751043651582e-06,
      "loss": 0.7098,
      "step": 430
    },
    {
      "epoch": 1.6148078725398314,
      "grad_norm": 2.810269832611084,
      "learning_rate": 1.0027861829824953e-06,
      "loss": 0.4501,
      "step": 431
    },
    {
      "epoch": 1.6185567010309279,
      "grad_norm": 2.2872512340545654,
      "learning_rate": 9.842507373684646e-07,
      "loss": 0.6829,
      "step": 432
    },
    {
      "epoch": 1.6223055295220243,
      "grad_norm": 2.7606618404388428,
      "learning_rate": 9.658694792042284e-07,
      "loss": 0.8641,
      "step": 433
    },
    {
      "epoch": 1.6260543580131208,
      "grad_norm": 2.647319793701172,
      "learning_rate": 9.476431142508097e-07,
      "loss": 0.684,
      "step": 434
    },
    {
      "epoch": 1.6298031865042173,
      "grad_norm": 2.3936805725097656,
      "learning_rate": 9.295723423220049e-07,
      "loss": 0.4689,
      "step": 435
    },
    {
      "epoch": 1.633552014995314,
      "grad_norm": 2.8151843547821045,
      "learning_rate": 9.116578572575091e-07,
      "loss": 0.7088,
      "step": 436
    },
    {
      "epoch": 1.6373008434864105,
      "grad_norm": 2.6258625984191895,
      "learning_rate": 8.939003468962726e-07,
      "loss": 0.7019,
      "step": 437
    },
    {
      "epoch": 1.6410496719775072,
      "grad_norm": 2.601217031478882,
      "learning_rate": 8.763004930500979e-07,
      "loss": 0.5881,
      "step": 438
    },
    {
      "epoch": 1.6447985004686037,
      "grad_norm": 2.325805187225342,
      "learning_rate": 8.58858971477457e-07,
      "loss": 0.6775,
      "step": 439
    },
    {
      "epoch": 1.6485473289597001,
      "grad_norm": 2.8295209407806396,
      "learning_rate": 8.415764518575415e-07,
      "loss": 0.7012,
      "step": 440
    },
    {
      "epoch": 1.6522961574507966,
      "grad_norm": 2.421982765197754,
      "learning_rate": 8.244535977645584e-07,
      "loss": 0.6516,
      "step": 441
    },
    {
      "epoch": 1.656044985941893,
      "grad_norm": 2.5072197914123535,
      "learning_rate": 8.074910666422475e-07,
      "loss": 0.7382,
      "step": 442
    },
    {
      "epoch": 1.6597938144329896,
      "grad_norm": 2.841768264770508,
      "learning_rate": 7.906895097786338e-07,
      "loss": 0.5838,
      "step": 443
    },
    {
      "epoch": 1.6635426429240863,
      "grad_norm": 2.738243818283081,
      "learning_rate": 7.740495722810271e-07,
      "loss": 0.7085,
      "step": 444
    },
    {
      "epoch": 1.6672914714151827,
      "grad_norm": 2.763880491256714,
      "learning_rate": 7.575718930512516e-07,
      "loss": 0.6332,
      "step": 445
    },
    {
      "epoch": 1.6710402999062794,
      "grad_norm": 2.3609251976013184,
      "learning_rate": 7.412571047611156e-07,
      "loss": 0.5682,
      "step": 446
    },
    {
      "epoch": 1.674789128397376,
      "grad_norm": 3.1101462841033936,
      "learning_rate": 7.25105833828113e-07,
      "loss": 0.8626,
      "step": 447
    },
    {
      "epoch": 1.6785379568884724,
      "grad_norm": 2.56254506111145,
      "learning_rate": 7.091187003913802e-07,
      "loss": 0.733,
      "step": 448
    },
    {
      "epoch": 1.6822867853795689,
      "grad_norm": 2.901610851287842,
      "learning_rate": 6.932963182878821e-07,
      "loss": 0.7838,
      "step": 449
    },
    {
      "epoch": 1.6860356138706654,
      "grad_norm": 2.393528699874878,
      "learning_rate": 6.776392950288397e-07,
      "loss": 0.6754,
      "step": 450
    },
    {
      "epoch": 1.6897844423617618,
      "grad_norm": 2.7382218837738037,
      "learning_rate": 6.621482317764105e-07,
      "loss": 0.7144,
      "step": 451
    },
    {
      "epoch": 1.6935332708528583,
      "grad_norm": 2.618114471435547,
      "learning_rate": 6.468237233206043e-07,
      "loss": 0.7003,
      "step": 452
    },
    {
      "epoch": 1.697282099343955,
      "grad_norm": 2.418599843978882,
      "learning_rate": 6.316663580564425e-07,
      "loss": 0.6957,
      "step": 453
    },
    {
      "epoch": 1.7010309278350515,
      "grad_norm": 2.5395448207855225,
      "learning_rate": 6.166767179613691e-07,
      "loss": 0.5987,
      "step": 454
    },
    {
      "epoch": 1.7047797563261482,
      "grad_norm": 2.5443437099456787,
      "learning_rate": 6.018553785729075e-07,
      "loss": 0.4977,
      "step": 455
    },
    {
      "epoch": 1.7085285848172447,
      "grad_norm": 2.4432942867279053,
      "learning_rate": 5.872029089665588e-07,
      "loss": 0.675,
      "step": 456
    },
    {
      "epoch": 1.7122774133083412,
      "grad_norm": 2.7112233638763428,
      "learning_rate": 5.727198717339511e-07,
      "loss": 0.7934,
      "step": 457
    },
    {
      "epoch": 1.7160262417994376,
      "grad_norm": 2.9048352241516113,
      "learning_rate": 5.584068229612422e-07,
      "loss": 0.7687,
      "step": 458
    },
    {
      "epoch": 1.7197750702905341,
      "grad_norm": 2.5611398220062256,
      "learning_rate": 5.442643122077673e-07,
      "loss": 0.516,
      "step": 459
    },
    {
      "epoch": 1.7235238987816306,
      "grad_norm": 2.6895134449005127,
      "learning_rate": 5.302928824849335e-07,
      "loss": 0.662,
      "step": 460
    },
    {
      "epoch": 1.7272727272727273,
      "grad_norm": 2.621039390563965,
      "learning_rate": 5.164930702353782e-07,
      "loss": 0.6618,
      "step": 461
    },
    {
      "epoch": 1.7310215557638238,
      "grad_norm": 2.5842714309692383,
      "learning_rate": 5.028654053123666e-07,
      "loss": 0.5826,
      "step": 462
    },
    {
      "epoch": 1.7347703842549205,
      "grad_norm": 2.4873173236846924,
      "learning_rate": 4.894104109594466e-07,
      "loss": 0.6776,
      "step": 463
    },
    {
      "epoch": 1.738519212746017,
      "grad_norm": 2.2931952476501465,
      "learning_rate": 4.7612860379036674e-07,
      "loss": 0.6094,
      "step": 464
    },
    {
      "epoch": 1.7422680412371134,
      "grad_norm": 2.699720859527588,
      "learning_rate": 4.6302049376922843e-07,
      "loss": 0.5347,
      "step": 465
    },
    {
      "epoch": 1.74601686972821,
      "grad_norm": 3.027738571166992,
      "learning_rate": 4.500865841909169e-07,
      "loss": 0.6739,
      "step": 466
    },
    {
      "epoch": 1.7497656982193064,
      "grad_norm": 2.582227945327759,
      "learning_rate": 4.373273716617682e-07,
      "loss": 0.547,
      "step": 467
    },
    {
      "epoch": 1.7535145267104029,
      "grad_norm": 3.154305934906006,
      "learning_rate": 4.247433460805067e-07,
      "loss": 0.7603,
      "step": 468
    },
    {
      "epoch": 1.7572633552014996,
      "grad_norm": 2.7173681259155273,
      "learning_rate": 4.123349906194357e-07,
      "loss": 0.6546,
      "step": 469
    },
    {
      "epoch": 1.761012183692596,
      "grad_norm": 2.6075384616851807,
      "learning_rate": 4.001027817058789e-07,
      "loss": 0.6589,
      "step": 470
    },
    {
      "epoch": 1.7647610121836927,
      "grad_norm": 2.7403595447540283,
      "learning_rate": 3.8804718900389673e-07,
      "loss": 0.7657,
      "step": 471
    },
    {
      "epoch": 1.7685098406747892,
      "grad_norm": 2.6404995918273926,
      "learning_rate": 3.7616867539624733e-07,
      "loss": 0.4362,
      "step": 472
    },
    {
      "epoch": 1.7722586691658857,
      "grad_norm": 2.6141321659088135,
      "learning_rate": 3.6446769696661445e-07,
      "loss": 0.5542,
      "step": 473
    },
    {
      "epoch": 1.7760074976569822,
      "grad_norm": 2.7053098678588867,
      "learning_rate": 3.5294470298209817e-07,
      "loss": 0.7215,
      "step": 474
    },
    {
      "epoch": 1.7797563261480787,
      "grad_norm": 2.3503482341766357,
      "learning_rate": 3.416001358759635e-07,
      "loss": 0.5368,
      "step": 475
    },
    {
      "epoch": 1.7835051546391751,
      "grad_norm": 2.7156739234924316,
      "learning_rate": 3.304344312306529e-07,
      "loss": 0.6804,
      "step": 476
    },
    {
      "epoch": 1.7872539831302718,
      "grad_norm": 2.5638582706451416,
      "learning_rate": 3.194480177610604e-07,
      "loss": 0.8309,
      "step": 477
    },
    {
      "epoch": 1.7910028116213683,
      "grad_norm": 2.8452188968658447,
      "learning_rate": 3.08641317298074e-07,
      "loss": 0.5907,
      "step": 478
    },
    {
      "epoch": 1.794751640112465,
      "grad_norm": 2.5471601486206055,
      "learning_rate": 2.980147447723775e-07,
      "loss": 0.6524,
      "step": 479
    },
    {
      "epoch": 1.7985004686035615,
      "grad_norm": 2.443708658218384,
      "learning_rate": 2.8756870819851736e-07,
      "loss": 0.5538,
      "step": 480
    },
    {
      "epoch": 1.802249297094658,
      "grad_norm": 2.7727556228637695,
      "learning_rate": 2.7730360865923954e-07,
      "loss": 0.6459,
      "step": 481
    },
    {
      "epoch": 1.8059981255857545,
      "grad_norm": 3.0080816745758057,
      "learning_rate": 2.672198402900883e-07,
      "loss": 0.7243,
      "step": 482
    },
    {
      "epoch": 1.809746954076851,
      "grad_norm": 2.2730140686035156,
      "learning_rate": 2.573177902642726e-07,
      "loss": 0.7811,
      "step": 483
    },
    {
      "epoch": 1.8134957825679474,
      "grad_norm": 2.5842466354370117,
      "learning_rate": 2.475978387778e-07,
      "loss": 0.595,
      "step": 484
    },
    {
      "epoch": 1.817244611059044,
      "grad_norm": 2.4688827991485596,
      "learning_rate": 2.380603590348829e-07,
      "loss": 0.6941,
      "step": 485
    },
    {
      "epoch": 1.8209934395501406,
      "grad_norm": 2.346224546432495,
      "learning_rate": 2.2870571723360212e-07,
      "loss": 0.7242,
      "step": 486
    },
    {
      "epoch": 1.824742268041237,
      "grad_norm": 2.721781015396118,
      "learning_rate": 2.1953427255185122e-07,
      "loss": 0.7004,
      "step": 487
    },
    {
      "epoch": 1.8284910965323338,
      "grad_norm": 2.539914846420288,
      "learning_rate": 2.1054637713354586e-07,
      "loss": 0.6182,
      "step": 488
    },
    {
      "epoch": 1.8322399250234302,
      "grad_norm": 2.6623215675354004,
      "learning_rate": 2.0174237607510138e-07,
      "loss": 0.4989,
      "step": 489
    },
    {
      "epoch": 1.8359887535145267,
      "grad_norm": 2.411862373352051,
      "learning_rate": 1.9312260741218114e-07,
      "loss": 0.4213,
      "step": 490
    },
    {
      "epoch": 1.8397375820056232,
      "grad_norm": 2.1537928581237793,
      "learning_rate": 1.8468740210672077e-07,
      "loss": 0.5353,
      "step": 491
    },
    {
      "epoch": 1.8434864104967197,
      "grad_norm": 3.083552598953247,
      "learning_rate": 1.7643708403422055e-07,
      "loss": 1.1005,
      "step": 492
    },
    {
      "epoch": 1.8472352389878162,
      "grad_norm": 2.445988178253174,
      "learning_rate": 1.6837196997130434e-07,
      "loss": 0.6158,
      "step": 493
    },
    {
      "epoch": 1.8509840674789129,
      "grad_norm": 2.336162567138672,
      "learning_rate": 1.6049236958356475e-07,
      "loss": 0.649,
      "step": 494
    },
    {
      "epoch": 1.8547328959700093,
      "grad_norm": 2.544160842895508,
      "learning_rate": 1.5279858541366876e-07,
      "loss": 0.7222,
      "step": 495
    },
    {
      "epoch": 1.858481724461106,
      "grad_norm": 2.5376548767089844,
      "learning_rate": 1.4529091286973994e-07,
      "loss": 0.7403,
      "step": 496
    },
    {
      "epoch": 1.8622305529522025,
      "grad_norm": 3.008063554763794,
      "learning_rate": 1.3796964021402072e-07,
      "loss": 0.7811,
      "step": 497
    },
    {
      "epoch": 1.865979381443299,
      "grad_norm": 2.640939712524414,
      "learning_rate": 1.3083504855180007e-07,
      "loss": 0.6864,
      "step": 498
    },
    {
      "epoch": 1.8697282099343955,
      "grad_norm": 2.6732051372528076,
      "learning_rate": 1.2388741182062348e-07,
      "loss": 0.6969,
      "step": 499
    },
    {
      "epoch": 1.873477038425492,
      "grad_norm": 2.688314199447632,
      "learning_rate": 1.1712699677977224e-07,
      "loss": 0.659,
      "step": 500
    },
    {
      "epoch": 1.8772258669165884,
      "grad_norm": 2.5495784282684326,
      "learning_rate": 1.1055406300002347e-07,
      "loss": 0.5468,
      "step": 501
    },
    {
      "epoch": 1.8809746954076851,
      "grad_norm": 2.6933047771453857,
      "learning_rate": 1.0416886285368188e-07,
      "loss": 0.5204,
      "step": 502
    },
    {
      "epoch": 1.8847235238987816,
      "grad_norm": 2.7378575801849365,
      "learning_rate": 9.797164150489035e-08,
      "loss": 0.5859,
      "step": 503
    },
    {
      "epoch": 1.8884723523898783,
      "grad_norm": 2.0356202125549316,
      "learning_rate": 9.1962636900218e-08,
      "loss": 0.3406,
      "step": 504
    },
    {
      "epoch": 1.8922211808809748,
      "grad_norm": 3.0371510982513428,
      "learning_rate": 8.614207975952083e-08,
      "loss": 0.7767,
      "step": 505
    },
    {
      "epoch": 1.8959700093720713,
      "grad_norm": 2.813620090484619,
      "learning_rate": 8.0510193567086e-08,
      "loss": 0.5616,
      "step": 506
    },
    {
      "epoch": 1.8997188378631678,
      "grad_norm": 2.557776689529419,
      "learning_rate": 7.5067194563051e-08,
      "loss": 0.5433,
      "step": 507
    },
    {
      "epoch": 1.9034676663542642,
      "grad_norm": 2.6855971813201904,
      "learning_rate": 6.981329173509909e-08,
      "loss": 0.6334,
      "step": 508
    },
    {
      "epoch": 1.9072164948453607,
      "grad_norm": 2.539196729660034,
      "learning_rate": 6.474868681043578e-08,
      "loss": 0.6286,
      "step": 509
    },
    {
      "epoch": 1.9109653233364574,
      "grad_norm": 2.1722402572631836,
      "learning_rate": 5.987357424804441e-08,
      "loss": 0.5581,
      "step": 510
    },
    {
      "epoch": 1.914714151827554,
      "grad_norm": 2.4900224208831787,
      "learning_rate": 5.518814123121885e-08,
      "loss": 0.6132,
      "step": 511
    },
    {
      "epoch": 1.9184629803186504,
      "grad_norm": 2.708784818649292,
      "learning_rate": 5.0692567660375334e-08,
      "loss": 0.6154,
      "step": 512
    },
    {
      "epoch": 1.922211808809747,
      "grad_norm": 2.573993444442749,
      "learning_rate": 4.638702614614854e-08,
      "loss": 0.6727,
      "step": 513
    },
    {
      "epoch": 1.9259606373008435,
      "grad_norm": 2.826658248901367,
      "learning_rate": 4.227168200276077e-08,
      "loss": 0.7038,
      "step": 514
    },
    {
      "epoch": 1.92970946579194,
      "grad_norm": 2.657594680786133,
      "learning_rate": 3.834669324167428e-08,
      "loss": 0.7022,
      "step": 515
    },
    {
      "epoch": 1.9334582942830365,
      "grad_norm": 2.5230019092559814,
      "learning_rate": 3.4612210565528323e-08,
      "loss": 0.5854,
      "step": 516
    },
    {
      "epoch": 1.937207122774133,
      "grad_norm": 2.966693162918091,
      "learning_rate": 3.10683773623488e-08,
      "loss": 0.6926,
      "step": 517
    },
    {
      "epoch": 1.9409559512652295,
      "grad_norm": 2.689201593399048,
      "learning_rate": 2.7715329700044315e-08,
      "loss": 0.7031,
      "step": 518
    },
    {
      "epoch": 1.9447047797563262,
      "grad_norm": 2.46077036857605,
      "learning_rate": 2.455319632118147e-08,
      "loss": 0.5426,
      "step": 519
    },
    {
      "epoch": 1.9484536082474226,
      "grad_norm": 2.4496285915374756,
      "learning_rate": 2.158209863804217e-08,
      "loss": 0.5547,
      "step": 520
    },
    {
      "epoch": 1.9522024367385193,
      "grad_norm": 2.5419907569885254,
      "learning_rate": 1.8802150727962876e-08,
      "loss": 0.5622,
      "step": 521
    },
    {
      "epoch": 1.9559512652296158,
      "grad_norm": 2.629879951477051,
      "learning_rate": 1.6213459328950355e-08,
      "loss": 0.6756,
      "step": 522
    },
    {
      "epoch": 1.9597000937207123,
      "grad_norm": 2.748396396636963,
      "learning_rate": 1.3816123835588835e-08,
      "loss": 0.5403,
      "step": 523
    },
    {
      "epoch": 1.9634489222118088,
      "grad_norm": 2.49360728263855,
      "learning_rate": 1.161023629522029e-08,
      "loss": 0.735,
      "step": 524
    },
    {
      "epoch": 1.9671977507029053,
      "grad_norm": 2.472134828567505,
      "learning_rate": 9.595881404411145e-09,
      "loss": 0.5665,
      "step": 525
    },
    {
      "epoch": 1.9709465791940017,
      "grad_norm": 2.7294187545776367,
      "learning_rate": 7.773136505700995e-09,
      "loss": 0.8548,
      "step": 526
    },
    {
      "epoch": 1.9746954076850984,
      "grad_norm": 2.79746675491333,
      "learning_rate": 6.142071584630538e-09,
      "loss": 0.7157,
      "step": 527
    },
    {
      "epoch": 1.978444236176195,
      "grad_norm": 2.4330995082855225,
      "learning_rate": 4.702749267057604e-09,
      "loss": 0.5992,
      "step": 528
    },
    {
      "epoch": 1.9821930646672916,
      "grad_norm": 2.6152493953704834,
      "learning_rate": 3.4552248167507576e-09,
      "loss": 0.6798,
      "step": 529
    },
    {
      "epoch": 1.985941893158388,
      "grad_norm": 2.38468861579895,
      "learning_rate": 2.3995461332676496e-09,
      "loss": 0.5985,
      "step": 530
    },
    {
      "epoch": 1.9896907216494846,
      "grad_norm": 2.6697652339935303,
      "learning_rate": 1.5357537501159425e-09,
      "loss": 0.5737,
      "step": 531
    },
    {
      "epoch": 1.993439550140581,
      "grad_norm": 2.0764174461364746,
      "learning_rate": 8.638808331973281e-10,
      "loss": 0.5335,
      "step": 532
    },
    {
      "epoch": 1.9971883786316775,
      "grad_norm": 2.559661626815796,
      "learning_rate": 3.8395317953354717e-10,
      "loss": 0.6069,
      "step": 533
    },
    {
      "epoch": 2.0,
      "grad_norm": 3.018411636352539,
      "learning_rate": 9.598921627607116e-11,
      "loss": 0.5042,
      "step": 534
    }
  ],
  "logging_steps": 1,
  "max_steps": 534,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.243263714603827e+16,
  "train_batch_size": 6,
  "trial_name": null,
  "trial_params": null
}