{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9997706947947719,
  "eval_steps": 500,
  "global_step": 2180,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00045861041045631735,
      "grad_norm": 500.8235978096947,
      "learning_rate": 0.0,
      "loss": 15.2987,
      "step": 1
    },
    {
      "epoch": 0.0009172208209126347,
      "grad_norm": 530.1425289034529,
      "learning_rate": 1.5151515151515153e-05,
      "loss": 15.3231,
      "step": 2
    },
    {
      "epoch": 0.001375831231368952,
      "grad_norm": 128.36622186975148,
      "learning_rate": 3.0303030303030306e-05,
      "loss": 12.8925,
      "step": 3
    },
    {
      "epoch": 0.0018344416418252694,
      "grad_norm": 105.65124474589697,
      "learning_rate": 4.545454545454546e-05,
      "loss": 12.3566,
      "step": 4
    },
    {
      "epoch": 0.002293052052281587,
      "grad_norm": 62.52417083275855,
      "learning_rate": 6.060606060606061e-05,
      "loss": 11.6828,
      "step": 5
    },
    {
      "epoch": 0.002751662462737904,
      "grad_norm": 29.058807499434394,
      "learning_rate": 7.575757575757576e-05,
      "loss": 11.1277,
      "step": 6
    },
    {
      "epoch": 0.0032102728731942215,
      "grad_norm": 22.772626084386864,
      "learning_rate": 9.090909090909092e-05,
      "loss": 11.0044,
      "step": 7
    },
    {
      "epoch": 0.0036688832836505388,
      "grad_norm": 16.546494361598675,
      "learning_rate": 0.00010606060606060606,
      "loss": 10.4259,
      "step": 8
    },
    {
      "epoch": 0.0041274936941068565,
      "grad_norm": 13.733717493630028,
      "learning_rate": 0.00012121212121212122,
      "loss": 10.2983,
      "step": 9
    },
    {
      "epoch": 0.004586104104563174,
      "grad_norm": 12.2634413701008,
      "learning_rate": 0.00013636363636363637,
      "loss": 9.8439,
      "step": 10
    },
    {
      "epoch": 0.005044714515019491,
      "grad_norm": 26.807044018765815,
      "learning_rate": 0.00015151515151515152,
      "loss": 9.7451,
      "step": 11
    },
    {
      "epoch": 0.005503324925475808,
      "grad_norm": 15.660203598228732,
      "learning_rate": 0.00016666666666666666,
      "loss": 9.4652,
      "step": 12
    },
    {
      "epoch": 0.005961935335932126,
      "grad_norm": 11.677060876167385,
      "learning_rate": 0.00018181818181818183,
      "loss": 9.279,
      "step": 13
    },
    {
      "epoch": 0.006420545746388443,
      "grad_norm": 8.539046229061668,
      "learning_rate": 0.00019696969696969695,
      "loss": 9.0412,
      "step": 14
    },
    {
      "epoch": 0.00687915615684476,
      "grad_norm": 9.989528698949144,
      "learning_rate": 0.00021212121212121213,
      "loss": 8.887,
      "step": 15
    },
    {
      "epoch": 0.0073377665673010775,
      "grad_norm": 7.039231847658708,
      "learning_rate": 0.00022727272727272727,
      "loss": 8.5555,
      "step": 16
    },
    {
      "epoch": 0.007796376977757395,
      "grad_norm": 6.982728324514236,
      "learning_rate": 0.00024242424242424245,
      "loss": 8.3578,
      "step": 17
    },
    {
      "epoch": 0.008254987388213713,
      "grad_norm": 6.227356112267731,
      "learning_rate": 0.00025757575757575756,
      "loss": 8.0937,
      "step": 18
    },
    {
      "epoch": 0.00871359779867003,
      "grad_norm": 4.212307470050734,
      "learning_rate": 0.00027272727272727274,
      "loss": 7.9205,
      "step": 19
    },
    {
      "epoch": 0.009172208209126348,
      "grad_norm": 4.024952588409275,
      "learning_rate": 0.0002878787878787879,
      "loss": 8.1321,
      "step": 20
    },
    {
      "epoch": 0.009630818619582664,
      "grad_norm": 3.661471415582712,
      "learning_rate": 0.00030303030303030303,
      "loss": 8.0055,
      "step": 21
    },
    {
      "epoch": 0.010089429030038982,
      "grad_norm": 3.8535937839321366,
      "learning_rate": 0.0003181818181818182,
      "loss": 7.7857,
      "step": 22
    },
    {
      "epoch": 0.010548039440495299,
      "grad_norm": 3.6068425335511085,
      "learning_rate": 0.0003333333333333333,
      "loss": 7.5285,
      "step": 23
    },
    {
      "epoch": 0.011006649850951617,
      "grad_norm": 2.7276801521148633,
      "learning_rate": 0.0003484848484848485,
      "loss": 7.5538,
      "step": 24
    },
    {
      "epoch": 0.011465260261407933,
      "grad_norm": 3.1055955669012487,
      "learning_rate": 0.00036363636363636367,
      "loss": 7.4361,
      "step": 25
    },
    {
      "epoch": 0.011923870671864251,
      "grad_norm": 2.711386908567338,
      "learning_rate": 0.0003787878787878788,
      "loss": 7.2681,
      "step": 26
    },
    {
      "epoch": 0.01238248108232057,
      "grad_norm": 2.312034806616609,
      "learning_rate": 0.0003939393939393939,
      "loss": 7.3909,
      "step": 27
    },
    {
      "epoch": 0.012841091492776886,
      "grad_norm": 2.3632524693525707,
      "learning_rate": 0.00040909090909090913,
      "loss": 7.2023,
      "step": 28
    },
    {
      "epoch": 0.013299701903233204,
      "grad_norm": 2.1972807650574313,
      "learning_rate": 0.00042424242424242425,
      "loss": 7.255,
      "step": 29
    },
    {
      "epoch": 0.01375831231368952,
      "grad_norm": 1.944272438326777,
      "learning_rate": 0.0004393939393939394,
      "loss": 6.9357,
      "step": 30
    },
    {
      "epoch": 0.014216922724145839,
      "grad_norm": 2.035183078091153,
      "learning_rate": 0.00045454545454545455,
      "loss": 7.0679,
      "step": 31
    },
    {
      "epoch": 0.014675533134602155,
      "grad_norm": 2.1880109658541276,
      "learning_rate": 0.0004696969696969697,
      "loss": 7.1796,
      "step": 32
    },
    {
      "epoch": 0.015134143545058473,
      "grad_norm": 2.2248451603632424,
      "learning_rate": 0.0004848484848484849,
      "loss": 7.0048,
      "step": 33
    },
    {
      "epoch": 0.01559275395551479,
      "grad_norm": 2.2195422922186676,
      "learning_rate": 0.0005,
      "loss": 6.985,
      "step": 34
    },
    {
      "epoch": 0.016051364365971106,
      "grad_norm": 3.0534102374479883,
      "learning_rate": 0.0005151515151515151,
      "loss": 6.8155,
      "step": 35
    },
    {
      "epoch": 0.016509974776427426,
      "grad_norm": 1.8565913764486222,
      "learning_rate": 0.0005303030303030302,
      "loss": 6.6784,
      "step": 36
    },
    {
      "epoch": 0.016968585186883742,
      "grad_norm": 1.4886586731961193,
      "learning_rate": 0.0005454545454545455,
      "loss": 6.6864,
      "step": 37
    },
    {
      "epoch": 0.01742719559734006,
      "grad_norm": 2.150601977772751,
      "learning_rate": 0.0005606060606060606,
      "loss": 6.6843,
      "step": 38
    },
    {
      "epoch": 0.017885806007796375,
      "grad_norm": 1.5518494789603754,
      "learning_rate": 0.0005757575757575758,
      "loss": 6.7124,
      "step": 39
    },
    {
      "epoch": 0.018344416418252695,
      "grad_norm": 1.522156353984075,
      "learning_rate": 0.0005909090909090909,
      "loss": 6.6466,
      "step": 40
    },
    {
      "epoch": 0.01880302682870901,
      "grad_norm": 1.295675164689553,
      "learning_rate": 0.0006060606060606061,
      "loss": 6.3354,
      "step": 41
    },
    {
      "epoch": 0.019261637239165328,
      "grad_norm": 1.274439750061726,
      "learning_rate": 0.0006212121212121212,
      "loss": 6.4965,
      "step": 42
    },
    {
      "epoch": 0.019720247649621648,
      "grad_norm": 1.3276713332742238,
      "learning_rate": 0.0006363636363636364,
      "loss": 6.5981,
      "step": 43
    },
    {
      "epoch": 0.020178858060077964,
      "grad_norm": 1.697975510667755,
      "learning_rate": 0.0006515151515151515,
      "loss": 6.3244,
      "step": 44
    },
    {
      "epoch": 0.02063746847053428,
      "grad_norm": 1.8151840076555448,
      "learning_rate": 0.0006666666666666666,
      "loss": 6.3094,
      "step": 45
    },
    {
      "epoch": 0.021096078880990597,
      "grad_norm": 1.07427168152187,
      "learning_rate": 0.0006818181818181818,
      "loss": 6.4739,
      "step": 46
    },
    {
      "epoch": 0.021554689291446917,
      "grad_norm": 1.269850000566983,
      "learning_rate": 0.000696969696969697,
      "loss": 6.2562,
      "step": 47
    },
    {
      "epoch": 0.022013299701903234,
      "grad_norm": 1.3386009839953275,
      "learning_rate": 0.0007121212121212122,
      "loss": 6.1672,
      "step": 48
    },
    {
      "epoch": 0.02247191011235955,
      "grad_norm": 1.4666976126851203,
      "learning_rate": 0.0007272727272727273,
      "loss": 6.5116,
      "step": 49
    },
    {
      "epoch": 0.022930520522815866,
      "grad_norm": 1.155954941914217,
      "learning_rate": 0.0007424242424242425,
      "loss": 6.3174,
      "step": 50
    },
    {
      "epoch": 0.023389130933272186,
      "grad_norm": 1.570646696445572,
      "learning_rate": 0.0007575757575757576,
      "loss": 6.3375,
      "step": 51
    },
    {
      "epoch": 0.023847741343728503,
      "grad_norm": 1.8737473597585088,
      "learning_rate": 0.0007727272727272727,
      "loss": 6.454,
      "step": 52
    },
    {
      "epoch": 0.02430635175418482,
      "grad_norm": 1.4202859351482426,
      "learning_rate": 0.0007878787878787878,
      "loss": 6.3634,
      "step": 53
    },
    {
      "epoch": 0.02476496216464114,
      "grad_norm": 1.5812064758302293,
      "learning_rate": 0.000803030303030303,
      "loss": 6.2789,
      "step": 54
    },
    {
      "epoch": 0.025223572575097455,
      "grad_norm": 1.5696759229534418,
      "learning_rate": 0.0008181818181818183,
      "loss": 6.1896,
      "step": 55
    },
    {
      "epoch": 0.025682182985553772,
      "grad_norm": 1.7924148279452325,
      "learning_rate": 0.0008333333333333334,
      "loss": 6.2494,
      "step": 56
    },
    {
      "epoch": 0.02614079339601009,
      "grad_norm": 1.4721929089408983,
      "learning_rate": 0.0008484848484848485,
      "loss": 6.1472,
      "step": 57
    },
    {
      "epoch": 0.026599403806466408,
      "grad_norm": 1.291472779675897,
      "learning_rate": 0.0008636363636363636,
      "loss": 6.2573,
      "step": 58
    },
    {
      "epoch": 0.027058014216922725,
      "grad_norm": 1.6431541946561006,
      "learning_rate": 0.0008787878787878789,
      "loss": 6.026,
      "step": 59
    },
    {
      "epoch": 0.02751662462737904,
      "grad_norm": 1.832738458214297,
      "learning_rate": 0.000893939393939394,
      "loss": 6.1022,
      "step": 60
    },
    {
      "epoch": 0.027975235037835357,
      "grad_norm": 1.3372295235946035,
      "learning_rate": 0.0009090909090909091,
      "loss": 5.9215,
      "step": 61
    },
    {
      "epoch": 0.028433845448291677,
      "grad_norm": 1.3317306751353744,
      "learning_rate": 0.0009242424242424242,
      "loss": 5.9651,
      "step": 62
    },
    {
      "epoch": 0.028892455858747994,
      "grad_norm": 1.476266645885569,
      "learning_rate": 0.0009393939393939394,
      "loss": 6.0256,
      "step": 63
    },
    {
      "epoch": 0.02935106626920431,
      "grad_norm": 1.7963486719522104,
      "learning_rate": 0.0009545454545454546,
      "loss": 6.1604,
      "step": 64
    },
    {
      "epoch": 0.029809676679660627,
      "grad_norm": 1.692196555654964,
      "learning_rate": 0.0009696969696969698,
      "loss": 5.923,
      "step": 65
    },
    {
      "epoch": 0.030268287090116947,
      "grad_norm": 1.604768599222416,
      "learning_rate": 0.000984848484848485,
      "loss": 5.9746,
      "step": 66
    },
    {
      "epoch": 0.030726897500573263,
      "grad_norm": 1.489260450267019,
      "learning_rate": 0.001,
      "loss": 5.9249,
      "step": 67
    },
    {
      "epoch": 0.03118550791102958,
      "grad_norm": 1.8602231038915897,
      "learning_rate": 0.0009999994478847943,
      "loss": 5.8767,
      "step": 68
    },
    {
      "epoch": 0.0316441183214859,
      "grad_norm": 1.906210515910984,
      "learning_rate": 0.0009999977915403962,
      "loss": 5.9991,
      "step": 69
    },
    {
      "epoch": 0.03210272873194221,
      "grad_norm": 1.4044955170143287,
      "learning_rate": 0.0009999950309704639,
      "loss": 5.795,
      "step": 70
    },
    {
      "epoch": 0.03256133914239853,
      "grad_norm": 1.6690474579938042,
      "learning_rate": 0.000999991166181094,
      "loss": 5.6973,
      "step": 71
    },
    {
      "epoch": 0.03301994955285485,
      "grad_norm": 1.7387563356253941,
      "learning_rate": 0.0009999861971808216,
      "loss": 5.9579,
      "step": 72
    },
    {
      "epoch": 0.033478559963311165,
      "grad_norm": 1.647119589617736,
      "learning_rate": 0.0009999801239806208,
      "loss": 5.6924,
      "step": 73
    },
    {
      "epoch": 0.033937170373767485,
      "grad_norm": 1.695485969756941,
      "learning_rate": 0.0009999729465939035,
      "loss": 5.8175,
      "step": 74
    },
    {
      "epoch": 0.034395780784223805,
      "grad_norm": 1.752456404567851,
      "learning_rate": 0.0009999646650365212,
      "loss": 5.841,
      "step": 75
    },
    {
      "epoch": 0.03485439119468012,
      "grad_norm": 1.3891848489883054,
      "learning_rate": 0.0009999552793267634,
      "loss": 5.8428,
      "step": 76
    },
    {
      "epoch": 0.03531300160513644,
      "grad_norm": 1.7302613855293794,
      "learning_rate": 0.0009999447894853577,
      "loss": 5.8052,
      "step": 77
    },
    {
      "epoch": 0.03577161201559275,
      "grad_norm": 1.5788537811855505,
      "learning_rate": 0.0009999331955354708,
      "loss": 5.7943,
      "step": 78
    },
    {
      "epoch": 0.03623022242604907,
      "grad_norm": 1.6404852512422425,
      "learning_rate": 0.0009999204975027073,
      "loss": 5.73,
      "step": 79
    },
    {
      "epoch": 0.03668883283650539,
      "grad_norm": 1.6452741960810733,
      "learning_rate": 0.0009999066954151103,
      "loss": 5.8108,
      "step": 80
    },
    {
      "epoch": 0.0371474432469617,
      "grad_norm": 1.3684269171327565,
      "learning_rate": 0.0009998917893031614,
      "loss": 5.6272,
      "step": 81
    },
    {
      "epoch": 0.03760605365741802,
      "grad_norm": 1.8014892274558039,
      "learning_rate": 0.0009998757791997801,
      "loss": 5.7708,
      "step": 82
    },
    {
      "epoch": 0.03806466406787434,
      "grad_norm": 1.85921630061907,
      "learning_rate": 0.0009998586651403238,
      "loss": 5.7237,
      "step": 83
    },
    {
      "epoch": 0.038523274478330656,
      "grad_norm": 1.571762814888667,
      "learning_rate": 0.0009998404471625885,
      "loss": 5.6299,
      "step": 84
    },
    {
      "epoch": 0.038981884888786976,
      "grad_norm": 1.8186234119610767,
      "learning_rate": 0.0009998211253068078,
      "loss": 5.8495,
      "step": 85
    },
    {
      "epoch": 0.039440495299243296,
      "grad_norm": 1.5970364793177365,
      "learning_rate": 0.0009998006996156535,
      "loss": 5.8326,
      "step": 86
    },
    {
      "epoch": 0.03989910570969961,
      "grad_norm": 1.63753111646841,
      "learning_rate": 0.0009997791701342347,
      "loss": 5.4621,
      "step": 87
    },
    {
      "epoch": 0.04035771612015593,
      "grad_norm": 1.6904866844189892,
      "learning_rate": 0.0009997565369100983,
      "loss": 5.5846,
      "step": 88
    },
    {
      "epoch": 0.04081632653061224,
      "grad_norm": 1.5419718056721496,
      "learning_rate": 0.0009997327999932291,
      "loss": 5.6292,
      "step": 89
    },
    {
      "epoch": 0.04127493694106856,
      "grad_norm": 1.536998169158985,
      "learning_rate": 0.000999707959436049,
      "loss": 5.7319,
      "step": 90
    },
    {
      "epoch": 0.04173354735152488,
      "grad_norm": 1.7236554703560536,
      "learning_rate": 0.0009996820152934176,
      "loss": 5.518,
      "step": 91
    },
    {
      "epoch": 0.042192157761981194,
      "grad_norm": 1.29119780794461,
      "learning_rate": 0.000999654967622631,
      "loss": 5.4312,
      "step": 92
    },
    {
      "epoch": 0.042650768172437514,
      "grad_norm": 1.6792950326927643,
      "learning_rate": 0.0009996268164834238,
      "loss": 5.4495,
      "step": 93
    },
    {
      "epoch": 0.043109378582893834,
      "grad_norm": 1.5716354795524257,
      "learning_rate": 0.000999597561937966,
      "loss": 5.5929,
      "step": 94
    },
    {
      "epoch": 0.04356798899335015,
      "grad_norm": 1.5087928475638015,
      "learning_rate": 0.0009995672040508656,
      "loss": 5.5869,
      "step": 95
    },
    {
      "epoch": 0.04402659940380647,
      "grad_norm": 1.4367797864007965,
      "learning_rate": 0.0009995357428891662,
      "loss": 5.6922,
      "step": 96
    },
    {
      "epoch": 0.04448520981426279,
      "grad_norm": 1.7761820643736554,
      "learning_rate": 0.0009995031785223491,
      "loss": 5.6367,
      "step": 97
    },
    {
      "epoch": 0.0449438202247191,
      "grad_norm": 1.5252360224068895,
      "learning_rate": 0.000999469511022331,
      "loss": 5.5637,
      "step": 98
    },
    {
      "epoch": 0.04540243063517542,
      "grad_norm": 1.5069449132548929,
      "learning_rate": 0.0009994347404634657,
      "loss": 5.5684,
      "step": 99
    },
    {
      "epoch": 0.04586104104563173,
      "grad_norm": 1.9537790166664204,
      "learning_rate": 0.0009993988669225423,
      "loss": 5.6334,
      "step": 100
    },
    {
      "epoch": 0.04631965145608805,
      "grad_norm": 1.691840829588729,
      "learning_rate": 0.000999361890478786,
      "loss": 5.6113,
      "step": 101
    },
    {
      "epoch": 0.04677826186654437,
      "grad_norm": 1.522137967954074,
      "learning_rate": 0.0009993238112138583,
      "loss": 5.288,
      "step": 102
    },
    {
      "epoch": 0.047236872277000685,
      "grad_norm": 1.7823864622107355,
      "learning_rate": 0.0009992846292118554,
      "loss": 5.5018,
      "step": 103
    },
    {
      "epoch": 0.047695482687457005,
      "grad_norm": 1.6744389882663637,
      "learning_rate": 0.000999244344559309,
      "loss": 5.5402,
      "step": 104
    },
    {
      "epoch": 0.048154093097913325,
      "grad_norm": 1.4687884191297234,
      "learning_rate": 0.0009992029573451869,
      "loss": 5.4721,
      "step": 105
    },
    {
      "epoch": 0.04861270350836964,
      "grad_norm": 1.588410984328697,
      "learning_rate": 0.0009991604676608905,
      "loss": 5.5704,
      "step": 106
    },
    {
      "epoch": 0.04907131391882596,
      "grad_norm": 1.7492512949087844,
      "learning_rate": 0.0009991168756002568,
      "loss": 5.3806,
      "step": 107
    },
    {
      "epoch": 0.04952992432928228,
      "grad_norm": 1.4481776924717384,
      "learning_rate": 0.0009990721812595574,
      "loss": 5.3923,
      "step": 108
    },
    {
      "epoch": 0.04998853473973859,
      "grad_norm": 1.6081933942958604,
      "learning_rate": 0.0009990263847374976,
      "loss": 5.4137,
      "step": 109
    },
    {
      "epoch": 0.05044714515019491,
      "grad_norm": 1.4956690519132552,
      "learning_rate": 0.0009989794861352173,
      "loss": 5.554,
      "step": 110
    },
    {
      "epoch": 0.050905755560651224,
      "grad_norm": 1.4841257193454869,
      "learning_rate": 0.0009989314855562905,
      "loss": 5.4901,
      "step": 111
    },
    {
      "epoch": 0.051364365971107544,
      "grad_norm": 1.5198287537792725,
      "learning_rate": 0.0009988823831067245,
      "loss": 5.419,
      "step": 112
    },
    {
      "epoch": 0.051822976381563864,
      "grad_norm": 1.5921167134243557,
      "learning_rate": 0.0009988321788949597,
      "loss": 5.3917,
      "step": 113
    },
    {
      "epoch": 0.05228158679202018,
      "grad_norm": 1.555424915521857,
      "learning_rate": 0.0009987808730318709,
      "loss": 5.4934,
      "step": 114
    },
    {
      "epoch": 0.052740197202476496,
      "grad_norm": 1.3883048664407023,
      "learning_rate": 0.0009987284656307644,
      "loss": 5.5121,
      "step": 115
    },
    {
      "epoch": 0.053198807612932816,
      "grad_norm": 1.5992350600304945,
      "learning_rate": 0.0009986749568073802,
      "loss": 5.4976,
      "step": 116
    },
    {
      "epoch": 0.05365741802338913,
      "grad_norm": 1.5531352515333243,
      "learning_rate": 0.0009986203466798905,
      "loss": 5.5721,
      "step": 117
    },
    {
      "epoch": 0.05411602843384545,
      "grad_norm": 1.3719788205197496,
      "learning_rate": 0.0009985646353688996,
      "loss": 5.4128,
      "step": 118
    },
    {
      "epoch": 0.05457463884430177,
      "grad_norm": 1.4997488074778824,
      "learning_rate": 0.0009985078229974437,
      "loss": 5.4278,
      "step": 119
    },
    {
      "epoch": 0.05503324925475808,
      "grad_norm": 1.569975483489946,
      "learning_rate": 0.0009984499096909905,
      "loss": 5.4434,
      "step": 120
    },
    {
      "epoch": 0.0554918596652144,
      "grad_norm": 1.6070084069037456,
      "learning_rate": 0.0009983908955774397,
      "loss": 5.3144,
      "step": 121
    },
    {
      "epoch": 0.055950470075670715,
      "grad_norm": 1.446026517437592,
      "learning_rate": 0.0009983307807871211,
      "loss": 5.392,
      "step": 122
    },
    {
      "epoch": 0.056409080486127035,
      "grad_norm": 1.5263986346323988,
      "learning_rate": 0.0009982695654527965,
      "loss": 5.3472,
      "step": 123
    },
    {
      "epoch": 0.056867690896583355,
      "grad_norm": 1.3599223817484805,
      "learning_rate": 0.0009982072497096571,
      "loss": 5.4112,
      "step": 124
    },
    {
      "epoch": 0.05732630130703967,
      "grad_norm": 1.6361725494415436,
      "learning_rate": 0.000998143833695325,
      "loss": 5.3697,
      "step": 125
    },
    {
      "epoch": 0.05778491171749599,
      "grad_norm": 1.543784072196881,
      "learning_rate": 0.0009980793175498517,
      "loss": 5.442,
      "step": 126
    },
    {
      "epoch": 0.05824352212795231,
      "grad_norm": 1.449407263214115,
      "learning_rate": 0.000998013701415719,
      "loss": 5.2642,
      "step": 127
    },
    {
      "epoch": 0.05870213253840862,
      "grad_norm": 1.4739175790593781,
      "learning_rate": 0.0009979469854378372,
      "loss": 5.3125,
      "step": 128
    },
    {
      "epoch": 0.05916074294886494,
      "grad_norm": 1.4564322986816842,
      "learning_rate": 0.000997879169763546,
      "loss": 5.3334,
      "step": 129
    },
    {
      "epoch": 0.05961935335932125,
      "grad_norm": 1.5920314838256873,
      "learning_rate": 0.000997810254542614,
      "loss": 5.3517,
      "step": 130
    },
    {
      "epoch": 0.06007796376977757,
      "grad_norm": 1.4278925861521121,
      "learning_rate": 0.0009977402399272374,
      "loss": 5.0874,
      "step": 131
    },
    {
      "epoch": 0.06053657418023389,
      "grad_norm": 1.5084927351235025,
      "learning_rate": 0.0009976691260720407,
      "loss": 5.3606,
      "step": 132
    },
    {
      "epoch": 0.060995184590690206,
      "grad_norm": 1.425357053961966,
      "learning_rate": 0.0009975969131340763,
      "loss": 5.3231,
      "step": 133
    },
    {
      "epoch": 0.061453795001146526,
      "grad_norm": 1.511765173758619,
      "learning_rate": 0.0009975236012728236,
      "loss": 5.4045,
      "step": 134
    },
    {
      "epoch": 0.061912405411602846,
      "grad_norm": 1.6739378440266066,
      "learning_rate": 0.0009974491906501886,
      "loss": 5.3774,
      "step": 135
    },
    {
      "epoch": 0.06237101582205916,
      "grad_norm": 1.4872377960431686,
      "learning_rate": 0.0009973736814305049,
      "loss": 5.3085,
      "step": 136
    },
    {
      "epoch": 0.06282962623251548,
      "grad_norm": 1.3446614991448431,
      "learning_rate": 0.0009972970737805312,
      "loss": 5.3471,
      "step": 137
    },
    {
      "epoch": 0.0632882366429718,
      "grad_norm": 1.491353830662239,
      "learning_rate": 0.0009972193678694525,
      "loss": 5.2869,
      "step": 138
    },
    {
      "epoch": 0.06374684705342812,
      "grad_norm": 1.4616243079099107,
      "learning_rate": 0.0009971405638688794,
      "loss": 5.146,
      "step": 139
    },
    {
      "epoch": 0.06420545746388442,
      "grad_norm": 1.3786315144473942,
      "learning_rate": 0.0009970606619528475,
      "loss": 5.2828,
      "step": 140
    },
    {
      "epoch": 0.06466406787434074,
      "grad_norm": 1.5581675642330386,
      "learning_rate": 0.000996979662297817,
      "loss": 5.2569,
      "step": 141
    },
    {
      "epoch": 0.06512267828479706,
      "grad_norm": 1.4723357767652112,
      "learning_rate": 0.0009968975650826721,
      "loss": 5.4027,
      "step": 142
    },
    {
      "epoch": 0.06558128869525338,
      "grad_norm": 1.4967686622334508,
      "learning_rate": 0.000996814370488722,
      "loss": 5.1972,
      "step": 143
    },
    {
      "epoch": 0.0660398991057097,
      "grad_norm": 1.4000885423479519,
      "learning_rate": 0.000996730078699698,
      "loss": 5.2827,
      "step": 144
    },
    {
      "epoch": 0.06649850951616602,
      "grad_norm": 1.4533526270508321,
      "learning_rate": 0.0009966446899017558,
      "loss": 5.266,
      "step": 145
    },
    {
      "epoch": 0.06695711992662233,
      "grad_norm": 1.54309759205673,
      "learning_rate": 0.0009965582042834728,
      "loss": 5.2799,
      "step": 146
    },
    {
      "epoch": 0.06741573033707865,
      "grad_norm": 1.6809092580532956,
      "learning_rate": 0.0009964706220358492,
      "loss": 5.2786,
      "step": 147
    },
    {
      "epoch": 0.06787434074753497,
      "grad_norm": 1.5938870076585887,
      "learning_rate": 0.000996381943352307,
      "loss": 5.277,
      "step": 148
    },
    {
      "epoch": 0.06833295115799129,
      "grad_norm": 1.579215976861361,
      "learning_rate": 0.0009962921684286896,
      "loss": 5.3313,
      "step": 149
    },
    {
      "epoch": 0.06879156156844761,
      "grad_norm": 1.450520320397999,
      "learning_rate": 0.0009962012974632614,
      "loss": 5.2198,
      "step": 150
    },
    {
      "epoch": 0.06925017197890392,
      "grad_norm": 1.6713278490108294,
      "learning_rate": 0.0009961093306567075,
      "loss": 5.2371,
      "step": 151
    },
    {
      "epoch": 0.06970878238936024,
      "grad_norm": 1.5677019810363055,
      "learning_rate": 0.0009960162682121328,
      "loss": 5.2428,
      "step": 152
    },
    {
      "epoch": 0.07016739279981656,
      "grad_norm": 1.5336077502442873,
      "learning_rate": 0.0009959221103350623,
      "loss": 5.027,
      "step": 153
    },
    {
      "epoch": 0.07062600321027288,
      "grad_norm": 1.555003304251843,
      "learning_rate": 0.0009958268572334394,
      "loss": 5.2364,
      "step": 154
    },
    {
      "epoch": 0.0710846136207292,
      "grad_norm": 1.4388050732358586,
      "learning_rate": 0.0009957305091176274,
      "loss": 5.2243,
      "step": 155
    },
    {
      "epoch": 0.0715432240311855,
      "grad_norm": 1.8928954810078222,
      "learning_rate": 0.0009956330662004075,
      "loss": 5.3076,
      "step": 156
    },
    {
      "epoch": 0.07200183444164182,
      "grad_norm": 1.4548087325115564,
      "learning_rate": 0.0009955345286969779,
      "loss": 5.2662,
      "step": 157
    },
    {
      "epoch": 0.07246044485209814,
      "grad_norm": 1.4931234999162524,
      "learning_rate": 0.0009954348968249551,
      "loss": 5.2859,
      "step": 158
    },
    {
      "epoch": 0.07291905526255446,
      "grad_norm": 1.5815637992860334,
      "learning_rate": 0.0009953341708043724,
      "loss": 5.253,
      "step": 159
    },
    {
      "epoch": 0.07337766567301078,
      "grad_norm": 1.556159936020389,
      "learning_rate": 0.0009952323508576793,
      "loss": 5.2385,
      "step": 160
    },
    {
      "epoch": 0.0738362760834671,
      "grad_norm": 1.7654592005772478,
      "learning_rate": 0.0009951294372097406,
      "loss": 5.2803,
      "step": 161
    },
    {
      "epoch": 0.0742948864939234,
      "grad_norm": 1.5160388283255,
      "learning_rate": 0.0009950254300878378,
      "loss": 5.0799,
      "step": 162
    },
    {
      "epoch": 0.07475349690437973,
      "grad_norm": 1.364541107642498,
      "learning_rate": 0.000994920329721666,
      "loss": 5.3125,
      "step": 163
    },
    {
      "epoch": 0.07521210731483605,
      "grad_norm": 1.3465113150754295,
      "learning_rate": 0.0009948141363433356,
      "loss": 5.13,
      "step": 164
    },
    {
      "epoch": 0.07567071772529237,
      "grad_norm": 1.5968670388058748,
      "learning_rate": 0.00099470685018737,
      "loss": 5.0501,
      "step": 165
    },
    {
      "epoch": 0.07612932813574869,
      "grad_norm": 1.4925833396750643,
      "learning_rate": 0.0009945984714907073,
      "loss": 5.2924,
      "step": 166
    },
    {
      "epoch": 0.07658793854620499,
      "grad_norm": 1.420530338389358,
      "learning_rate": 0.000994489000492697,
      "loss": 5.3658,
      "step": 167
    },
    {
      "epoch": 0.07704654895666131,
      "grad_norm": 1.4386161728681028,
      "learning_rate": 0.0009943784374351016,
      "loss": 5.0137,
      "step": 168
    },
    {
      "epoch": 0.07750515936711763,
      "grad_norm": 1.421791410216098,
      "learning_rate": 0.0009942667825620951,
      "loss": 4.9307,
      "step": 169
    },
    {
      "epoch": 0.07796376977757395,
      "grad_norm": 1.431314390880854,
      "learning_rate": 0.0009941540361202634,
      "loss": 5.223,
      "step": 170
    },
    {
      "epoch": 0.07842238018803027,
      "grad_norm": 1.7016078169229105,
      "learning_rate": 0.0009940401983586022,
      "loss": 5.2077,
      "step": 171
    },
    {
      "epoch": 0.07888099059848659,
      "grad_norm": 1.363691673282607,
      "learning_rate": 0.000993925269528518,
      "loss": 5.3011,
      "step": 172
    },
    {
      "epoch": 0.0793396010089429,
      "grad_norm": 1.3482736431057105,
      "learning_rate": 0.0009938092498838265,
      "loss": 5.2797,
      "step": 173
    },
    {
      "epoch": 0.07979821141939922,
      "grad_norm": 1.6150766235681897,
      "learning_rate": 0.0009936921396807524,
      "loss": 5.0613,
      "step": 174
    },
    {
      "epoch": 0.08025682182985554,
      "grad_norm": 1.4493152632540773,
      "learning_rate": 0.0009935739391779292,
      "loss": 5.1108,
      "step": 175
    },
    {
      "epoch": 0.08071543224031186,
      "grad_norm": 1.3529772465275312,
      "learning_rate": 0.000993454648636398,
      "loss": 5.1434,
      "step": 176
    },
    {
      "epoch": 0.08117404265076818,
      "grad_norm": 1.6258837298220254,
      "learning_rate": 0.0009933342683196074,
      "loss": 5.1592,
      "step": 177
    },
    {
      "epoch": 0.08163265306122448,
      "grad_norm": 1.396306385354903,
      "learning_rate": 0.0009932127984934125,
      "loss": 5.3019,
      "step": 178
    },
    {
      "epoch": 0.0820912634716808,
      "grad_norm": 1.6093989076505226,
      "learning_rate": 0.0009930902394260745,
      "loss": 5.1071,
      "step": 179
    },
    {
      "epoch": 0.08254987388213712,
      "grad_norm": 1.7692716220679523,
      "learning_rate": 0.0009929665913882607,
      "loss": 5.122,
      "step": 180
    },
    {
      "epoch": 0.08300848429259344,
      "grad_norm": 1.3984263002756976,
      "learning_rate": 0.0009928418546530425,
      "loss": 5.2333,
      "step": 181
    },
    {
      "epoch": 0.08346709470304976,
      "grad_norm": 1.5304497846642438,
      "learning_rate": 0.0009927160294958964,
      "loss": 5.1822,
      "step": 182
    },
    {
      "epoch": 0.08392570511350608,
      "grad_norm": 1.3284483389792503,
      "learning_rate": 0.000992589116194702,
      "loss": 5.061,
      "step": 183
    },
    {
      "epoch": 0.08438431552396239,
      "grad_norm": 1.3859921365051426,
      "learning_rate": 0.000992461115029743,
      "loss": 5.1642,
      "step": 184
    },
    {
      "epoch": 0.08484292593441871,
      "grad_norm": 1.5225089042991213,
      "learning_rate": 0.000992332026283704,
      "loss": 5.1369,
      "step": 185
    },
    {
      "epoch": 0.08530153634487503,
      "grad_norm": 1.5019431975772009,
      "learning_rate": 0.0009922018502416736,
      "loss": 5.0734,
      "step": 186
    },
    {
      "epoch": 0.08576014675533135,
      "grad_norm": 1.3916466628680189,
      "learning_rate": 0.0009920705871911395,
      "loss": 5.1101,
      "step": 187
    },
    {
      "epoch": 0.08621875716578767,
      "grad_norm": 1.5827392606840116,
      "learning_rate": 0.0009919382374219915,
      "loss": 5.0452,
      "step": 188
    },
    {
      "epoch": 0.08667736757624397,
      "grad_norm": 1.3840872309420276,
      "learning_rate": 0.0009918048012265187,
      "loss": 5.217,
      "step": 189
    },
    {
      "epoch": 0.0871359779867003,
      "grad_norm": 1.3202890726979277,
      "learning_rate": 0.0009916702788994097,
      "loss": 5.0911,
      "step": 190
    },
    {
      "epoch": 0.08759458839715661,
      "grad_norm": 1.447452080127782,
      "learning_rate": 0.0009915346707377519,
      "loss": 5.2523,
      "step": 191
    },
    {
      "epoch": 0.08805319880761293,
      "grad_norm": 1.5715855970029937,
      "learning_rate": 0.0009913979770410305,
      "loss": 5.2517,
      "step": 192
    },
    {
      "epoch": 0.08851180921806925,
      "grad_norm": 1.409185638203683,
      "learning_rate": 0.0009912601981111285,
      "loss": 5.2432,
      "step": 193
    },
    {
      "epoch": 0.08897041962852557,
      "grad_norm": 1.602834131574966,
      "learning_rate": 0.0009911213342523248,
      "loss": 5.18,
      "step": 194
    },
    {
      "epoch": 0.08942903003898188,
      "grad_norm": 1.4850769190342132,
      "learning_rate": 0.000990981385771295,
      "loss": 5.1421,
      "step": 195
    },
    {
      "epoch": 0.0898876404494382,
      "grad_norm": 1.6729844959662046,
      "learning_rate": 0.00099084035297711,
      "loss": 5.1511,
      "step": 196
    },
    {
      "epoch": 0.09034625085989452,
      "grad_norm": 1.4327840766354687,
      "learning_rate": 0.000990698236181235,
      "loss": 5.035,
      "step": 197
    },
    {
      "epoch": 0.09080486127035084,
      "grad_norm": 1.6134524670186894,
      "learning_rate": 0.0009905550356975293,
      "loss": 5.1142,
      "step": 198
    },
    {
      "epoch": 0.09126347168080716,
      "grad_norm": 1.6059390073131903,
      "learning_rate": 0.0009904107518422457,
      "loss": 5.0963,
      "step": 199
    },
    {
      "epoch": 0.09172208209126347,
      "grad_norm": 1.5036286865820736,
      "learning_rate": 0.0009902653849340295,
      "loss": 5.1975,
      "step": 200
    },
    {
      "epoch": 0.09218069250171979,
      "grad_norm": 1.51744297093448,
      "learning_rate": 0.0009901189352939177,
      "loss": 5.1859,
      "step": 201
    },
    {
      "epoch": 0.0926393029121761,
      "grad_norm": 1.5850325055279009,
      "learning_rate": 0.0009899714032453387,
      "loss": 5.0212,
      "step": 202
    },
    {
      "epoch": 0.09309791332263243,
      "grad_norm": 1.5105397533271556,
      "learning_rate": 0.000989822789114111,
      "loss": 5.0111,
      "step": 203
    },
    {
      "epoch": 0.09355652373308875,
      "grad_norm": 1.6862792769924022,
      "learning_rate": 0.0009896730932284434,
      "loss": 5.1736,
      "step": 204
    },
    {
      "epoch": 0.09401513414354506,
      "grad_norm": 1.4714414749291926,
      "learning_rate": 0.0009895223159189332,
      "loss": 5.1407,
      "step": 205
    },
    {
      "epoch": 0.09447374455400137,
      "grad_norm": 1.6174941048706324,
      "learning_rate": 0.0009893704575185663,
      "loss": 5.1093,
      "step": 206
    },
    {
      "epoch": 0.09493235496445769,
      "grad_norm": 1.3885463317156945,
      "learning_rate": 0.000989217518362716,
      "loss": 5.1668,
      "step": 207
    },
    {
      "epoch": 0.09539096537491401,
      "grad_norm": 1.4962134475234805,
      "learning_rate": 0.0009890634987891425,
      "loss": 5.2263,
      "step": 208
    },
    {
      "epoch": 0.09584957578537033,
      "grad_norm": 1.5030741984621436,
      "learning_rate": 0.0009889083991379917,
      "loss": 5.2433,
      "step": 209
    },
    {
      "epoch": 0.09630818619582665,
      "grad_norm": 1.428333348606357,
      "learning_rate": 0.0009887522197517954,
      "loss": 5.2611,
      "step": 210
    },
    {
      "epoch": 0.09676679660628296,
      "grad_norm": 1.32357703332451,
      "learning_rate": 0.0009885949609754693,
      "loss": 5.1428,
      "step": 211
    },
    {
      "epoch": 0.09722540701673928,
      "grad_norm": 1.5006213527950498,
      "learning_rate": 0.000988436623156314,
      "loss": 4.9511,
      "step": 212
    },
    {
      "epoch": 0.0976840174271956,
      "grad_norm": 1.3704780204071454,
      "learning_rate": 0.0009882772066440114,
      "loss": 4.8037,
      "step": 213
    },
    {
      "epoch": 0.09814262783765192,
      "grad_norm": 1.3711425726368134,
      "learning_rate": 0.0009881167117906276,
      "loss": 5.1257,
      "step": 214
    },
    {
      "epoch": 0.09860123824810824,
      "grad_norm": 1.3707758977949238,
      "learning_rate": 0.0009879551389506084,
      "loss": 5.2023,
      "step": 215
    },
    {
      "epoch": 0.09905984865856456,
      "grad_norm": 1.6392677255924117,
      "learning_rate": 0.0009877924884807814,
      "loss": 5.1024,
      "step": 216
    },
    {
      "epoch": 0.09951845906902086,
      "grad_norm": 1.3275193369113956,
      "learning_rate": 0.000987628760740354,
      "loss": 5.0945,
      "step": 217
    },
    {
      "epoch": 0.09997706947947718,
      "grad_norm": 1.4180131949928432,
      "learning_rate": 0.0009874639560909118,
      "loss": 4.983,
      "step": 218
    },
    {
      "epoch": 0.1004356798899335,
      "grad_norm": 1.7393584150019836,
      "learning_rate": 0.0009872980748964202,
      "loss": 4.9676,
      "step": 219
    },
    {
      "epoch": 0.10089429030038982,
      "grad_norm": 1.549752057244901,
      "learning_rate": 0.000987131117523221,
      "loss": 5.0521,
      "step": 220
    },
    {
      "epoch": 0.10135290071084614,
      "grad_norm": 1.4403540251951896,
      "learning_rate": 0.000986963084340033,
      "loss": 4.9813,
      "step": 221
    },
    {
      "epoch": 0.10181151112130245,
      "grad_norm": 1.751055358798769,
      "learning_rate": 0.0009867939757179508,
      "loss": 5.0719,
      "step": 222
    },
    {
      "epoch": 0.10227012153175877,
      "grad_norm": 1.328474358582597,
      "learning_rate": 0.0009866237920304443,
      "loss": 4.9842,
      "step": 223
    },
    {
      "epoch": 0.10272873194221509,
      "grad_norm": 1.280231351473476,
      "learning_rate": 0.0009864525336533577,
      "loss": 4.9983,
      "step": 224
    },
    {
      "epoch": 0.10318734235267141,
      "grad_norm": 1.5705547134100584,
      "learning_rate": 0.000986280200964908,
      "loss": 5.0618,
      "step": 225
    },
    {
      "epoch": 0.10364595276312773,
      "grad_norm": 1.5653960652110968,
      "learning_rate": 0.0009861067943456856,
      "loss": 5.1115,
      "step": 226
    },
    {
      "epoch": 0.10410456317358405,
      "grad_norm": 1.2338779201578804,
      "learning_rate": 0.000985932314178652,
      "loss": 4.9369,
      "step": 227
    },
    {
      "epoch": 0.10456317358404035,
      "grad_norm": 1.2874937156586375,
      "learning_rate": 0.00098575676084914,
      "loss": 5.1417,
      "step": 228
    },
    {
      "epoch": 0.10502178399449667,
      "grad_norm": 1.3398416000604758,
      "learning_rate": 0.0009855801347448518,
      "loss": 4.9879,
      "step": 229
    },
    {
      "epoch": 0.10548039440495299,
      "grad_norm": 1.2492989302639697,
      "learning_rate": 0.0009854024362558596,
      "loss": 5.0279,
      "step": 230
    },
    {
      "epoch": 0.10593900481540931,
      "grad_norm": 1.383605853371909,
      "learning_rate": 0.0009852236657746035,
      "loss": 4.8481,
      "step": 231
    },
    {
      "epoch": 0.10639761522586563,
      "grad_norm": 1.5190174930781977,
      "learning_rate": 0.0009850438236958911,
      "loss": 5.1486,
      "step": 232
    },
    {
      "epoch": 0.10685622563632194,
      "grad_norm": 1.4552310543213898,
      "learning_rate": 0.0009848629104168966,
      "loss": 5.1803,
      "step": 233
    },
    {
      "epoch": 0.10731483604677826,
      "grad_norm": 1.3670851001538669,
      "learning_rate": 0.00098468092633716,
      "loss": 5.0475,
      "step": 234
    },
    {
      "epoch": 0.10777344645723458,
      "grad_norm": 1.4104988205107645,
      "learning_rate": 0.0009844978718585855,
      "loss": 5.039,
      "step": 235
    },
    {
      "epoch": 0.1082320568676909,
      "grad_norm": 1.5531130083146936,
      "learning_rate": 0.0009843137473854423,
      "loss": 5.0075,
      "step": 236
    },
    {
      "epoch": 0.10869066727814722,
      "grad_norm": 1.4795241840289297,
      "learning_rate": 0.000984128553324362,
      "loss": 5.0136,
      "step": 237
    },
    {
      "epoch": 0.10914927768860354,
      "grad_norm": 1.6555422730566642,
      "learning_rate": 0.0009839422900843383,
      "loss": 5.1997,
      "step": 238
    },
    {
      "epoch": 0.10960788809905984,
      "grad_norm": 1.2880863033522636,
      "learning_rate": 0.0009837549580767261,
      "loss": 4.952,
      "step": 239
    },
    {
      "epoch": 0.11006649850951616,
      "grad_norm": 1.328615702409932,
      "learning_rate": 0.0009835665577152411,
      "loss": 5.0828,
      "step": 240
    },
    {
      "epoch": 0.11052510891997248,
      "grad_norm": 1.433950409711686,
      "learning_rate": 0.000983377089415958,
      "loss": 5.1355,
      "step": 241
    },
    {
      "epoch": 0.1109837193304288,
      "grad_norm": 1.4304378291589837,
      "learning_rate": 0.0009831865535973102,
      "loss": 5.0595,
      "step": 242
    },
    {
      "epoch": 0.11144232974088512,
      "grad_norm": 1.230905201118474,
      "learning_rate": 0.0009829949506800885,
      "loss": 5.0261,
      "step": 243
    },
    {
      "epoch": 0.11190094015134143,
      "grad_norm": 1.27148954109162,
      "learning_rate": 0.0009828022810874405,
      "loss": 4.8725,
      "step": 244
    },
    {
      "epoch": 0.11235955056179775,
      "grad_norm": 1.3164467600184036,
      "learning_rate": 0.0009826085452448693,
      "loss": 5.0485,
      "step": 245
    },
    {
      "epoch": 0.11281816097225407,
      "grad_norm": 1.3824245681908924,
      "learning_rate": 0.000982413743580233,
      "loss": 4.9107,
      "step": 246
    },
    {
      "epoch": 0.11327677138271039,
      "grad_norm": 1.5741449509164165,
      "learning_rate": 0.0009822178765237436,
      "loss": 5.0937,
      "step": 247
    },
    {
      "epoch": 0.11373538179316671,
      "grad_norm": 1.3629557991680756,
      "learning_rate": 0.0009820209445079654,
      "loss": 4.9812,
      "step": 248
    },
    {
      "epoch": 0.11419399220362302,
      "grad_norm": 1.3382989128169784,
      "learning_rate": 0.0009818229479678158,
      "loss": 4.9219,
      "step": 249
    },
    {
      "epoch": 0.11465260261407934,
      "grad_norm": 1.2317282437132013,
      "learning_rate": 0.0009816238873405615,
      "loss": 5.0736,
      "step": 250
    },
    {
      "epoch": 0.11511121302453566,
      "grad_norm": 1.2477564929088334,
      "learning_rate": 0.0009814237630658207,
      "loss": 5.0939,
      "step": 251
    },
    {
      "epoch": 0.11556982343499198,
      "grad_norm": 1.3570021536172399,
      "learning_rate": 0.00098122257558556,
      "loss": 5.1593,
      "step": 252
    },
    {
      "epoch": 0.1160284338454483,
      "grad_norm": 1.3439478146419286,
      "learning_rate": 0.0009810203253440937,
      "loss": 5.1636,
      "step": 253
    },
    {
      "epoch": 0.11648704425590461,
      "grad_norm": 1.2609537251514615,
      "learning_rate": 0.0009808170127880837,
      "loss": 5.0711,
      "step": 254
    },
    {
      "epoch": 0.11694565466636092,
      "grad_norm": 1.2808352946280326,
      "learning_rate": 0.000980612638366538,
      "loss": 4.9418,
      "step": 255
    },
    {
      "epoch": 0.11740426507681724,
      "grad_norm": 1.3026951029003901,
      "learning_rate": 0.0009804072025308096,
      "loss": 4.9872,
      "step": 256
    },
    {
      "epoch": 0.11786287548727356,
      "grad_norm": 1.2506394836696617,
      "learning_rate": 0.000980200705734595,
      "loss": 4.8908,
      "step": 257
    },
    {
      "epoch": 0.11832148589772988,
      "grad_norm": 1.4627087509306085,
      "learning_rate": 0.0009799931484339344,
      "loss": 4.9864,
      "step": 258
    },
    {
      "epoch": 0.1187800963081862,
      "grad_norm": 1.3581525784429678,
      "learning_rate": 0.0009797845310872103,
      "loss": 4.8975,
      "step": 259
    },
    {
      "epoch": 0.1192387067186425,
      "grad_norm": 1.34225952162814,
      "learning_rate": 0.0009795748541551457,
      "loss": 5.1611,
      "step": 260
    },
    {
      "epoch": 0.11969731712909883,
      "grad_norm": 1.314905423232624,
      "learning_rate": 0.000979364118100804,
      "loss": 5.2406,
      "step": 261
    },
    {
      "epoch": 0.12015592753955515,
      "grad_norm": 1.3987597834711132,
      "learning_rate": 0.0009791523233895875,
      "loss": 5.0278,
      "step": 262
    },
    {
      "epoch": 0.12061453795001147,
      "grad_norm": 1.2873462153748099,
      "learning_rate": 0.0009789394704892364,
      "loss": 4.8972,
      "step": 263
    },
    {
      "epoch": 0.12107314836046779,
      "grad_norm": 1.6230910440946535,
      "learning_rate": 0.0009787255598698282,
      "loss": 5.2198,
      "step": 264
    },
    {
      "epoch": 0.1215317587709241,
      "grad_norm": 1.3586771094363017,
      "learning_rate": 0.0009785105920037758,
      "loss": 4.9581,
      "step": 265
    },
    {
      "epoch": 0.12199036918138041,
      "grad_norm": 1.1988577260451987,
      "learning_rate": 0.0009782945673658275,
      "loss": 4.9701,
      "step": 266
    },
    {
      "epoch": 0.12244897959183673,
      "grad_norm": 1.4089920040566026,
      "learning_rate": 0.0009780774864330654,
      "loss": 4.9676,
      "step": 267
    },
    {
      "epoch": 0.12290759000229305,
      "grad_norm": 1.429652811630171,
      "learning_rate": 0.000977859349684904,
      "loss": 5.0087,
      "step": 268
    },
    {
      "epoch": 0.12336620041274937,
      "grad_norm": 1.4660897461601616,
      "learning_rate": 0.00097764015760309,
      "loss": 4.965,
      "step": 269
    },
    {
      "epoch": 0.12382481082320569,
      "grad_norm": 1.4594388047265134,
      "learning_rate": 0.0009774199106717004,
      "loss": 4.8121,
      "step": 270
    },
    {
      "epoch": 0.124283421233662,
      "grad_norm": 1.2972728387608525,
      "learning_rate": 0.0009771986093771417,
      "loss": 4.887,
      "step": 271
    },
    {
      "epoch": 0.12474203164411832,
      "grad_norm": 1.3305111038894375,
      "learning_rate": 0.0009769762542081496,
      "loss": 4.9439,
      "step": 272
    },
    {
      "epoch": 0.12520064205457465,
      "grad_norm": 1.4848067107584129,
      "learning_rate": 0.000976752845655786,
      "loss": 4.8049,
      "step": 273
    },
    {
      "epoch": 0.12565925246503096,
      "grad_norm": 1.3208057424624866,
      "learning_rate": 0.0009765283842134411,
      "loss": 4.9654,
      "step": 274
    },
    {
      "epoch": 0.12611786287548726,
      "grad_norm": 1.3448498128508568,
      "learning_rate": 0.0009763028703768282,
      "loss": 4.9599,
      "step": 275
    },
    {
      "epoch": 0.1265764732859436,
      "grad_norm": 1.2693661435676187,
      "learning_rate": 0.0009760763046439862,
      "loss": 4.9818,
      "step": 276
    },
    {
      "epoch": 0.1270350836963999,
      "grad_norm": 1.2216893347428637,
      "learning_rate": 0.0009758486875152766,
      "loss": 4.9507,
      "step": 277
    },
    {
      "epoch": 0.12749369410685624,
      "grad_norm": 1.320676156050438,
      "learning_rate": 0.0009756200194933829,
      "loss": 5.0115,
      "step": 278
    },
    {
      "epoch": 0.12795230451731254,
      "grad_norm": 1.3390487072055126,
      "learning_rate": 0.0009753903010833094,
      "loss": 4.9455,
      "step": 279
    },
    {
      "epoch": 0.12841091492776885,
      "grad_norm": 1.2144396591298865,
      "learning_rate": 0.0009751595327923803,
      "loss": 5.1665,
      "step": 280
    },
    {
      "epoch": 0.12886952533822518,
      "grad_norm": 1.2985174718219066,
      "learning_rate": 0.0009749277151302382,
      "loss": 5.0882,
      "step": 281
    },
    {
      "epoch": 0.1293281357486815,
      "grad_norm": 1.2977275071511523,
      "learning_rate": 0.0009746948486088435,
      "loss": 4.9673,
      "step": 282
    },
    {
      "epoch": 0.12978674615913782,
      "grad_norm": 1.3056338355960142,
      "learning_rate": 0.0009744609337424727,
      "loss": 4.9567,
      "step": 283
    },
    {
      "epoch": 0.13024535656959413,
      "grad_norm": 1.179614564554638,
      "learning_rate": 0.0009742259710477177,
      "loss": 5.0762,
      "step": 284
    },
    {
      "epoch": 0.13070396698005043,
      "grad_norm": 1.484132010247195,
      "learning_rate": 0.0009739899610434841,
      "loss": 4.9205,
      "step": 285
    },
    {
      "epoch": 0.13116257739050677,
      "grad_norm": 1.3982835090820156,
      "learning_rate": 0.0009737529042509913,
      "loss": 4.8264,
      "step": 286
    },
    {
      "epoch": 0.13162118780096307,
      "grad_norm": 1.2284761537721443,
      "learning_rate": 0.0009735148011937693,
      "loss": 4.946,
      "step": 287
    },
    {
      "epoch": 0.1320797982114194,
      "grad_norm": 1.5088899523141377,
      "learning_rate": 0.00097327565239766,
      "loss": 4.976,
      "step": 288
    },
    {
      "epoch": 0.13253840862187571,
      "grad_norm": 1.2579631574511398,
      "learning_rate": 0.0009730354583908136,
      "loss": 4.9742,
      "step": 289
    },
    {
      "epoch": 0.13299701903233205,
      "grad_norm": 1.2573997423026015,
      "learning_rate": 0.0009727942197036895,
      "loss": 4.9442,
      "step": 290
    },
    {
      "epoch": 0.13345562944278835,
      "grad_norm": 1.2509388393101417,
      "learning_rate": 0.0009725519368690539,
      "loss": 4.9074,
      "step": 291
    },
    {
      "epoch": 0.13391423985324466,
      "grad_norm": 1.2045780516852502,
      "learning_rate": 0.0009723086104219787,
      "loss": 4.9314,
      "step": 292
    },
    {
      "epoch": 0.134372850263701,
      "grad_norm": 1.2481504063057227,
      "learning_rate": 0.0009720642408998409,
      "loss": 4.8446,
      "step": 293
    },
    {
      "epoch": 0.1348314606741573,
      "grad_norm": 1.2274733285290946,
      "learning_rate": 0.0009718188288423211,
      "loss": 4.9651,
      "step": 294
    },
    {
      "epoch": 0.13529007108461363,
      "grad_norm": 1.2391866952136221,
      "learning_rate": 0.0009715723747914022,
      "loss": 5.041,
      "step": 295
    },
    {
      "epoch": 0.13574868149506994,
      "grad_norm": 1.1742656712665467,
      "learning_rate": 0.0009713248792913685,
      "loss": 4.9224,
      "step": 296
    },
    {
      "epoch": 0.13620729190552625,
      "grad_norm": 1.3905000684104418,
      "learning_rate": 0.0009710763428888037,
      "loss": 5.0376,
      "step": 297
    },
    {
      "epoch": 0.13666590231598258,
      "grad_norm": 1.264482798094133,
      "learning_rate": 0.0009708267661325909,
      "loss": 5.0683,
      "step": 298
    },
    {
      "epoch": 0.13712451272643889,
      "grad_norm": 1.393718888914881,
      "learning_rate": 0.0009705761495739107,
      "loss": 4.8544,
      "step": 299
    },
    {
      "epoch": 0.13758312313689522,
      "grad_norm": 1.279169083538889,
      "learning_rate": 0.0009703244937662399,
      "loss": 4.9465,
      "step": 300
    },
    {
      "epoch": 0.13804173354735153,
      "grad_norm": 1.2198945785690876,
      "learning_rate": 0.0009700717992653505,
      "loss": 5.0768,
      "step": 301
    },
    {
      "epoch": 0.13850034395780783,
      "grad_norm": 1.2943111399169975,
      "learning_rate": 0.0009698180666293083,
      "loss": 4.9876,
      "step": 302
    },
    {
      "epoch": 0.13895895436826416,
      "grad_norm": 1.313798945687526,
      "learning_rate": 0.000969563296418472,
      "loss": 4.8754,
      "step": 303
    },
    {
      "epoch": 0.13941756477872047,
      "grad_norm": 1.2085820300440804,
      "learning_rate": 0.0009693074891954914,
      "loss": 4.9144,
      "step": 304
    },
    {
      "epoch": 0.1398761751891768,
      "grad_norm": 1.281017062045696,
      "learning_rate": 0.0009690506455253072,
      "loss": 4.8007,
      "step": 305
    },
    {
      "epoch": 0.1403347855996331,
      "grad_norm": 1.3200030528515496,
      "learning_rate": 0.0009687927659751481,
      "loss": 4.927,
      "step": 306
    },
    {
      "epoch": 0.14079339601008942,
      "grad_norm": 1.3118374903692733,
      "learning_rate": 0.0009685338511145312,
      "loss": 5.0351,
      "step": 307
    },
    {
      "epoch": 0.14125200642054575,
      "grad_norm": 1.2979477845998362,
      "learning_rate": 0.0009682739015152598,
      "loss": 4.9517,
      "step": 308
    },
    {
      "epoch": 0.14171061683100206,
      "grad_norm": 1.38047699037995,
      "learning_rate": 0.0009680129177514226,
      "loss": 4.9806,
      "step": 309
    },
    {
      "epoch": 0.1421692272414584,
      "grad_norm": 1.1949659769312908,
      "learning_rate": 0.0009677509003993915,
      "loss": 4.9377,
      "step": 310
    },
    {
      "epoch": 0.1426278376519147,
      "grad_norm": 1.300547132673643,
      "learning_rate": 0.0009674878500378221,
      "loss": 4.8501,
      "step": 311
    },
    {
      "epoch": 0.143086448062371,
      "grad_norm": 1.2380802001856297,
      "learning_rate": 0.0009672237672476505,
      "loss": 4.8392,
      "step": 312
    },
    {
      "epoch": 0.14354505847282734,
      "grad_norm": 1.24869561684098,
      "learning_rate": 0.0009669586526120935,
      "loss": 4.9414,
      "step": 313
    },
    {
      "epoch": 0.14400366888328364,
      "grad_norm": 1.2661889381327558,
      "learning_rate": 0.0009666925067166459,
      "loss": 4.9871,
      "step": 314
    },
    {
      "epoch": 0.14446227929373998,
      "grad_norm": 1.3188936055330538,
      "learning_rate": 0.000966425330149081,
      "loss": 4.8147,
      "step": 315
    },
    {
      "epoch": 0.14492088970419628,
      "grad_norm": 1.2535599404130728,
      "learning_rate": 0.0009661571234994475,
      "loss": 4.9327,
      "step": 316
    },
    {
      "epoch": 0.14537950011465262,
      "grad_norm": 1.3850534971135746,
      "learning_rate": 0.0009658878873600691,
      "loss": 5.1543,
      "step": 317
    },
    {
      "epoch": 0.14583811052510892,
      "grad_norm": 1.2666309848859247,
      "learning_rate": 0.0009656176223255438,
      "loss": 4.9261,
      "step": 318
    },
    {
      "epoch": 0.14629672093556523,
      "grad_norm": 1.19284941707845,
      "learning_rate": 0.000965346328992741,
      "loss": 4.9163,
      "step": 319
    },
    {
      "epoch": 0.14675533134602156,
      "grad_norm": 1.394277934299874,
      "learning_rate": 0.0009650740079608014,
      "loss": 4.8612,
      "step": 320
    },
    {
      "epoch": 0.14721394175647787,
      "grad_norm": 1.1283199312283065,
      "learning_rate": 0.0009648006598311353,
      "loss": 4.797,
      "step": 321
    },
    {
      "epoch": 0.1476725521669342,
      "grad_norm": 1.216916351667639,
      "learning_rate": 0.0009645262852074214,
      "loss": 4.9231,
      "step": 322
    },
    {
      "epoch": 0.1481311625773905,
      "grad_norm": 1.2152695239350666,
      "learning_rate": 0.0009642508846956053,
      "loss": 4.8606,
      "step": 323
    },
    {
      "epoch": 0.1485897729878468,
      "grad_norm": 1.3216426091597269,
      "learning_rate": 0.0009639744589038983,
      "loss": 5.0458,
      "step": 324
    },
    {
      "epoch": 0.14904838339830315,
      "grad_norm": 1.2274030558488738,
      "learning_rate": 0.0009636970084427759,
      "loss": 4.9767,
      "step": 325
    },
    {
      "epoch": 0.14950699380875945,
      "grad_norm": 1.1744767061773043,
      "learning_rate": 0.0009634185339249766,
      "loss": 4.864,
      "step": 326
    },
    {
      "epoch": 0.1499656042192158,
      "grad_norm": 1.2184363536237404,
      "learning_rate": 0.0009631390359655003,
      "loss": 4.7819,
      "step": 327
    },
    {
      "epoch": 0.1504242146296721,
      "grad_norm": 1.2159515853853755,
      "learning_rate": 0.0009628585151816074,
      "loss": 4.8658,
      "step": 328
    },
    {
      "epoch": 0.1508828250401284,
      "grad_norm": 1.0793293449302956,
      "learning_rate": 0.0009625769721928172,
      "loss": 4.7535,
      "step": 329
    },
    {
      "epoch": 0.15134143545058473,
      "grad_norm": 1.2409545534600503,
      "learning_rate": 0.0009622944076209061,
      "loss": 4.965,
      "step": 330
    },
    {
      "epoch": 0.15180004586104104,
      "grad_norm": 1.1815517029524967,
      "learning_rate": 0.0009620108220899071,
      "loss": 4.9136,
      "step": 331
    },
    {
      "epoch": 0.15225865627149737,
      "grad_norm": 1.191407713341987,
      "learning_rate": 0.0009617262162261075,
      "loss": 4.9001,
      "step": 332
    },
    {
      "epoch": 0.15271726668195368,
      "grad_norm": 1.27022011040861,
      "learning_rate": 0.0009614405906580486,
      "loss": 4.8963,
      "step": 333
    },
    {
      "epoch": 0.15317587709240998,
      "grad_norm": 1.1610490962874633,
      "learning_rate": 0.000961153946016523,
      "loss": 4.7165,
      "step": 334
    },
    {
      "epoch": 0.15363448750286632,
      "grad_norm": 1.203023105802456,
      "learning_rate": 0.000960866282934574,
      "loss": 4.8401,
      "step": 335
    },
    {
      "epoch": 0.15409309791332262,
      "grad_norm": 1.1818212087429278,
      "learning_rate": 0.0009605776020474945,
      "loss": 4.9058,
      "step": 336
    },
    {
      "epoch": 0.15455170832377896,
      "grad_norm": 1.1430686313521128,
      "learning_rate": 0.0009602879039928249,
      "loss": 4.9335,
      "step": 337
    },
    {
      "epoch": 0.15501031873423526,
      "grad_norm": 1.2425465934467872,
      "learning_rate": 0.0009599971894103521,
      "loss": 4.9437,
      "step": 338
    },
    {
      "epoch": 0.1554689291446916,
      "grad_norm": 1.195460703020855,
      "learning_rate": 0.0009597054589421077,
      "loss": 4.8697,
      "step": 339
    },
    {
      "epoch": 0.1559275395551479,
      "grad_norm": 1.224385043256869,
      "learning_rate": 0.0009594127132323669,
      "loss": 4.9258,
      "step": 340
    },
    {
      "epoch": 0.1563861499656042,
      "grad_norm": 1.3161131832572537,
      "learning_rate": 0.0009591189529276474,
      "loss": 4.8403,
      "step": 341
    },
    {
      "epoch": 0.15684476037606054,
      "grad_norm": 1.1756800040823898,
      "learning_rate": 0.0009588241786767072,
      "loss": 4.6657,
      "step": 342
    },
    {
      "epoch": 0.15730337078651685,
      "grad_norm": 1.3753692648587865,
      "learning_rate": 0.0009585283911305436,
      "loss": 4.9049,
      "step": 343
    },
    {
      "epoch": 0.15776198119697318,
      "grad_norm": 1.2400991118253437,
      "learning_rate": 0.000958231590942392,
      "loss": 4.9722,
      "step": 344
    },
    {
      "epoch": 0.1582205916074295,
      "grad_norm": 1.2600968938164434,
      "learning_rate": 0.0009579337787677238,
      "loss": 5.0056,
      "step": 345
    },
    {
      "epoch": 0.1586792020178858,
      "grad_norm": 1.2106463666851053,
      "learning_rate": 0.0009576349552642456,
      "loss": 4.649,
      "step": 346
    },
    {
      "epoch": 0.15913781242834213,
      "grad_norm": 1.378293578737432,
      "learning_rate": 0.0009573351210918975,
      "loss": 4.9612,
      "step": 347
    },
    {
      "epoch": 0.15959642283879844,
      "grad_norm": 1.1789117877443365,
      "learning_rate": 0.0009570342769128514,
      "loss": 4.7929,
      "step": 348
    },
    {
      "epoch": 0.16005503324925477,
      "grad_norm": 1.2872761776035453,
      "learning_rate": 0.0009567324233915099,
      "loss": 5.0087,
      "step": 349
    },
    {
      "epoch": 0.16051364365971107,
      "grad_norm": 1.391553962294976,
      "learning_rate": 0.0009564295611945047,
      "loss": 4.7731,
      "step": 350
    },
    {
      "epoch": 0.16097225407016738,
      "grad_norm": 1.2083928667158645,
      "learning_rate": 0.000956125690990695,
      "loss": 5.109,
      "step": 351
    },
    {
      "epoch": 0.16143086448062371,
      "grad_norm": 1.0889325417376887,
      "learning_rate": 0.0009558208134511665,
      "loss": 4.6748,
      "step": 352
    },
    {
      "epoch": 0.16188947489108002,
      "grad_norm": 1.175764356219315,
      "learning_rate": 0.0009555149292492289,
      "loss": 4.8482,
      "step": 353
    },
    {
      "epoch": 0.16234808530153635,
      "grad_norm": 1.2020190965005082,
      "learning_rate": 0.0009552080390604159,
      "loss": 4.8386,
      "step": 354
    },
    {
      "epoch": 0.16280669571199266,
      "grad_norm": 1.2362488243620078,
      "learning_rate": 0.0009549001435624823,
      "loss": 4.8488,
      "step": 355
    },
    {
      "epoch": 0.16326530612244897,
      "grad_norm": 1.305824843374311,
      "learning_rate": 0.0009545912434354029,
      "loss": 4.9478,
      "step": 356
    },
    {
      "epoch": 0.1637239165329053,
      "grad_norm": 1.3624165964136437,
      "learning_rate": 0.0009542813393613721,
      "loss": 4.8491,
      "step": 357
    },
    {
      "epoch": 0.1641825269433616,
      "grad_norm": 1.275051867900283,
      "learning_rate": 0.0009539704320248006,
      "loss": 4.6465,
      "step": 358
    },
    {
      "epoch": 0.16464113735381794,
      "grad_norm": 1.2849077757034413,
      "learning_rate": 0.0009536585221123151,
      "loss": 5.1152,
      "step": 359
    },
    {
      "epoch": 0.16509974776427425,
      "grad_norm": 1.3639483541181663,
      "learning_rate": 0.0009533456103127565,
      "loss": 4.8532,
      "step": 360
    },
    {
      "epoch": 0.16555835817473058,
      "grad_norm": 1.1229564688408953,
      "learning_rate": 0.000953031697317178,
      "loss": 4.9097,
      "step": 361
    },
    {
      "epoch": 0.16601696858518689,
      "grad_norm": 1.179124334575757,
      "learning_rate": 0.0009527167838188445,
      "loss": 4.9069,
      "step": 362
    },
    {
      "epoch": 0.1664755789956432,
      "grad_norm": 1.245110982243652,
      "learning_rate": 0.0009524008705132299,
      "loss": 4.7615,
      "step": 363
    },
    {
      "epoch": 0.16693418940609953,
      "grad_norm": 1.1323273162338836,
      "learning_rate": 0.0009520839580980166,
      "loss": 4.6172,
      "step": 364
    },
    {
      "epoch": 0.16739279981655583,
      "grad_norm": 1.2481197543349178,
      "learning_rate": 0.0009517660472730929,
      "loss": 4.8993,
      "step": 365
    },
    {
      "epoch": 0.16785141022701217,
      "grad_norm": 1.2054933763017088,
      "learning_rate": 0.0009514471387405526,
      "loss": 4.7432,
      "step": 366
    },
    {
      "epoch": 0.16831002063746847,
      "grad_norm": 1.2199872962200347,
      "learning_rate": 0.0009511272332046926,
      "loss": 4.8756,
      "step": 367
    },
    {
      "epoch": 0.16876863104792478,
      "grad_norm": 1.1885140686574114,
      "learning_rate": 0.0009508063313720119,
      "loss": 4.8135,
      "step": 368
    },
    {
      "epoch": 0.1692272414583811,
      "grad_norm": 1.289225739922563,
      "learning_rate": 0.0009504844339512095,
      "loss": 4.9991,
      "step": 369
    },
    {
      "epoch": 0.16968585186883742,
      "grad_norm": 1.1725107074082917,
      "learning_rate": 0.0009501615416531835,
      "loss": 4.8058,
      "step": 370
    },
    {
      "epoch": 0.17014446227929375,
      "grad_norm": 1.2209348687494346,
      "learning_rate": 0.0009498376551910285,
      "loss": 4.8854,
      "step": 371
    },
    {
      "epoch": 0.17060307268975006,
      "grad_norm": 1.1690963019787017,
      "learning_rate": 0.0009495127752800352,
      "loss": 4.9347,
      "step": 372
    },
    {
      "epoch": 0.17106168310020636,
      "grad_norm": 1.3826833835764778,
      "learning_rate": 0.0009491869026376882,
      "loss": 4.8435,
      "step": 373
    },
    {
      "epoch": 0.1715202935106627,
      "grad_norm": 1.219139203563034,
      "learning_rate": 0.0009488600379836648,
      "loss": 4.7652,
      "step": 374
    },
    {
      "epoch": 0.171978903921119,
      "grad_norm": 1.2993977387557027,
      "learning_rate": 0.0009485321820398321,
      "loss": 4.9064,
      "step": 375
    },
    {
      "epoch": 0.17243751433157534,
      "grad_norm": 1.1926983026189613,
      "learning_rate": 0.0009482033355302475,
      "loss": 4.8168,
      "step": 376
    },
    {
      "epoch": 0.17289612474203164,
      "grad_norm": 1.2487642229499785,
      "learning_rate": 0.0009478734991811556,
      "loss": 4.9777,
      "step": 377
    },
    {
      "epoch": 0.17335473515248795,
      "grad_norm": 1.1554203488218102,
      "learning_rate": 0.0009475426737209871,
      "loss": 4.7707,
      "step": 378
    },
    {
      "epoch": 0.17381334556294428,
      "grad_norm": 1.4328223668552231,
      "learning_rate": 0.000947210859880357,
      "loss": 4.8267,
      "step": 379
    },
    {
      "epoch": 0.1742719559734006,
      "grad_norm": 1.1426314317054624,
      "learning_rate": 0.0009468780583920631,
      "loss": 4.7296,
      "step": 380
    },
    {
      "epoch": 0.17473056638385692,
      "grad_norm": 1.2347562463724606,
      "learning_rate": 0.0009465442699910846,
      "loss": 4.9696,
      "step": 381
    },
    {
      "epoch": 0.17518917679431323,
      "grad_norm": 1.2745846403997165,
      "learning_rate": 0.0009462094954145801,
      "loss": 4.8009,
      "step": 382
    },
    {
      "epoch": 0.17564778720476956,
      "grad_norm": 1.235954066122123,
      "learning_rate": 0.0009458737354018859,
      "loss": 4.9097,
      "step": 383
    },
    {
      "epoch": 0.17610639761522587,
      "grad_norm": 1.1316320632305195,
      "learning_rate": 0.000945536990694515,
      "loss": 4.8537,
      "step": 384
    },
    {
      "epoch": 0.17656500802568217,
      "grad_norm": 1.2371959713787706,
      "learning_rate": 0.0009451992620361551,
      "loss": 4.7752,
      "step": 385
    },
    {
      "epoch": 0.1770236184361385,
      "grad_norm": 1.1775677287879704,
      "learning_rate": 0.0009448605501726664,
      "loss": 4.9249,
      "step": 386
    },
    {
      "epoch": 0.17748222884659481,
      "grad_norm": 1.1839540990559874,
      "learning_rate": 0.000944520855852081,
      "loss": 4.6169,
      "step": 387
    },
    {
      "epoch": 0.17794083925705115,
      "grad_norm": 1.2254315480411488,
      "learning_rate": 0.0009441801798246002,
      "loss": 4.8624,
      "step": 388
    },
    {
      "epoch": 0.17839944966750745,
      "grad_norm": 1.2919202735534043,
      "learning_rate": 0.0009438385228425939,
      "loss": 5.0522,
      "step": 389
    },
    {
      "epoch": 0.17885806007796376,
      "grad_norm": 1.200014539192207,
      "learning_rate": 0.0009434958856605982,
      "loss": 4.915,
      "step": 390
    },
    {
      "epoch": 0.1793166704884201,
      "grad_norm": 1.338871897432486,
      "learning_rate": 0.0009431522690353137,
      "loss": 4.7695,
      "step": 391
    },
    {
      "epoch": 0.1797752808988764,
      "grad_norm": 1.2257942281076104,
      "learning_rate": 0.0009428076737256044,
      "loss": 4.8602,
      "step": 392
    },
    {
      "epoch": 0.18023389130933273,
      "grad_norm": 1.2265033009086583,
      "learning_rate": 0.0009424621004924954,
      "loss": 4.6949,
      "step": 393
    },
    {
      "epoch": 0.18069250171978904,
      "grad_norm": 1.2476320002509662,
      "learning_rate": 0.0009421155500991719,
      "loss": 5.0161,
      "step": 394
    },
    {
      "epoch": 0.18115111213024535,
      "grad_norm": 1.183352952641163,
      "learning_rate": 0.0009417680233109767,
      "loss": 4.6925,
      "step": 395
    },
    {
      "epoch": 0.18160972254070168,
      "grad_norm": 1.1176263999112626,
      "learning_rate": 0.000941419520895409,
      "loss": 4.7631,
      "step": 396
    },
    {
      "epoch": 0.18206833295115799,
      "grad_norm": 1.201300438149257,
      "learning_rate": 0.0009410700436221229,
      "loss": 4.7675,
      "step": 397
    },
    {
      "epoch": 0.18252694336161432,
      "grad_norm": 1.175983197748345,
      "learning_rate": 0.0009407195922629252,
      "loss": 4.6006,
      "step": 398
    },
    {
      "epoch": 0.18298555377207062,
      "grad_norm": 1.1023567816020134,
      "learning_rate": 0.000940368167591774,
      "loss": 4.6853,
      "step": 399
    },
    {
      "epoch": 0.18344416418252693,
      "grad_norm": 1.3077442671524,
      "learning_rate": 0.0009400157703847769,
      "loss": 4.7302,
      "step": 400
    },
    {
      "epoch": 0.18390277459298326,
      "grad_norm": 1.2254794344277589,
      "learning_rate": 0.0009396624014201895,
      "loss": 4.7892,
      "step": 401
    },
    {
      "epoch": 0.18436138500343957,
      "grad_norm": 1.234727893555762,
      "learning_rate": 0.000939308061478413,
      "loss": 4.7958,
      "step": 402
    },
    {
      "epoch": 0.1848199954138959,
      "grad_norm": 1.3122641821309018,
      "learning_rate": 0.0009389527513419935,
      "loss": 4.8132,
      "step": 403
    },
    {
      "epoch": 0.1852786058243522,
      "grad_norm": 1.2472074608004775,
      "learning_rate": 0.0009385964717956195,
      "loss": 4.6504,
      "step": 404
    },
    {
      "epoch": 0.18573721623480852,
      "grad_norm": 1.238276264258856,
      "learning_rate": 0.0009382392236261201,
      "loss": 4.7575,
      "step": 405
    },
    {
      "epoch": 0.18619582664526485,
      "grad_norm": 1.2534577950725028,
      "learning_rate": 0.0009378810076224644,
      "loss": 4.7177,
      "step": 406
    },
    {
      "epoch": 0.18665443705572116,
      "grad_norm": 1.205271117307525,
      "learning_rate": 0.0009375218245757582,
      "loss": 4.686,
      "step": 407
    },
    {
      "epoch": 0.1871130474661775,
      "grad_norm": 1.2341926393223217,
      "learning_rate": 0.0009371616752792432,
      "loss": 4.7172,
      "step": 408
    },
    {
      "epoch": 0.1875716578766338,
      "grad_norm": 1.3106670104278353,
      "learning_rate": 0.0009368005605282949,
      "loss": 4.7099,
      "step": 409
    },
    {
      "epoch": 0.18803026828709013,
      "grad_norm": 1.275668352770907,
      "learning_rate": 0.0009364384811204212,
      "loss": 4.7871,
      "step": 410
    },
    {
      "epoch": 0.18848887869754644,
      "grad_norm": 1.3339297002755484,
      "learning_rate": 0.00093607543785526,
      "loss": 4.8057,
      "step": 411
    },
    {
      "epoch": 0.18894748910800274,
      "grad_norm": 1.4136744027221881,
      "learning_rate": 0.0009357114315345787,
      "loss": 4.7367,
      "step": 412
    },
    {
      "epoch": 0.18940609951845908,
      "grad_norm": 1.177588373911629,
      "learning_rate": 0.0009353464629622705,
      "loss": 4.6703,
      "step": 413
    },
    {
      "epoch": 0.18986470992891538,
      "grad_norm": 1.209289646278639,
      "learning_rate": 0.0009349805329443544,
      "loss": 4.6406,
      "step": 414
    },
    {
      "epoch": 0.19032332033937172,
      "grad_norm": 1.3841772416670355,
      "learning_rate": 0.0009346136422889724,
      "loss": 4.6975,
      "step": 415
    },
    {
      "epoch": 0.19078193074982802,
      "grad_norm": 1.1462265891338494,
      "learning_rate": 0.0009342457918063882,
      "loss": 4.6941,
      "step": 416
    },
    {
      "epoch": 0.19124054116028433,
      "grad_norm": 1.1980809030902486,
      "learning_rate": 0.0009338769823089853,
      "loss": 4.6861,
      "step": 417
    },
    {
      "epoch": 0.19169915157074066,
      "grad_norm": 1.3327045445027812,
      "learning_rate": 0.0009335072146112648,
      "loss": 4.8545,
      "step": 418
    },
    {
      "epoch": 0.19215776198119697,
      "grad_norm": 1.2666311988510415,
      "learning_rate": 0.0009331364895298444,
      "loss": 4.4241,
      "step": 419
    },
    {
      "epoch": 0.1926163723916533,
      "grad_norm": 1.3418403650368294,
      "learning_rate": 0.0009327648078834559,
      "loss": 4.7763,
      "step": 420
    },
    {
      "epoch": 0.1930749828021096,
      "grad_norm": 1.3831752345794146,
      "learning_rate": 0.0009323921704929434,
      "loss": 4.6932,
      "step": 421
    },
    {
      "epoch": 0.1935335932125659,
      "grad_norm": 1.3278626393188422,
      "learning_rate": 0.0009320185781812623,
      "loss": 4.4976,
      "step": 422
    },
    {
      "epoch": 0.19399220362302225,
      "grad_norm": 1.400017777158702,
      "learning_rate": 0.0009316440317734762,
      "loss": 4.6445,
      "step": 423
    },
    {
      "epoch": 0.19445081403347855,
      "grad_norm": 1.3718529837955453,
      "learning_rate": 0.0009312685320967565,
      "loss": 4.8624,
      "step": 424
    },
    {
      "epoch": 0.1949094244439349,
      "grad_norm": 1.2919581870855115,
      "learning_rate": 0.0009308920799803793,
      "loss": 4.6193,
      "step": 425
    },
    {
      "epoch": 0.1953680348543912,
      "grad_norm": 1.3131386857656369,
      "learning_rate": 0.0009305146762557246,
      "loss": 4.6983,
      "step": 426
    },
    {
      "epoch": 0.1958266452648475,
      "grad_norm": 1.3153112644224645,
      "learning_rate": 0.0009301363217562736,
      "loss": 4.6751,
      "step": 427
    },
    {
      "epoch": 0.19628525567530383,
      "grad_norm": 1.3162466017474532,
      "learning_rate": 0.0009297570173176074,
      "loss": 4.6836,
      "step": 428
    },
    {
      "epoch": 0.19674386608576014,
      "grad_norm": 1.331198650251798,
      "learning_rate": 0.000929376763777405,
      "loss": 4.6916,
      "step": 429
    },
    {
      "epoch": 0.19720247649621647,
      "grad_norm": 1.3668831561444377,
      "learning_rate": 0.0009289955619754413,
      "loss": 4.7362,
      "step": 430
    },
    {
      "epoch": 0.19766108690667278,
      "grad_norm": 1.2412852055730121,
      "learning_rate": 0.0009286134127535859,
      "loss": 4.5877,
      "step": 431
    },
    {
      "epoch": 0.1981196973171291,
      "grad_norm": 1.2072918990099037,
      "learning_rate": 0.0009282303169558,
      "loss": 4.6405,
      "step": 432
    },
    {
      "epoch": 0.19857830772758542,
      "grad_norm": 1.3037414418627227,
      "learning_rate": 0.0009278462754281359,
      "loss": 4.7009,
      "step": 433
    },
    {
      "epoch": 0.19903691813804172,
      "grad_norm": 1.2247296879574305,
      "learning_rate": 0.0009274612890187342,
      "loss": 4.6489,
      "step": 434
    },
    {
      "epoch": 0.19949552854849806,
      "grad_norm": 1.3231280851260596,
      "learning_rate": 0.0009270753585778222,
      "loss": 4.4362,
      "step": 435
    },
    {
      "epoch": 0.19995413895895436,
      "grad_norm": 1.2418581987942559,
      "learning_rate": 0.0009266884849577124,
      "loss": 4.5499,
      "step": 436
    },
    {
      "epoch": 0.2004127493694107,
      "grad_norm": 1.294447632220281,
      "learning_rate": 0.0009263006690127998,
      "loss": 4.6676,
      "step": 437
    },
    {
      "epoch": 0.200871359779867,
      "grad_norm": 1.261856214073854,
      "learning_rate": 0.0009259119115995609,
      "loss": 4.4883,
      "step": 438
    },
    {
      "epoch": 0.2013299701903233,
      "grad_norm": 1.37251041386146,
      "learning_rate": 0.0009255222135765511,
      "loss": 4.666,
      "step": 439
    },
    {
      "epoch": 0.20178858060077964,
      "grad_norm": 1.2867902279023755,
      "learning_rate": 0.0009251315758044032,
      "loss": 4.6267,
      "step": 440
    },
    {
      "epoch": 0.20224719101123595,
      "grad_norm": 1.290310610910457,
      "learning_rate": 0.0009247399991458255,
      "loss": 4.5622,
      "step": 441
    },
    {
      "epoch": 0.20270580142169228,
      "grad_norm": 1.2273441067324191,
      "learning_rate": 0.0009243474844655996,
      "loss": 4.4213,
      "step": 442
    },
    {
      "epoch": 0.2031644118321486,
      "grad_norm": 1.1830219640006625,
      "learning_rate": 0.0009239540326305791,
      "loss": 4.689,
      "step": 443
    },
    {
      "epoch": 0.2036230222426049,
      "grad_norm": 1.2994309933555794,
      "learning_rate": 0.0009235596445096864,
      "loss": 4.6328,
      "step": 444
    },
    {
      "epoch": 0.20408163265306123,
      "grad_norm": 1.3295092948809324,
      "learning_rate": 0.0009231643209739127,
      "loss": 4.5755,
      "step": 445
    },
    {
      "epoch": 0.20454024306351754,
      "grad_norm": 1.1409192816085072,
      "learning_rate": 0.0009227680628963145,
      "loss": 4.6169,
      "step": 446
    },
    {
      "epoch": 0.20499885347397387,
      "grad_norm": 1.4461863872186738,
      "learning_rate": 0.000922370871152012,
      "loss": 4.635,
      "step": 447
    },
    {
      "epoch": 0.20545746388443017,
      "grad_norm": 1.3084516056633826,
      "learning_rate": 0.0009219727466181877,
      "loss": 4.5676,
      "step": 448
    },
    {
      "epoch": 0.20591607429488648,
      "grad_norm": 1.270894184635985,
      "learning_rate": 0.0009215736901740841,
      "loss": 4.6732,
      "step": 449
    },
    {
      "epoch": 0.20637468470534281,
      "grad_norm": 1.3192960174021482,
      "learning_rate": 0.0009211737027010016,
      "loss": 4.5502,
      "step": 450
    },
    {
      "epoch": 0.20683329511579912,
      "grad_norm": 1.201087913835216,
      "learning_rate": 0.0009207727850822971,
      "loss": 4.6429,
      "step": 451
    },
    {
      "epoch": 0.20729190552625545,
      "grad_norm": 1.2190161859072517,
      "learning_rate": 0.0009203709382033814,
      "loss": 4.4715,
      "step": 452
    },
    {
      "epoch": 0.20775051593671176,
      "grad_norm": 1.1946873783310576,
      "learning_rate": 0.0009199681629517173,
      "loss": 4.4923,
      "step": 453
    },
    {
      "epoch": 0.2082091263471681,
      "grad_norm": 1.2174522914059893,
      "learning_rate": 0.0009195644602168184,
      "loss": 4.4989,
      "step": 454
    },
    {
      "epoch": 0.2086677367576244,
      "grad_norm": 1.293409857892351,
      "learning_rate": 0.0009191598308902464,
      "loss": 4.5353,
      "step": 455
    },
    {
      "epoch": 0.2091263471680807,
      "grad_norm": 1.2408232139602717,
      "learning_rate": 0.0009187542758656091,
      "loss": 4.3718,
      "step": 456
    },
    {
      "epoch": 0.20958495757853704,
      "grad_norm": 1.2024063248986145,
      "learning_rate": 0.0009183477960385591,
      "loss": 4.5791,
      "step": 457
    },
    {
      "epoch": 0.21004356798899335,
      "grad_norm": 1.2533843508151217,
      "learning_rate": 0.0009179403923067912,
      "loss": 4.3455,
      "step": 458
    },
    {
      "epoch": 0.21050217839944968,
      "grad_norm": 1.3259189277402823,
      "learning_rate": 0.0009175320655700406,
      "loss": 4.4051,
      "step": 459
    },
    {
      "epoch": 0.21096078880990599,
      "grad_norm": 1.2954666507852552,
      "learning_rate": 0.0009171228167300805,
      "loss": 4.5125,
      "step": 460
    },
    {
      "epoch": 0.2114193992203623,
      "grad_norm": 1.2477344000877666,
      "learning_rate": 0.0009167126466907215,
      "loss": 4.5161,
      "step": 461
    },
    {
      "epoch": 0.21187800963081863,
      "grad_norm": 1.2122987725230499,
      "learning_rate": 0.0009163015563578074,
      "loss": 4.7014,
      "step": 462
    },
    {
      "epoch": 0.21233662004127493,
      "grad_norm": 1.2016051263155831,
      "learning_rate": 0.0009158895466392158,
      "loss": 4.4286,
      "step": 463
    },
    {
      "epoch": 0.21279523045173127,
      "grad_norm": 1.2474167610661386,
      "learning_rate": 0.0009154766184448535,
      "loss": 4.2447,
      "step": 464
    },
    {
      "epoch": 0.21325384086218757,
      "grad_norm": 1.3062377523858775,
      "learning_rate": 0.0009150627726866568,
      "loss": 4.3421,
      "step": 465
    },
    {
      "epoch": 0.21371245127264388,
      "grad_norm": 1.29316435152623,
      "learning_rate": 0.000914648010278587,
      "loss": 4.5971,
      "step": 466
    },
    {
      "epoch": 0.2141710616831002,
      "grad_norm": 1.3263034909743527,
      "learning_rate": 0.0009142323321366315,
      "loss": 4.6431,
      "step": 467
    },
    {
      "epoch": 0.21462967209355652,
      "grad_norm": 1.2065761133980615,
      "learning_rate": 0.0009138157391787986,
      "loss": 4.394,
      "step": 468
    },
    {
      "epoch": 0.21508828250401285,
      "grad_norm": 1.2124137478436883,
      "learning_rate": 0.0009133982323251177,
      "loss": 4.6036,
      "step": 469
    },
    {
      "epoch": 0.21554689291446916,
      "grad_norm": 1.1546038648632737,
      "learning_rate": 0.0009129798124976365,
      "loss": 4.5034,
      "step": 470
    },
    {
      "epoch": 0.21600550332492546,
      "grad_norm": 1.2259757527653632,
      "learning_rate": 0.0009125604806204187,
      "loss": 4.2603,
      "step": 471
    },
    {
      "epoch": 0.2164641137353818,
      "grad_norm": 1.2008265851893363,
      "learning_rate": 0.0009121402376195421,
      "loss": 4.5321,
      "step": 472
    },
    {
      "epoch": 0.2169227241458381,
      "grad_norm": 1.147802580436936,
      "learning_rate": 0.0009117190844230972,
      "loss": 4.3386,
      "step": 473
    },
    {
      "epoch": 0.21738133455629444,
      "grad_norm": 1.1746536508188052,
      "learning_rate": 0.0009112970219611841,
      "loss": 4.4297,
      "step": 474
    },
    {
      "epoch": 0.21783994496675074,
      "grad_norm": 1.1466508903649502,
      "learning_rate": 0.0009108740511659115,
      "loss": 4.4338,
      "step": 475
    },
    {
      "epoch": 0.21829855537720708,
      "grad_norm": 1.1738382154973588,
      "learning_rate": 0.0009104501729713935,
      "loss": 4.366,
      "step": 476
    },
    {
      "epoch": 0.21875716578766338,
      "grad_norm": 1.2440285684614694,
      "learning_rate": 0.0009100253883137488,
      "loss": 4.4605,
      "step": 477
    },
    {
      "epoch": 0.2192157761981197,
      "grad_norm": 1.2854807257114729,
      "learning_rate": 0.0009095996981310974,
      "loss": 4.4282,
      "step": 478
    },
    {
      "epoch": 0.21967438660857602,
      "grad_norm": 1.175611567689229,
      "learning_rate": 0.0009091731033635596,
      "loss": 4.4131,
      "step": 479
    },
    {
      "epoch": 0.22013299701903233,
      "grad_norm": 1.2506540472347665,
      "learning_rate": 0.0009087456049532529,
      "loss": 4.4381,
      "step": 480
    },
    {
      "epoch": 0.22059160742948866,
      "grad_norm": 1.184427604979077,
      "learning_rate": 0.0009083172038442914,
      "loss": 4.2899,
      "step": 481
    },
    {
      "epoch": 0.22105021783994497,
      "grad_norm": 1.1984965692495606,
      "learning_rate": 0.0009078879009827817,
      "loss": 4.4617,
      "step": 482
    },
    {
      "epoch": 0.22150882825040127,
      "grad_norm": 1.1074058276734418,
      "learning_rate": 0.0009074576973168223,
      "loss": 4.3417,
      "step": 483
    },
    {
      "epoch": 0.2219674386608576,
      "grad_norm": 1.2216087262762645,
      "learning_rate": 0.0009070265937965015,
      "loss": 4.4027,
      "step": 484
    },
    {
      "epoch": 0.2224260490713139,
      "grad_norm": 1.1485631612064169,
      "learning_rate": 0.0009065945913738942,
      "loss": 4.4045,
      "step": 485
    },
    {
      "epoch": 0.22288465948177025,
      "grad_norm": 1.112411286685907,
      "learning_rate": 0.0009061616910030609,
      "loss": 4.3131,
      "step": 486
    },
    {
      "epoch": 0.22334326989222655,
      "grad_norm": 1.0835970551777978,
      "learning_rate": 0.0009057278936400453,
      "loss": 4.3783,
      "step": 487
    },
    {
      "epoch": 0.22380188030268286,
      "grad_norm": 1.1224243794002595,
      "learning_rate": 0.0009052932002428715,
      "loss": 4.4078,
      "step": 488
    },
    {
      "epoch": 0.2242604907131392,
      "grad_norm": 1.169532483007231,
      "learning_rate": 0.0009048576117715435,
      "loss": 4.3475,
      "step": 489
    },
    {
      "epoch": 0.2247191011235955,
      "grad_norm": 1.1653105482612955,
      "learning_rate": 0.0009044211291880407,
      "loss": 4.1715,
      "step": 490
    },
    {
      "epoch": 0.22517771153405183,
      "grad_norm": 1.1490974346677685,
      "learning_rate": 0.000903983753456318,
      "loss": 4.3779,
      "step": 491
    },
    {
      "epoch": 0.22563632194450814,
      "grad_norm": 1.1657273744556513,
      "learning_rate": 0.0009035454855423026,
      "loss": 4.304,
      "step": 492
    },
    {
      "epoch": 0.22609493235496445,
      "grad_norm": 1.1006962599036976,
      "learning_rate": 0.0009031063264138922,
      "loss": 4.3758,
      "step": 493
    },
    {
      "epoch": 0.22655354276542078,
      "grad_norm": 1.1858925578588162,
      "learning_rate": 0.0009026662770409522,
      "loss": 4.4575,
      "step": 494
    },
    {
      "epoch": 0.22701215317587708,
      "grad_norm": 1.1434375420337357,
      "learning_rate": 0.0009022253383953147,
      "loss": 4.3312,
      "step": 495
    },
    {
      "epoch": 0.22747076358633342,
      "grad_norm": 1.0755295498158388,
      "learning_rate": 0.0009017835114507753,
      "loss": 4.343,
      "step": 496
    },
    {
      "epoch": 0.22792937399678972,
      "grad_norm": 1.1721069297162419,
      "learning_rate": 0.0009013407971830914,
      "loss": 4.5632,
      "step": 497
    },
    {
      "epoch": 0.22838798440724603,
      "grad_norm": 1.1647421441707269,
      "learning_rate": 0.0009008971965699801,
      "loss": 4.4018,
      "step": 498
    },
    {
      "epoch": 0.22884659481770236,
      "grad_norm": 1.1035713430702432,
      "learning_rate": 0.0009004527105911163,
      "loss": 4.3141,
      "step": 499
    },
    {
      "epoch": 0.22930520522815867,
      "grad_norm": 1.1195196147468476,
      "learning_rate": 0.0009000073402281295,
      "loss": 4.2632,
      "step": 500
    },
    {
      "epoch": 0.229763815638615,
      "grad_norm": 1.119396799536596,
      "learning_rate": 0.0008995610864646028,
      "loss": 4.2405,
      "step": 501
    },
    {
      "epoch": 0.2302224260490713,
      "grad_norm": 1.1080689091634248,
      "learning_rate": 0.0008991139502860703,
      "loss": 4.1858,
      "step": 502
    },
    {
      "epoch": 0.23068103645952764,
      "grad_norm": 1.1334052195946764,
      "learning_rate": 0.0008986659326800146,
      "loss": 4.315,
      "step": 503
    },
    {
      "epoch": 0.23113964686998395,
      "grad_norm": 1.1562584074462194,
      "learning_rate": 0.0008982170346358651,
      "loss": 4.2802,
      "step": 504
    },
    {
      "epoch": 0.23159825728044026,
      "grad_norm": 1.118713326042857,
      "learning_rate": 0.0008977672571449956,
      "loss": 4.5768,
      "step": 505
    },
    {
      "epoch": 0.2320568676908966,
      "grad_norm": 1.1656627656262073,
      "learning_rate": 0.0008973166012007217,
      "loss": 4.2393,
      "step": 506
    },
    {
      "epoch": 0.2325154781013529,
      "grad_norm": 1.1868418074110105,
      "learning_rate": 0.0008968650677982998,
      "loss": 4.3134,
      "step": 507
    },
    {
      "epoch": 0.23297408851180923,
      "grad_norm": 1.0230452713220064,
      "learning_rate": 0.0008964126579349236,
      "loss": 4.2201,
      "step": 508
    },
    {
      "epoch": 0.23343269892226554,
      "grad_norm": 1.1466554736629961,
      "learning_rate": 0.0008959593726097226,
      "loss": 4.2408,
      "step": 509
    },
    {
      "epoch": 0.23389130933272184,
      "grad_norm": 1.1999913337948418,
      "learning_rate": 0.0008955052128237596,
      "loss": 4.4559,
      "step": 510
    },
    {
      "epoch": 0.23434991974317818,
      "grad_norm": 1.084825250603039,
      "learning_rate": 0.0008950501795800288,
      "loss": 4.2339,
      "step": 511
    },
    {
      "epoch": 0.23480853015363448,
      "grad_norm": 1.0200620371214644,
      "learning_rate": 0.0008945942738834532,
      "loss": 4.3438,
      "step": 512
    },
    {
      "epoch": 0.23526714056409082,
      "grad_norm": 1.1121615685347008,
      "learning_rate": 0.0008941374967408826,
      "loss": 4.3041,
      "step": 513
    },
    {
      "epoch": 0.23572575097454712,
      "grad_norm": 1.0964785066086764,
      "learning_rate": 0.0008936798491610916,
      "loss": 4.308,
      "step": 514
    },
    {
      "epoch": 0.23618436138500343,
      "grad_norm": 1.0511266967473323,
      "learning_rate": 0.0008932213321547768,
      "loss": 4.2828,
      "step": 515
    },
    {
      "epoch": 0.23664297179545976,
      "grad_norm": 1.1794731201131547,
      "learning_rate": 0.0008927619467345554,
      "loss": 4.2632,
      "step": 516
    },
    {
      "epoch": 0.23710158220591607,
      "grad_norm": 1.0974498895033717,
      "learning_rate": 0.0008923016939149615,
      "loss": 4.1129,
      "step": 517
    },
    {
      "epoch": 0.2375601926163724,
      "grad_norm": 1.0614747445976656,
      "learning_rate": 0.0008918405747124458,
      "loss": 4.2661,
      "step": 518
    },
    {
      "epoch": 0.2380188030268287,
      "grad_norm": 1.034346531475499,
      "learning_rate": 0.0008913785901453721,
      "loss": 4.2532,
      "step": 519
    },
    {
      "epoch": 0.238477413437285,
      "grad_norm": 1.1957325781566033,
      "learning_rate": 0.000890915741234015,
      "loss": 4.3519,
      "step": 520
    },
    {
      "epoch": 0.23893602384774135,
      "grad_norm": 1.1981949657005182,
      "learning_rate": 0.0008904520290005582,
      "loss": 4.4609,
      "step": 521
    },
    {
      "epoch": 0.23939463425819765,
      "grad_norm": 1.0516125039374238,
      "learning_rate": 0.000889987454469092,
      "loss": 4.2103,
      "step": 522
    },
    {
      "epoch": 0.239853244668654,
      "grad_norm": 1.1115019596118134,
      "learning_rate": 0.0008895220186656111,
      "loss": 4.1803,
      "step": 523
    },
    {
      "epoch": 0.2403118550791103,
      "grad_norm": 1.1690059222296931,
      "learning_rate": 0.0008890557226180122,
      "loss": 4.1815,
      "step": 524
    },
    {
      "epoch": 0.24077046548956663,
      "grad_norm": 1.1935343356575163,
      "learning_rate": 0.0008885885673560921,
      "loss": 4.3824,
      "step": 525
    },
    {
      "epoch": 0.24122907590002293,
      "grad_norm": 1.1191048566484874,
      "learning_rate": 0.0008881205539115444,
      "loss": 4.2452,
      "step": 526
    },
    {
      "epoch": 0.24168768631047924,
      "grad_norm": 1.189615832789953,
      "learning_rate": 0.0008876516833179589,
      "loss": 4.1762,
      "step": 527
    },
    {
      "epoch": 0.24214629672093557,
      "grad_norm": 1.183839728742806,
      "learning_rate": 0.0008871819566108177,
      "loss": 4.2307,
      "step": 528
    },
    {
      "epoch": 0.24260490713139188,
      "grad_norm": 1.0038823883768353,
      "learning_rate": 0.000886711374827494,
      "loss": 4.3608,
      "step": 529
    },
    {
      "epoch": 0.2430635175418482,
      "grad_norm": 1.2347446405775153,
      "learning_rate": 0.0008862399390072491,
      "loss": 4.2988,
      "step": 530
    },
    {
      "epoch": 0.24352212795230452,
      "grad_norm": 1.1240656618594,
      "learning_rate": 0.0008857676501912305,
      "loss": 4.2847,
      "step": 531
    },
    {
      "epoch": 0.24398073836276082,
      "grad_norm": 1.0532502156181462,
      "learning_rate": 0.0008852945094224697,
      "loss": 4.3115,
      "step": 532
    },
    {
      "epoch": 0.24443934877321716,
      "grad_norm": 1.1155896794146767,
      "learning_rate": 0.0008848205177458795,
      "loss": 4.1284,
      "step": 533
    },
    {
      "epoch": 0.24489795918367346,
      "grad_norm": 1.1544917026119283,
      "learning_rate": 0.0008843456762082518,
      "loss": 4.2477,
      "step": 534
    },
    {
      "epoch": 0.2453565695941298,
      "grad_norm": 1.1044969191816651,
      "learning_rate": 0.0008838699858582557,
      "loss": 4.3947,
      "step": 535
    },
    {
      "epoch": 0.2458151800045861,
      "grad_norm": 1.11789402845311,
      "learning_rate": 0.0008833934477464347,
      "loss": 4.3273,
      "step": 536
    },
    {
      "epoch": 0.2462737904150424,
      "grad_norm": 1.0516588727695435,
      "learning_rate": 0.0008829160629252045,
      "loss": 4.3527,
      "step": 537
    },
    {
      "epoch": 0.24673240082549874,
      "grad_norm": 1.030828999609917,
      "learning_rate": 0.0008824378324488509,
      "loss": 4.3111,
      "step": 538
    },
    {
      "epoch": 0.24719101123595505,
      "grad_norm": 1.0204468019856126,
      "learning_rate": 0.0008819587573735268,
      "loss": 4.0818,
      "step": 539
    },
    {
      "epoch": 0.24764962164641138,
      "grad_norm": 1.0720264453812243,
      "learning_rate": 0.0008814788387572513,
      "loss": 4.2501,
      "step": 540
    },
    {
      "epoch": 0.2481082320568677,
      "grad_norm": 1.109165933798806,
      "learning_rate": 0.0008809980776599053,
      "loss": 4.1074,
      "step": 541
    },
    {
      "epoch": 0.248566842467324,
      "grad_norm": 1.0424313942393764,
      "learning_rate": 0.0008805164751432312,
      "loss": 4.3814,
      "step": 542
    },
    {
      "epoch": 0.24902545287778033,
      "grad_norm": 1.0128395415612328,
      "learning_rate": 0.0008800340322708292,
      "loss": 4.2029,
      "step": 543
    },
    {
      "epoch": 0.24948406328823663,
      "grad_norm": 1.0742515541616493,
      "learning_rate": 0.0008795507501081555,
      "loss": 4.1352,
      "step": 544
    },
    {
      "epoch": 0.24994267369869297,
      "grad_norm": 1.1803037150323459,
      "learning_rate": 0.0008790666297225196,
      "loss": 4.3352,
      "step": 545
    },
    {
      "epoch": 0.2504012841091493,
      "grad_norm": 1.158062842350423,
      "learning_rate": 0.0008785816721830829,
      "loss": 4.3672,
      "step": 546
    },
    {
      "epoch": 0.2508598945196056,
      "grad_norm": 1.0956776453649357,
      "learning_rate": 0.0008780958785608546,
      "loss": 4.2164,
      "step": 547
    },
    {
      "epoch": 0.2513185049300619,
      "grad_norm": 1.1151333627868287,
      "learning_rate": 0.0008776092499286912,
      "loss": 4.1957,
      "step": 548
    },
    {
      "epoch": 0.2517771153405182,
      "grad_norm": 1.1123889676819365,
      "learning_rate": 0.0008771217873612929,
      "loss": 4.4252,
      "step": 549
    },
    {
      "epoch": 0.2522357257509745,
      "grad_norm": 0.9941429637476723,
      "learning_rate": 0.0008766334919352017,
      "loss": 4.2068,
      "step": 550
    },
    {
      "epoch": 0.2526943361614309,
      "grad_norm": 1.034840015024924,
      "learning_rate": 0.0008761443647287987,
      "loss": 4.2517,
      "step": 551
    },
    {
      "epoch": 0.2531529465718872,
      "grad_norm": 1.0896412975858896,
      "learning_rate": 0.0008756544068223026,
      "loss": 4.2092,
      "step": 552
    },
    {
      "epoch": 0.2536115569823435,
      "grad_norm": 1.0934775362224043,
      "learning_rate": 0.0008751636192977659,
      "loss": 4.3747,
      "step": 553
    },
    {
      "epoch": 0.2540701673927998,
      "grad_norm": 1.0576406469235335,
      "learning_rate": 0.0008746720032390737,
      "loss": 4.2482,
      "step": 554
    },
    {
      "epoch": 0.2545287778032561,
      "grad_norm": 1.0076055804088937,
      "learning_rate": 0.0008741795597319408,
      "loss": 4.2989,
      "step": 555
    },
    {
      "epoch": 0.2549873882137125,
      "grad_norm": 1.071486529213565,
      "learning_rate": 0.0008736862898639095,
      "loss": 4.3028,
      "step": 556
    },
    {
      "epoch": 0.2554459986241688,
      "grad_norm": 1.011480662875392,
      "learning_rate": 0.0008731921947243468,
      "loss": 4.3002,
      "step": 557
    },
    {
      "epoch": 0.2559046090346251,
      "grad_norm": 0.9654458470413098,
      "learning_rate": 0.0008726972754044427,
      "loss": 4.2676,
      "step": 558
    },
    {
      "epoch": 0.2563632194450814,
      "grad_norm": 1.0356582499507772,
      "learning_rate": 0.0008722015329972069,
      "loss": 4.2743,
      "step": 559
    },
    {
      "epoch": 0.2568218298555377,
      "grad_norm": 1.047498906297585,
      "learning_rate": 0.0008717049685974672,
      "loss": 4.1702,
      "step": 560
    },
    {
      "epoch": 0.25728044026599406,
      "grad_norm": 1.056147430750636,
      "learning_rate": 0.0008712075833018665,
      "loss": 4.1461,
      "step": 561
    },
    {
      "epoch": 0.25773905067645037,
      "grad_norm": 1.0562757423504503,
      "learning_rate": 0.0008707093782088608,
      "loss": 4.2059,
      "step": 562
    },
    {
      "epoch": 0.25819766108690667,
      "grad_norm": 1.0371406464253377,
      "learning_rate": 0.0008702103544187167,
      "loss": 4.2526,
      "step": 563
    },
    {
      "epoch": 0.258656271497363,
      "grad_norm": 1.1119206157292203,
      "learning_rate": 0.0008697105130335085,
      "loss": 4.3078,
      "step": 564
    },
    {
      "epoch": 0.2591148819078193,
      "grad_norm": 1.103562419456214,
      "learning_rate": 0.0008692098551571164,
      "loss": 4.262,
      "step": 565
    },
    {
      "epoch": 0.25957349231827564,
      "grad_norm": 1.0974676396086707,
      "learning_rate": 0.0008687083818952235,
      "loss": 4.2658,
      "step": 566
    },
    {
      "epoch": 0.26003210272873195,
      "grad_norm": 1.2517409472163399,
      "learning_rate": 0.0008682060943553143,
      "loss": 4.2598,
      "step": 567
    },
    {
      "epoch": 0.26049071313918826,
      "grad_norm": 1.0433783182418965,
      "learning_rate": 0.0008677029936466707,
      "loss": 4.4366,
      "step": 568
    },
    {
      "epoch": 0.26094932354964456,
      "grad_norm": 1.096870000772142,
      "learning_rate": 0.0008671990808803711,
      "loss": 4.2739,
      "step": 569
    },
    {
      "epoch": 0.26140793396010087,
      "grad_norm": 1.06261921837877,
      "learning_rate": 0.0008666943571692871,
      "loss": 4.3157,
      "step": 570
    },
    {
      "epoch": 0.26186654437055723,
      "grad_norm": 1.0781422074644758,
      "learning_rate": 0.0008661888236280813,
      "loss": 4.019,
      "step": 571
    },
    {
      "epoch": 0.26232515478101354,
      "grad_norm": 1.0252831202872519,
      "learning_rate": 0.0008656824813732045,
      "loss": 4.2196,
      "step": 572
    },
    {
      "epoch": 0.26278376519146984,
      "grad_norm": 0.9688275300427008,
      "learning_rate": 0.000865175331522894,
      "loss": 4.0755,
      "step": 573
    },
    {
      "epoch": 0.26324237560192615,
      "grad_norm": 1.0225323658489383,
      "learning_rate": 0.0008646673751971703,
      "loss": 4.0792,
      "step": 574
    },
    {
      "epoch": 0.26370098601238245,
      "grad_norm": 1.053597194429668,
      "learning_rate": 0.000864158613517835,
      "loss": 4.3191,
      "step": 575
    },
    {
      "epoch": 0.2641595964228388,
      "grad_norm": 0.9116462378157888,
      "learning_rate": 0.0008636490476084681,
      "loss": 4.2123,
      "step": 576
    },
    {
      "epoch": 0.2646182068332951,
      "grad_norm": 1.0798714916072572,
      "learning_rate": 0.0008631386785944264,
      "loss": 4.3433,
      "step": 577
    },
    {
      "epoch": 0.26507681724375143,
      "grad_norm": 1.0462343943486259,
      "learning_rate": 0.0008626275076028397,
      "loss": 4.2935,
      "step": 578
    },
    {
      "epoch": 0.26553542765420773,
      "grad_norm": 1.0937610659720705,
      "learning_rate": 0.0008621155357626091,
      "loss": 4.1661,
      "step": 579
    },
    {
      "epoch": 0.2659940380646641,
      "grad_norm": 1.0433437578681346,
      "learning_rate": 0.0008616027642044042,
      "loss": 4.3693,
      "step": 580
    },
    {
      "epoch": 0.2664526484751204,
      "grad_norm": 1.0537332372453507,
      "learning_rate": 0.000861089194060661,
      "loss": 4.1704,
      "step": 581
    },
    {
      "epoch": 0.2669112588855767,
      "grad_norm": 1.0822110323677163,
      "learning_rate": 0.000860574826465579,
      "loss": 4.3492,
      "step": 582
    },
    {
      "epoch": 0.267369869296033,
      "grad_norm": 1.071792579403557,
      "learning_rate": 0.0008600596625551191,
      "loss": 4.2584,
      "step": 583
    },
    {
      "epoch": 0.2678284797064893,
      "grad_norm": 1.0047505479818484,
      "learning_rate": 0.0008595437034670006,
      "loss": 4.2164,
      "step": 584
    },
    {
      "epoch": 0.2682870901169457,
      "grad_norm": 1.0242366423488012,
      "learning_rate": 0.0008590269503406985,
      "loss": 4.2782,
      "step": 585
    },
    {
      "epoch": 0.268745700527402,
      "grad_norm": 0.9829709535946065,
      "learning_rate": 0.0008585094043174423,
      "loss": 4.274,
      "step": 586
    },
    {
      "epoch": 0.2692043109378583,
      "grad_norm": 1.0270566543689745,
      "learning_rate": 0.0008579910665402118,
      "loss": 4.3344,
      "step": 587
    },
    {
      "epoch": 0.2696629213483146,
      "grad_norm": 0.9475466316322786,
      "learning_rate": 0.000857471938153736,
      "loss": 4.2631,
      "step": 588
    },
    {
      "epoch": 0.2701215317587709,
      "grad_norm": 0.9877616172110739,
      "learning_rate": 0.0008569520203044892,
      "loss": 4.0978,
      "step": 589
    },
    {
      "epoch": 0.27058014216922727,
      "grad_norm": 1.0418886304840747,
      "learning_rate": 0.0008564313141406901,
      "loss": 4.2527,
      "step": 590
    },
    {
      "epoch": 0.2710387525796836,
      "grad_norm": 1.0118360897822969,
      "learning_rate": 0.0008559098208122973,
      "loss": 4.2336,
      "step": 591
    },
    {
      "epoch": 0.2714973629901399,
      "grad_norm": 1.0079566123412265,
      "learning_rate": 0.0008553875414710089,
      "loss": 4.0794,
      "step": 592
    },
    {
      "epoch": 0.2719559734005962,
      "grad_norm": 1.0414195641348358,
      "learning_rate": 0.0008548644772702579,
      "loss": 4.3232,
      "step": 593
    },
    {
      "epoch": 0.2724145838110525,
      "grad_norm": 1.0991076621544467,
      "learning_rate": 0.0008543406293652116,
      "loss": 4.2263,
      "step": 594
    },
    {
      "epoch": 0.27287319422150885,
      "grad_norm": 0.9536232144511817,
      "learning_rate": 0.0008538159989127671,
      "loss": 4.1953,
      "step": 595
    },
    {
      "epoch": 0.27333180463196516,
      "grad_norm": 0.955138430319968,
      "learning_rate": 0.0008532905870715505,
      "loss": 4.2505,
      "step": 596
    },
    {
      "epoch": 0.27379041504242146,
      "grad_norm": 0.954474457794096,
      "learning_rate": 0.0008527643950019131,
      "loss": 4.203,
      "step": 597
    },
    {
      "epoch": 0.27424902545287777,
      "grad_norm": 1.02607604279233,
      "learning_rate": 0.0008522374238659296,
      "loss": 4.1389,
      "step": 598
    },
    {
      "epoch": 0.2747076358633341,
      "grad_norm": 1.078611418798867,
      "learning_rate": 0.0008517096748273951,
      "loss": 4.2319,
      "step": 599
    },
    {
      "epoch": 0.27516624627379044,
      "grad_norm": 1.030626981649032,
      "learning_rate": 0.0008511811490518227,
      "loss": 4.3074,
      "step": 600
    },
    {
      "epoch": 0.27562485668424674,
      "grad_norm": 1.0132257402225566,
      "learning_rate": 0.0008506518477064405,
      "loss": 4.1439,
      "step": 601
    },
    {
      "epoch": 0.27608346709470305,
      "grad_norm": 1.1200898303488376,
      "learning_rate": 0.0008501217719601903,
      "loss": 4.292,
      "step": 602
    },
    {
      "epoch": 0.27654207750515936,
      "grad_norm": 0.9798911585689026,
      "learning_rate": 0.0008495909229837233,
      "loss": 4.1472,
      "step": 603
    },
    {
      "epoch": 0.27700068791561566,
      "grad_norm": 1.0321737319216875,
      "learning_rate": 0.000849059301949399,
      "loss": 4.1569,
      "step": 604
    },
    {
      "epoch": 0.277459298326072,
      "grad_norm": 0.9795998555013009,
      "learning_rate": 0.0008485269100312812,
      "loss": 4.2006,
      "step": 605
    },
    {
      "epoch": 0.27791790873652833,
      "grad_norm": 0.9886537997486151,
      "learning_rate": 0.0008479937484051368,
      "loss": 4.2544,
      "step": 606
    },
    {
      "epoch": 0.27837651914698464,
      "grad_norm": 0.9741470089670419,
      "learning_rate": 0.0008474598182484323,
      "loss": 4.152,
      "step": 607
    },
    {
      "epoch": 0.27883512955744094,
      "grad_norm": 1.008631912336084,
      "learning_rate": 0.0008469251207403317,
      "loss": 4.1816,
      "step": 608
    },
    {
      "epoch": 0.27929373996789725,
      "grad_norm": 1.0036452748290987,
      "learning_rate": 0.0008463896570616934,
      "loss": 4.2824,
      "step": 609
    },
    {
      "epoch": 0.2797523503783536,
      "grad_norm": 0.9804024688194676,
      "learning_rate": 0.0008458534283950678,
      "loss": 4.193,
      "step": 610
    },
    {
      "epoch": 0.2802109607888099,
      "grad_norm": 0.9852623142961642,
      "learning_rate": 0.0008453164359246952,
      "loss": 4.1201,
      "step": 611
    },
    {
      "epoch": 0.2806695711992662,
      "grad_norm": 1.0193361236300191,
      "learning_rate": 0.0008447786808365022,
      "loss": 4.3419,
      "step": 612
    },
    {
      "epoch": 0.2811281816097225,
      "grad_norm": 0.9788949591882863,
      "learning_rate": 0.0008442401643181,
      "loss": 4.2153,
      "step": 613
    },
    {
      "epoch": 0.28158679202017883,
      "grad_norm": 1.0091657544564596,
      "learning_rate": 0.0008437008875587811,
      "loss": 4.1247,
      "step": 614
    },
    {
      "epoch": 0.2820454024306352,
      "grad_norm": 0.9875574415701251,
      "learning_rate": 0.0008431608517495171,
      "loss": 4.0808,
      "step": 615
    },
    {
      "epoch": 0.2825040128410915,
      "grad_norm": 0.9666122688946135,
      "learning_rate": 0.0008426200580829561,
      "loss": 4.1669,
      "step": 616
    },
    {
      "epoch": 0.2829626232515478,
      "grad_norm": 0.9451888364108172,
      "learning_rate": 0.0008420785077534195,
      "loss": 4.1828,
      "step": 617
    },
    {
      "epoch": 0.2834212336620041,
      "grad_norm": 0.9653476092519604,
      "learning_rate": 0.0008415362019569001,
      "loss": 4.1532,
      "step": 618
    },
    {
      "epoch": 0.2838798440724604,
      "grad_norm": 0.9376786543572818,
      "learning_rate": 0.0008409931418910591,
      "loss": 4.2305,
      "step": 619
    },
    {
      "epoch": 0.2843384544829168,
      "grad_norm": 0.9976349202521874,
      "learning_rate": 0.0008404493287552232,
      "loss": 4.2156,
      "step": 620
    },
    {
      "epoch": 0.2847970648933731,
      "grad_norm": 0.9460955034085381,
      "learning_rate": 0.0008399047637503825,
      "loss": 4.1386,
      "step": 621
    },
    {
      "epoch": 0.2852556753038294,
      "grad_norm": 1.019668834270955,
      "learning_rate": 0.0008393594480791875,
      "loss": 4.2276,
      "step": 622
    },
    {
      "epoch": 0.2857142857142857,
      "grad_norm": 0.9998635784800967,
      "learning_rate": 0.0008388133829459463,
      "loss": 4.3359,
      "step": 623
    },
    {
      "epoch": 0.286172896124742,
      "grad_norm": 1.0129588863800796,
      "learning_rate": 0.0008382665695566227,
      "loss": 4.2148,
      "step": 624
    },
    {
      "epoch": 0.28663150653519837,
      "grad_norm": 0.9660079622369675,
      "learning_rate": 0.0008377190091188324,
      "loss": 4.1426,
      "step": 625
    },
    {
      "epoch": 0.28709011694565467,
      "grad_norm": 0.9709216904143456,
      "learning_rate": 0.0008371707028418413,
      "loss": 4.2791,
      "step": 626
    },
    {
      "epoch": 0.287548727356111,
      "grad_norm": 0.95825847696404,
      "learning_rate": 0.0008366216519365621,
      "loss": 4.0546,
      "step": 627
    },
    {
      "epoch": 0.2880073377665673,
      "grad_norm": 1.007562766257185,
      "learning_rate": 0.0008360718576155525,
      "loss": 4.1232,
      "step": 628
    },
    {
      "epoch": 0.28846594817702365,
      "grad_norm": 1.0100065230810145,
      "learning_rate": 0.0008355213210930118,
      "loss": 4.3806,
      "step": 629
    },
    {
      "epoch": 0.28892455858747995,
      "grad_norm": 0.9886168967281,
      "learning_rate": 0.0008349700435847778,
      "loss": 4.1769,
      "step": 630
    },
    {
      "epoch": 0.28938316899793626,
      "grad_norm": 0.9484086111624377,
      "learning_rate": 0.0008344180263083256,
      "loss": 4.004,
      "step": 631
    },
    {
      "epoch": 0.28984177940839256,
      "grad_norm": 0.9454390691326322,
      "learning_rate": 0.000833865270482764,
      "loss": 4.1348,
      "step": 632
    },
    {
      "epoch": 0.29030038981884887,
      "grad_norm": 1.0501722606067898,
      "learning_rate": 0.0008333117773288324,
      "loss": 4.0418,
      "step": 633
    },
    {
      "epoch": 0.29075900022930523,
      "grad_norm": 0.926706438781146,
      "learning_rate": 0.0008327575480688985,
      "loss": 4.1457,
      "step": 634
    },
    {
      "epoch": 0.29121761063976154,
      "grad_norm": 0.8925422905421097,
      "learning_rate": 0.000832202583926956,
      "loss": 4.1702,
      "step": 635
    },
    {
      "epoch": 0.29167622105021784,
      "grad_norm": 0.9792256705809717,
      "learning_rate": 0.0008316468861286217,
      "loss": 4.2398,
      "step": 636
    },
    {
      "epoch": 0.29213483146067415,
      "grad_norm": 0.9910118152009887,
      "learning_rate": 0.0008310904559011323,
      "loss": 4.1623,
      "step": 637
    },
    {
      "epoch": 0.29259344187113046,
      "grad_norm": 1.002701651232777,
      "learning_rate": 0.0008305332944733419,
      "loss": 4.2165,
      "step": 638
    },
    {
      "epoch": 0.2930520522815868,
      "grad_norm": 1.0355866239294436,
      "learning_rate": 0.0008299754030757202,
      "loss": 4.2983,
      "step": 639
    },
    {
      "epoch": 0.2935106626920431,
      "grad_norm": 1.0550434642127773,
      "learning_rate": 0.0008294167829403481,
      "loss": 4.1418,
      "step": 640
    },
    {
      "epoch": 0.29396927310249943,
      "grad_norm": 1.023020355373199,
      "learning_rate": 0.0008288574353009164,
      "loss": 4.0991,
      "step": 641
    },
    {
      "epoch": 0.29442788351295573,
      "grad_norm": 1.0000911602177687,
      "learning_rate": 0.0008282973613927225,
      "loss": 4.2299,
      "step": 642
    },
    {
      "epoch": 0.29488649392341204,
      "grad_norm": 1.0169632957029322,
      "learning_rate": 0.0008277365624526675,
      "loss": 4.0346,
      "step": 643
    },
    {
      "epoch": 0.2953451043338684,
      "grad_norm": 1.079026810831671,
      "learning_rate": 0.0008271750397192541,
      "loss": 4.1175,
      "step": 644
    },
    {
      "epoch": 0.2958037147443247,
      "grad_norm": 1.0177002955186487,
      "learning_rate": 0.0008266127944325832,
      "loss": 4.1956,
      "step": 645
    },
    {
      "epoch": 0.296262325154781,
      "grad_norm": 0.9990181549879754,
      "learning_rate": 0.0008260498278343513,
      "loss": 4.1873,
      "step": 646
    },
    {
      "epoch": 0.2967209355652373,
      "grad_norm": 0.9473254863268945,
      "learning_rate": 0.0008254861411678485,
      "loss": 4.0586,
      "step": 647
    },
    {
      "epoch": 0.2971795459756936,
      "grad_norm": 0.9768611880333975,
      "learning_rate": 0.0008249217356779544,
      "loss": 4.1441,
      "step": 648
    },
    {
      "epoch": 0.29763815638615,
      "grad_norm": 0.9616053419855922,
      "learning_rate": 0.0008243566126111363,
      "loss": 4.3226,
      "step": 649
    },
    {
      "epoch": 0.2980967667966063,
      "grad_norm": 1.0342667667264818,
      "learning_rate": 0.0008237907732154466,
      "loss": 4.1462,
      "step": 650
    },
    {
      "epoch": 0.2985553772070626,
      "grad_norm": 0.9416784708090612,
      "learning_rate": 0.0008232242187405194,
      "loss": 4.1766,
      "step": 651
    },
    {
      "epoch": 0.2990139876175189,
      "grad_norm": 1.0233468018266214,
      "learning_rate": 0.000822656950437568,
      "loss": 4.1834,
      "step": 652
    },
    {
      "epoch": 0.2994725980279752,
      "grad_norm": 0.9928445930504993,
      "learning_rate": 0.0008220889695593823,
      "loss": 4.1943,
      "step": 653
    },
    {
      "epoch": 0.2999312084384316,
      "grad_norm": 0.9507839255738049,
      "learning_rate": 0.0008215202773603259,
      "loss": 4.1127,
      "step": 654
    },
    {
      "epoch": 0.3003898188488879,
      "grad_norm": 0.9983364537086458,
      "learning_rate": 0.0008209508750963328,
      "loss": 4.209,
      "step": 655
    },
    {
      "epoch": 0.3008484292593442,
      "grad_norm": 1.157597390648076,
      "learning_rate": 0.0008203807640249062,
      "loss": 4.3538,
      "step": 656
    },
    {
      "epoch": 0.3013070396698005,
      "grad_norm": 1.0471033049311222,
      "learning_rate": 0.0008198099454051136,
      "loss": 4.1357,
      "step": 657
    },
    {
      "epoch": 0.3017656500802568,
      "grad_norm": 0.9511192804635336,
      "learning_rate": 0.0008192384204975857,
      "loss": 4.1614,
      "step": 658
    },
    {
      "epoch": 0.30222426049071316,
      "grad_norm": 0.9573613808966338,
      "learning_rate": 0.000818666190564513,
      "loss": 4.0566,
      "step": 659
    },
    {
      "epoch": 0.30268287090116947,
      "grad_norm": 1.0345158138795825,
      "learning_rate": 0.0008180932568696426,
      "loss": 4.1018,
      "step": 660
    },
    {
      "epoch": 0.30314148131162577,
      "grad_norm": 1.1255850661362268,
      "learning_rate": 0.0008175196206782764,
      "loss": 4.0723,
      "step": 661
    },
    {
      "epoch": 0.3036000917220821,
      "grad_norm": 1.0335784223289346,
      "learning_rate": 0.0008169452832572675,
      "loss": 4.3366,
      "step": 662
    },
    {
      "epoch": 0.3040587021325384,
      "grad_norm": 0.9411279249105933,
      "learning_rate": 0.0008163702458750173,
      "loss": 4.2092,
      "step": 663
    },
    {
      "epoch": 0.30451731254299474,
      "grad_norm": 1.0478921346462295,
      "learning_rate": 0.0008157945098014734,
      "loss": 4.2181,
      "step": 664
    },
    {
      "epoch": 0.30497592295345105,
      "grad_norm": 0.9261329728181534,
      "learning_rate": 0.0008152180763081267,
      "loss": 4.2864,
      "step": 665
    },
    {
      "epoch": 0.30543453336390736,
      "grad_norm": 0.9648673193674436,
      "learning_rate": 0.0008146409466680076,
      "loss": 4.1844,
      "step": 666
    },
    {
      "epoch": 0.30589314377436366,
      "grad_norm": 1.0419172664045746,
      "learning_rate": 0.0008140631221556845,
      "loss": 4.1795,
      "step": 667
    },
    {
      "epoch": 0.30635175418481997,
      "grad_norm": 0.9475080047292253,
      "learning_rate": 0.0008134846040472599,
      "loss": 4.1389,
      "step": 668
    },
    {
      "epoch": 0.30681036459527633,
      "grad_norm": 0.9717898373495044,
      "learning_rate": 0.0008129053936203688,
      "loss": 4.1443,
      "step": 669
    },
    {
      "epoch": 0.30726897500573264,
      "grad_norm": 0.9925154473460774,
      "learning_rate": 0.0008123254921541745,
      "loss": 4.134,
      "step": 670
    },
    {
      "epoch": 0.30772758541618894,
      "grad_norm": 0.9879807992757266,
      "learning_rate": 0.0008117449009293668,
      "loss": 4.1983,
      "step": 671
    },
    {
      "epoch": 0.30818619582664525,
      "grad_norm": 0.9923775896318972,
      "learning_rate": 0.0008111636212281586,
      "loss": 4.1304,
      "step": 672
    },
    {
      "epoch": 0.3086448062371016,
      "grad_norm": 0.9786305041273194,
      "learning_rate": 0.0008105816543342833,
      "loss": 4.1726,
      "step": 673
    },
    {
      "epoch": 0.3091034166475579,
      "grad_norm": 0.9491595423446998,
      "learning_rate": 0.0008099990015329919,
      "loss": 4.143,
      "step": 674
    },
    {
      "epoch": 0.3095620270580142,
      "grad_norm": 0.8610576001530647,
      "learning_rate": 0.0008094156641110504,
      "loss": 4.3804,
      "step": 675
    },
    {
      "epoch": 0.31002063746847053,
      "grad_norm": 0.8933563507380835,
      "learning_rate": 0.0008088316433567369,
      "loss": 4.2938,
      "step": 676
    },
    {
      "epoch": 0.31047924787892683,
      "grad_norm": 1.0482070303029112,
      "learning_rate": 0.0008082469405598378,
      "loss": 4.1355,
      "step": 677
    },
    {
      "epoch": 0.3109378582893832,
      "grad_norm": 0.918070038827949,
      "learning_rate": 0.0008076615570116468,
      "loss": 4.3004,
      "step": 678
    },
    {
      "epoch": 0.3113964686998395,
      "grad_norm": 0.866348019486383,
      "learning_rate": 0.0008070754940049603,
      "loss": 4.1112,
      "step": 679
    },
    {
      "epoch": 0.3118550791102958,
      "grad_norm": 0.959406065632039,
      "learning_rate": 0.0008064887528340756,
      "loss": 4.3229,
      "step": 680
    },
    {
      "epoch": 0.3123136895207521,
      "grad_norm": 0.9037034562475015,
      "learning_rate": 0.0008059013347947874,
      "loss": 4.0909,
      "step": 681
    },
    {
      "epoch": 0.3127722999312084,
      "grad_norm": 0.8972379819483837,
      "learning_rate": 0.0008053132411843857,
      "loss": 4.2071,
      "step": 682
    },
    {
      "epoch": 0.3132309103416648,
      "grad_norm": 0.9761069532270543,
      "learning_rate": 0.0008047244733016521,
      "loss": 3.9979,
      "step": 683
    },
    {
      "epoch": 0.3136895207521211,
      "grad_norm": 0.9124608388876452,
      "learning_rate": 0.0008041350324468573,
      "loss": 4.233,
      "step": 684
    },
    {
      "epoch": 0.3141481311625774,
      "grad_norm": 0.9250326449684395,
      "learning_rate": 0.0008035449199217583,
      "loss": 4.1482,
      "step": 685
    },
    {
      "epoch": 0.3146067415730337,
      "grad_norm": 0.9547281955256928,
      "learning_rate": 0.0008029541370295957,
      "loss": 4.1966,
      "step": 686
    },
    {
      "epoch": 0.31506535198349,
      "grad_norm": 0.9240157393548275,
      "learning_rate": 0.0008023626850750903,
      "loss": 4.2381,
      "step": 687
    },
    {
      "epoch": 0.31552396239394637,
      "grad_norm": 0.9300102757471376,
      "learning_rate": 0.0008017705653644406,
      "loss": 3.9716,
      "step": 688
    },
    {
      "epoch": 0.3159825728044027,
      "grad_norm": 0.9392378698259768,
      "learning_rate": 0.0008011777792053195,
      "loss": 4.2083,
      "step": 689
    },
    {
      "epoch": 0.316441183214859,
      "grad_norm": 0.9393256917878479,
      "learning_rate": 0.0008005843279068725,
      "loss": 4.2904,
      "step": 690
    },
    {
      "epoch": 0.3168997936253153,
      "grad_norm": 0.9414620961644443,
      "learning_rate": 0.000799990212779713,
      "loss": 4.1789,
      "step": 691
    },
    {
      "epoch": 0.3173584040357716,
      "grad_norm": 0.9145549610121856,
      "learning_rate": 0.0007993954351359214,
      "loss": 4.3598,
      "step": 692
    },
    {
      "epoch": 0.31781701444622795,
      "grad_norm": 0.9476444189987323,
      "learning_rate": 0.0007987999962890406,
      "loss": 4.0796,
      "step": 693
    },
    {
      "epoch": 0.31827562485668426,
      "grad_norm": 0.9321698948604105,
      "learning_rate": 0.0007982038975540742,
      "loss": 4.0122,
      "step": 694
    },
    {
      "epoch": 0.31873423526714056,
      "grad_norm": 0.9518419158720559,
      "learning_rate": 0.0007976071402474826,
      "loss": 4.2314,
      "step": 695
    },
    {
      "epoch": 0.31919284567759687,
      "grad_norm": 0.9423632911280964,
      "learning_rate": 0.0007970097256871811,
      "loss": 4.2214,
      "step": 696
    },
    {
      "epoch": 0.3196514560880532,
      "grad_norm": 0.8837644684251644,
      "learning_rate": 0.0007964116551925364,
      "loss": 4.0494,
      "step": 697
    },
    {
      "epoch": 0.32011006649850954,
      "grad_norm": 0.9679106813266904,
      "learning_rate": 0.0007958129300843637,
      "loss": 4.0097,
      "step": 698
    },
    {
      "epoch": 0.32056867690896584,
      "grad_norm": 0.9529408453850848,
      "learning_rate": 0.0007952135516849239,
      "loss": 4.2246,
      "step": 699
    },
    {
      "epoch": 0.32102728731942215,
      "grad_norm": 0.906721630313311,
      "learning_rate": 0.0007946135213179207,
      "loss": 4.3204,
      "step": 700
    },
    {
      "epoch": 0.32148589772987846,
      "grad_norm": 0.9063926085640105,
      "learning_rate": 0.0007940128403084977,
      "loss": 4.1144,
      "step": 701
    },
    {
      "epoch": 0.32194450814033476,
      "grad_norm": 0.9300814708447666,
      "learning_rate": 0.0007934115099832355,
      "loss": 4.1381,
      "step": 702
    },
    {
      "epoch": 0.3224031185507911,
      "grad_norm": 0.9661220121745884,
      "learning_rate": 0.0007928095316701483,
      "loss": 4.1958,
      "step": 703
    },
    {
      "epoch": 0.32286172896124743,
      "grad_norm": 0.9548941627174306,
      "learning_rate": 0.0007922069066986819,
      "loss": 4.0778,
      "step": 704
    },
    {
      "epoch": 0.32332033937170374,
      "grad_norm": 0.9924139868185463,
      "learning_rate": 0.0007916036363997097,
      "loss": 4.1899,
      "step": 705
    },
    {
      "epoch": 0.32377894978216004,
      "grad_norm": 0.9833657989464873,
      "learning_rate": 0.0007909997221055308,
      "loss": 4.1402,
      "step": 706
    },
    {
      "epoch": 0.32423756019261635,
      "grad_norm": 1.1785651636728007,
      "learning_rate": 0.0007903951651498658,
      "loss": 4.1281,
      "step": 707
    },
    {
      "epoch": 0.3246961706030727,
      "grad_norm": 0.938148431247077,
      "learning_rate": 0.0007897899668678557,
      "loss": 4.2137,
      "step": 708
    },
    {
      "epoch": 0.325154781013529,
      "grad_norm": 0.9851233810777071,
      "learning_rate": 0.0007891841285960566,
      "loss": 4.0711,
      "step": 709
    },
    {
      "epoch": 0.3256133914239853,
      "grad_norm": 1.0330988816962603,
      "learning_rate": 0.0007885776516724388,
      "loss": 4.0497,
      "step": 710
    },
    {
      "epoch": 0.3260720018344416,
      "grad_norm": 1.077504091955506,
      "learning_rate": 0.0007879705374363831,
      "loss": 4.1547,
      "step": 711
    },
    {
      "epoch": 0.32653061224489793,
      "grad_norm": 0.883880577828925,
      "learning_rate": 0.000787362787228677,
      "loss": 4.2037,
      "step": 712
    },
    {
      "epoch": 0.3269892226553543,
      "grad_norm": 0.9313149022761364,
      "learning_rate": 0.0007867544023915134,
      "loss": 3.9817,
      "step": 713
    },
    {
      "epoch": 0.3274478330658106,
      "grad_norm": 1.012094604608198,
      "learning_rate": 0.0007861453842684861,
      "loss": 4.1033,
      "step": 714
    },
    {
      "epoch": 0.3279064434762669,
      "grad_norm": 1.0210255979364244,
      "learning_rate": 0.0007855357342045882,
      "loss": 4.0044,
      "step": 715
    },
    {
      "epoch": 0.3283650538867232,
      "grad_norm": 0.9471011279515731,
      "learning_rate": 0.0007849254535462074,
      "loss": 4.1256,
      "step": 716
    },
    {
      "epoch": 0.3288236642971795,
      "grad_norm": 0.9524076255738089,
      "learning_rate": 0.0007843145436411252,
      "loss": 4.0156,
      "step": 717
    },
    {
      "epoch": 0.3292822747076359,
      "grad_norm": 1.0727766544185602,
      "learning_rate": 0.0007837030058385117,
      "loss": 4.4152,
      "step": 718
    },
    {
      "epoch": 0.3297408851180922,
      "grad_norm": 0.9925270349091415,
      "learning_rate": 0.0007830908414889246,
      "loss": 4.1546,
      "step": 719
    },
    {
      "epoch": 0.3301994955285485,
      "grad_norm": 1.0090169741481252,
      "learning_rate": 0.0007824780519443046,
      "loss": 4.0295,
      "step": 720
    },
    {
      "epoch": 0.3306581059390048,
      "grad_norm": 0.9271807745515079,
      "learning_rate": 0.0007818646385579735,
      "loss": 4.1158,
      "step": 721
    },
    {
      "epoch": 0.33111671634946116,
      "grad_norm": 1.009534720872593,
      "learning_rate": 0.0007812506026846307,
      "loss": 4.3044,
      "step": 722
    },
    {
      "epoch": 0.33157532675991747,
      "grad_norm": 1.0086874481383514,
      "learning_rate": 0.0007806359456803504,
      "loss": 4.1115,
      "step": 723
    },
    {
      "epoch": 0.33203393717037377,
      "grad_norm": 0.9095373018887838,
      "learning_rate": 0.0007800206689025785,
      "loss": 4.298,
      "step": 724
    },
    {
      "epoch": 0.3324925475808301,
      "grad_norm": 0.9438069266201196,
      "learning_rate": 0.0007794047737101297,
      "loss": 4.1768,
      "step": 725
    },
    {
      "epoch": 0.3329511579912864,
      "grad_norm": 0.9083140654126173,
      "learning_rate": 0.0007787882614631843,
      "loss": 4.0931,
      "step": 726
    },
    {
      "epoch": 0.33340976840174275,
      "grad_norm": 0.893515035135105,
      "learning_rate": 0.0007781711335232856,
      "loss": 4.165,
      "step": 727
    },
    {
      "epoch": 0.33386837881219905,
      "grad_norm": 1.036964953778352,
      "learning_rate": 0.0007775533912533363,
      "loss": 4.3251,
      "step": 728
    },
    {
      "epoch": 0.33432698922265536,
      "grad_norm": 0.9594434825936703,
      "learning_rate": 0.0007769350360175962,
      "loss": 4.2536,
      "step": 729
    },
    {
      "epoch": 0.33478559963311166,
      "grad_norm": 0.9112908388825509,
      "learning_rate": 0.0007763160691816784,
      "loss": 3.9897,
      "step": 730
    },
    {
      "epoch": 0.33524421004356797,
      "grad_norm": 1.0042619515700617,
      "learning_rate": 0.000775696492112547,
      "loss": 4.0837,
      "step": 731
    },
    {
      "epoch": 0.33570282045402433,
      "grad_norm": 0.9981820356516703,
      "learning_rate": 0.0007750763061785137,
      "loss": 4.1941,
      "step": 732
    },
    {
      "epoch": 0.33616143086448064,
      "grad_norm": 0.9759412415391815,
      "learning_rate": 0.000774455512749235,
      "loss": 4.1184,
      "step": 733
    },
    {
      "epoch": 0.33662004127493694,
      "grad_norm": 0.9389179994678325,
      "learning_rate": 0.0007738341131957085,
      "loss": 4.1269,
      "step": 734
    },
    {
      "epoch": 0.33707865168539325,
      "grad_norm": 0.9644978343323549,
      "learning_rate": 0.000773212108890271,
      "loss": 3.9484,
      "step": 735
    },
    {
      "epoch": 0.33753726209584956,
      "grad_norm": 0.9110252985429655,
      "learning_rate": 0.0007725895012065947,
      "loss": 4.0839,
      "step": 736
    },
    {
      "epoch": 0.3379958725063059,
      "grad_norm": 0.9891115571913581,
      "learning_rate": 0.0007719662915196844,
      "loss": 4.0794,
      "step": 737
    },
    {
      "epoch": 0.3384544829167622,
      "grad_norm": 0.9504803816034868,
      "learning_rate": 0.0007713424812058736,
      "loss": 4.2143,
      "step": 738
    },
    {
      "epoch": 0.33891309332721853,
      "grad_norm": 0.9434727606544564,
      "learning_rate": 0.0007707180716428237,
      "loss": 4.2379,
      "step": 739
    },
    {
      "epoch": 0.33937170373767483,
      "grad_norm": 0.9010921639375071,
      "learning_rate": 0.0007700930642095184,
      "loss": 4.092,
      "step": 740
    },
    {
      "epoch": 0.33983031414813114,
      "grad_norm": 0.8856390314165904,
      "learning_rate": 0.0007694674602862621,
      "loss": 4.1475,
      "step": 741
    },
    {
      "epoch": 0.3402889245585875,
      "grad_norm": 0.9243830305694336,
      "learning_rate": 0.0007688412612546769,
      "loss": 3.9874,
      "step": 742
    },
    {
      "epoch": 0.3407475349690438,
      "grad_norm": 0.8743670432813972,
      "learning_rate": 0.0007682144684976983,
      "loss": 4.0017,
      "step": 743
    },
    {
      "epoch": 0.3412061453795001,
      "grad_norm": 0.9110536300703093,
      "learning_rate": 0.0007675870833995739,
      "loss": 4.2329,
      "step": 744
    },
    {
      "epoch": 0.3416647557899564,
      "grad_norm": 0.9145617685508306,
      "learning_rate": 0.0007669591073458592,
      "loss": 4.0014,
      "step": 745
    },
    {
      "epoch": 0.3421233662004127,
      "grad_norm": 0.943220115436867,
      "learning_rate": 0.0007663305417234146,
      "loss": 4.1057,
      "step": 746
    },
    {
      "epoch": 0.3425819766108691,
      "grad_norm": 0.9393007631728266,
      "learning_rate": 0.0007657013879204022,
      "loss": 3.9819,
      "step": 747
    },
    {
      "epoch": 0.3430405870213254,
      "grad_norm": 0.9345118460508727,
      "learning_rate": 0.0007650716473262842,
      "loss": 4.0863,
      "step": 748
    },
    {
      "epoch": 0.3434991974317817,
      "grad_norm": 0.9164497459069919,
      "learning_rate": 0.0007644413213318177,
      "loss": 4.1091,
      "step": 749
    },
    {
      "epoch": 0.343957807842238,
      "grad_norm": 0.9783629404895698,
      "learning_rate": 0.0007638104113290531,
      "loss": 4.1203,
      "step": 750
    },
    {
      "epoch": 0.3444164182526943,
      "grad_norm": 0.9728307366415673,
      "learning_rate": 0.0007631789187113303,
      "loss": 4.0753,
      "step": 751
    },
    {
      "epoch": 0.3448750286631507,
      "grad_norm": 0.9226953530686045,
      "learning_rate": 0.000762546844873276,
      "loss": 4.3393,
      "step": 752
    },
    {
      "epoch": 0.345333639073607,
      "grad_norm": 0.9664135322242463,
      "learning_rate": 0.0007619141912108007,
      "loss": 4.0978,
      "step": 753
    },
    {
      "epoch": 0.3457922494840633,
      "grad_norm": 0.8962050451542138,
      "learning_rate": 0.000761280959121095,
      "loss": 4.123,
      "step": 754
    },
    {
      "epoch": 0.3462508598945196,
      "grad_norm": 0.8978685230755128,
      "learning_rate": 0.0007606471500026273,
      "loss": 4.0724,
      "step": 755
    },
    {
      "epoch": 0.3467094703049759,
      "grad_norm": 0.8978137131339885,
      "learning_rate": 0.0007600127652551401,
      "loss": 4.1622,
      "step": 756
    },
    {
      "epoch": 0.34716808071543226,
      "grad_norm": 0.8759137551597629,
      "learning_rate": 0.0007593778062796472,
      "loss": 4.3434,
      "step": 757
    },
    {
      "epoch": 0.34762669112588857,
      "grad_norm": 0.9182175294109279,
      "learning_rate": 0.000758742274478431,
      "loss": 4.1192,
      "step": 758
    },
    {
      "epoch": 0.34808530153634487,
      "grad_norm": 0.8767150890910086,
      "learning_rate": 0.0007581061712550381,
      "loss": 4.1801,
      "step": 759
    },
    {
      "epoch": 0.3485439119468012,
      "grad_norm": 0.8992583127561219,
      "learning_rate": 0.0007574694980142779,
      "loss": 4.168,
      "step": 760
    },
    {
      "epoch": 0.3490025223572575,
      "grad_norm": 0.9144735578652025,
      "learning_rate": 0.0007568322561622183,
      "loss": 4.1272,
      "step": 761
    },
    {
      "epoch": 0.34946113276771384,
      "grad_norm": 0.8960520225344586,
      "learning_rate": 0.0007561944471061826,
      "loss": 3.887,
      "step": 762
    },
    {
      "epoch": 0.34991974317817015,
      "grad_norm": 0.9376653982331855,
      "learning_rate": 0.0007555560722547475,
      "loss": 4.3041,
      "step": 763
    },
    {
      "epoch": 0.35037835358862646,
      "grad_norm": 0.892179181908615,
      "learning_rate": 0.0007549171330177387,
      "loss": 4.0804,
      "step": 764
    },
    {
      "epoch": 0.35083696399908276,
      "grad_norm": 0.9127506999808277,
      "learning_rate": 0.0007542776308062285,
      "loss": 3.9208,
      "step": 765
    },
    {
      "epoch": 0.3512955744095391,
      "grad_norm": 0.9119483506445285,
      "learning_rate": 0.0007536375670325325,
      "loss": 4.1947,
      "step": 766
    },
    {
      "epoch": 0.35175418481999543,
      "grad_norm": 0.8977440326840286,
      "learning_rate": 0.0007529969431102063,
      "loss": 4.1386,
      "step": 767
    },
    {
      "epoch": 0.35221279523045174,
      "grad_norm": 0.9390554414149044,
      "learning_rate": 0.000752355760454043,
      "loss": 4.1742,
      "step": 768
    },
    {
      "epoch": 0.35267140564090804,
      "grad_norm": 0.860753202162299,
      "learning_rate": 0.0007517140204800693,
      "loss": 4.1637,
      "step": 769
    },
    {
      "epoch": 0.35313001605136435,
      "grad_norm": 0.8991813803944004,
      "learning_rate": 0.0007510717246055425,
      "loss": 4.1124,
      "step": 770
    },
    {
      "epoch": 0.3535886264618207,
      "grad_norm": 0.9782179838535272,
      "learning_rate": 0.0007504288742489482,
      "loss": 4.0772,
      "step": 771
    },
    {
      "epoch": 0.354047236872277,
      "grad_norm": 0.8553018363563852,
      "learning_rate": 0.0007497854708299963,
      "loss": 4.0505,
      "step": 772
    },
    {
      "epoch": 0.3545058472827333,
      "grad_norm": 0.8931717857374777,
      "learning_rate": 0.0007491415157696178,
      "loss": 4.1502,
      "step": 773
    },
    {
      "epoch": 0.35496445769318963,
      "grad_norm": 0.8772287700207598,
      "learning_rate": 0.0007484970104899623,
      "loss": 4.1406,
      "step": 774
    },
    {
      "epoch": 0.35542306810364593,
      "grad_norm": 0.8754490611136192,
      "learning_rate": 0.0007478519564143945,
      "loss": 4.125,
      "step": 775
    },
    {
      "epoch": 0.3558816785141023,
      "grad_norm": 0.9580005179816752,
      "learning_rate": 0.000747206354967491,
      "loss": 4.0161,
      "step": 776
    },
    {
      "epoch": 0.3563402889245586,
      "grad_norm": 0.8608408277163018,
      "learning_rate": 0.0007465602075750373,
      "loss": 4.1791,
      "step": 777
    },
    {
      "epoch": 0.3567988993350149,
      "grad_norm": 0.8600608364789372,
      "learning_rate": 0.0007459135156640247,
      "loss": 4.1079,
      "step": 778
    },
    {
      "epoch": 0.3572575097454712,
      "grad_norm": 0.9236518740123941,
      "learning_rate": 0.0007452662806626468,
      "loss": 3.933,
      "step": 779
    },
    {
      "epoch": 0.3577161201559275,
      "grad_norm": 0.9741527977662501,
      "learning_rate": 0.0007446185040002967,
      "loss": 4.1659,
      "step": 780
    },
    {
      "epoch": 0.3581747305663839,
      "grad_norm": 0.8882723588821434,
      "learning_rate": 0.0007439701871075642,
      "loss": 4.1213,
      "step": 781
    },
    {
      "epoch": 0.3586333409768402,
      "grad_norm": 0.9212503486454555,
      "learning_rate": 0.0007433213314162313,
      "loss": 4.1043,
      "step": 782
    },
    {
      "epoch": 0.3590919513872965,
      "grad_norm": 0.8467451241339713,
      "learning_rate": 0.0007426719383592705,
      "loss": 3.9531,
      "step": 783
    },
    {
      "epoch": 0.3595505617977528,
      "grad_norm": 0.8673286607696294,
      "learning_rate": 0.000742022009370841,
      "loss": 4.1264,
      "step": 784
    },
    {
      "epoch": 0.3600091722082091,
      "grad_norm": 0.9238892377757724,
      "learning_rate": 0.0007413715458862855,
      "loss": 4.0718,
      "step": 785
    },
    {
      "epoch": 0.36046778261866547,
      "grad_norm": 0.8686168914301998,
      "learning_rate": 0.0007407205493421272,
      "loss": 3.9348,
      "step": 786
    },
    {
      "epoch": 0.3609263930291218,
      "grad_norm": 0.8865703729599965,
      "learning_rate": 0.0007400690211760661,
      "loss": 3.8762,
      "step": 787
    },
    {
      "epoch": 0.3613850034395781,
      "grad_norm": 0.9512893234134777,
      "learning_rate": 0.0007394169628269771,
      "loss": 3.9912,
      "step": 788
    },
    {
      "epoch": 0.3618436138500344,
      "grad_norm": 0.9187113705326728,
      "learning_rate": 0.0007387643757349051,
      "loss": 4.0646,
      "step": 789
    },
    {
      "epoch": 0.3623022242604907,
      "grad_norm": 0.9512504460979749,
      "learning_rate": 0.0007381112613410635,
      "loss": 4.2507,
      "step": 790
    },
    {
      "epoch": 0.36276083467094705,
      "grad_norm": 0.9490555183341,
      "learning_rate": 0.0007374576210878298,
      "loss": 4.0359,
      "step": 791
    },
    {
      "epoch": 0.36321944508140336,
      "grad_norm": 0.8984101021777058,
      "learning_rate": 0.0007368034564187425,
      "loss": 4.1653,
      "step": 792
    },
    {
      "epoch": 0.36367805549185966,
      "grad_norm": 1.050110135997148,
      "learning_rate": 0.0007361487687784989,
      "loss": 4.3608,
      "step": 793
    },
    {
      "epoch": 0.36413666590231597,
      "grad_norm": 0.993994516135655,
      "learning_rate": 0.0007354935596129513,
      "loss": 4.0845,
      "step": 794
    },
    {
      "epoch": 0.3645952763127723,
      "grad_norm": 0.9596337340942739,
      "learning_rate": 0.000734837830369103,
      "loss": 3.9939,
      "step": 795
    },
    {
      "epoch": 0.36505388672322864,
      "grad_norm": 0.8344652031598586,
      "learning_rate": 0.0007341815824951066,
      "loss": 4.0337,
      "step": 796
    },
    {
      "epoch": 0.36551249713368494,
      "grad_norm": 0.9121756616353818,
      "learning_rate": 0.0007335248174402597,
      "loss": 3.9813,
      "step": 797
    },
    {
      "epoch": 0.36597110754414125,
      "grad_norm": 0.9540839447417431,
      "learning_rate": 0.0007328675366550023,
      "loss": 4.1086,
      "step": 798
    },
    {
      "epoch": 0.36642971795459756,
      "grad_norm": 0.9156755261346214,
      "learning_rate": 0.0007322097415909134,
      "loss": 3.958,
      "step": 799
    },
    {
      "epoch": 0.36688832836505386,
      "grad_norm": 0.8853244679729705,
      "learning_rate": 0.0007315514337007071,
      "loss": 4.0251,
      "step": 800
    },
    {
      "epoch": 0.3673469387755102,
      "grad_norm": 0.9435117403038707,
      "learning_rate": 0.0007308926144382312,
      "loss": 4.0596,
      "step": 801
    },
    {
      "epoch": 0.36780554918596653,
      "grad_norm": 0.9769968073214329,
      "learning_rate": 0.0007302332852584619,
      "loss": 4.1734,
      "step": 802
    },
    {
      "epoch": 0.36826415959642284,
      "grad_norm": 0.9190823764923192,
      "learning_rate": 0.0007295734476175018,
      "loss": 4.0289,
      "step": 803
    },
    {
      "epoch": 0.36872277000687914,
      "grad_norm": 0.8997310161240203,
      "learning_rate": 0.0007289131029725768,
      "loss": 4.0291,
      "step": 804
    },
    {
      "epoch": 0.36918138041733545,
      "grad_norm": 0.936881833121541,
      "learning_rate": 0.0007282522527820319,
      "loss": 4.0582,
      "step": 805
    },
    {
      "epoch": 0.3696399908277918,
      "grad_norm": 0.8025136908807218,
      "learning_rate": 0.000727590898505329,
      "loss": 3.9592,
      "step": 806
    },
    {
      "epoch": 0.3700986012382481,
      "grad_norm": 0.889435851076379,
      "learning_rate": 0.0007269290416030429,
      "loss": 4.0593,
      "step": 807
    },
    {
      "epoch": 0.3705572116487044,
      "grad_norm": 0.894015191968085,
      "learning_rate": 0.000726266683536859,
      "loss": 4.093,
      "step": 808
    },
    {
      "epoch": 0.3710158220591607,
      "grad_norm": 0.871418247274549,
      "learning_rate": 0.0007256038257695687,
      "loss": 3.9954,
      "step": 809
    },
    {
      "epoch": 0.37147443246961703,
      "grad_norm": 0.9382733830670867,
      "learning_rate": 0.0007249404697650678,
      "loss": 4.1267,
      "step": 810
    },
    {
      "epoch": 0.3719330428800734,
      "grad_norm": 0.936215104808343,
      "learning_rate": 0.0007242766169883518,
      "loss": 4.0927,
      "step": 811
    },
    {
      "epoch": 0.3723916532905297,
      "grad_norm": 0.956748302146746,
      "learning_rate": 0.0007236122689055138,
      "loss": 4.189,
      "step": 812
    },
    {
      "epoch": 0.372850263700986,
      "grad_norm": 0.8793653317040636,
      "learning_rate": 0.0007229474269837401,
      "loss": 3.9827,
      "step": 813
    },
    {
      "epoch": 0.3733088741114423,
      "grad_norm": 1.003682292455992,
      "learning_rate": 0.0007222820926913085,
      "loss": 4.1762,
      "step": 814
    },
    {
      "epoch": 0.3737674845218987,
      "grad_norm": 0.8922058196074301,
      "learning_rate": 0.0007216162674975833,
      "loss": 4.0533,
      "step": 815
    },
    {
      "epoch": 0.374226094932355,
      "grad_norm": 0.9054349024352683,
      "learning_rate": 0.0007209499528730138,
      "loss": 3.9775,
      "step": 816
    },
    {
      "epoch": 0.3746847053428113,
      "grad_norm": 0.8849471060834453,
      "learning_rate": 0.0007202831502891294,
      "loss": 4.2775,
      "step": 817
    },
    {
      "epoch": 0.3751433157532676,
      "grad_norm": 0.8453041312681328,
      "learning_rate": 0.0007196158612185375,
      "loss": 3.9064,
      "step": 818
    },
    {
      "epoch": 0.3756019261637239,
      "grad_norm": 0.9065288533972647,
      "learning_rate": 0.0007189480871349201,
      "loss": 3.9548,
      "step": 819
    },
    {
      "epoch": 0.37606053657418026,
      "grad_norm": 0.8695047537304627,
      "learning_rate": 0.0007182798295130299,
      "loss": 4.0415,
      "step": 820
    },
    {
      "epoch": 0.37651914698463657,
      "grad_norm": 0.9351294752102677,
      "learning_rate": 0.0007176110898286878,
      "loss": 4.0756,
      "step": 821
    },
    {
      "epoch": 0.37697775739509287,
      "grad_norm": 0.892216754837046,
      "learning_rate": 0.0007169418695587791,
      "loss": 3.8962,
      "step": 822
    },
    {
      "epoch": 0.3774363678055492,
      "grad_norm": 0.867983015801153,
      "learning_rate": 0.0007162721701812506,
      "loss": 3.8516,
      "step": 823
    },
    {
      "epoch": 0.3778949782160055,
      "grad_norm": 0.9237504955725637,
      "learning_rate": 0.0007156019931751072,
      "loss": 3.9752,
      "step": 824
    },
    {
      "epoch": 0.37835358862646185,
      "grad_norm": 0.8631773833323212,
      "learning_rate": 0.0007149313400204082,
      "loss": 4.2331,
      "step": 825
    },
    {
      "epoch": 0.37881219903691815,
      "grad_norm": 0.9096076154933478,
      "learning_rate": 0.0007142602121982653,
      "loss": 4.1303,
      "step": 826
    },
    {
      "epoch": 0.37927080944737446,
      "grad_norm": 0.9428076088076401,
      "learning_rate": 0.0007135886111908379,
      "loss": 4.0051,
      "step": 827
    },
    {
      "epoch": 0.37972941985783076,
      "grad_norm": 0.9844279364891863,
      "learning_rate": 0.0007129165384813303,
      "loss": 4.161,
      "step": 828
    },
    {
      "epoch": 0.38018803026828707,
      "grad_norm": 0.8789910739126864,
      "learning_rate": 0.0007122439955539888,
      "loss": 4.1874,
      "step": 829
    },
    {
      "epoch": 0.38064664067874343,
      "grad_norm": 0.9877976800690257,
      "learning_rate": 0.0007115709838940983,
      "loss": 3.9438,
      "step": 830
    },
    {
      "epoch": 0.38110525108919974,
      "grad_norm": 0.9006145836442371,
      "learning_rate": 0.0007108975049879785,
      "loss": 4.147,
      "step": 831
    },
    {
      "epoch": 0.38156386149965604,
      "grad_norm": 0.9089975707322127,
      "learning_rate": 0.0007102235603229814,
      "loss": 4.1923,
      "step": 832
    },
    {
      "epoch": 0.38202247191011235,
      "grad_norm": 0.8725279801073029,
      "learning_rate": 0.000709549151387487,
      "loss": 4.1896,
      "step": 833
    },
    {
      "epoch": 0.38248108232056865,
      "grad_norm": 0.8983652008032615,
      "learning_rate": 0.0007088742796709013,
      "loss": 4.1488,
      "step": 834
    },
    {
      "epoch": 0.382939692731025,
      "grad_norm": 0.8865579362650281,
      "learning_rate": 0.000708198946663652,
      "loss": 4.173,
      "step": 835
    },
    {
      "epoch": 0.3833983031414813,
      "grad_norm": 0.8478183080526649,
      "learning_rate": 0.0007075231538571856,
      "loss": 4.2164,
      "step": 836
    },
    {
      "epoch": 0.38385691355193763,
      "grad_norm": 0.8592109881295178,
      "learning_rate": 0.0007068469027439641,
      "loss": 4.0245,
      "step": 837
    },
    {
      "epoch": 0.38431552396239393,
      "grad_norm": 0.9163673311312518,
      "learning_rate": 0.0007061701948174613,
      "loss": 3.9001,
      "step": 838
    },
    {
      "epoch": 0.38477413437285024,
      "grad_norm": 0.9896695954805433,
      "learning_rate": 0.0007054930315721606,
      "loss": 4.0036,
      "step": 839
    },
    {
      "epoch": 0.3852327447833066,
      "grad_norm": 0.9151104967372045,
      "learning_rate": 0.0007048154145035501,
      "loss": 4.0423,
      "step": 840
    },
    {
      "epoch": 0.3856913551937629,
      "grad_norm": 0.9226160369535424,
      "learning_rate": 0.0007041373451081207,
      "loss": 3.952,
      "step": 841
    },
    {
      "epoch": 0.3861499656042192,
      "grad_norm": 0.9525577589675372,
      "learning_rate": 0.0007034588248833621,
      "loss": 4.2398,
      "step": 842
    },
    {
      "epoch": 0.3866085760146755,
      "grad_norm": 0.9121585200668445,
      "learning_rate": 0.0007027798553277595,
      "loss": 4.1135,
      "step": 843
    },
    {
      "epoch": 0.3870671864251318,
      "grad_norm": 0.9366512290951872,
      "learning_rate": 0.0007021004379407909,
      "loss": 3.9709,
      "step": 844
    },
    {
      "epoch": 0.3875257968355882,
      "grad_norm": 0.9220154208793143,
      "learning_rate": 0.0007014205742229227,
      "loss": 4.0729,
      "step": 845
    },
    {
      "epoch": 0.3879844072460445,
      "grad_norm": 0.8867373328005244,
      "learning_rate": 0.0007007402656756072,
      "loss": 4.2256,
      "step": 846
    },
    {
      "epoch": 0.3884430176565008,
      "grad_norm": 0.9382278299330966,
      "learning_rate": 0.0007000595138012797,
      "loss": 4.1712,
      "step": 847
    },
    {
      "epoch": 0.3889016280669571,
      "grad_norm": 1.0075122960000837,
      "learning_rate": 0.0006993783201033535,
      "loss": 4.1025,
      "step": 848
    },
    {
      "epoch": 0.3893602384774134,
      "grad_norm": 0.9830387834808473,
      "learning_rate": 0.0006986966860862182,
      "loss": 4.0289,
      "step": 849
    },
    {
      "epoch": 0.3898188488878698,
      "grad_norm": 0.8861102753044039,
      "learning_rate": 0.000698014613255236,
      "loss": 3.9855,
      "step": 850
    },
    {
      "epoch": 0.3902774592983261,
      "grad_norm": 0.9597172432653165,
      "learning_rate": 0.0006973321031167382,
      "loss": 4.1562,
      "step": 851
    },
    {
      "epoch": 0.3907360697087824,
      "grad_norm": 0.9833427818785277,
      "learning_rate": 0.0006966491571780216,
      "loss": 3.9268,
      "step": 852
    },
    {
      "epoch": 0.3911946801192387,
      "grad_norm": 0.9071899607884865,
      "learning_rate": 0.0006959657769473453,
      "loss": 4.1692,
      "step": 853
    },
    {
      "epoch": 0.391653290529695,
      "grad_norm": 0.9355847384888029,
      "learning_rate": 0.000695281963933928,
      "loss": 4.0642,
      "step": 854
    },
    {
      "epoch": 0.39211190094015136,
      "grad_norm": 0.8961423749294206,
      "learning_rate": 0.0006945977196479438,
      "loss": 4.236,
      "step": 855
    },
    {
      "epoch": 0.39257051135060766,
      "grad_norm": 0.9292337964430353,
      "learning_rate": 0.0006939130456005196,
      "loss": 4.0941,
      "step": 856
    },
    {
      "epoch": 0.39302912176106397,
      "grad_norm": 0.9320051843994537,
      "learning_rate": 0.0006932279433037311,
      "loss": 4.0434,
      "step": 857
    },
    {
      "epoch": 0.3934877321715203,
      "grad_norm": 0.91466574269659,
      "learning_rate": 0.0006925424142705997,
      "loss": 4.0824,
      "step": 858
    },
    {
      "epoch": 0.39394634258197664,
      "grad_norm": 0.905645334987164,
      "learning_rate": 0.0006918564600150896,
      "loss": 4.0016,
      "step": 859
    },
    {
      "epoch": 0.39440495299243294,
      "grad_norm": 0.9111279720533533,
      "learning_rate": 0.0006911700820521042,
      "loss": 4.1468,
      "step": 860
    },
    {
      "epoch": 0.39486356340288925,
      "grad_norm": 0.9814234688448878,
      "learning_rate": 0.0006904832818974818,
      "loss": 4.3309,
      "step": 861
    },
    {
      "epoch": 0.39532217381334556,
      "grad_norm": 0.909463736919346,
      "learning_rate": 0.0006897960610679939,
      "loss": 4.1301,
      "step": 862
    },
    {
      "epoch": 0.39578078422380186,
      "grad_norm": 0.8953558845976561,
      "learning_rate": 0.0006891084210813407,
      "loss": 4.1306,
      "step": 863
    },
    {
      "epoch": 0.3962393946342582,
      "grad_norm": 0.8569902825148525,
      "learning_rate": 0.0006884203634561483,
      "loss": 4.056,
      "step": 864
    },
    {
      "epoch": 0.39669800504471453,
      "grad_norm": 0.8896406520917401,
      "learning_rate": 0.0006877318897119651,
      "loss": 3.9683,
      "step": 865
    },
    {
      "epoch": 0.39715661545517084,
      "grad_norm": 0.883866545141719,
      "learning_rate": 0.0006870430013692579,
      "loss": 4.0373,
      "step": 866
    },
    {
      "epoch": 0.39761522586562714,
      "grad_norm": 0.8609244948173354,
      "learning_rate": 0.0006863536999494101,
      "loss": 4.1765,
      "step": 867
    },
    {
      "epoch": 0.39807383627608345,
      "grad_norm": 0.9238353375119052,
      "learning_rate": 0.0006856639869747167,
      "loss": 3.8996,
      "step": 868
    },
    {
      "epoch": 0.3985324466865398,
      "grad_norm": 0.9608495931812143,
      "learning_rate": 0.0006849738639683818,
      "loss": 4.0776,
      "step": 869
    },
    {
      "epoch": 0.3989910570969961,
      "grad_norm": 0.9003935920393377,
      "learning_rate": 0.000684283332454515,
      "loss": 4.0179,
      "step": 870
    },
    {
      "epoch": 0.3994496675074524,
      "grad_norm": 1.000607514153423,
      "learning_rate": 0.0006835923939581281,
      "loss": 4.2776,
      "step": 871
    },
    {
      "epoch": 0.3999082779179087,
      "grad_norm": 0.9920012184673828,
      "learning_rate": 0.0006829010500051318,
      "loss": 4.0388,
      "step": 872
    },
    {
      "epoch": 0.40036688832836503,
      "grad_norm": 0.8806573356968799,
      "learning_rate": 0.0006822093021223321,
      "loss": 4.089,
      "step": 873
    },
    {
      "epoch": 0.4008254987388214,
      "grad_norm": 0.8517647027964018,
      "learning_rate": 0.0006815171518374268,
      "loss": 3.9851,
      "step": 874
    },
    {
      "epoch": 0.4012841091492777,
      "grad_norm": 0.9400675250566718,
      "learning_rate": 0.0006808246006790031,
      "loss": 4.0313,
      "step": 875
    },
    {
      "epoch": 0.401742719559734,
      "grad_norm": 0.9411115457135417,
      "learning_rate": 0.0006801316501765329,
      "loss": 4.1841,
      "step": 876
    },
    {
      "epoch": 0.4022013299701903,
      "grad_norm": 0.9474872752554954,
      "learning_rate": 0.0006794383018603704,
      "loss": 4.1593,
      "step": 877
    },
    {
      "epoch": 0.4026599403806466,
      "grad_norm": 0.8919629443632937,
      "learning_rate": 0.0006787445572617481,
      "loss": 4.0836,
      "step": 878
    },
    {
      "epoch": 0.403118550791103,
      "grad_norm": 0.8644240944056273,
      "learning_rate": 0.0006780504179127734,
      "loss": 4.057,
      "step": 879
    },
    {
      "epoch": 0.4035771612015593,
      "grad_norm": 0.9219573598376047,
      "learning_rate": 0.0006773558853464265,
      "loss": 3.974,
      "step": 880
    },
    {
      "epoch": 0.4040357716120156,
      "grad_norm": 0.8625821793579006,
      "learning_rate": 0.000676660961096555,
      "loss": 4.031,
      "step": 881
    },
    {
      "epoch": 0.4044943820224719,
      "grad_norm": 0.8655909843165139,
      "learning_rate": 0.000675965646697872,
      "loss": 4.042,
      "step": 882
    },
    {
      "epoch": 0.4049529924329282,
      "grad_norm": 0.9023004576693187,
      "learning_rate": 0.0006752699436859519,
      "loss": 3.951,
      "step": 883
    },
    {
      "epoch": 0.40541160284338457,
      "grad_norm": 0.9090225650426957,
      "learning_rate": 0.0006745738535972279,
      "loss": 4.07,
      "step": 884
    },
    {
      "epoch": 0.4058702132538409,
      "grad_norm": 0.8757012943705063,
      "learning_rate": 0.0006738773779689874,
      "loss": 3.9811,
      "step": 885
    },
    {
      "epoch": 0.4063288236642972,
      "grad_norm": 0.9182649631639688,
      "learning_rate": 0.0006731805183393696,
      "loss": 4.0037,
      "step": 886
    },
    {
      "epoch": 0.4067874340747535,
      "grad_norm": 0.9288454715465556,
      "learning_rate": 0.0006724832762473618,
      "loss": 3.9968,
      "step": 887
    },
    {
      "epoch": 0.4072460444852098,
      "grad_norm": 0.9408804520282358,
      "learning_rate": 0.0006717856532327956,
      "loss": 3.9723,
      "step": 888
    },
    {
      "epoch": 0.40770465489566615,
      "grad_norm": 0.9683719553982374,
      "learning_rate": 0.0006710876508363444,
      "loss": 4.1106,
      "step": 889
    },
    {
      "epoch": 0.40816326530612246,
      "grad_norm": 0.9586505172790123,
      "learning_rate": 0.0006703892705995189,
      "loss": 4.1046,
      "step": 890
    },
    {
      "epoch": 0.40862187571657876,
      "grad_norm": 0.8935499519472804,
      "learning_rate": 0.0006696905140646647,
      "loss": 3.9823,
      "step": 891
    },
    {
      "epoch": 0.40908048612703507,
      "grad_norm": 0.9145377280896959,
      "learning_rate": 0.0006689913827749581,
      "loss": 4.0551,
      "step": 892
    },
    {
      "epoch": 0.4095390965374914,
      "grad_norm": 1.0543324380602936,
      "learning_rate": 0.0006682918782744032,
      "loss": 3.7134,
      "step": 893
    },
    {
      "epoch": 0.40999770694794774,
      "grad_norm": 0.9033643196315053,
      "learning_rate": 0.0006675920021078282,
      "loss": 4.3136,
      "step": 894
    },
    {
      "epoch": 0.41045631735840404,
      "grad_norm": 0.8631365350540151,
      "learning_rate": 0.0006668917558208823,
      "loss": 4.0138,
      "step": 895
    },
    {
      "epoch": 0.41091492776886035,
      "grad_norm": 0.9256381621421464,
      "learning_rate": 0.0006661911409600321,
      "loss": 4.0647,
      "step": 896
    },
    {
      "epoch": 0.41137353817931666,
      "grad_norm": 0.8788779259462295,
      "learning_rate": 0.0006654901590725577,
      "loss": 4.0831,
      "step": 897
    },
    {
      "epoch": 0.41183214858977296,
      "grad_norm": 0.8722680095775824,
      "learning_rate": 0.0006647888117065507,
      "loss": 3.9645,
      "step": 898
    },
    {
      "epoch": 0.4122907590002293,
      "grad_norm": 0.8785044949268789,
      "learning_rate": 0.0006640871004109086,
      "loss": 3.9681,
      "step": 899
    },
    {
      "epoch": 0.41274936941068563,
      "grad_norm": 0.87234415158041,
      "learning_rate": 0.000663385026735334,
      "loss": 4.1561,
      "step": 900
    },
    {
      "epoch": 0.41320797982114194,
      "grad_norm": 0.8929301153212248,
      "learning_rate": 0.0006626825922303285,
      "loss": 4.0581,
      "step": 901
    },
    {
      "epoch": 0.41366659023159824,
      "grad_norm": 0.8534719992864876,
      "learning_rate": 0.0006619797984471915,
      "loss": 4.1023,
      "step": 902
    },
    {
      "epoch": 0.41412520064205455,
      "grad_norm": 0.9114197149496868,
      "learning_rate": 0.0006612766469380158,
      "loss": 4.3292,
      "step": 903
    },
    {
      "epoch": 0.4145838110525109,
      "grad_norm": 0.8866044931383896,
      "learning_rate": 0.0006605731392556833,
      "loss": 4.0935,
      "step": 904
    },
    {
      "epoch": 0.4150424214629672,
      "grad_norm": 0.867644649423686,
      "learning_rate": 0.0006598692769538637,
      "loss": 4.1427,
      "step": 905
    },
    {
      "epoch": 0.4155010318734235,
      "grad_norm": 0.8662435472794747,
      "learning_rate": 0.0006591650615870091,
      "loss": 4.1169,
      "step": 906
    },
    {
      "epoch": 0.4159596422838798,
      "grad_norm": 0.8798807860720519,
      "learning_rate": 0.0006584604947103514,
      "loss": 4.0089,
      "step": 907
    },
    {
      "epoch": 0.4164182526943362,
      "grad_norm": 0.9102655153830563,
      "learning_rate": 0.0006577555778798993,
      "loss": 3.9404,
      "step": 908
    },
    {
      "epoch": 0.4168768631047925,
      "grad_norm": 0.8613170425195118,
      "learning_rate": 0.0006570503126524336,
      "loss": 4.0219,
      "step": 909
    },
    {
      "epoch": 0.4173354735152488,
      "grad_norm": 0.9274134638042333,
      "learning_rate": 0.0006563447005855054,
      "loss": 4.0304,
      "step": 910
    },
    {
      "epoch": 0.4177940839257051,
      "grad_norm": 0.9094105457129602,
      "learning_rate": 0.000655638743237431,
      "loss": 4.1475,
      "step": 911
    },
    {
      "epoch": 0.4182526943361614,
      "grad_norm": 0.8835494238176675,
      "learning_rate": 0.0006549324421672894,
      "loss": 4.0793,
      "step": 912
    },
    {
      "epoch": 0.4187113047466178,
      "grad_norm": 0.8790525010073729,
      "learning_rate": 0.0006542257989349194,
      "loss": 4.1361,
      "step": 913
    },
    {
      "epoch": 0.4191699151570741,
      "grad_norm": 0.917420400528014,
      "learning_rate": 0.0006535188151009142,
      "loss": 4.1625,
      "step": 914
    },
    {
      "epoch": 0.4196285255675304,
      "grad_norm": 0.8655393557558054,
      "learning_rate": 0.0006528114922266204,
      "loss": 4.095,
      "step": 915
    },
    {
      "epoch": 0.4200871359779867,
      "grad_norm": 0.8487909109939115,
      "learning_rate": 0.0006521038318741327,
      "loss": 4.0309,
      "step": 916
    },
    {
      "epoch": 0.420545746388443,
      "grad_norm": 0.8048106548939055,
      "learning_rate": 0.0006513958356062912,
      "loss": 3.9567,
      "step": 917
    },
    {
      "epoch": 0.42100435679889936,
      "grad_norm": 0.9154598175577628,
      "learning_rate": 0.0006506875049866781,
      "loss": 4.0964,
      "step": 918
    },
    {
      "epoch": 0.42146296720935567,
      "grad_norm": 0.8583399139152715,
      "learning_rate": 0.0006499788415796137,
      "loss": 4.0087,
      "step": 919
    },
    {
      "epoch": 0.42192157761981197,
      "grad_norm": 0.8193801309854472,
      "learning_rate": 0.0006492698469501532,
      "loss": 4.0601,
      "step": 920
    },
    {
      "epoch": 0.4223801880302683,
      "grad_norm": 0.8363152298407556,
      "learning_rate": 0.0006485605226640837,
      "loss": 4.0882,
      "step": 921
    },
    {
      "epoch": 0.4228387984407246,
      "grad_norm": 0.8400862673538964,
      "learning_rate": 0.00064785087028792,
      "loss": 3.915,
      "step": 922
    },
    {
      "epoch": 0.42329740885118095,
      "grad_norm": 0.8905708674033345,
      "learning_rate": 0.0006471408913889019,
      "loss": 4.0087,
      "step": 923
    },
    {
      "epoch": 0.42375601926163725,
      "grad_norm": 0.8772190974083053,
      "learning_rate": 0.0006464305875349892,
      "loss": 3.9426,
      "step": 924
    },
    {
      "epoch": 0.42421462967209356,
      "grad_norm": 0.8622237361374623,
      "learning_rate": 0.000645719960294861,
      "loss": 3.8795,
      "step": 925
    },
    {
      "epoch": 0.42467324008254986,
      "grad_norm": 0.9314636581215623,
      "learning_rate": 0.0006450090112379092,
      "loss": 3.9746,
      "step": 926
    },
    {
      "epoch": 0.42513185049300617,
      "grad_norm": 0.9001172161594793,
      "learning_rate": 0.0006442977419342371,
      "loss": 4.0407,
      "step": 927
    },
    {
      "epoch": 0.42559046090346253,
      "grad_norm": 0.8635637702426278,
      "learning_rate": 0.000643586153954655,
      "loss": 4.0068,
      "step": 928
    },
    {
      "epoch": 0.42604907131391884,
      "grad_norm": 0.9254812013911382,
      "learning_rate": 0.0006428742488706772,
      "loss": 3.8913,
      "step": 929
    },
    {
      "epoch": 0.42650768172437514,
      "grad_norm": 0.865163527068074,
      "learning_rate": 0.0006421620282545182,
      "loss": 4.1661,
      "step": 930
    },
    {
      "epoch": 0.42696629213483145,
      "grad_norm": 0.8977064537803436,
      "learning_rate": 0.0006414494936790892,
      "loss": 3.9844,
      "step": 931
    },
    {
      "epoch": 0.42742490254528775,
      "grad_norm": 0.8636255702287435,
      "learning_rate": 0.0006407366467179951,
      "loss": 4.1191,
      "step": 932
    },
    {
      "epoch": 0.4278835129557441,
      "grad_norm": 0.8560182225300311,
      "learning_rate": 0.0006400234889455301,
      "loss": 4.1002,
      "step": 933
    },
    {
      "epoch": 0.4283421233662004,
      "grad_norm": 0.8597329903149598,
      "learning_rate": 0.0006393100219366755,
      "loss": 3.9454,
      "step": 934
    },
    {
      "epoch": 0.42880073377665673,
      "grad_norm": 0.8943695610417158,
      "learning_rate": 0.0006385962472670953,
      "loss": 4.057,
      "step": 935
    },
    {
      "epoch": 0.42925934418711303,
      "grad_norm": 0.8027943344532441,
      "learning_rate": 0.0006378821665131328,
      "loss": 4.1838,
      "step": 936
    },
    {
      "epoch": 0.42971795459756934,
      "grad_norm": 0.8232632919402921,
      "learning_rate": 0.0006371677812518072,
      "loss": 4.0753,
      "step": 937
    },
    {
      "epoch": 0.4301765650080257,
      "grad_norm": 0.8224456611533565,
      "learning_rate": 0.0006364530930608107,
      "loss": 4.0781,
      "step": 938
    },
    {
      "epoch": 0.430635175418482,
      "grad_norm": 0.8504123970577474,
      "learning_rate": 0.0006357381035185038,
      "loss": 4.1247,
      "step": 939
    },
    {
      "epoch": 0.4310937858289383,
      "grad_norm": 0.9035456750027031,
      "learning_rate": 0.0006350228142039131,
      "loss": 4.0668,
      "step": 940
    },
    {
      "epoch": 0.4315523962393946,
      "grad_norm": 0.8117713423333338,
      "learning_rate": 0.000634307226696727,
      "loss": 4.2163,
      "step": 941
    },
    {
      "epoch": 0.4320110066498509,
      "grad_norm": 0.8370684544738802,
      "learning_rate": 0.0006335913425772926,
      "loss": 3.8496,
      "step": 942
    },
    {
      "epoch": 0.4324696170603073,
      "grad_norm": 0.869176317403616,
      "learning_rate": 0.0006328751634266117,
      "loss": 4.062,
      "step": 943
    },
    {
      "epoch": 0.4329282274707636,
      "grad_norm": 0.924132888095462,
      "learning_rate": 0.0006321586908263382,
      "loss": 4.1133,
      "step": 944
    },
    {
      "epoch": 0.4333868378812199,
      "grad_norm": 0.8487305584882137,
      "learning_rate": 0.0006314419263587732,
      "loss": 4.007,
      "step": 945
    },
    {
      "epoch": 0.4338454482916762,
      "grad_norm": 0.9641814632605741,
      "learning_rate": 0.0006307248716068637,
      "loss": 3.7564,
      "step": 946
    },
    {
      "epoch": 0.4343040587021325,
      "grad_norm": 0.9058294211839414,
      "learning_rate": 0.0006300075281541964,
      "loss": 4.29,
      "step": 947
    },
    {
      "epoch": 0.4347626691125889,
      "grad_norm": 0.9408251425104568,
      "learning_rate": 0.0006292898975849966,
      "loss": 4.037,
      "step": 948
    },
    {
      "epoch": 0.4352212795230452,
      "grad_norm": 0.8761143118911825,
      "learning_rate": 0.000628571981484123,
      "loss": 4.2461,
      "step": 949
    },
    {
      "epoch": 0.4356798899335015,
      "grad_norm": 0.8930831462319441,
      "learning_rate": 0.0006278537814370654,
      "loss": 4.054,
      "step": 950
    },
    {
      "epoch": 0.4361385003439578,
      "grad_norm": 0.8416105819886343,
      "learning_rate": 0.0006271352990299406,
      "loss": 3.9646,
      "step": 951
    },
    {
      "epoch": 0.43659711075441415,
      "grad_norm": 0.8706256898338477,
      "learning_rate": 0.0006264165358494885,
      "loss": 4.1123,
      "step": 952
    },
    {
      "epoch": 0.43705572116487046,
      "grad_norm": 0.8872517509925854,
      "learning_rate": 0.0006256974934830694,
      "loss": 3.9452,
      "step": 953
    },
    {
      "epoch": 0.43751433157532676,
      "grad_norm": 0.8742788679474678,
      "learning_rate": 0.0006249781735186606,
      "loss": 4.0424,
      "step": 954
    },
    {
      "epoch": 0.43797294198578307,
      "grad_norm": 0.8808710970807023,
      "learning_rate": 0.0006242585775448518,
      "loss": 3.9219,
      "step": 955
    },
    {
      "epoch": 0.4384315523962394,
      "grad_norm": 0.8961289063124063,
      "learning_rate": 0.0006235387071508427,
      "loss": 4.0925,
      "step": 956
    },
    {
      "epoch": 0.43889016280669574,
      "grad_norm": 0.8849807339872224,
      "learning_rate": 0.0006228185639264384,
      "loss": 3.9507,
      "step": 957
    },
    {
      "epoch": 0.43934877321715204,
      "grad_norm": 0.9038518285276887,
      "learning_rate": 0.0006220981494620475,
      "loss": 4.1506,
      "step": 958
    },
    {
      "epoch": 0.43980738362760835,
      "grad_norm": 0.856432383971463,
      "learning_rate": 0.000621377465348677,
      "loss": 3.9403,
      "step": 959
    },
    {
      "epoch": 0.44026599403806466,
      "grad_norm": 0.8829891069853008,
      "learning_rate": 0.0006206565131779293,
      "loss": 4.0918,
      "step": 960
    },
    {
      "epoch": 0.44072460444852096,
      "grad_norm": 0.938943594376205,
      "learning_rate": 0.0006199352945419994,
      "loss": 4.0403,
      "step": 961
    },
    {
      "epoch": 0.4411832148589773,
      "grad_norm": 0.8684609748071331,
      "learning_rate": 0.00061921381103367,
      "loss": 4.051,
      "step": 962
    },
    {
      "epoch": 0.44164182526943363,
      "grad_norm": 0.8826917424889226,
      "learning_rate": 0.0006184920642463094,
      "loss": 4.1128,
      "step": 963
    },
    {
      "epoch": 0.44210043567988994,
      "grad_norm": 0.8800640723930493,
      "learning_rate": 0.0006177700557738672,
      "loss": 3.9331,
      "step": 964
    },
    {
      "epoch": 0.44255904609034624,
      "grad_norm": 0.8643806731187181,
      "learning_rate": 0.0006170477872108706,
      "loss": 4.0493,
      "step": 965
    },
    {
      "epoch": 0.44301765650080255,
      "grad_norm": 0.8702396845512184,
      "learning_rate": 0.0006163252601524216,
      "loss": 4.1185,
      "step": 966
    },
    {
      "epoch": 0.4434762669112589,
      "grad_norm": 0.825498918119337,
      "learning_rate": 0.0006156024761941925,
      "loss": 4.088,
      "step": 967
    },
    {
      "epoch": 0.4439348773217152,
      "grad_norm": 0.8264240980105254,
      "learning_rate": 0.000614879436932424,
      "loss": 3.958,
      "step": 968
    },
    {
      "epoch": 0.4443934877321715,
      "grad_norm": 0.8410292555775948,
      "learning_rate": 0.0006141561439639196,
      "loss": 3.9983,
      "step": 969
    },
    {
      "epoch": 0.4448520981426278,
      "grad_norm": 0.8449779913587135,
      "learning_rate": 0.0006134325988860433,
      "loss": 3.9912,
      "step": 970
    },
    {
      "epoch": 0.44531070855308413,
      "grad_norm": 0.8237387917351962,
      "learning_rate": 0.0006127088032967165,
      "loss": 4.0167,
      "step": 971
    },
    {
      "epoch": 0.4457693189635405,
      "grad_norm": 0.8731575758087284,
      "learning_rate": 0.0006119847587944131,
      "loss": 4.0582,
      "step": 972
    },
    {
      "epoch": 0.4462279293739968,
      "grad_norm": 0.8913155976646492,
      "learning_rate": 0.0006112604669781572,
      "loss": 4.262,
      "step": 973
    },
    {
      "epoch": 0.4466865397844531,
      "grad_norm": 0.8122108898206493,
      "learning_rate": 0.0006105359294475188,
      "loss": 3.9709,
      "step": 974
    },
    {
      "epoch": 0.4471451501949094,
      "grad_norm": 0.8706819807696702,
      "learning_rate": 0.0006098111478026107,
      "loss": 4.1933,
      "step": 975
    },
    {
      "epoch": 0.4476037606053657,
      "grad_norm": 0.8469071107557692,
      "learning_rate": 0.0006090861236440848,
      "loss": 3.9775,
      "step": 976
    },
    {
      "epoch": 0.4480623710158221,
      "grad_norm": 0.890280460599693,
      "learning_rate": 0.0006083608585731282,
      "loss": 3.9942,
      "step": 977
    },
    {
      "epoch": 0.4485209814262784,
      "grad_norm": 0.8296911964675399,
      "learning_rate": 0.0006076353541914609,
      "loss": 4.1019,
      "step": 978
    },
    {
      "epoch": 0.4489795918367347,
      "grad_norm": 0.8458512082409888,
      "learning_rate": 0.0006069096121013307,
      "loss": 3.99,
      "step": 979
    },
    {
      "epoch": 0.449438202247191,
      "grad_norm": 0.8345350771727671,
      "learning_rate": 0.0006061836339055105,
      "loss": 4.0207,
      "step": 980
    },
    {
      "epoch": 0.4498968126576473,
      "grad_norm": 0.8692184836628772,
      "learning_rate": 0.0006054574212072948,
      "loss": 4.0159,
      "step": 981
    },
    {
      "epoch": 0.45035542306810367,
      "grad_norm": 0.8829747935409601,
      "learning_rate": 0.0006047309756104958,
      "loss": 4.0272,
      "step": 982
    },
    {
      "epoch": 0.45081403347856,
      "grad_norm": 0.8757713387389586,
      "learning_rate": 0.00060400429871944,
      "loss": 3.9547,
      "step": 983
    },
    {
      "epoch": 0.4512726438890163,
      "grad_norm": 0.8938648805904006,
      "learning_rate": 0.0006032773921389654,
      "loss": 4.0744,
      "step": 984
    },
    {
      "epoch": 0.4517312542994726,
      "grad_norm": 0.854885725679199,
      "learning_rate": 0.0006025502574744162,
      "loss": 4.0089,
      "step": 985
    },
    {
      "epoch": 0.4521898647099289,
      "grad_norm": 0.9094125667617234,
      "learning_rate": 0.000601822896331641,
      "loss": 3.9849,
      "step": 986
    },
    {
      "epoch": 0.45264847512038525,
      "grad_norm": 0.8915744283293024,
      "learning_rate": 0.0006010953103169883,
      "loss": 4.1104,
      "step": 987
    },
    {
      "epoch": 0.45310708553084156,
      "grad_norm": 0.8470383877551108,
      "learning_rate": 0.0006003675010373034,
      "loss": 4.0706,
      "step": 988
    },
    {
      "epoch": 0.45356569594129786,
      "grad_norm": 0.8048399834576214,
      "learning_rate": 0.0005996394700999246,
      "loss": 3.9363,
      "step": 989
    },
    {
      "epoch": 0.45402430635175417,
      "grad_norm": 0.8683408127086488,
      "learning_rate": 0.0005989112191126794,
      "loss": 3.8773,
      "step": 990
    },
    {
      "epoch": 0.4544829167622105,
      "grad_norm": 0.8373712469949989,
      "learning_rate": 0.0005981827496838822,
      "loss": 4.0722,
      "step": 991
    },
    {
      "epoch": 0.45494152717266684,
      "grad_norm": 0.9056011083258807,
      "learning_rate": 0.0005974540634223286,
      "loss": 4.1113,
      "step": 992
    },
    {
      "epoch": 0.45540013758312314,
      "grad_norm": 0.883051061467609,
      "learning_rate": 0.0005967251619372939,
      "loss": 4.1126,
      "step": 993
    },
    {
      "epoch": 0.45585874799357945,
      "grad_norm": 0.8878152711681365,
      "learning_rate": 0.0005959960468385284,
      "loss": 3.9594,
      "step": 994
    },
    {
      "epoch": 0.45631735840403576,
      "grad_norm": 0.8506078761060287,
      "learning_rate": 0.0005952667197362542,
      "loss": 4.0666,
      "step": 995
    },
    {
      "epoch": 0.45677596881449206,
      "grad_norm": 0.8883402666820127,
      "learning_rate": 0.0005945371822411621,
      "loss": 3.8944,
      "step": 996
    },
    {
      "epoch": 0.4572345792249484,
      "grad_norm": 0.8735917916870409,
      "learning_rate": 0.0005938074359644063,
      "loss": 4.0507,
      "step": 997
    },
    {
      "epoch": 0.45769318963540473,
      "grad_norm": 0.8614444683825694,
      "learning_rate": 0.0005930774825176034,
      "loss": 4.0153,
      "step": 998
    },
    {
      "epoch": 0.45815180004586104,
      "grad_norm": 0.9011607140270347,
      "learning_rate": 0.0005923473235128268,
      "loss": 3.9533,
      "step": 999
    },
    {
      "epoch": 0.45861041045631734,
      "grad_norm": 0.8687988860433068,
      "learning_rate": 0.0005916169605626042,
      "loss": 4.0451,
      "step": 1000
    },
    {
      "epoch": 0.4590690208667737,
      "grad_norm": 0.8341914306292794,
      "learning_rate": 0.0005908863952799134,
      "loss": 3.9902,
      "step": 1001
    },
    {
      "epoch": 0.45952763127723,
      "grad_norm": 0.8480068739093757,
      "learning_rate": 0.0005901556292781793,
      "loss": 3.9116,
      "step": 1002
    },
    {
      "epoch": 0.4599862416876863,
      "grad_norm": 0.8567063810165959,
      "learning_rate": 0.0005894246641712698,
      "loss": 4.1067,
      "step": 1003
    },
    {
      "epoch": 0.4604448520981426,
      "grad_norm": 0.8471598982083012,
      "learning_rate": 0.0005886935015734931,
      "loss": 3.9683,
      "step": 1004
    },
    {
      "epoch": 0.4609034625085989,
      "grad_norm": 0.785490619574275,
      "learning_rate": 0.0005879621430995928,
      "loss": 4.079,
      "step": 1005
    },
    {
      "epoch": 0.4613620729190553,
      "grad_norm": 0.8783042692590106,
      "learning_rate": 0.0005872305903647455,
      "loss": 3.9658,
      "step": 1006
    },
    {
      "epoch": 0.4618206833295116,
      "grad_norm": 0.8341713121833756,
      "learning_rate": 0.0005864988449845569,
      "loss": 4.0678,
      "step": 1007
    },
    {
      "epoch": 0.4622792937399679,
      "grad_norm": 0.8718670630072387,
      "learning_rate": 0.0005857669085750578,
      "loss": 4.0895,
      "step": 1008
    },
    {
      "epoch": 0.4627379041504242,
      "grad_norm": 0.8379273070350968,
      "learning_rate": 0.0005850347827527013,
      "loss": 3.9229,
      "step": 1009
    },
    {
      "epoch": 0.4631965145608805,
      "grad_norm": 0.8502570489823545,
      "learning_rate": 0.0005843024691343584,
      "loss": 3.9565,
      "step": 1010
    },
    {
      "epoch": 0.4636551249713369,
      "grad_norm": 0.825450432012477,
      "learning_rate": 0.000583569969337315,
      "loss": 3.9706,
      "step": 1011
    },
    {
      "epoch": 0.4641137353817932,
      "grad_norm": 0.8392582274492076,
      "learning_rate": 0.0005828372849792686,
      "loss": 4.0408,
      "step": 1012
    },
    {
      "epoch": 0.4645723457922495,
      "grad_norm": 0.8739397432437715,
      "learning_rate": 0.0005821044176783234,
      "loss": 4.001,
      "step": 1013
    },
    {
      "epoch": 0.4650309562027058,
      "grad_norm": 0.865479628020138,
      "learning_rate": 0.0005813713690529886,
      "loss": 4.0853,
      "step": 1014
    },
    {
      "epoch": 0.4654895666131621,
      "grad_norm": 0.8353965852501258,
      "learning_rate": 0.0005806381407221729,
      "loss": 3.9867,
      "step": 1015
    },
    {
      "epoch": 0.46594817702361846,
      "grad_norm": 0.8498284767164144,
      "learning_rate": 0.0005799047343051826,
      "loss": 4.0612,
      "step": 1016
    },
    {
      "epoch": 0.46640678743407477,
      "grad_norm": 0.8668597943019036,
      "learning_rate": 0.0005791711514217171,
      "loss": 4.0874,
      "step": 1017
    },
    {
      "epoch": 0.46686539784453107,
      "grad_norm": 0.8407042489156414,
      "learning_rate": 0.0005784373936918654,
      "loss": 4.0608,
      "step": 1018
    },
    {
      "epoch": 0.4673240082549874,
      "grad_norm": 0.8834153017293637,
      "learning_rate": 0.0005777034627361025,
      "loss": 3.85,
      "step": 1019
    },
    {
      "epoch": 0.4677826186654437,
      "grad_norm": 0.9503215615750892,
      "learning_rate": 0.0005769693601752864,
      "loss": 4.1368,
      "step": 1020
    },
    {
      "epoch": 0.46824122907590005,
      "grad_norm": 0.834158323414524,
      "learning_rate": 0.0005762350876306537,
      "loss": 4.0571,
      "step": 1021
    },
    {
      "epoch": 0.46869983948635635,
      "grad_norm": 0.8602960381360125,
      "learning_rate": 0.0005755006467238168,
      "loss": 4.0275,
      "step": 1022
    },
    {
      "epoch": 0.46915844989681266,
      "grad_norm": 0.932500518621388,
      "learning_rate": 0.0005747660390767593,
      "loss": 3.9778,
      "step": 1023
    },
    {
      "epoch": 0.46961706030726896,
      "grad_norm": 0.879346194360602,
      "learning_rate": 0.0005740312663118338,
      "loss": 3.9397,
      "step": 1024
    },
    {
      "epoch": 0.47007567071772527,
      "grad_norm": 0.8781696167824469,
      "learning_rate": 0.0005732963300517568,
      "loss": 4.1832,
      "step": 1025
    },
    {
      "epoch": 0.47053428112818163,
      "grad_norm": 0.8562884505172144,
      "learning_rate": 0.0005725612319196064,
      "loss": 4.0231,
      "step": 1026
    },
    {
      "epoch": 0.47099289153863794,
      "grad_norm": 0.8587788781977856,
      "learning_rate": 0.000571825973538818,
      "loss": 4.0183,
      "step": 1027
    },
    {
      "epoch": 0.47145150194909424,
      "grad_norm": 0.7919865600959863,
      "learning_rate": 0.0005710905565331811,
      "loss": 4.0722,
      "step": 1028
    },
    {
      "epoch": 0.47191011235955055,
      "grad_norm": 0.8372255461209692,
      "learning_rate": 0.0005703549825268353,
      "loss": 3.9727,
      "step": 1029
    },
    {
      "epoch": 0.47236872277000685,
      "grad_norm": 0.8061639622036475,
      "learning_rate": 0.0005696192531442667,
      "loss": 4.1971,
      "step": 1030
    },
    {
      "epoch": 0.4728273331804632,
      "grad_norm": 0.8316404033009281,
      "learning_rate": 0.000568883370010305,
      "loss": 4.3679,
      "step": 1031
    },
    {
      "epoch": 0.4732859435909195,
      "grad_norm": 0.8705581934615126,
      "learning_rate": 0.0005681473347501192,
      "loss": 3.9357,
      "step": 1032
    },
    {
      "epoch": 0.47374455400137583,
      "grad_norm": 0.8466603462249649,
      "learning_rate": 0.0005674111489892144,
      "loss": 4.0354,
      "step": 1033
    },
    {
      "epoch": 0.47420316441183213,
      "grad_norm": 0.8190852811642318,
      "learning_rate": 0.0005666748143534282,
      "loss": 4.003,
      "step": 1034
    },
    {
      "epoch": 0.47466177482228844,
      "grad_norm": 0.9114802205205319,
      "learning_rate": 0.0005659383324689266,
      "loss": 4.0447,
      "step": 1035
    },
    {
      "epoch": 0.4751203852327448,
      "grad_norm": 0.9499313516149608,
      "learning_rate": 0.0005652017049622007,
      "loss": 4.0206,
      "step": 1036
    },
    {
      "epoch": 0.4755789956432011,
      "grad_norm": 0.8570928083356836,
      "learning_rate": 0.0005644649334600641,
      "loss": 4.1545,
      "step": 1037
    },
    {
      "epoch": 0.4760376060536574,
      "grad_norm": 0.8354432441903566,
      "learning_rate": 0.0005637280195896474,
      "loss": 4.1353,
      "step": 1038
    },
    {
      "epoch": 0.4764962164641137,
      "grad_norm": 0.8240883620120258,
      "learning_rate": 0.0005629909649783961,
      "loss": 4.264,
      "step": 1039
    },
    {
      "epoch": 0.47695482687457,
      "grad_norm": 0.9204525677246502,
      "learning_rate": 0.0005622537712540664,
      "loss": 3.9849,
      "step": 1040
    },
    {
      "epoch": 0.4774134372850264,
      "grad_norm": 0.8618096521561374,
      "learning_rate": 0.0005615164400447218,
      "loss": 4.0507,
      "step": 1041
    },
    {
      "epoch": 0.4778720476954827,
      "grad_norm": 0.8276839808245064,
      "learning_rate": 0.0005607789729787294,
      "loss": 3.8848,
      "step": 1042
    },
    {
      "epoch": 0.478330658105939,
      "grad_norm": 0.89772950139979,
      "learning_rate": 0.0005600413716847564,
      "loss": 3.9601,
      "step": 1043
    },
    {
      "epoch": 0.4787892685163953,
      "grad_norm": 0.8465680776993538,
      "learning_rate": 0.000559303637791766,
      "loss": 3.9736,
      "step": 1044
    },
    {
      "epoch": 0.47924787892685167,
      "grad_norm": 0.8499621628963232,
      "learning_rate": 0.0005585657729290151,
      "loss": 4.0117,
      "step": 1045
    },
    {
      "epoch": 0.479706489337308,
      "grad_norm": 0.8419295679909813,
      "learning_rate": 0.000557827778726049,
      "loss": 3.9294,
      "step": 1046
    },
    {
      "epoch": 0.4801650997477643,
      "grad_norm": 0.840591692829678,
      "learning_rate": 0.0005570896568126993,
      "loss": 3.9573,
      "step": 1047
    },
    {
      "epoch": 0.4806237101582206,
      "grad_norm": 0.9011706231414043,
      "learning_rate": 0.0005563514088190788,
      "loss": 3.9234,
      "step": 1048
    },
    {
      "epoch": 0.4810823205686769,
      "grad_norm": 0.9199920159831172,
      "learning_rate": 0.0005556130363755798,
      "loss": 3.7514,
      "step": 1049
    },
    {
      "epoch": 0.48154093097913325,
      "grad_norm": 0.8777453170076515,
      "learning_rate": 0.0005548745411128688,
      "loss": 3.9543,
      "step": 1050
    },
    {
      "epoch": 0.48199954138958956,
      "grad_norm": 0.9563777117786407,
      "learning_rate": 0.0005541359246618835,
      "loss": 4.1124,
      "step": 1051
    },
    {
      "epoch": 0.48245815180004586,
      "grad_norm": 0.9223793764461932,
      "learning_rate": 0.0005533971886538293,
      "loss": 4.0643,
      "step": 1052
    },
    {
      "epoch": 0.48291676221050217,
      "grad_norm": 0.8719430621838554,
      "learning_rate": 0.000552658334720176,
      "loss": 3.9334,
      "step": 1053
    },
    {
      "epoch": 0.4833753726209585,
      "grad_norm": 0.9329803610990207,
      "learning_rate": 0.0005519193644926535,
      "loss": 3.9369,
      "step": 1054
    },
    {
      "epoch": 0.48383398303141484,
      "grad_norm": 0.9061194885448639,
      "learning_rate": 0.0005511802796032485,
      "loss": 3.977,
      "step": 1055
    },
    {
      "epoch": 0.48429259344187114,
      "grad_norm": 0.8349774598834837,
      "learning_rate": 0.0005504410816842009,
      "loss": 3.9549,
      "step": 1056
    },
    {
      "epoch": 0.48475120385232745,
      "grad_norm": 0.7901177703071475,
      "learning_rate": 0.0005497017723680009,
      "loss": 3.9918,
      "step": 1057
    },
    {
      "epoch": 0.48520981426278376,
      "grad_norm": 0.8152545084064261,
      "learning_rate": 0.0005489623532873836,
      "loss": 4.0189,
      "step": 1058
    },
    {
      "epoch": 0.48566842467324006,
      "grad_norm": 0.8446022066189335,
      "learning_rate": 0.0005482228260753273,
      "loss": 4.1246,
      "step": 1059
    },
    {
      "epoch": 0.4861270350836964,
      "grad_norm": 0.8984048124773267,
      "learning_rate": 0.0005474831923650488,
      "loss": 3.9122,
      "step": 1060
    },
    {
      "epoch": 0.48658564549415273,
      "grad_norm": 0.928219542881913,
      "learning_rate": 0.00054674345379,
      "loss": 4.0126,
      "step": 1061
    },
    {
      "epoch": 0.48704425590460904,
      "grad_norm": 0.840389623779675,
      "learning_rate": 0.000546003611983865,
      "loss": 3.9273,
      "step": 1062
    },
    {
      "epoch": 0.48750286631506534,
      "grad_norm": 0.8381470619053049,
      "learning_rate": 0.0005452636685805552,
      "loss": 4.2545,
      "step": 1063
    },
    {
      "epoch": 0.48796147672552165,
      "grad_norm": 0.9356771654139964,
      "learning_rate": 0.0005445236252142066,
      "loss": 4.0204,
      "step": 1064
    },
    {
      "epoch": 0.488420087135978,
      "grad_norm": 0.9392163016339402,
      "learning_rate": 0.000543783483519176,
      "loss": 3.9173,
      "step": 1065
    },
    {
      "epoch": 0.4888786975464343,
      "grad_norm": 0.8192649999288789,
      "learning_rate": 0.0005430432451300374,
      "loss": 3.9058,
      "step": 1066
    },
    {
      "epoch": 0.4893373079568906,
      "grad_norm": 0.8167936661259081,
      "learning_rate": 0.0005423029116815781,
      "loss": 4.0162,
      "step": 1067
    },
    {
      "epoch": 0.4897959183673469,
      "grad_norm": 0.9178246032377416,
      "learning_rate": 0.0005415624848087959,
      "loss": 4.1363,
      "step": 1068
    },
    {
      "epoch": 0.49025452877780323,
      "grad_norm": 0.8700541900002499,
      "learning_rate": 0.000540821966146894,
      "loss": 3.9273,
      "step": 1069
    },
    {
      "epoch": 0.4907131391882596,
      "grad_norm": 0.8700713478455369,
      "learning_rate": 0.0005400813573312793,
      "loss": 4.0409,
      "step": 1070
    },
    {
      "epoch": 0.4911717495987159,
      "grad_norm": 0.8085295831978601,
      "learning_rate": 0.0005393406599975572,
      "loss": 4.0784,
      "step": 1071
    },
    {
      "epoch": 0.4916303600091722,
      "grad_norm": 0.8654545335446551,
      "learning_rate": 0.0005385998757815287,
      "loss": 3.9376,
      "step": 1072
    },
    {
      "epoch": 0.4920889704196285,
      "grad_norm": 0.8570726996324632,
      "learning_rate": 0.0005378590063191867,
      "loss": 4.051,
      "step": 1073
    },
    {
      "epoch": 0.4925475808300848,
      "grad_norm": 0.919262310448744,
      "learning_rate": 0.0005371180532467124,
      "loss": 3.9643,
      "step": 1074
    },
    {
      "epoch": 0.4930061912405412,
      "grad_norm": 0.8697461371036209,
      "learning_rate": 0.000536377018200472,
      "loss": 3.9995,
      "step": 1075
    },
    {
      "epoch": 0.4934648016509975,
      "grad_norm": 0.8458737367543695,
      "learning_rate": 0.0005356359028170118,
      "loss": 4.066,
      "step": 1076
    },
    {
      "epoch": 0.4939234120614538,
      "grad_norm": 0.9611042806448151,
      "learning_rate": 0.0005348947087330564,
      "loss": 4.1452,
      "step": 1077
    },
    {
      "epoch": 0.4943820224719101,
      "grad_norm": 0.8843351952909304,
      "learning_rate": 0.0005341534375855037,
      "loss": 3.9811,
      "step": 1078
    },
    {
      "epoch": 0.4948406328823664,
      "grad_norm": 0.8139570741562949,
      "learning_rate": 0.0005334120910114222,
      "loss": 3.9328,
      "step": 1079
    },
    {
      "epoch": 0.49529924329282277,
      "grad_norm": 0.8843702791942254,
      "learning_rate": 0.0005326706706480467,
      "loss": 3.8931,
      "step": 1080
    },
    {
      "epoch": 0.49575785370327907,
      "grad_norm": 0.8307175097973042,
      "learning_rate": 0.0005319291781327749,
      "loss": 4.0505,
      "step": 1081
    },
    {
      "epoch": 0.4962164641137354,
      "grad_norm": 0.8902021646635564,
      "learning_rate": 0.0005311876151031642,
      "loss": 3.9474,
      "step": 1082
    },
    {
      "epoch": 0.4966750745241917,
      "grad_norm": 0.8878195563428339,
      "learning_rate": 0.0005304459831969274,
      "loss": 3.9783,
      "step": 1083
    },
    {
      "epoch": 0.497133684934648,
      "grad_norm": 0.8020839184778524,
      "learning_rate": 0.0005297042840519294,
      "loss": 3.9183,
      "step": 1084
    },
    {
      "epoch": 0.49759229534510435,
      "grad_norm": 0.9306541559426623,
      "learning_rate": 0.0005289625193061838,
      "loss": 3.9399,
      "step": 1085
    },
    {
      "epoch": 0.49805090575556066,
      "grad_norm": 0.8408356611443711,
      "learning_rate": 0.0005282206905978489,
      "loss": 3.9321,
      "step": 1086
    },
    {
      "epoch": 0.49850951616601696,
      "grad_norm": 0.8837895886013931,
      "learning_rate": 0.0005274787995652246,
      "loss": 3.9376,
      "step": 1087
    },
    {
      "epoch": 0.49896812657647327,
      "grad_norm": 0.8760083664826684,
      "learning_rate": 0.000526736847846748,
      "loss": 4.0478,
      "step": 1088
    },
    {
      "epoch": 0.4994267369869296,
      "grad_norm": 0.828092995904643,
      "learning_rate": 0.0005259948370809901,
      "loss": 4.048,
      "step": 1089
    },
    {
      "epoch": 0.49988534739738594,
      "grad_norm": 0.7757576676070874,
      "learning_rate": 0.0005252527689066533,
      "loss": 4.0403,
      "step": 1090
    },
    {
      "epoch": 0.5003439578078422,
      "grad_norm": 0.8653981530632717,
      "learning_rate": 0.0005245106449625654,
      "loss": 3.9373,
      "step": 1091
    },
    {
      "epoch": 0.5008025682182986,
      "grad_norm": 0.8782364780772528,
      "learning_rate": 0.0005237684668876785,
      "loss": 4.1542,
      "step": 1092
    },
    {
      "epoch": 0.5012611786287549,
      "grad_norm": 0.8965428812179721,
      "learning_rate": 0.0005230262363210637,
      "loss": 4.118,
      "step": 1093
    },
    {
      "epoch": 0.5017197890392112,
      "grad_norm": 0.8427958717542839,
      "learning_rate": 0.0005222839549019079,
      "loss": 3.9433,
      "step": 1094
    },
    {
      "epoch": 0.5021783994496675,
      "grad_norm": 0.8231339830398618,
      "learning_rate": 0.0005215416242695108,
      "loss": 4.099,
      "step": 1095
    },
    {
      "epoch": 0.5026370098601238,
      "grad_norm": 0.9294221242849144,
      "learning_rate": 0.0005207992460632804,
      "loss": 3.8767,
      "step": 1096
    },
    {
      "epoch": 0.5030956202705802,
      "grad_norm": 0.9057334476490516,
      "learning_rate": 0.0005200568219227299,
      "loss": 4.0529,
      "step": 1097
    },
    {
      "epoch": 0.5035542306810364,
      "grad_norm": 0.8251150298143428,
      "learning_rate": 0.000519314353487474,
      "loss": 4.0522,
      "step": 1098
    },
    {
      "epoch": 0.5040128410914928,
      "grad_norm": 0.8279543981511379,
      "learning_rate": 0.0005185718423972251,
      "loss": 4.0995,
      "step": 1099
    },
    {
      "epoch": 0.504471451501949,
      "grad_norm": 0.8013377401783501,
      "learning_rate": 0.0005178292902917898,
      "loss": 4.0543,
      "step": 1100
    },
    {
      "epoch": 0.5049300619124054,
      "grad_norm": 0.870114338513729,
      "learning_rate": 0.0005170866988110656,
      "loss": 3.8467,
      "step": 1101
    },
    {
      "epoch": 0.5053886723228618,
      "grad_norm": 0.8235509031702154,
      "learning_rate": 0.0005163440695950362,
      "loss": 3.9447,
      "step": 1102
    },
    {
      "epoch": 0.505847282733318,
      "grad_norm": 0.8526864705454564,
      "learning_rate": 0.0005156014042837695,
      "loss": 4.1007,
      "step": 1103
    },
    {
      "epoch": 0.5063058931437744,
      "grad_norm": 0.8442871877714265,
      "learning_rate": 0.0005148587045174128,
      "loss": 3.9812,
      "step": 1104
    },
    {
      "epoch": 0.5067645035542306,
      "grad_norm": 0.8384061360700004,
      "learning_rate": 0.0005141159719361891,
      "loss": 3.9873,
      "step": 1105
    },
    {
      "epoch": 0.507223113964687,
      "grad_norm": 0.9451566623871935,
      "learning_rate": 0.0005133732081803945,
      "loss": 3.9361,
      "step": 1106
    },
    {
      "epoch": 0.5076817243751434,
      "grad_norm": 0.8811765861389518,
      "learning_rate": 0.0005126304148903936,
      "loss": 4.1209,
      "step": 1107
    },
    {
      "epoch": 0.5081403347855996,
      "grad_norm": 0.9084505825909959,
      "learning_rate": 0.0005118875937066161,
      "loss": 4.0307,
      "step": 1108
    },
    {
      "epoch": 0.508598945196056,
      "grad_norm": 0.8518590533793241,
      "learning_rate": 0.0005111447462695537,
      "loss": 3.9565,
      "step": 1109
    },
    {
      "epoch": 0.5090575556065122,
      "grad_norm": 0.8482789395398891,
      "learning_rate": 0.0005104018742197557,
      "loss": 3.8905,
      "step": 1110
    },
    {
      "epoch": 0.5095161660169686,
      "grad_norm": 0.9710231057892567,
      "learning_rate": 0.0005096589791978261,
      "loss": 4.0829,
      "step": 1111
    },
    {
      "epoch": 0.509974776427425,
      "grad_norm": 0.8448309031856932,
      "learning_rate": 0.0005089160628444192,
      "loss": 4.0799,
      "step": 1112
    },
    {
      "epoch": 0.5104333868378812,
      "grad_norm": 0.8221046721758846,
      "learning_rate": 0.0005081731268002371,
      "loss": 3.8419,
      "step": 1113
    },
    {
      "epoch": 0.5108919972483376,
      "grad_norm": 0.8189481569486858,
      "learning_rate": 0.0005074301727060243,
      "loss": 3.8641,
      "step": 1114
    },
    {
      "epoch": 0.5113506076587938,
      "grad_norm": 0.9113271205551557,
      "learning_rate": 0.0005066872022025663,
      "loss": 4.1454,
      "step": 1115
    },
    {
      "epoch": 0.5118092180692502,
      "grad_norm": 0.8850074689111833,
      "learning_rate": 0.0005059442169306844,
      "loss": 4.0467,
      "step": 1116
    },
    {
      "epoch": 0.5122678284797065,
      "grad_norm": 0.8151222420349856,
      "learning_rate": 0.0005052012185312321,
      "loss": 4.0308,
      "step": 1117
    },
    {
      "epoch": 0.5127264388901628,
      "grad_norm": 0.8448819295965957,
      "learning_rate": 0.0005044582086450925,
      "loss": 3.9768,
      "step": 1118
    },
    {
      "epoch": 0.5131850493006191,
      "grad_norm": 0.8940121068789406,
      "learning_rate": 0.0005037151889131737,
      "loss": 3.8691,
      "step": 1119
    },
    {
      "epoch": 0.5136436597110754,
      "grad_norm": 0.8682769928827484,
      "learning_rate": 0.0005029721609764059,
      "loss": 3.978,
      "step": 1120
    },
    {
      "epoch": 0.5141022701215318,
      "grad_norm": 0.8787395316523117,
      "learning_rate": 0.000502229126475737,
      "loss": 3.7947,
      "step": 1121
    },
    {
      "epoch": 0.5145608805319881,
      "grad_norm": 0.8754770554260577,
      "learning_rate": 0.0005014860870521293,
      "loss": 4.077,
      "step": 1122
    },
    {
      "epoch": 0.5150194909424444,
      "grad_norm": 0.8749504197711012,
      "learning_rate": 0.0005007430443465569,
      "loss": 4.0335,
      "step": 1123
    },
    {
      "epoch": 0.5154781013529007,
      "grad_norm": 0.8574031393854339,
      "learning_rate": 0.0005,
      "loss": 4.0137,
      "step": 1124
    },
    {
      "epoch": 0.515936711763357,
      "grad_norm": 0.8625177830784013,
      "learning_rate": 0.0004992569556534432,
      "loss": 4.0217,
      "step": 1125
    },
    {
      "epoch": 0.5163953221738133,
      "grad_norm": 0.893806521568333,
      "learning_rate": 0.0004985139129478707,
      "loss": 3.8867,
      "step": 1126
    },
    {
      "epoch": 0.5168539325842697,
      "grad_norm": 0.8419475860281654,
      "learning_rate": 0.0004977708735242633,
      "loss": 4.0457,
      "step": 1127
    },
    {
      "epoch": 0.517312542994726,
      "grad_norm": 0.8992481684450351,
      "learning_rate": 0.0004970278390235942,
      "loss": 4.0023,
      "step": 1128
    },
    {
      "epoch": 0.5177711534051823,
      "grad_norm": 0.8549692418980576,
      "learning_rate": 0.0004962848110868262,
      "loss": 3.9208,
      "step": 1129
    },
    {
      "epoch": 0.5182297638156386,
      "grad_norm": 0.8867678240081562,
      "learning_rate": 0.0004955417913549074,
      "loss": 4.1398,
      "step": 1130
    },
    {
      "epoch": 0.5186883742260949,
      "grad_norm": 0.9286978963383549,
      "learning_rate": 0.0004947987814687679,
      "loss": 3.8979,
      "step": 1131
    },
    {
      "epoch": 0.5191469846365513,
      "grad_norm": 0.8713353280724242,
      "learning_rate": 0.0004940557830693157,
      "loss": 3.9575,
      "step": 1132
    },
    {
      "epoch": 0.5196055950470075,
      "grad_norm": 0.8180757975258407,
      "learning_rate": 0.0004933127977974338,
      "loss": 3.9657,
      "step": 1133
    },
    {
      "epoch": 0.5200642054574639,
      "grad_norm": 0.9614526377773037,
      "learning_rate": 0.0004925698272939757,
      "loss": 3.9823,
      "step": 1134
    },
    {
      "epoch": 0.5205228158679202,
      "grad_norm": 0.9418817136613216,
      "learning_rate": 0.0004918268731997632,
      "loss": 3.9394,
      "step": 1135
    },
    {
      "epoch": 0.5209814262783765,
      "grad_norm": 0.9007536677015185,
      "learning_rate": 0.0004910839371555809,
      "loss": 3.9215,
      "step": 1136
    },
    {
      "epoch": 0.5214400366888329,
      "grad_norm": 0.7813979205524939,
      "learning_rate": 0.0004903410208021739,
      "loss": 3.858,
      "step": 1137
    },
    {
      "epoch": 0.5218986470992891,
      "grad_norm": 0.8891145871513453,
      "learning_rate": 0.0004895981257802443,
      "loss": 3.9055,
      "step": 1138
    },
    {
      "epoch": 0.5223572575097455,
      "grad_norm": 0.9088204490542745,
      "learning_rate": 0.0004888552537304463,
      "loss": 3.8276,
      "step": 1139
    },
    {
      "epoch": 0.5228158679202017,
      "grad_norm": 0.9388973719674987,
      "learning_rate": 0.00048811240629338394,
      "loss": 4.0116,
      "step": 1140
    },
    {
      "epoch": 0.5232744783306581,
      "grad_norm": 0.8565345473544737,
      "learning_rate": 0.00048736958510960663,
      "loss": 4.0762,
      "step": 1141
    },
    {
      "epoch": 0.5237330887411145,
      "grad_norm": 0.8436823592623858,
      "learning_rate": 0.00048662679181960564,
      "loss": 3.974,
      "step": 1142
    },
    {
      "epoch": 0.5241916991515707,
      "grad_norm": 0.8414592043542444,
      "learning_rate": 0.00048588402806381094,
      "loss": 3.8969,
      "step": 1143
    },
    {
      "epoch": 0.5246503095620271,
      "grad_norm": 0.9050185941521767,
      "learning_rate": 0.0004851412954825874,
      "loss": 3.9479,
      "step": 1144
    },
    {
      "epoch": 0.5251089199724833,
      "grad_norm": 0.8761693262347168,
      "learning_rate": 0.00048439859571623034,
      "loss": 4.0762,
      "step": 1145
    },
    {
      "epoch": 0.5255675303829397,
      "grad_norm": 0.8080408606387931,
      "learning_rate": 0.00048365593040496373,
      "loss": 3.9741,
      "step": 1146
    },
    {
      "epoch": 0.526026140793396,
      "grad_norm": 0.8429974289250153,
      "learning_rate": 0.00048291330118893443,
      "loss": 4.0051,
      "step": 1147
    },
    {
      "epoch": 0.5264847512038523,
      "grad_norm": 0.8844084171234445,
      "learning_rate": 0.0004821707097082102,
      "loss": 4.1623,
      "step": 1148
    },
    {
      "epoch": 0.5269433616143087,
      "grad_norm": 0.904201814112372,
      "learning_rate": 0.0004814281576027749,
      "loss": 4.0346,
      "step": 1149
    },
    {
      "epoch": 0.5274019720247649,
      "grad_norm": 0.887353058944347,
      "learning_rate": 0.000480685646512526,
      "loss": 4.1331,
      "step": 1150
    },
    {
      "epoch": 0.5278605824352213,
      "grad_norm": 0.8467208322671712,
      "learning_rate": 0.00047994317807727025,
      "loss": 3.812,
      "step": 1151
    },
    {
      "epoch": 0.5283191928456776,
      "grad_norm": 0.8621799223748493,
      "learning_rate": 0.00047920075393671974,
      "loss": 4.0505,
      "step": 1152
    },
    {
      "epoch": 0.5287778032561339,
      "grad_norm": 0.8399249824893646,
      "learning_rate": 0.0004784583757304893,
      "loss": 4.0783,
      "step": 1153
    },
    {
      "epoch": 0.5292364136665902,
      "grad_norm": 0.8407610405489785,
      "learning_rate": 0.00047771604509809214,
      "loss": 4.0412,
      "step": 1154
    },
    {
      "epoch": 0.5296950240770465,
      "grad_norm": 0.8252710183532881,
      "learning_rate": 0.0004769737636789364,
      "loss": 3.8091,
      "step": 1155
    },
    {
      "epoch": 0.5301536344875029,
      "grad_norm": 0.8121862037655869,
      "learning_rate": 0.00047623153311232157,
      "loss": 3.8296,
      "step": 1156
    },
    {
      "epoch": 0.5306122448979592,
      "grad_norm": 0.8249778094115953,
      "learning_rate": 0.0004754893550374346,
      "loss": 3.8243,
      "step": 1157
    },
    {
      "epoch": 0.5310708553084155,
      "grad_norm": 0.8403648367844845,
      "learning_rate": 0.00047474723109334685,
      "loss": 3.992,
      "step": 1158
    },
    {
      "epoch": 0.5315294657188718,
      "grad_norm": 0.891817158749492,
      "learning_rate": 0.00047400516291900993,
      "loss": 4.0832,
      "step": 1159
    },
    {
      "epoch": 0.5319880761293282,
      "grad_norm": 0.8560943808851792,
      "learning_rate": 0.0004732631521532522,
      "loss": 3.9179,
      "step": 1160
    },
    {
      "epoch": 0.5324466865397844,
      "grad_norm": 0.8265495372740086,
      "learning_rate": 0.0004725212004347755,
      "loss": 3.9602,
      "step": 1161
    },
    {
      "epoch": 0.5329052969502408,
      "grad_norm": 0.8384963549293463,
      "learning_rate": 0.00047177930940215095,
      "loss": 3.8966,
      "step": 1162
    },
    {
      "epoch": 0.533363907360697,
      "grad_norm": 0.8386311770826589,
      "learning_rate": 0.00047103748069381624,
      "loss": 4.1186,
      "step": 1163
    },
    {
      "epoch": 0.5338225177711534,
      "grad_norm": 0.8277240698637819,
      "learning_rate": 0.0004702957159480707,
      "loss": 4.1893,
      "step": 1164
    },
    {
      "epoch": 0.5342811281816098,
      "grad_norm": 0.8676111959893974,
      "learning_rate": 0.00046955401680307267,
      "loss": 3.987,
      "step": 1165
    },
    {
      "epoch": 0.534739738592066,
      "grad_norm": 0.83557137098514,
      "learning_rate": 0.0004688123848968359,
      "loss": 3.9719,
      "step": 1166
    },
    {
      "epoch": 0.5351983490025224,
      "grad_norm": 0.8416848622607069,
      "learning_rate": 0.00046807082186722516,
      "loss": 4.0152,
      "step": 1167
    },
    {
      "epoch": 0.5356569594129786,
      "grad_norm": 0.8432537593697849,
      "learning_rate": 0.0004673293293519535,
      "loss": 4.0699,
      "step": 1168
    },
    {
      "epoch": 0.536115569823435,
      "grad_norm": 0.8504134577023877,
      "learning_rate": 0.00046658790898857806,
      "loss": 3.8952,
      "step": 1169
    },
    {
      "epoch": 0.5365741802338914,
      "grad_norm": 0.8807391110492246,
      "learning_rate": 0.0004658465624144963,
      "loss": 4.1078,
      "step": 1170
    },
    {
      "epoch": 0.5370327906443476,
      "grad_norm": 0.7790258012023324,
      "learning_rate": 0.0004651052912669438,
      "loss": 3.8394,
      "step": 1171
    },
    {
      "epoch": 0.537491401054804,
      "grad_norm": 0.8691947885309586,
      "learning_rate": 0.0004643640971829883,
      "loss": 4.0201,
      "step": 1172
    },
    {
      "epoch": 0.5379500114652602,
      "grad_norm": 0.8331822783374527,
      "learning_rate": 0.0004636229817995281,
      "loss": 3.7469,
      "step": 1173
    },
    {
      "epoch": 0.5384086218757166,
      "grad_norm": 0.8483557292662469,
      "learning_rate": 0.0004628819467532876,
      "loss": 4.0653,
      "step": 1174
    },
    {
      "epoch": 0.538867232286173,
      "grad_norm": 0.8740475675246197,
      "learning_rate": 0.00046214099368081335,
      "loss": 3.8731,
      "step": 1175
    },
    {
      "epoch": 0.5393258426966292,
      "grad_norm": 0.9242725689382597,
      "learning_rate": 0.0004614001242184714,
      "loss": 4.0119,
      "step": 1176
    },
    {
      "epoch": 0.5397844531070856,
      "grad_norm": 0.924502894747377,
      "learning_rate": 0.000460659340002443,
      "loss": 3.9695,
      "step": 1177
    },
    {
      "epoch": 0.5402430635175418,
      "grad_norm": 0.8846879113405902,
      "learning_rate": 0.00045991864266872073,
      "loss": 3.9258,
      "step": 1178
    },
    {
      "epoch": 0.5407016739279982,
      "grad_norm": 0.812532745788098,
      "learning_rate": 0.00045917803385310595,
      "loss": 4.0241,
      "step": 1179
    },
    {
      "epoch": 0.5411602843384545,
      "grad_norm": 0.8329060473895894,
      "learning_rate": 0.00045843751519120417,
      "loss": 4.139,
      "step": 1180
    },
    {
      "epoch": 0.5416188947489108,
      "grad_norm": 0.8314439470442567,
      "learning_rate": 0.00045769708831842193,
      "loss": 3.9954,
      "step": 1181
    },
    {
      "epoch": 0.5420775051593671,
      "grad_norm": 0.8622129616710019,
      "learning_rate": 0.00045695675486996266,
      "loss": 3.8488,
      "step": 1182
    },
    {
      "epoch": 0.5425361155698234,
      "grad_norm": 0.8822597151285329,
      "learning_rate": 0.00045621651648082405,
      "loss": 3.8348,
      "step": 1183
    },
    {
      "epoch": 0.5429947259802798,
      "grad_norm": 0.8560823739387852,
      "learning_rate": 0.00045547637478579356,
      "loss": 4.0716,
      "step": 1184
    },
    {
      "epoch": 0.5434533363907361,
      "grad_norm": 0.8461577853981957,
      "learning_rate": 0.0004547363314194449,
      "loss": 4.1308,
      "step": 1185
    },
    {
      "epoch": 0.5439119468011924,
      "grad_norm": 0.8632594489887655,
      "learning_rate": 0.000453996388016135,
      "loss": 3.9755,
      "step": 1186
    },
    {
      "epoch": 0.5443705572116487,
      "grad_norm": 0.8801980969990155,
      "learning_rate": 0.0004532565462099999,
      "loss": 4.0643,
      "step": 1187
    },
    {
      "epoch": 0.544829167622105,
      "grad_norm": 0.878398989683914,
      "learning_rate": 0.0004525168076349513,
      "loss": 4.0772,
      "step": 1188
    },
    {
      "epoch": 0.5452877780325613,
      "grad_norm": 0.8750635562000197,
      "learning_rate": 0.0004517771739246729,
      "loss": 3.9621,
      "step": 1189
    },
    {
      "epoch": 0.5457463884430177,
      "grad_norm": 0.8375723101897348,
      "learning_rate": 0.0004510376467126165,
      "loss": 3.9933,
      "step": 1190
    },
    {
      "epoch": 0.546204998853474,
      "grad_norm": 0.8298254228662598,
      "learning_rate": 0.0004502982276319992,
      "loss": 3.8355,
      "step": 1191
    },
    {
      "epoch": 0.5466636092639303,
      "grad_norm": 0.8258112031084085,
      "learning_rate": 0.0004495589183157991,
      "loss": 3.9649,
      "step": 1192
    },
    {
      "epoch": 0.5471222196743866,
      "grad_norm": 0.8025327357314109,
      "learning_rate": 0.0004488197203967517,
      "loss": 3.9643,
      "step": 1193
    },
    {
      "epoch": 0.5475808300848429,
      "grad_norm": 0.8759797315742215,
      "learning_rate": 0.0004480806355073467,
      "loss": 3.9098,
      "step": 1194
    },
    {
      "epoch": 0.5480394404952993,
      "grad_norm": 0.8158102677840855,
      "learning_rate": 0.000447341665279824,
      "loss": 3.9912,
      "step": 1195
    },
    {
      "epoch": 0.5484980509057555,
      "grad_norm": 0.8506741056503226,
      "learning_rate": 0.0004466028113461708,
      "loss": 3.941,
      "step": 1196
    },
    {
      "epoch": 0.5489566613162119,
      "grad_norm": 0.8345663507471519,
      "learning_rate": 0.0004458640753381167,
      "loss": 3.9967,
      "step": 1197
    },
    {
      "epoch": 0.5494152717266682,
      "grad_norm": 0.8358761091395189,
      "learning_rate": 0.0004451254588871313,
      "loss": 3.9712,
      "step": 1198
    },
    {
      "epoch": 0.5498738821371245,
      "grad_norm": 0.8294254407262581,
      "learning_rate": 0.0004443869636244203,
      "loss": 3.9365,
      "step": 1199
    },
    {
      "epoch": 0.5503324925475809,
      "grad_norm": 0.8200071311413815,
      "learning_rate": 0.0004436485911809212,
      "loss": 3.9801,
      "step": 1200
    },
    {
      "epoch": 0.5507911029580371,
      "grad_norm": 0.8547736734241981,
      "learning_rate": 0.00044291034318730087,
      "loss": 3.9181,
      "step": 1201
    },
    {
      "epoch": 0.5512497133684935,
      "grad_norm": 0.9455390636013008,
      "learning_rate": 0.0004421722212739511,
      "loss": 3.9401,
      "step": 1202
    },
    {
      "epoch": 0.5517083237789497,
      "grad_norm": 0.8069074867038949,
      "learning_rate": 0.0004414342270709848,
      "loss": 3.9625,
      "step": 1203
    },
    {
      "epoch": 0.5521669341894061,
      "grad_norm": 0.8000055347377829,
      "learning_rate": 0.00044069636220823397,
      "loss": 4.0717,
      "step": 1204
    },
    {
      "epoch": 0.5526255445998625,
      "grad_norm": 0.8311536788257395,
      "learning_rate": 0.0004399586283152437,
      "loss": 3.8968,
      "step": 1205
    },
    {
      "epoch": 0.5530841550103187,
      "grad_norm": 0.8637351009802011,
      "learning_rate": 0.0004392210270212706,
      "loss": 4.0995,
      "step": 1206
    },
    {
      "epoch": 0.5535427654207751,
      "grad_norm": 0.8763248775475909,
      "learning_rate": 0.00043848355995527825,
      "loss": 3.9201,
      "step": 1207
    },
    {
      "epoch": 0.5540013758312313,
      "grad_norm": 0.8300236426032449,
      "learning_rate": 0.00043774622874593374,
      "loss": 3.917,
      "step": 1208
    },
    {
      "epoch": 0.5544599862416877,
      "grad_norm": 0.7980479669283372,
      "learning_rate": 0.000437009035021604,
      "loss": 4.0205,
      "step": 1209
    },
    {
      "epoch": 0.554918596652144,
      "grad_norm": 0.8165024493660378,
      "learning_rate": 0.00043627198041035274,
      "loss": 4.0448,
      "step": 1210
    },
    {
      "epoch": 0.5553772070626003,
      "grad_norm": 0.8650027884386023,
      "learning_rate": 0.00043553506653993597,
      "loss": 3.8349,
      "step": 1211
    },
    {
      "epoch": 0.5558358174730567,
      "grad_norm": 0.8522031751613686,
      "learning_rate": 0.0004347982950377992,
      "loss": 3.8574,
      "step": 1212
    },
    {
      "epoch": 0.5562944278835129,
      "grad_norm": 0.8378395938116246,
      "learning_rate": 0.0004340616675310735,
      "loss": 3.831,
      "step": 1213
    },
    {
      "epoch": 0.5567530382939693,
      "grad_norm": 0.9587699515426824,
      "learning_rate": 0.00043332518564657193,
      "loss": 4.2553,
      "step": 1214
    },
    {
      "epoch": 0.5572116487044256,
      "grad_norm": 0.8497425969678125,
      "learning_rate": 0.0004325888510107856,
      "loss": 3.8955,
      "step": 1215
    },
    {
      "epoch": 0.5576702591148819,
      "grad_norm": 0.8462403347113711,
      "learning_rate": 0.0004318526652498809,
      "loss": 4.0575,
      "step": 1216
    },
    {
      "epoch": 0.5581288695253382,
      "grad_norm": 0.8683350574873998,
      "learning_rate": 0.00043111662998969523,
      "loss": 4.0847,
      "step": 1217
    },
    {
      "epoch": 0.5585874799357945,
      "grad_norm": 0.951513888850931,
      "learning_rate": 0.0004303807468557335,
      "loss": 3.9846,
      "step": 1218
    },
    {
      "epoch": 0.5590460903462509,
      "grad_norm": 0.8706539068905234,
      "learning_rate": 0.0004296450174731648,
      "loss": 3.8757,
      "step": 1219
    },
    {
      "epoch": 0.5595047007567072,
      "grad_norm": 0.8350528100426509,
      "learning_rate": 0.0004289094434668188,
      "loss": 4.0425,
      "step": 1220
    },
    {
      "epoch": 0.5599633111671635,
      "grad_norm": 0.8257121592136712,
      "learning_rate": 0.00042817402646118185,
      "loss": 3.9614,
      "step": 1221
    },
    {
      "epoch": 0.5604219215776198,
      "grad_norm": 0.8538395803710843,
      "learning_rate": 0.0004274387680803936,
      "loss": 3.9283,
      "step": 1222
    },
    {
      "epoch": 0.5608805319880761,
      "grad_norm": 0.845742377533164,
      "learning_rate": 0.00042670366994824327,
      "loss": 3.9544,
      "step": 1223
    },
    {
      "epoch": 0.5613391423985324,
      "grad_norm": 0.8149247544920447,
      "learning_rate": 0.0004259687336881663,
      "loss": 4.0099,
      "step": 1224
    },
    {
      "epoch": 0.5617977528089888,
      "grad_norm": 0.8604762685699942,
      "learning_rate": 0.0004252339609232408,
      "loss": 3.932,
      "step": 1225
    },
    {
      "epoch": 0.562256363219445,
      "grad_norm": 0.8208349528282588,
      "learning_rate": 0.0004244993532761834,
      "loss": 3.9773,
      "step": 1226
    },
    {
      "epoch": 0.5627149736299014,
      "grad_norm": 0.8196277113098684,
      "learning_rate": 0.00042376491236934634,
      "loss": 3.7994,
      "step": 1227
    },
    {
      "epoch": 0.5631735840403577,
      "grad_norm": 0.8323142483794499,
      "learning_rate": 0.0004230306398247136,
      "loss": 4.0727,
      "step": 1228
    },
    {
      "epoch": 0.563632194450814,
      "grad_norm": 0.8237889695324633,
      "learning_rate": 0.0004222965372638976,
      "loss": 4.1567,
      "step": 1229
    },
    {
      "epoch": 0.5640908048612704,
      "grad_norm": 0.8525767391767366,
      "learning_rate": 0.0004215626063081348,
      "loss": 3.9466,
      "step": 1230
    },
    {
      "epoch": 0.5645494152717266,
      "grad_norm": 0.862391607347089,
      "learning_rate": 0.000420828848578283,
      "loss": 4.2662,
      "step": 1231
    },
    {
      "epoch": 0.565008025682183,
      "grad_norm": 0.8284619754631404,
      "learning_rate": 0.0004200952656948175,
      "loss": 4.0492,
      "step": 1232
    },
    {
      "epoch": 0.5654666360926393,
      "grad_norm": 0.8172949899872005,
      "learning_rate": 0.0004193618592778272,
      "loss": 4.0185,
      "step": 1233
    },
    {
      "epoch": 0.5659252465030956,
      "grad_norm": 0.8152399042131192,
      "learning_rate": 0.0004186286309470116,
      "loss": 4.0692,
      "step": 1234
    },
    {
      "epoch": 0.566383856913552,
      "grad_norm": 0.878134824179865,
      "learning_rate": 0.0004178955823216767,
      "loss": 4.1236,
      "step": 1235
    },
    {
      "epoch": 0.5668424673240082,
      "grad_norm": 0.8046516464287378,
      "learning_rate": 0.00041716271502073137,
      "loss": 3.7751,
      "step": 1236
    },
    {
      "epoch": 0.5673010777344646,
      "grad_norm": 0.8061645779527581,
      "learning_rate": 0.000416430030662685,
      "loss": 3.8675,
      "step": 1237
    },
    {
      "epoch": 0.5677596881449208,
      "grad_norm": 0.8949644138890044,
      "learning_rate": 0.00041569753086564173,
      "loss": 4.0644,
      "step": 1238
    },
    {
      "epoch": 0.5682182985553772,
      "grad_norm": 0.8537624035653519,
      "learning_rate": 0.0004149652172472988,
      "loss": 3.9553,
      "step": 1239
    },
    {
      "epoch": 0.5686769089658336,
      "grad_norm": 0.8444291958518464,
      "learning_rate": 0.00041423309142494234,
      "loss": 3.9873,
      "step": 1240
    },
    {
      "epoch": 0.5691355193762898,
      "grad_norm": 0.8136052325267931,
      "learning_rate": 0.0004135011550154433,
      "loss": 4.0298,
      "step": 1241
    },
    {
      "epoch": 0.5695941297867462,
      "grad_norm": 0.8320501514535497,
      "learning_rate": 0.0004127694096352546,
      "loss": 4.0148,
      "step": 1242
    },
    {
      "epoch": 0.5700527401972024,
      "grad_norm": 1.0139813152975106,
      "learning_rate": 0.00041203785690040743,
      "loss": 3.9841,
      "step": 1243
    },
    {
      "epoch": 0.5705113506076588,
      "grad_norm": 0.8649032410674123,
      "learning_rate": 0.00041130649842650694,
      "loss": 4.1028,
      "step": 1244
    },
    {
      "epoch": 0.5709699610181151,
      "grad_norm": 0.8285487455571592,
      "learning_rate": 0.00041057533582873016,
      "loss": 4.0456,
      "step": 1245
    },
    {
      "epoch": 0.5714285714285714,
      "grad_norm": 0.8709677040527968,
      "learning_rate": 0.0004098443707218208,
      "loss": 4.0917,
      "step": 1246
    },
    {
      "epoch": 0.5718871818390278,
      "grad_norm": 0.9159418078369724,
      "learning_rate": 0.00040911360472008673,
      "loss": 3.9295,
      "step": 1247
    },
    {
      "epoch": 0.572345792249484,
      "grad_norm": 0.8579664011078851,
      "learning_rate": 0.0004083830394373959,
      "loss": 3.9863,
      "step": 1248
    },
    {
      "epoch": 0.5728044026599404,
      "grad_norm": 0.8474765798123399,
      "learning_rate": 0.00040765267648717324,
      "loss": 3.8885,
      "step": 1249
    },
    {
      "epoch": 0.5732630130703967,
      "grad_norm": 0.8022816702952272,
      "learning_rate": 0.00040692251748239677,
      "loss": 3.7851,
      "step": 1250
    },
    {
      "epoch": 0.573721623480853,
      "grad_norm": 0.8831288047456819,
      "learning_rate": 0.00040619256403559383,
      "loss": 4.0344,
      "step": 1251
    },
    {
      "epoch": 0.5741802338913093,
      "grad_norm": 0.8416732879359443,
      "learning_rate": 0.000405462817758838,
      "loss": 3.8869,
      "step": 1252
    },
    {
      "epoch": 0.5746388443017657,
      "grad_norm": 0.8376905504186051,
      "learning_rate": 0.0004047332802637457,
      "loss": 4.0011,
      "step": 1253
    },
    {
      "epoch": 0.575097454712222,
      "grad_norm": 0.9545958643184378,
      "learning_rate": 0.00040400395316147157,
      "loss": 4.1321,
      "step": 1254
    },
    {
      "epoch": 0.5755560651226783,
      "grad_norm": 0.8114511994804052,
      "learning_rate": 0.00040327483806270627,
      "loss": 3.8668,
      "step": 1255
    },
    {
      "epoch": 0.5760146755331346,
      "grad_norm": 0.850847114381483,
      "learning_rate": 0.0004025459365776715,
      "loss": 4.0757,
      "step": 1256
    },
    {
      "epoch": 0.5764732859435909,
      "grad_norm": 0.9846510939895753,
      "learning_rate": 0.00040181725031611794,
      "loss": 3.8824,
      "step": 1257
    },
    {
      "epoch": 0.5769318963540473,
      "grad_norm": 0.8522424993180225,
      "learning_rate": 0.0004010887808873206,
      "loss": 3.9494,
      "step": 1258
    },
    {
      "epoch": 0.5773905067645035,
      "grad_norm": 0.9319016073062735,
      "learning_rate": 0.00040036052990007553,
      "loss": 3.933,
      "step": 1259
    },
    {
      "epoch": 0.5778491171749599,
      "grad_norm": 0.8031515968335533,
      "learning_rate": 0.0003996324989626967,
      "loss": 3.9472,
      "step": 1260
    },
    {
      "epoch": 0.5783077275854162,
      "grad_norm": 0.8520939529106164,
      "learning_rate": 0.00039890468968301166,
      "loss": 4.0569,
      "step": 1261
    },
    {
      "epoch": 0.5787663379958725,
      "grad_norm": 0.8990852249778584,
      "learning_rate": 0.0003981771036683591,
      "loss": 3.9004,
      "step": 1262
    },
    {
      "epoch": 0.5792249484063289,
      "grad_norm": 0.8797208915398257,
      "learning_rate": 0.00039744974252558385,
      "loss": 4.0539,
      "step": 1263
    },
    {
      "epoch": 0.5796835588167851,
      "grad_norm": 0.8328135088834862,
      "learning_rate": 0.00039672260786103463,
      "loss": 4.0939,
      "step": 1264
    },
    {
      "epoch": 0.5801421692272415,
      "grad_norm": 0.8497773078496057,
      "learning_rate": 0.00039599570128055994,
      "loss": 3.8653,
      "step": 1265
    },
    {
      "epoch": 0.5806007796376977,
      "grad_norm": 0.9181625557298596,
      "learning_rate": 0.0003952690243895044,
      "loss": 4.2324,
      "step": 1266
    },
    {
      "epoch": 0.5810593900481541,
      "grad_norm": 0.8313428592644312,
      "learning_rate": 0.0003945425787927054,
      "loss": 3.8478,
      "step": 1267
    },
    {
      "epoch": 0.5815180004586105,
      "grad_norm": 0.8017565515812612,
      "learning_rate": 0.00039381636609448975,
      "loss": 4.0165,
      "step": 1268
    },
    {
      "epoch": 0.5819766108690667,
      "grad_norm": 0.7646945419394551,
      "learning_rate": 0.0003930903878986693,
      "loss": 4.1066,
      "step": 1269
    },
    {
      "epoch": 0.5824352212795231,
      "grad_norm": 0.815390261536426,
      "learning_rate": 0.00039236464580853916,
      "loss": 4.0051,
      "step": 1270
    },
    {
      "epoch": 0.5828938316899793,
      "grad_norm": 0.8411704858898817,
      "learning_rate": 0.0003916391414268718,
      "loss": 4.0872,
      "step": 1271
    },
    {
      "epoch": 0.5833524421004357,
      "grad_norm": 0.7817640753022312,
      "learning_rate": 0.00039091387635591536,
      "loss": 3.8483,
      "step": 1272
    },
    {
      "epoch": 0.583811052510892,
      "grad_norm": 0.8244387355109192,
      "learning_rate": 0.0003901888521973894,
      "loss": 4.0572,
      "step": 1273
    },
    {
      "epoch": 0.5842696629213483,
      "grad_norm": 0.8080495169593723,
      "learning_rate": 0.0003894640705524813,
      "loss": 3.7881,
      "step": 1274
    },
    {
      "epoch": 0.5847282733318047,
      "grad_norm": 0.8242247878436546,
      "learning_rate": 0.00038873953302184284,
      "loss": 3.8981,
      "step": 1275
    },
    {
      "epoch": 0.5851868837422609,
      "grad_norm": 0.8823349547775333,
      "learning_rate": 0.000388015241205587,
      "loss": 4.0719,
      "step": 1276
    },
    {
      "epoch": 0.5856454941527173,
      "grad_norm": 0.8750763669812404,
      "learning_rate": 0.00038729119670328355,
      "loss": 4.0646,
      "step": 1277
    },
    {
      "epoch": 0.5861041045631736,
      "grad_norm": 0.8949900857459375,
      "learning_rate": 0.00038656740111395665,
      "loss": 4.0738,
      "step": 1278
    },
    {
      "epoch": 0.5865627149736299,
      "grad_norm": 0.8358875091480471,
      "learning_rate": 0.00038584385603608053,
      "loss": 4.1011,
      "step": 1279
    },
    {
      "epoch": 0.5870213253840862,
      "grad_norm": 0.7769835029127443,
      "learning_rate": 0.00038512056306757615,
      "loss": 3.9139,
      "step": 1280
    },
    {
      "epoch": 0.5874799357945425,
      "grad_norm": 0.7830184162208818,
      "learning_rate": 0.0003843975238058075,
      "loss": 4.0092,
      "step": 1281
    },
    {
      "epoch": 0.5879385462049989,
      "grad_norm": 0.8226764495340454,
      "learning_rate": 0.00038367473984757863,
      "loss": 4.0716,
      "step": 1282
    },
    {
      "epoch": 0.5883971566154552,
      "grad_norm": 0.8400954720383131,
      "learning_rate": 0.0003829522127891296,
      "loss": 3.7537,
      "step": 1283
    },
    {
      "epoch": 0.5888557670259115,
      "grad_norm": 0.8735883447628784,
      "learning_rate": 0.0003822299442261329,
      "loss": 4.0253,
      "step": 1284
    },
    {
      "epoch": 0.5893143774363678,
      "grad_norm": 0.8131951604418847,
      "learning_rate": 0.00038150793575369063,
      "loss": 3.8848,
      "step": 1285
    },
    {
      "epoch": 0.5897729878468241,
      "grad_norm": 0.834477472040288,
      "learning_rate": 0.0003807861889663299,
      "loss": 3.9107,
      "step": 1286
    },
    {
      "epoch": 0.5902315982572804,
      "grad_norm": 0.8508746522696186,
      "learning_rate": 0.0003800647054580006,
      "loss": 3.9297,
      "step": 1287
    },
    {
      "epoch": 0.5906902086677368,
      "grad_norm": 0.7833604716994917,
      "learning_rate": 0.00037934348682207064,
      "loss": 3.838,
      "step": 1288
    },
    {
      "epoch": 0.591148819078193,
      "grad_norm": 0.8865149981775038,
      "learning_rate": 0.00037862253465132306,
      "loss": 4.0483,
      "step": 1289
    },
    {
      "epoch": 0.5916074294886494,
      "grad_norm": 0.8686399418748056,
      "learning_rate": 0.00037790185053795245,
      "loss": 4.0237,
      "step": 1290
    },
    {
      "epoch": 0.5920660398991057,
      "grad_norm": 0.8443224408820185,
      "learning_rate": 0.0003771814360735616,
      "loss": 3.9299,
      "step": 1291
    },
    {
      "epoch": 0.592524650309562,
      "grad_norm": 0.8416881280830705,
      "learning_rate": 0.00037646129284915755,
      "loss": 3.9242,
      "step": 1292
    },
    {
      "epoch": 0.5929832607200184,
      "grad_norm": 0.8385742001140908,
      "learning_rate": 0.00037574142245514825,
      "loss": 3.9911,
      "step": 1293
    },
    {
      "epoch": 0.5934418711304746,
      "grad_norm": 0.8221901334554023,
      "learning_rate": 0.0003750218264813393,
      "loss": 4.0395,
      "step": 1294
    },
    {
      "epoch": 0.593900481540931,
      "grad_norm": 0.8216184415993678,
      "learning_rate": 0.0003743025065169305,
      "loss": 4.1645,
      "step": 1295
    },
    {
      "epoch": 0.5943590919513873,
      "grad_norm": 0.8575310854779062,
      "learning_rate": 0.0003735834641505116,
      "loss": 3.9099,
      "step": 1296
    },
    {
      "epoch": 0.5948177023618436,
      "grad_norm": 0.7924677712063152,
      "learning_rate": 0.00037286470097005954,
      "loss": 3.8936,
      "step": 1297
    },
    {
      "epoch": 0.5952763127723,
      "grad_norm": 0.8514747440437649,
      "learning_rate": 0.0003721462185629347,
      "loss": 3.9909,
      "step": 1298
    },
    {
      "epoch": 0.5957349231827562,
      "grad_norm": 0.8361702535369052,
      "learning_rate": 0.00037142801851587707,
      "loss": 4.0794,
      "step": 1299
    },
    {
      "epoch": 0.5961935335932126,
      "grad_norm": 0.810362819376,
      "learning_rate": 0.00037071010241500357,
      "loss": 3.9495,
      "step": 1300
    },
    {
      "epoch": 0.5966521440036688,
      "grad_norm": 0.773509720900754,
      "learning_rate": 0.00036999247184580383,
      "loss": 3.9917,
      "step": 1301
    },
    {
      "epoch": 0.5971107544141252,
      "grad_norm": 0.8179347676784767,
      "learning_rate": 0.00036927512839313636,
      "loss": 3.8919,
      "step": 1302
    },
    {
      "epoch": 0.5975693648245816,
      "grad_norm": 0.822213138391178,
      "learning_rate": 0.0003685580736412268,
      "loss": 3.8966,
      "step": 1303
    },
    {
      "epoch": 0.5980279752350378,
      "grad_norm": 0.8176273489467315,
      "learning_rate": 0.000367841309173662,
      "loss": 3.8177,
      "step": 1304
    },
    {
      "epoch": 0.5984865856454942,
      "grad_norm": 0.8032279214425556,
      "learning_rate": 0.0003671248365733883,
      "loss": 3.8235,
      "step": 1305
    },
    {
      "epoch": 0.5989451960559504,
      "grad_norm": 0.7969195354045244,
      "learning_rate": 0.0003664086574227075,
      "loss": 3.9842,
      "step": 1306
    },
    {
      "epoch": 0.5994038064664068,
      "grad_norm": 0.817956483568161,
      "learning_rate": 0.000365692773303273,
      "loss": 3.9314,
      "step": 1307
    },
    {
      "epoch": 0.5998624168768631,
      "grad_norm": 0.8498833647507577,
      "learning_rate": 0.00036497718579608696,
      "loss": 4.0059,
      "step": 1308
    },
    {
      "epoch": 0.6003210272873194,
      "grad_norm": 0.9687182417401565,
      "learning_rate": 0.0003642618964814964,
      "loss": 4.1365,
      "step": 1309
    },
    {
      "epoch": 0.6007796376977758,
      "grad_norm": 0.8278489518717294,
      "learning_rate": 0.00036354690693918946,
      "loss": 4.0058,
      "step": 1310
    },
    {
      "epoch": 0.601238248108232,
      "grad_norm": 0.8501325969318744,
      "learning_rate": 0.00036283221874819284,
      "loss": 4.0191,
      "step": 1311
    },
    {
      "epoch": 0.6016968585186884,
      "grad_norm": 0.8090641788684051,
      "learning_rate": 0.0003621178334868672,
      "loss": 3.9987,
      "step": 1312
    },
    {
      "epoch": 0.6021554689291447,
      "grad_norm": 0.8215705469154732,
      "learning_rate": 0.00036140375273290476,
      "loss": 3.809,
      "step": 1313
    },
    {
      "epoch": 0.602614079339601,
      "grad_norm": 0.8666920223743241,
      "learning_rate": 0.0003606899780633245,
      "loss": 3.806,
      "step": 1314
    },
    {
      "epoch": 0.6030726897500573,
      "grad_norm": 0.8477181599351774,
      "learning_rate": 0.0003599765110544699,
      "loss": 3.8822,
      "step": 1315
    },
    {
      "epoch": 0.6035313001605136,
      "grad_norm": 0.837521401385877,
      "learning_rate": 0.0003592633532820052,
      "loss": 3.9921,
      "step": 1316
    },
    {
      "epoch": 0.60398991057097,
      "grad_norm": 0.7933669984062443,
      "learning_rate": 0.0003585505063209109,
      "loss": 4.0473,
      "step": 1317
    },
    {
      "epoch": 0.6044485209814263,
      "grad_norm": 0.8508458542421583,
      "learning_rate": 0.00035783797174548194,
      "loss": 3.8319,
      "step": 1318
    },
    {
      "epoch": 0.6049071313918826,
      "grad_norm": 0.8972544775076465,
      "learning_rate": 0.00035712575112932277,
      "loss": 4.1701,
      "step": 1319
    },
    {
      "epoch": 0.6053657418023389,
      "grad_norm": 0.8306145314890927,
      "learning_rate": 0.000356413846045345,
      "loss": 3.8975,
      "step": 1320
    },
    {
      "epoch": 0.6058243522127952,
      "grad_norm": 0.8451207354827946,
      "learning_rate": 0.000355702258065763,
      "loss": 3.8793,
      "step": 1321
    },
    {
      "epoch": 0.6062829626232515,
      "grad_norm": 0.8107753626947296,
      "learning_rate": 0.0003549909887620909,
      "loss": 4.1118,
      "step": 1322
    },
    {
      "epoch": 0.6067415730337079,
      "grad_norm": 0.8067083396730332,
      "learning_rate": 0.00035428003970513914,
      "loss": 4.0526,
      "step": 1323
    },
    {
      "epoch": 0.6072001834441642,
      "grad_norm": 0.7664387150327846,
      "learning_rate": 0.00035356941246501085,
      "loss": 3.9799,
      "step": 1324
    },
    {
      "epoch": 0.6076587938546205,
      "grad_norm": 0.8347833302311147,
      "learning_rate": 0.0003528591086110984,
      "loss": 3.8075,
      "step": 1325
    },
    {
      "epoch": 0.6081174042650768,
      "grad_norm": 0.8081926082753758,
      "learning_rate": 0.00035214912971208,
      "loss": 3.6647,
      "step": 1326
    },
    {
      "epoch": 0.6085760146755331,
      "grad_norm": 0.8102611423034609,
      "learning_rate": 0.0003514394773359163,
      "loss": 3.9554,
      "step": 1327
    },
    {
      "epoch": 0.6090346250859895,
      "grad_norm": 0.8322017174919772,
      "learning_rate": 0.0003507301530498469,
      "loss": 3.9402,
      "step": 1328
    },
    {
      "epoch": 0.6094932354964457,
      "grad_norm": 0.8404120025396397,
      "learning_rate": 0.00035002115842038646,
      "loss": 3.9699,
      "step": 1329
    },
    {
      "epoch": 0.6099518459069021,
      "grad_norm": 0.8729159528485936,
      "learning_rate": 0.00034931249501332195,
      "loss": 4.0461,
      "step": 1330
    },
    {
      "epoch": 0.6104104563173584,
      "grad_norm": 0.8430349426624364,
      "learning_rate": 0.00034860416439370885,
      "loss": 3.9875,
      "step": 1331
    },
    {
      "epoch": 0.6108690667278147,
      "grad_norm": 0.7830147216365495,
      "learning_rate": 0.0003478961681258674,
      "loss": 3.9356,
      "step": 1332
    },
    {
      "epoch": 0.6113276771382711,
      "grad_norm": 0.8008242328086939,
      "learning_rate": 0.0003471885077733796,
      "loss": 3.9974,
      "step": 1333
    },
    {
      "epoch": 0.6117862875487273,
      "grad_norm": 0.8541382421681679,
      "learning_rate": 0.0003464811848990859,
      "loss": 4.0749,
      "step": 1334
    },
    {
      "epoch": 0.6122448979591837,
      "grad_norm": 0.8292780490059963,
      "learning_rate": 0.00034577420106508063,
      "loss": 3.9356,
      "step": 1335
    },
    {
      "epoch": 0.6127035083696399,
      "grad_norm": 0.8713076065970108,
      "learning_rate": 0.0003450675578327105,
      "loss": 3.974,
      "step": 1336
    },
    {
      "epoch": 0.6131621187800963,
      "grad_norm": 0.8842333962290365,
      "learning_rate": 0.000344361256762569,
      "loss": 3.8322,
      "step": 1337
    },
    {
      "epoch": 0.6136207291905527,
      "grad_norm": 0.8842758626835941,
      "learning_rate": 0.00034365529941449456,
      "loss": 3.9059,
      "step": 1338
    },
    {
      "epoch": 0.6140793396010089,
      "grad_norm": 0.8196903344366039,
      "learning_rate": 0.0003429496873475664,
      "loss": 3.8271,
      "step": 1339
    },
    {
      "epoch": 0.6145379500114653,
      "grad_norm": 0.8055047447008723,
      "learning_rate": 0.0003422444221201009,
      "loss": 3.9938,
      "step": 1340
    },
    {
      "epoch": 0.6149965604219215,
      "grad_norm": 0.8033413694943682,
      "learning_rate": 0.0003415395052896487,
      "loss": 3.9691,
      "step": 1341
    },
    {
      "epoch": 0.6154551708323779,
      "grad_norm": 0.8721252914796155,
      "learning_rate": 0.0003408349384129912,
      "loss": 3.9324,
      "step": 1342
    },
    {
      "epoch": 0.6159137812428342,
      "grad_norm": 0.8480375529360689,
      "learning_rate": 0.00034013072304613643,
      "loss": 4.1227,
      "step": 1343
    },
    {
      "epoch": 0.6163723916532905,
      "grad_norm": 0.8339872795697629,
      "learning_rate": 0.00033942686074431674,
      "loss": 3.9767,
      "step": 1344
    },
    {
      "epoch": 0.6168310020637469,
      "grad_norm": 0.8008855603479714,
      "learning_rate": 0.0003387233530619843,
      "loss": 4.1845,
      "step": 1345
    },
    {
      "epoch": 0.6172896124742032,
      "grad_norm": 0.8004036137968406,
      "learning_rate": 0.0003380202015528084,
      "loss": 4.0332,
      "step": 1346
    },
    {
      "epoch": 0.6177482228846595,
      "grad_norm": 0.8212474549943031,
      "learning_rate": 0.0003373174077696715,
      "loss": 3.8809,
      "step": 1347
    },
    {
      "epoch": 0.6182068332951158,
      "grad_norm": 0.7876000682205,
      "learning_rate": 0.0003366149732646661,
      "loss": 3.8963,
      "step": 1348
    },
    {
      "epoch": 0.6186654437055721,
      "grad_norm": 0.76296745309948,
      "learning_rate": 0.00033591289958909143,
      "loss": 3.8485,
      "step": 1349
    },
    {
      "epoch": 0.6191240541160284,
      "grad_norm": 0.8915688403205311,
      "learning_rate": 0.00033521118829344954,
      "loss": 4.2008,
      "step": 1350
    },
    {
      "epoch": 0.6195826645264848,
      "grad_norm": 0.7939916068770287,
      "learning_rate": 0.0003345098409274423,
      "loss": 3.8705,
      "step": 1351
    },
    {
      "epoch": 0.6200412749369411,
      "grad_norm": 0.8041860238308928,
      "learning_rate": 0.00033380885903996796,
      "loss": 4.046,
      "step": 1352
    },
    {
      "epoch": 0.6204998853473974,
      "grad_norm": 0.8442900290440644,
      "learning_rate": 0.00033310824417911766,
      "loss": 4.0102,
      "step": 1353
    },
    {
      "epoch": 0.6209584957578537,
      "grad_norm": 0.8518124003719916,
      "learning_rate": 0.00033240799789217184,
      "loss": 3.8491,
      "step": 1354
    },
    {
      "epoch": 0.62141710616831,
      "grad_norm": 0.845837064063259,
      "learning_rate": 0.00033170812172559694,
      "loss": 3.9176,
      "step": 1355
    },
    {
      "epoch": 0.6218757165787664,
      "grad_norm": 0.7947864707117716,
      "learning_rate": 0.000331008617225042,
      "loss": 3.8385,
      "step": 1356
    },
    {
      "epoch": 0.6223343269892226,
      "grad_norm": 0.8980452577967142,
      "learning_rate": 0.0003303094859353355,
      "loss": 3.7779,
      "step": 1357
    },
    {
      "epoch": 0.622792937399679,
      "grad_norm": 0.8544078741711488,
      "learning_rate": 0.0003296107294004812,
      "loss": 3.9623,
      "step": 1358
    },
    {
      "epoch": 0.6232515478101353,
      "grad_norm": 0.8196200561886481,
      "learning_rate": 0.0003289123491636559,
      "loss": 4.1429,
      "step": 1359
    },
    {
      "epoch": 0.6237101582205916,
      "grad_norm": 0.8271064581571431,
      "learning_rate": 0.00032821434676720443,
      "loss": 4.0204,
      "step": 1360
    },
    {
      "epoch": 0.624168768631048,
      "grad_norm": 0.8278812981044426,
      "learning_rate": 0.00032751672375263836,
      "loss": 4.019,
      "step": 1361
    },
    {
      "epoch": 0.6246273790415042,
      "grad_norm": 0.8579073565572565,
      "learning_rate": 0.0003268194816606305,
      "loss": 3.9361,
      "step": 1362
    },
    {
      "epoch": 0.6250859894519606,
      "grad_norm": 0.8301957041547844,
      "learning_rate": 0.00032612262203101267,
      "loss": 3.7968,
      "step": 1363
    },
    {
      "epoch": 0.6255445998624168,
      "grad_norm": 0.7802270155144487,
      "learning_rate": 0.00032542614640277225,
      "loss": 3.8278,
      "step": 1364
    },
    {
      "epoch": 0.6260032102728732,
      "grad_norm": 0.8296255078207849,
      "learning_rate": 0.0003247300563140481,
      "loss": 4.0348,
      "step": 1365
    },
    {
      "epoch": 0.6264618206833296,
      "grad_norm": 0.8431682675702794,
      "learning_rate": 0.00032403435330212807,
      "loss": 3.8385,
      "step": 1366
    },
    {
      "epoch": 0.6269204310937858,
      "grad_norm": 0.820015148640858,
      "learning_rate": 0.00032333903890344515,
      "loss": 4.0285,
      "step": 1367
    },
    {
      "epoch": 0.6273790415042422,
      "grad_norm": 0.7976640576681971,
      "learning_rate": 0.00032264411465357333,
      "loss": 3.8047,
      "step": 1368
    },
    {
      "epoch": 0.6278376519146984,
      "grad_norm": 0.7990656295043932,
      "learning_rate": 0.00032194958208722654,
      "loss": 3.816,
      "step": 1369
    },
    {
      "epoch": 0.6282962623251548,
      "grad_norm": 0.8371264105962292,
      "learning_rate": 0.00032125544273825204,
      "loss": 4.1368,
      "step": 1370
    },
    {
      "epoch": 0.6287548727356111,
      "grad_norm": 0.7879086393094584,
      "learning_rate": 0.0003205616981396297,
      "loss": 3.8058,
      "step": 1371
    },
    {
      "epoch": 0.6292134831460674,
      "grad_norm": 0.8842164976565197,
      "learning_rate": 0.00031986834982346713,
      "loss": 3.9065,
      "step": 1372
    },
    {
      "epoch": 0.6296720935565238,
      "grad_norm": 0.8480102699625841,
      "learning_rate": 0.00031917539932099694,
      "loss": 4.0234,
      "step": 1373
    },
    {
      "epoch": 0.63013070396698,
      "grad_norm": 0.8075245342244977,
      "learning_rate": 0.00031848284816257336,
      "loss": 3.9266,
      "step": 1374
    },
    {
      "epoch": 0.6305893143774364,
      "grad_norm": 0.8415283725821179,
      "learning_rate": 0.0003177906978776682,
      "loss": 4.0715,
      "step": 1375
    },
    {
      "epoch": 0.6310479247878927,
      "grad_norm": 0.8137065174893241,
      "learning_rate": 0.0003170989499948683,
      "loss": 3.982,
      "step": 1376
    },
    {
      "epoch": 0.631506535198349,
      "grad_norm": 0.8527820190172468,
      "learning_rate": 0.0003164076060418719,
      "loss": 3.8203,
      "step": 1377
    },
    {
      "epoch": 0.6319651456088053,
      "grad_norm": 0.8053408662283964,
      "learning_rate": 0.000315716667545485,
      "loss": 3.9858,
      "step": 1378
    },
    {
      "epoch": 0.6324237560192616,
      "grad_norm": 0.7918623672042598,
      "learning_rate": 0.00031502613603161836,
      "loss": 3.9217,
      "step": 1379
    },
    {
      "epoch": 0.632882366429718,
      "grad_norm": 0.8003886798673967,
      "learning_rate": 0.00031433601302528335,
      "loss": 3.8349,
      "step": 1380
    },
    {
      "epoch": 0.6333409768401743,
      "grad_norm": 0.8363614597630419,
      "learning_rate": 0.00031364630005058995,
      "loss": 3.9762,
      "step": 1381
    },
    {
      "epoch": 0.6337995872506306,
      "grad_norm": 0.8100711046647334,
      "learning_rate": 0.0003129569986307422,
      "loss": 3.8362,
      "step": 1382
    },
    {
      "epoch": 0.6342581976610869,
      "grad_norm": 0.8293576816595525,
      "learning_rate": 0.00031226811028803515,
      "loss": 4.0536,
      "step": 1383
    },
    {
      "epoch": 0.6347168080715432,
      "grad_norm": 0.8672471681034112,
      "learning_rate": 0.00031157963654385173,
      "loss": 4.0704,
      "step": 1384
    },
    {
      "epoch": 0.6351754184819995,
      "grad_norm": 0.835271417045549,
      "learning_rate": 0.0003108915789186592,
      "loss": 3.8952,
      "step": 1385
    },
    {
      "epoch": 0.6356340288924559,
      "grad_norm": 0.8455786305036637,
      "learning_rate": 0.00031020393893200604,
      "loss": 3.9725,
      "step": 1386
    },
    {
      "epoch": 0.6360926393029122,
      "grad_norm": 0.8080627199603889,
      "learning_rate": 0.00030951671810251823,
      "loss": 3.9322,
      "step": 1387
    },
    {
      "epoch": 0.6365512497133685,
      "grad_norm": 0.85556368404487,
      "learning_rate": 0.0003088299179478959,
      "loss": 4.0657,
      "step": 1388
    },
    {
      "epoch": 0.6370098601238248,
      "grad_norm": 0.8542625859918379,
      "learning_rate": 0.0003081435399849104,
      "loss": 3.9495,
      "step": 1389
    },
    {
      "epoch": 0.6374684705342811,
      "grad_norm": 0.8234623626447184,
      "learning_rate": 0.0003074575857294004,
      "loss": 3.8727,
      "step": 1390
    },
    {
      "epoch": 0.6379270809447375,
      "grad_norm": 0.8132119203386518,
      "learning_rate": 0.0003067720566962691,
      "loss": 4.0405,
      "step": 1391
    },
    {
      "epoch": 0.6383856913551937,
      "grad_norm": 0.8093416563483133,
      "learning_rate": 0.0003060869543994806,
      "loss": 3.9443,
      "step": 1392
    },
    {
      "epoch": 0.6388443017656501,
      "grad_norm": 0.8519533967927642,
      "learning_rate": 0.0003054022803520562,
      "loss": 3.8569,
      "step": 1393
    },
    {
      "epoch": 0.6393029121761064,
      "grad_norm": 0.7986293113367736,
      "learning_rate": 0.0003047180360660721,
      "loss": 3.8858,
      "step": 1394
    },
    {
      "epoch": 0.6397615225865627,
      "grad_norm": 0.8506955769758677,
      "learning_rate": 0.00030403422305265475,
      "loss": 3.9661,
      "step": 1395
    },
    {
      "epoch": 0.6402201329970191,
      "grad_norm": 0.8427209567043839,
      "learning_rate": 0.0003033508428219785,
      "loss": 4.0351,
      "step": 1396
    },
    {
      "epoch": 0.6406787434074753,
      "grad_norm": 0.8175392334482335,
      "learning_rate": 0.00030266789688326184,
      "loss": 3.9257,
      "step": 1397
    },
    {
      "epoch": 0.6411373538179317,
      "grad_norm": 0.7807241690828611,
      "learning_rate": 0.00030198538674476393,
      "loss": 3.7777,
      "step": 1398
    },
    {
      "epoch": 0.6415959642283879,
      "grad_norm": 0.799333942421373,
      "learning_rate": 0.00030130331391378185,
      "loss": 3.808,
      "step": 1399
    },
    {
      "epoch": 0.6420545746388443,
      "grad_norm": 0.8239822709469106,
      "learning_rate": 0.0003006216798966468,
      "loss": 3.8881,
      "step": 1400
    },
    {
      "epoch": 0.6425131850493007,
      "grad_norm": 0.7956341933771285,
      "learning_rate": 0.00029994048619872034,
      "loss": 3.8324,
      "step": 1401
    },
    {
      "epoch": 0.6429717954597569,
      "grad_norm": 0.8407846899051726,
      "learning_rate": 0.0002992597343243927,
      "loss": 4.1117,
      "step": 1402
    },
    {
      "epoch": 0.6434304058702133,
      "grad_norm": 0.868063623735232,
      "learning_rate": 0.0002985794257770773,
      "loss": 4.0779,
      "step": 1403
    },
    {
      "epoch": 0.6438890162806695,
      "grad_norm": 0.8334125215436817,
      "learning_rate": 0.0002978995620592092,
      "loss": 3.9518,
      "step": 1404
    },
    {
      "epoch": 0.6443476266911259,
      "grad_norm": 0.7978359203953094,
      "learning_rate": 0.0002972201446722405,
      "loss": 3.9042,
      "step": 1405
    },
    {
      "epoch": 0.6448062371015822,
      "grad_norm": 0.8168277405353133,
      "learning_rate": 0.00029654117511663803,
      "loss": 3.7799,
      "step": 1406
    },
    {
      "epoch": 0.6452648475120385,
      "grad_norm": 0.8166124654382816,
      "learning_rate": 0.0002958626548918795,
      "loss": 4.1326,
      "step": 1407
    },
    {
      "epoch": 0.6457234579224949,
      "grad_norm": 0.8266231339304663,
      "learning_rate": 0.00029518458549645014,
      "loss": 3.8746,
      "step": 1408
    },
    {
      "epoch": 0.6461820683329511,
      "grad_norm": 0.8352879854125377,
      "learning_rate": 0.00029450696842783954,
      "loss": 3.9037,
      "step": 1409
    },
    {
      "epoch": 0.6466406787434075,
      "grad_norm": 0.8371776849432822,
      "learning_rate": 0.00029382980518253865,
      "loss": 4.0251,
      "step": 1410
    },
    {
      "epoch": 0.6470992891538638,
      "grad_norm": 0.8270189700200532,
      "learning_rate": 0.00029315309725603595,
      "loss": 3.949,
      "step": 1411
    },
    {
      "epoch": 0.6475578995643201,
      "grad_norm": 0.8688755166851149,
      "learning_rate": 0.00029247684614281446,
      "loss": 3.9394,
      "step": 1412
    },
    {
      "epoch": 0.6480165099747764,
      "grad_norm": 0.8261556871780721,
      "learning_rate": 0.0002918010533363481,
      "loss": 3.8524,
      "step": 1413
    },
    {
      "epoch": 0.6484751203852327,
      "grad_norm": 0.839533261636269,
      "learning_rate": 0.0002911257203290987,
      "loss": 4.0629,
      "step": 1414
    },
    {
      "epoch": 0.6489337307956891,
      "grad_norm": 0.8042014132981041,
      "learning_rate": 0.00029045084861251314,
      "loss": 4.0263,
      "step": 1415
    },
    {
      "epoch": 0.6493923412061454,
      "grad_norm": 0.841065042769581,
      "learning_rate": 0.00028977643967701897,
      "loss": 3.9148,
      "step": 1416
    },
    {
      "epoch": 0.6498509516166017,
      "grad_norm": 0.8390195431204432,
      "learning_rate": 0.00028910249501202156,
      "loss": 3.9941,
      "step": 1417
    },
    {
      "epoch": 0.650309562027058,
      "grad_norm": 0.8440703047863256,
      "learning_rate": 0.00028842901610590165,
      "loss": 4.0652,
      "step": 1418
    },
    {
      "epoch": 0.6507681724375143,
      "grad_norm": 0.8439622491753982,
      "learning_rate": 0.00028775600444601123,
      "loss": 3.8639,
      "step": 1419
    },
    {
      "epoch": 0.6512267828479706,
      "grad_norm": 0.8360349286982197,
      "learning_rate": 0.00028708346151866973,
      "loss": 4.0693,
      "step": 1420
    },
    {
      "epoch": 0.651685393258427,
      "grad_norm": 0.8086284760145629,
      "learning_rate": 0.0002864113888091622,
      "loss": 3.9646,
      "step": 1421
    },
    {
      "epoch": 0.6521440036688833,
      "grad_norm": 0.8712664160963192,
      "learning_rate": 0.0002857397878017348,
      "loss": 3.885,
      "step": 1422
    },
    {
      "epoch": 0.6526026140793396,
      "grad_norm": 0.8178095031435751,
      "learning_rate": 0.00028506865997959173,
      "loss": 4.0602,
      "step": 1423
    },
    {
      "epoch": 0.6530612244897959,
      "grad_norm": 0.8478157620164236,
      "learning_rate": 0.000284398006824893,
      "loss": 3.8159,
      "step": 1424
    },
    {
      "epoch": 0.6535198349002522,
      "grad_norm": 0.82815935827227,
      "learning_rate": 0.00028372782981874963,
      "loss": 3.8842,
      "step": 1425
    },
    {
      "epoch": 0.6539784453107086,
      "grad_norm": 0.789944719144222,
      "learning_rate": 0.00028305813044122096,
      "loss": 3.8922,
      "step": 1426
    },
    {
      "epoch": 0.6544370557211648,
      "grad_norm": 0.8286276570490952,
      "learning_rate": 0.0002823889101713122,
      "loss": 4.019,
      "step": 1427
    },
    {
      "epoch": 0.6548956661316212,
      "grad_norm": 0.8691586168441466,
      "learning_rate": 0.0002817201704869701,
      "loss": 4.2205,
      "step": 1428
    },
    {
      "epoch": 0.6553542765420775,
      "grad_norm": 0.8325325872566529,
      "learning_rate": 0.00028105191286508,
      "loss": 3.8638,
      "step": 1429
    },
    {
      "epoch": 0.6558128869525338,
      "grad_norm": 0.8696316237624052,
      "learning_rate": 0.00028038413878146245,
      "loss": 3.8282,
      "step": 1430
    },
    {
      "epoch": 0.6562714973629902,
      "grad_norm": 0.8386106923288569,
      "learning_rate": 0.00027971684971087073,
      "loss": 3.9456,
      "step": 1431
    },
    {
      "epoch": 0.6567301077734464,
      "grad_norm": 0.819392632840561,
      "learning_rate": 0.00027905004712698643,
      "loss": 4.0068,
      "step": 1432
    },
    {
      "epoch": 0.6571887181839028,
      "grad_norm": 0.8345965404527053,
      "learning_rate": 0.0002783837325024167,
      "loss": 4.0318,
      "step": 1433
    },
    {
      "epoch": 0.657647328594359,
      "grad_norm": 0.9253633311349919,
      "learning_rate": 0.00027771790730869153,
      "loss": 3.8889,
      "step": 1434
    },
    {
      "epoch": 0.6581059390048154,
      "grad_norm": 0.9129217002669905,
      "learning_rate": 0.0002770525730162599,
      "loss": 3.8426,
      "step": 1435
    },
    {
      "epoch": 0.6585645494152718,
      "grad_norm": 0.8400125788505369,
      "learning_rate": 0.00027638773109448645,
      "loss": 4.0646,
      "step": 1436
    },
    {
      "epoch": 0.659023159825728,
      "grad_norm": 0.8173705206052762,
      "learning_rate": 0.00027572338301164824,
      "loss": 3.8229,
      "step": 1437
    },
    {
      "epoch": 0.6594817702361844,
      "grad_norm": 0.8312932032480269,
      "learning_rate": 0.0002750595302349324,
      "loss": 3.8665,
      "step": 1438
    },
    {
      "epoch": 0.6599403806466407,
      "grad_norm": 0.8597310369069138,
      "learning_rate": 0.00027439617423043145,
      "loss": 3.8416,
      "step": 1439
    },
    {
      "epoch": 0.660398991057097,
      "grad_norm": 0.8451204682478889,
      "learning_rate": 0.00027373331646314114,
      "loss": 3.8594,
      "step": 1440
    },
    {
      "epoch": 0.6608576014675533,
      "grad_norm": 0.8867961743662639,
      "learning_rate": 0.0002730709583969572,
      "loss": 4.1597,
      "step": 1441
    },
    {
      "epoch": 0.6613162118780096,
      "grad_norm": 0.8430481271873226,
      "learning_rate": 0.0002724091014946711,
      "loss": 3.9273,
      "step": 1442
    },
    {
      "epoch": 0.661774822288466,
      "grad_norm": 0.8183650185019431,
      "learning_rate": 0.00027174774721796824,
      "loss": 3.7777,
      "step": 1443
    },
    {
      "epoch": 0.6622334326989223,
      "grad_norm": 0.9032698895583491,
      "learning_rate": 0.0002710868970274232,
      "loss": 3.9008,
      "step": 1444
    },
    {
      "epoch": 0.6626920431093786,
      "grad_norm": 0.8074320955257098,
      "learning_rate": 0.0002704265523824982,
      "loss": 4.0728,
      "step": 1445
    },
    {
      "epoch": 0.6631506535198349,
      "grad_norm": 0.8191861756325085,
      "learning_rate": 0.00026976671474153826,
      "loss": 3.9045,
      "step": 1446
    },
    {
      "epoch": 0.6636092639302912,
      "grad_norm": 0.8536546512634279,
      "learning_rate": 0.00026910738556176886,
      "loss": 3.9931,
      "step": 1447
    },
    {
      "epoch": 0.6640678743407475,
      "grad_norm": 0.8135728799073785,
      "learning_rate": 0.0002684485662992929,
      "loss": 4.0188,
      "step": 1448
    },
    {
      "epoch": 0.6645264847512039,
      "grad_norm": 0.8307110283766342,
      "learning_rate": 0.0002677902584090869,
      "loss": 3.7911,
      "step": 1449
    },
    {
      "epoch": 0.6649850951616602,
      "grad_norm": 0.829923648765451,
      "learning_rate": 0.00026713246334499774,
      "loss": 3.8853,
      "step": 1450
    },
    {
      "epoch": 0.6654437055721165,
      "grad_norm": 0.793260355416037,
      "learning_rate": 0.00026647518255974023,
      "loss": 3.8239,
      "step": 1451
    },
    {
      "epoch": 0.6659023159825728,
      "grad_norm": 0.8840191117569759,
      "learning_rate": 0.0002658184175048934,
      "loss": 3.9196,
      "step": 1452
    },
    {
      "epoch": 0.6663609263930291,
      "grad_norm": 0.8471034561350809,
      "learning_rate": 0.00026516216963089694,
      "loss": 4.0411,
      "step": 1453
    },
    {
      "epoch": 0.6668195368034855,
      "grad_norm": 0.866577370365908,
      "learning_rate": 0.0002645064403870488,
      "loss": 3.9413,
      "step": 1454
    },
    {
      "epoch": 0.6672781472139417,
      "grad_norm": 0.8450445362250235,
      "learning_rate": 0.0002638512312215011,
      "loss": 3.9415,
      "step": 1455
    },
    {
      "epoch": 0.6677367576243981,
      "grad_norm": 0.8589997268725659,
      "learning_rate": 0.0002631965435812575,
      "loss": 3.9368,
      "step": 1456
    },
    {
      "epoch": 0.6681953680348544,
      "grad_norm": 0.8386950224594587,
      "learning_rate": 0.00026254237891217046,
      "loss": 3.9714,
      "step": 1457
    },
    {
      "epoch": 0.6686539784453107,
      "grad_norm": 0.8688536395201032,
      "learning_rate": 0.0002618887386589367,
      "loss": 3.874,
      "step": 1458
    },
    {
      "epoch": 0.6691125888557671,
      "grad_norm": 0.8363855617271021,
      "learning_rate": 0.0002612356242650949,
      "loss": 3.7517,
      "step": 1459
    },
    {
      "epoch": 0.6695711992662233,
      "grad_norm": 0.9400027450760177,
      "learning_rate": 0.0002605830371730229,
      "loss": 4.026,
      "step": 1460
    },
    {
      "epoch": 0.6700298096766797,
      "grad_norm": 0.8849489727844753,
      "learning_rate": 0.0002599309788239339,
      "loss": 3.8523,
      "step": 1461
    },
    {
      "epoch": 0.6704884200871359,
      "grad_norm": 0.8197217821054994,
      "learning_rate": 0.00025927945065787306,
      "loss": 3.9098,
      "step": 1462
    },
    {
      "epoch": 0.6709470304975923,
      "grad_norm": 0.8517913631119876,
      "learning_rate": 0.0002586284541137145,
      "loss": 3.9338,
      "step": 1463
    },
    {
      "epoch": 0.6714056409080487,
      "grad_norm": 0.8359686720909615,
      "learning_rate": 0.00025797799062915905,
      "loss": 3.8887,
      "step": 1464
    },
    {
      "epoch": 0.6718642513185049,
      "grad_norm": 0.8753574255461343,
      "learning_rate": 0.00025732806164072966,
      "loss": 3.9846,
      "step": 1465
    },
    {
      "epoch": 0.6723228617289613,
      "grad_norm": 0.9033455876087808,
      "learning_rate": 0.00025667866858376874,
      "loss": 4.0503,
      "step": 1466
    },
    {
      "epoch": 0.6727814721394175,
      "grad_norm": 0.9184651458114076,
      "learning_rate": 0.0002560298128924358,
      "loss": 4.0447,
      "step": 1467
    },
    {
      "epoch": 0.6732400825498739,
      "grad_norm": 0.8504115182056106,
      "learning_rate": 0.0002553814959997032,
      "loss": 3.7914,
      "step": 1468
    },
    {
      "epoch": 0.6736986929603302,
      "grad_norm": 0.8150377333767564,
      "learning_rate": 0.00025473371933735334,
      "loss": 3.9877,
      "step": 1469
    },
    {
      "epoch": 0.6741573033707865,
      "grad_norm": 0.8270726529502985,
      "learning_rate": 0.00025408648433597534,
      "loss": 3.7834,
      "step": 1470
    },
    {
      "epoch": 0.6746159137812429,
      "grad_norm": 0.7912942428045493,
      "learning_rate": 0.00025343979242496283,
      "loss": 4.0675,
      "step": 1471
    },
    {
      "epoch": 0.6750745241916991,
      "grad_norm": 0.8266714363432616,
      "learning_rate": 0.00025279364503250925,
      "loss": 3.9023,
      "step": 1472
    },
    {
      "epoch": 0.6755331346021555,
      "grad_norm": 0.9217007892175163,
      "learning_rate": 0.0002521480435856056,
      "loss": 3.7979,
      "step": 1473
    },
    {
      "epoch": 0.6759917450126118,
      "grad_norm": 0.8557680120414126,
      "learning_rate": 0.0002515029895100378,
      "loss": 4.0426,
      "step": 1474
    },
    {
      "epoch": 0.6764503554230681,
      "grad_norm": 0.7845854680238283,
      "learning_rate": 0.0002508584842303822,
      "loss": 3.8629,
      "step": 1475
    },
    {
      "epoch": 0.6769089658335244,
      "grad_norm": 0.9547960419652226,
      "learning_rate": 0.0002502145291700038,
      "loss": 3.9595,
      "step": 1476
    },
    {
      "epoch": 0.6773675762439807,
      "grad_norm": 0.8623121945474526,
      "learning_rate": 0.0002495711257510517,
      "loss": 3.9328,
      "step": 1477
    },
    {
      "epoch": 0.6778261866544371,
      "grad_norm": 0.8339917924649397,
      "learning_rate": 0.0002489282753944575,
      "loss": 3.8459,
      "step": 1478
    },
    {
      "epoch": 0.6782847970648934,
      "grad_norm": 0.8386942365434303,
      "learning_rate": 0.00024828597951993093,
      "loss": 3.9373,
      "step": 1479
    },
    {
      "epoch": 0.6787434074753497,
      "grad_norm": 0.8212250107362843,
      "learning_rate": 0.00024764423954595706,
      "loss": 3.7922,
      "step": 1480
    },
    {
      "epoch": 0.679202017885806,
      "grad_norm": 0.8308469883170753,
      "learning_rate": 0.0002470030568897938,
      "loss": 3.9716,
      "step": 1481
    },
    {
      "epoch": 0.6796606282962623,
      "grad_norm": 0.8160966486900366,
      "learning_rate": 0.00024636243296746773,
      "loss": 3.8376,
      "step": 1482
    },
    {
      "epoch": 0.6801192387067186,
      "grad_norm": 0.8465855952055958,
      "learning_rate": 0.0002457223691937716,
      "loss": 3.9195,
      "step": 1483
    },
    {
      "epoch": 0.680577849117175,
      "grad_norm": 0.8391360347796458,
      "learning_rate": 0.0002450828669822613,
      "loss": 3.8893,
      "step": 1484
    },
    {
      "epoch": 0.6810364595276313,
      "grad_norm": 0.8120080009600139,
      "learning_rate": 0.00024444392774525253,
      "loss": 4.0687,
      "step": 1485
    },
    {
      "epoch": 0.6814950699380876,
      "grad_norm": 0.8594728961368204,
      "learning_rate": 0.00024380555289381733,
      "loss": 3.8973,
      "step": 1486
    },
    {
      "epoch": 0.6819536803485439,
      "grad_norm": 0.7977985486448946,
      "learning_rate": 0.00024316774383778184,
      "loss": 3.9134,
      "step": 1487
    },
    {
      "epoch": 0.6824122907590002,
      "grad_norm": 0.861209741769707,
      "learning_rate": 0.0002425305019857222,
      "loss": 3.9696,
      "step": 1488
    },
    {
      "epoch": 0.6828709011694566,
      "grad_norm": 0.8626880970610618,
      "learning_rate": 0.00024189382874496184,
      "loss": 4.0094,
      "step": 1489
    },
    {
      "epoch": 0.6833295115799128,
      "grad_norm": 0.8587476632418379,
      "learning_rate": 0.00024125772552156916,
      "loss": 4.0161,
      "step": 1490
    },
    {
      "epoch": 0.6837881219903692,
      "grad_norm": 0.8256170507131546,
      "learning_rate": 0.00024062219372035292,
      "loss": 3.9606,
      "step": 1491
    },
    {
      "epoch": 0.6842467324008255,
      "grad_norm": 0.8488851058081343,
      "learning_rate": 0.00023998723474486007,
      "loss": 4.0055,
      "step": 1492
    },
    {
      "epoch": 0.6847053428112818,
      "grad_norm": 0.8580077299842647,
      "learning_rate": 0.00023935284999737272,
      "loss": 3.9667,
      "step": 1493
    },
    {
      "epoch": 0.6851639532217382,
      "grad_norm": 0.8242673509289121,
      "learning_rate": 0.00023871904087890505,
      "loss": 3.9865,
      "step": 1494
    },
    {
      "epoch": 0.6856225636321944,
      "grad_norm": 0.8510710907143796,
      "learning_rate": 0.00023808580878919945,
      "loss": 3.8534,
      "step": 1495
    },
    {
      "epoch": 0.6860811740426508,
      "grad_norm": 0.8217926103303747,
      "learning_rate": 0.00023745315512672398,
      "loss": 3.8909,
      "step": 1496
    },
    {
      "epoch": 0.686539784453107,
      "grad_norm": 0.8317764853696693,
      "learning_rate": 0.0002368210812886698,
      "loss": 3.9653,
      "step": 1497
    },
    {
      "epoch": 0.6869983948635634,
      "grad_norm": 0.8277093003103875,
      "learning_rate": 0.0002361895886709471,
      "loss": 3.904,
      "step": 1498
    },
    {
      "epoch": 0.6874570052740198,
      "grad_norm": 0.8521281417661036,
      "learning_rate": 0.0002355586786681823,
      "loss": 3.8844,
      "step": 1499
    },
    {
      "epoch": 0.687915615684476,
      "grad_norm": 0.8265494584008372,
      "learning_rate": 0.00023492835267371575,
      "loss": 3.991,
      "step": 1500
    },
    {
      "epoch": 0.6883742260949324,
      "grad_norm": 0.8005936996128341,
      "learning_rate": 0.0002342986120795978,
      "loss": 3.9413,
      "step": 1501
    },
    {
      "epoch": 0.6888328365053886,
      "grad_norm": 0.8563356247738724,
      "learning_rate": 0.0002336694582765857,
      "loss": 4.0419,
      "step": 1502
    },
    {
      "epoch": 0.689291446915845,
      "grad_norm": 0.8086007801501258,
      "learning_rate": 0.00023304089265414085,
      "loss": 3.9987,
      "step": 1503
    },
    {
      "epoch": 0.6897500573263013,
      "grad_norm": 0.8414184320939578,
      "learning_rate": 0.00023241291660042613,
      "loss": 3.9461,
      "step": 1504
    },
    {
      "epoch": 0.6902086677367576,
      "grad_norm": 0.8387410990041797,
      "learning_rate": 0.00023178553150230186,
      "loss": 3.9583,
      "step": 1505
    },
    {
      "epoch": 0.690667278147214,
      "grad_norm": 0.8370608767974049,
      "learning_rate": 0.00023115873874532324,
      "loss": 4.0422,
      "step": 1506
    },
    {
      "epoch": 0.6911258885576702,
      "grad_norm": 0.8518286776240952,
      "learning_rate": 0.00023053253971373796,
      "loss": 3.9145,
      "step": 1507
    },
    {
      "epoch": 0.6915844989681266,
      "grad_norm": 0.8226359469583736,
      "learning_rate": 0.00022990693579048166,
      "loss": 3.9657,
      "step": 1508
    },
    {
      "epoch": 0.6920431093785829,
      "grad_norm": 0.8416452979433746,
      "learning_rate": 0.00022928192835717644,
      "loss": 3.925,
      "step": 1509
    },
    {
      "epoch": 0.6925017197890392,
      "grad_norm": 0.8085130392672399,
      "learning_rate": 0.00022865751879412634,
      "loss": 4.0533,
      "step": 1510
    },
    {
      "epoch": 0.6929603301994955,
      "grad_norm": 0.8146597422348393,
      "learning_rate": 0.00022803370848031585,
      "loss": 3.9866,
      "step": 1511
    },
    {
      "epoch": 0.6934189406099518,
      "grad_norm": 0.8114255626360013,
      "learning_rate": 0.00022741049879340542,
      "loss": 3.853,
      "step": 1512
    },
    {
      "epoch": 0.6938775510204082,
      "grad_norm": 0.8560348052157918,
      "learning_rate": 0.00022678789110972897,
      "loss": 3.9134,
      "step": 1513
    },
    {
      "epoch": 0.6943361614308645,
      "grad_norm": 0.8618270395179488,
      "learning_rate": 0.00022616588680429155,
      "loss": 3.9875,
      "step": 1514
    },
    {
      "epoch": 0.6947947718413208,
      "grad_norm": 0.8621956123367567,
      "learning_rate": 0.00022554448725076526,
      "loss": 3.896,
      "step": 1515
    },
    {
      "epoch": 0.6952533822517771,
      "grad_norm": 0.8112849907757043,
      "learning_rate": 0.0002249236938214863,
      "loss": 3.9351,
      "step": 1516
    },
    {
      "epoch": 0.6957119926622334,
      "grad_norm": 0.8360645228877108,
      "learning_rate": 0.00022430350788745296,
      "loss": 4.1373,
      "step": 1517
    },
    {
      "epoch": 0.6961706030726897,
      "grad_norm": 0.8162284850328998,
      "learning_rate": 0.00022368393081832166,
      "loss": 3.9655,
      "step": 1518
    },
    {
      "epoch": 0.6966292134831461,
      "grad_norm": 0.8196422248234556,
      "learning_rate": 0.00022306496398240383,
      "loss": 3.8683,
      "step": 1519
    },
    {
      "epoch": 0.6970878238936024,
      "grad_norm": 0.8788690657167267,
      "learning_rate": 0.00022244660874666373,
      "loss": 3.9495,
      "step": 1520
    },
    {
      "epoch": 0.6975464343040587,
      "grad_norm": 0.8600788458203402,
      "learning_rate": 0.00022182886647671452,
      "loss": 3.8465,
      "step": 1521
    },
    {
      "epoch": 0.698005044714515,
      "grad_norm": 0.8153449734067152,
      "learning_rate": 0.0002212117385368157,
      "loss": 3.859,
      "step": 1522
    },
    {
      "epoch": 0.6984636551249713,
      "grad_norm": 0.8119625290556672,
      "learning_rate": 0.00022059522628987038,
      "loss": 3.8825,
      "step": 1523
    },
    {
      "epoch": 0.6989222655354277,
      "grad_norm": 0.818414340301475,
      "learning_rate": 0.00021997933109742162,
      "loss": 3.9449,
      "step": 1524
    },
    {
      "epoch": 0.6993808759458839,
      "grad_norm": 0.8749812227388654,
      "learning_rate": 0.00021936405431964969,
      "loss": 3.7651,
      "step": 1525
    },
    {
      "epoch": 0.6998394863563403,
      "grad_norm": 0.8529994967696798,
      "learning_rate": 0.00021874939731536926,
      "loss": 3.8536,
      "step": 1526
    },
    {
      "epoch": 0.7002980967667966,
      "grad_norm": 0.7815983645278257,
      "learning_rate": 0.00021813536144202656,
      "loss": 3.7956,
      "step": 1527
    },
    {
      "epoch": 0.7007567071772529,
      "grad_norm": 0.8561782545596429,
      "learning_rate": 0.00021752194805569553,
      "loss": 3.8256,
      "step": 1528
    },
    {
      "epoch": 0.7012153175877093,
      "grad_norm": 0.8771737987650479,
      "learning_rate": 0.0002169091585110754,
      "loss": 3.841,
      "step": 1529
    },
    {
      "epoch": 0.7016739279981655,
      "grad_norm": 0.8747136007072696,
      "learning_rate": 0.00021629699416148828,
      "loss": 4.0614,
      "step": 1530
    },
    {
      "epoch": 0.7021325384086219,
      "grad_norm": 0.8681978303888083,
      "learning_rate": 0.000215685456358875,
      "loss": 4.1174,
      "step": 1531
    },
    {
      "epoch": 0.7025911488190782,
      "grad_norm": 0.8065729378387347,
      "learning_rate": 0.00021507454645379258,
      "loss": 4.0124,
      "step": 1532
    },
    {
      "epoch": 0.7030497592295345,
      "grad_norm": 0.7814894863467946,
      "learning_rate": 0.00021446426579541184,
      "loss": 3.8218,
      "step": 1533
    },
    {
      "epoch": 0.7035083696399909,
      "grad_norm": 0.8190089389944872,
      "learning_rate": 0.00021385461573151387,
      "loss": 4.0097,
      "step": 1534
    },
    {
      "epoch": 0.7039669800504471,
      "grad_norm": 0.8751372238786197,
      "learning_rate": 0.00021324559760848677,
      "loss": 3.7915,
      "step": 1535
    },
    {
      "epoch": 0.7044255904609035,
      "grad_norm": 0.8457891586208977,
      "learning_rate": 0.00021263721277132303,
      "loss": 3.8935,
      "step": 1536
    },
    {
      "epoch": 0.7048842008713598,
      "grad_norm": 0.8417487221971869,
      "learning_rate": 0.0002120294625636171,
      "loss": 3.9907,
      "step": 1537
    },
    {
      "epoch": 0.7053428112818161,
      "grad_norm": 0.8776140467736573,
      "learning_rate": 0.0002114223483275613,
      "loss": 4.0125,
      "step": 1538
    },
    {
      "epoch": 0.7058014216922724,
      "grad_norm": 0.8610531598904658,
      "learning_rate": 0.0002108158714039435,
      "loss": 4.1747,
      "step": 1539
    },
    {
      "epoch": 0.7062600321027287,
      "grad_norm": 0.826544547686025,
      "learning_rate": 0.00021021003313214455,
      "loss": 3.9858,
      "step": 1540
    },
    {
      "epoch": 0.7067186425131851,
      "grad_norm": 0.8314199370615432,
      "learning_rate": 0.00020960483485013432,
      "loss": 3.9208,
      "step": 1541
    },
    {
      "epoch": 0.7071772529236414,
      "grad_norm": 0.8400947461878123,
      "learning_rate": 0.0002090002778944694,
      "loss": 3.9568,
      "step": 1542
    },
    {
      "epoch": 0.7076358633340977,
      "grad_norm": 0.8553173394422553,
      "learning_rate": 0.00020839636360029025,
      "loss": 3.9466,
      "step": 1543
    },
    {
      "epoch": 0.708094473744554,
      "grad_norm": 0.8588477150788186,
      "learning_rate": 0.00020779309330131818,
      "loss": 3.8812,
      "step": 1544
    },
    {
      "epoch": 0.7085530841550103,
      "grad_norm": 0.8275269939189922,
      "learning_rate": 0.00020719046832985184,
      "loss": 3.8681,
      "step": 1545
    },
    {
      "epoch": 0.7090116945654666,
      "grad_norm": 0.8253472715894828,
      "learning_rate": 0.0002065884900167646,
      "loss": 4.0638,
      "step": 1546
    },
    {
      "epoch": 0.709470304975923,
      "grad_norm": 0.8275823330522397,
      "learning_rate": 0.0002059871596915024,
      "loss": 4.1548,
      "step": 1547
    },
    {
      "epoch": 0.7099289153863793,
      "grad_norm": 0.8217311552625174,
      "learning_rate": 0.0002053864786820795,
      "loss": 3.74,
      "step": 1548
    },
    {
      "epoch": 0.7103875257968356,
      "grad_norm": 0.838877748072078,
      "learning_rate": 0.00020478644831507627,
      "loss": 4.0433,
      "step": 1549
    },
    {
      "epoch": 0.7108461362072919,
      "grad_norm": 0.8319714210159528,
      "learning_rate": 0.00020418706991563634,
      "loss": 3.9419,
      "step": 1550
    },
    {
      "epoch": 0.7113047466177482,
      "grad_norm": 0.8336552991372955,
      "learning_rate": 0.00020358834480746363,
      "loss": 3.9818,
      "step": 1551
    },
    {
      "epoch": 0.7117633570282046,
      "grad_norm": 0.8324146120738223,
      "learning_rate": 0.0002029902743128188,
      "loss": 4.0192,
      "step": 1552
    },
    {
      "epoch": 0.7122219674386608,
      "grad_norm": 0.8424559422817726,
      "learning_rate": 0.0002023928597525174,
      "loss": 3.9035,
      "step": 1553
    },
    {
      "epoch": 0.7126805778491172,
      "grad_norm": 0.8188647371920655,
      "learning_rate": 0.00020179610244592595,
      "loss": 3.9411,
      "step": 1554
    },
    {
      "epoch": 0.7131391882595735,
      "grad_norm": 0.8664545065651408,
      "learning_rate": 0.00020120000371095937,
      "loss": 3.8243,
      "step": 1555
    },
    {
      "epoch": 0.7135977986700298,
      "grad_norm": 0.8004070628904555,
      "learning_rate": 0.0002006045648640787,
      "loss": 3.8505,
      "step": 1556
    },
    {
      "epoch": 0.7140564090804862,
      "grad_norm": 0.8327900427470745,
      "learning_rate": 0.00020000978722028713,
      "loss": 4.112,
      "step": 1557
    },
    {
      "epoch": 0.7145150194909424,
      "grad_norm": 0.8165064498109851,
      "learning_rate": 0.00019941567209312767,
      "loss": 3.901,
      "step": 1558
    },
    {
      "epoch": 0.7149736299013988,
      "grad_norm": 0.8772119114240009,
      "learning_rate": 0.00019882222079468036,
      "loss": 3.8415,
      "step": 1559
    },
    {
      "epoch": 0.715432240311855,
      "grad_norm": 0.8135541742058268,
      "learning_rate": 0.0001982294346355595,
      "loss": 3.9247,
      "step": 1560
    },
    {
      "epoch": 0.7158908507223114,
      "grad_norm": 0.8310922735771817,
      "learning_rate": 0.00019763731492490976,
      "loss": 3.8925,
      "step": 1561
    },
    {
      "epoch": 0.7163494611327678,
      "grad_norm": 0.852726727517768,
      "learning_rate": 0.00019704586297040422,
      "loss": 3.9403,
      "step": 1562
    },
    {
      "epoch": 0.716808071543224,
      "grad_norm": 0.855678234677389,
      "learning_rate": 0.0001964550800782417,
      "loss": 3.8529,
      "step": 1563
    },
    {
      "epoch": 0.7172666819536804,
      "grad_norm": 0.8215005530887987,
      "learning_rate": 0.00019586496755314288,
      "loss": 3.8234,
      "step": 1564
    },
    {
      "epoch": 0.7177252923641366,
      "grad_norm": 0.7996814647998756,
      "learning_rate": 0.00019527552669834798,
      "loss": 3.9184,
      "step": 1565
    },
    {
      "epoch": 0.718183902774593,
      "grad_norm": 0.8417094203478128,
      "learning_rate": 0.0001946867588156142,
      "loss": 3.7569,
      "step": 1566
    },
    {
      "epoch": 0.7186425131850493,
      "grad_norm": 0.8405296219725839,
      "learning_rate": 0.00019409866520521258,
      "loss": 3.9181,
      "step": 1567
    },
    {
      "epoch": 0.7191011235955056,
      "grad_norm": 0.8524006005221206,
      "learning_rate": 0.00019351124716592455,
      "loss": 3.7515,
      "step": 1568
    },
    {
      "epoch": 0.719559734005962,
      "grad_norm": 0.8875872474880201,
      "learning_rate": 0.0001929245059950397,
      "loss": 4.0947,
      "step": 1569
    },
    {
      "epoch": 0.7200183444164182,
      "grad_norm": 0.8930356527219394,
      "learning_rate": 0.0001923384429883533,
      "loss": 3.7333,
      "step": 1570
    },
    {
      "epoch": 0.7204769548268746,
      "grad_norm": 0.8238203845331978,
      "learning_rate": 0.00019175305944016237,
      "loss": 3.987,
      "step": 1571
    },
    {
      "epoch": 0.7209355652373309,
      "grad_norm": 0.8883726758634649,
      "learning_rate": 0.00019116835664326326,
      "loss": 3.8915,
      "step": 1572
    },
    {
      "epoch": 0.7213941756477872,
      "grad_norm": 0.8249567829831989,
      "learning_rate": 0.0001905843358889497,
      "loss": 3.916,
      "step": 1573
    },
    {
      "epoch": 0.7218527860582435,
      "grad_norm": 0.8451038769272609,
      "learning_rate": 0.00019000099846700836,
      "loss": 3.9107,
      "step": 1574
    },
    {
      "epoch": 0.7223113964686998,
      "grad_norm": 0.8491857088959867,
      "learning_rate": 0.00018941834566571692,
      "loss": 3.7993,
      "step": 1575
    },
    {
      "epoch": 0.7227700068791562,
      "grad_norm": 0.81703266382779,
      "learning_rate": 0.00018883637877184145,
      "loss": 3.9819,
      "step": 1576
    },
    {
      "epoch": 0.7232286172896125,
      "grad_norm": 0.8890548695566171,
      "learning_rate": 0.00018825509907063325,
      "loss": 3.9344,
      "step": 1577
    },
    {
      "epoch": 0.7236872277000688,
      "grad_norm": 0.7955995018776147,
      "learning_rate": 0.00018767450784582557,
      "loss": 3.9188,
      "step": 1578
    },
    {
      "epoch": 0.7241458381105251,
      "grad_norm": 0.8258950263348376,
      "learning_rate": 0.00018709460637963122,
      "loss": 3.8597,
      "step": 1579
    },
    {
      "epoch": 0.7246044485209814,
      "grad_norm": 0.8509576967217473,
      "learning_rate": 0.00018651539595274013,
      "loss": 3.8647,
      "step": 1580
    },
    {
      "epoch": 0.7250630589314377,
      "grad_norm": 0.8028527796683224,
      "learning_rate": 0.00018593687784431578,
      "loss": 3.7651,
      "step": 1581
    },
    {
      "epoch": 0.7255216693418941,
      "grad_norm": 0.8348404627323313,
      "learning_rate": 0.00018535905333199248,
      "loss": 3.9025,
      "step": 1582
    },
    {
      "epoch": 0.7259802797523504,
      "grad_norm": 0.8954550101979705,
      "learning_rate": 0.0001847819236918733,
      "loss": 4.1266,
      "step": 1583
    },
    {
      "epoch": 0.7264388901628067,
      "grad_norm": 0.8483420536214036,
      "learning_rate": 0.00018420549019852655,
      "loss": 4.0076,
      "step": 1584
    },
    {
      "epoch": 0.726897500573263,
      "grad_norm": 0.8470941176091954,
      "learning_rate": 0.00018362975412498266,
      "loss": 3.8452,
      "step": 1585
    },
    {
      "epoch": 0.7273561109837193,
      "grad_norm": 0.8335773845154771,
      "learning_rate": 0.00018305471674273261,
      "loss": 3.8766,
      "step": 1586
    },
    {
      "epoch": 0.7278147213941757,
      "grad_norm": 0.8448234532857356,
      "learning_rate": 0.0001824803793217237,
      "loss": 3.9579,
      "step": 1587
    },
    {
      "epoch": 0.7282733318046319,
      "grad_norm": 0.828396498511952,
      "learning_rate": 0.00018190674313035737,
      "loss": 3.9883,
      "step": 1588
    },
    {
      "epoch": 0.7287319422150883,
      "grad_norm": 0.8164565795803416,
      "learning_rate": 0.00018133380943548716,
      "loss": 3.8133,
      "step": 1589
    },
    {
      "epoch": 0.7291905526255446,
      "grad_norm": 0.8601340848204636,
      "learning_rate": 0.00018076157950241452,
      "loss": 3.9492,
      "step": 1590
    },
    {
      "epoch": 0.7296491630360009,
      "grad_norm": 0.9050414506740856,
      "learning_rate": 0.00018019005459488652,
      "loss": 4.0503,
      "step": 1591
    },
    {
      "epoch": 0.7301077734464573,
      "grad_norm": 0.8579254474699252,
      "learning_rate": 0.00017961923597509388,
      "loss": 3.8389,
      "step": 1592
    },
    {
      "epoch": 0.7305663838569135,
      "grad_norm": 0.8368385043249761,
      "learning_rate": 0.00017904912490366722,
      "loss": 4.0391,
      "step": 1593
    },
    {
      "epoch": 0.7310249942673699,
      "grad_norm": 0.8422814628962383,
      "learning_rate": 0.00017847972263967433,
      "loss": 3.9168,
      "step": 1594
    },
    {
      "epoch": 0.7314836046778261,
      "grad_norm": 0.8778586966249694,
      "learning_rate": 0.0001779110304406177,
      "loss": 3.7822,
      "step": 1595
    },
    {
      "epoch": 0.7319422150882825,
      "grad_norm": 0.9137982323703012,
      "learning_rate": 0.000177343049562432,
      "loss": 3.9703,
      "step": 1596
    },
    {
      "epoch": 0.7324008254987389,
      "grad_norm": 0.8359496191518861,
      "learning_rate": 0.0001767757812594807,
      "loss": 3.8372,
      "step": 1597
    },
    {
      "epoch": 0.7328594359091951,
      "grad_norm": 0.8663844262234691,
      "learning_rate": 0.0001762092267845534,
      "loss": 3.8599,
      "step": 1598
    },
    {
      "epoch": 0.7333180463196515,
      "grad_norm": 0.8690594978188074,
      "learning_rate": 0.00017564338738886365,
      "loss": 4.0494,
      "step": 1599
    },
    {
      "epoch": 0.7337766567301077,
      "grad_norm": 0.8085411744025421,
      "learning_rate": 0.0001750782643220457,
      "loss": 4.1033,
      "step": 1600
    },
    {
      "epoch": 0.7342352671405641,
      "grad_norm": 0.8359991631871402,
      "learning_rate": 0.00017451385883215166,
      "loss": 3.725,
      "step": 1601
    },
    {
      "epoch": 0.7346938775510204,
      "grad_norm": 0.8207723744824981,
      "learning_rate": 0.00017395017216564863,
      "loss": 3.9142,
      "step": 1602
    },
    {
      "epoch": 0.7351524879614767,
      "grad_norm": 0.8277681049497209,
      "learning_rate": 0.00017338720556741687,
      "loss": 4.0738,
      "step": 1603
    },
    {
      "epoch": 0.7356110983719331,
      "grad_norm": 0.8372971988651867,
      "learning_rate": 0.00017282496028074606,
      "loss": 3.9153,
      "step": 1604
    },
    {
      "epoch": 0.7360697087823893,
      "grad_norm": 0.8002732026019794,
      "learning_rate": 0.00017226343754733254,
      "loss": 3.8659,
      "step": 1605
    },
    {
      "epoch": 0.7365283191928457,
      "grad_norm": 0.8268686888883537,
      "learning_rate": 0.00017170263860727769,
      "loss": 3.8991,
      "step": 1606
    },
    {
      "epoch": 0.736986929603302,
      "grad_norm": 0.8295380150679361,
      "learning_rate": 0.0001711425646990838,
      "loss": 3.9817,
      "step": 1607
    },
    {
      "epoch": 0.7374455400137583,
      "grad_norm": 0.8765708168590606,
      "learning_rate": 0.00017058321705965202,
      "loss": 3.921,
      "step": 1608
    },
    {
      "epoch": 0.7379041504242146,
      "grad_norm": 0.8418948695659235,
      "learning_rate": 0.0001700245969242798,
      "loss": 3.8767,
      "step": 1609
    },
    {
      "epoch": 0.7383627608346709,
      "grad_norm": 0.8729525329513275,
      "learning_rate": 0.00016946670552665804,
      "loss": 4.023,
      "step": 1610
    },
    {
      "epoch": 0.7388213712451273,
      "grad_norm": 0.8382015490572661,
      "learning_rate": 0.00016890954409886795,
      "loss": 3.9322,
      "step": 1611
    },
    {
      "epoch": 0.7392799816555836,
      "grad_norm": 0.8967820864783295,
      "learning_rate": 0.00016835311387137836,
      "loss": 3.9153,
      "step": 1612
    },
    {
      "epoch": 0.7397385920660399,
      "grad_norm": 0.8520255320067875,
      "learning_rate": 0.0001677974160730441,
      "loss": 3.9628,
      "step": 1613
    },
    {
      "epoch": 0.7401972024764962,
      "grad_norm": 0.9490015488812981,
      "learning_rate": 0.00016724245193110176,
      "loss": 3.9486,
      "step": 1614
    },
    {
      "epoch": 0.7406558128869525,
      "grad_norm": 0.8647462876081115,
      "learning_rate": 0.00016668822267116784,
      "loss": 3.8917,
      "step": 1615
    },
    {
      "epoch": 0.7411144232974088,
      "grad_norm": 0.8294616131978336,
      "learning_rate": 0.00016613472951723597,
      "loss": 3.849,
      "step": 1616
    },
    {
      "epoch": 0.7415730337078652,
      "grad_norm": 0.8511360565105168,
      "learning_rate": 0.00016558197369167434,
      "loss": 4.0139,
      "step": 1617
    },
    {
      "epoch": 0.7420316441183215,
      "grad_norm": 0.817415765727012,
      "learning_rate": 0.00016502995641522216,
      "loss": 3.9718,
      "step": 1618
    },
    {
      "epoch": 0.7424902545287778,
      "grad_norm": 0.8376278128859767,
      "learning_rate": 0.00016447867890698843,
      "loss": 4.0064,
      "step": 1619
    },
    {
      "epoch": 0.7429488649392341,
      "grad_norm": 0.8111186693613259,
      "learning_rate": 0.00016392814238444753,
      "loss": 3.8117,
      "step": 1620
    },
    {
      "epoch": 0.7434074753496904,
      "grad_norm": 0.8591321135953793,
      "learning_rate": 0.00016337834806343782,
      "loss": 3.8225,
      "step": 1621
    },
    {
      "epoch": 0.7438660857601468,
      "grad_norm": 0.8963885198928919,
      "learning_rate": 0.0001628292971581588,
      "loss": 4.0286,
      "step": 1622
    },
    {
      "epoch": 0.744324696170603,
      "grad_norm": 0.8561324229837618,
      "learning_rate": 0.00016228099088116772,
      "loss": 4.0088,
      "step": 1623
    },
    {
      "epoch": 0.7447833065810594,
      "grad_norm": 0.8537425294990139,
      "learning_rate": 0.00016173343044337734,
      "loss": 3.7388,
      "step": 1624
    },
    {
      "epoch": 0.7452419169915158,
      "grad_norm": 0.8229903877397557,
      "learning_rate": 0.00016118661705405356,
      "loss": 4.0096,
      "step": 1625
    },
    {
      "epoch": 0.745700527401972,
      "grad_norm": 0.8145835995508554,
      "learning_rate": 0.00016064055192081255,
      "loss": 3.8475,
      "step": 1626
    },
    {
      "epoch": 0.7461591378124284,
      "grad_norm": 0.8314202120605467,
      "learning_rate": 0.00016009523624961757,
      "loss": 3.7322,
      "step": 1627
    },
    {
      "epoch": 0.7466177482228846,
      "grad_norm": 0.7991245840062375,
      "learning_rate": 0.00015955067124477678,
      "loss": 3.8986,
      "step": 1628
    },
    {
      "epoch": 0.747076358633341,
      "grad_norm": 0.8589248876456193,
      "learning_rate": 0.000159006858108941,
      "loss": 3.925,
      "step": 1629
    },
    {
      "epoch": 0.7475349690437973,
      "grad_norm": 0.8633325540500519,
      "learning_rate": 0.00015846379804310002,
      "loss": 3.8413,
      "step": 1630
    },
    {
      "epoch": 0.7479935794542536,
      "grad_norm": 0.8560486555158954,
      "learning_rate": 0.00015792149224658054,
      "loss": 4.0387,
      "step": 1631
    },
    {
      "epoch": 0.74845218986471,
      "grad_norm": 0.8520853833493139,
      "learning_rate": 0.00015737994191704385,
      "loss": 3.8369,
      "step": 1632
    },
    {
      "epoch": 0.7489108002751662,
      "grad_norm": 0.8922079894929041,
      "learning_rate": 0.0001568391482504829,
      "loss": 3.7835,
      "step": 1633
    },
    {
      "epoch": 0.7493694106856226,
      "grad_norm": 0.8341831075463805,
      "learning_rate": 0.00015629911244121903,
      "loss": 4.1251,
      "step": 1634
    },
    {
      "epoch": 0.7498280210960789,
      "grad_norm": 0.8491356959740831,
      "learning_rate": 0.0001557598356819,
      "loss": 3.8885,
      "step": 1635
    },
    {
      "epoch": 0.7502866315065352,
      "grad_norm": 0.8773033291370399,
      "learning_rate": 0.00015522131916349786,
      "loss": 3.9548,
      "step": 1636
    },
    {
      "epoch": 0.7507452419169915,
      "grad_norm": 0.9366190978828319,
      "learning_rate": 0.00015468356407530493,
      "loss": 4.0864,
      "step": 1637
    },
    {
      "epoch": 0.7512038523274478,
      "grad_norm": 0.8550570100784317,
      "learning_rate": 0.00015414657160493217,
      "loss": 3.9843,
      "step": 1638
    },
    {
      "epoch": 0.7516624627379042,
      "grad_norm": 0.8298326550097367,
      "learning_rate": 0.00015361034293830673,
      "loss": 3.954,
      "step": 1639
    },
    {
      "epoch": 0.7521210731483605,
      "grad_norm": 0.8423544656183941,
      "learning_rate": 0.00015307487925966844,
      "loss": 3.9991,
      "step": 1640
    },
    {
      "epoch": 0.7525796835588168,
      "grad_norm": 0.8307841958444293,
      "learning_rate": 0.00015254018175156776,
      "loss": 3.9316,
      "step": 1641
    },
    {
      "epoch": 0.7530382939692731,
      "grad_norm": 0.8352424467704046,
      "learning_rate": 0.0001520062515948632,
      "loss": 3.7273,
      "step": 1642
    },
    {
      "epoch": 0.7534969043797294,
      "grad_norm": 0.8857477344604218,
      "learning_rate": 0.0001514730899687189,
      "loss": 3.9646,
      "step": 1643
    },
    {
      "epoch": 0.7539555147901857,
      "grad_norm": 0.8325163799237432,
      "learning_rate": 0.00015094069805060122,
      "loss": 3.9241,
      "step": 1644
    },
    {
      "epoch": 0.7544141252006421,
      "grad_norm": 0.8622501578537539,
      "learning_rate": 0.00015040907701627666,
      "loss": 3.7732,
      "step": 1645
    },
    {
      "epoch": 0.7548727356110984,
      "grad_norm": 0.8279486718130658,
      "learning_rate": 0.00014987822803980976,
      "loss": 3.9722,
      "step": 1646
    },
    {
      "epoch": 0.7553313460215547,
      "grad_norm": 0.8470918302373529,
      "learning_rate": 0.00014934815229355965,
      "loss": 3.8545,
      "step": 1647
    },
    {
      "epoch": 0.755789956432011,
      "grad_norm": 0.8546558008230275,
      "learning_rate": 0.00014881885094817748,
      "loss": 4.0816,
      "step": 1648
    },
    {
      "epoch": 0.7562485668424673,
      "grad_norm": 0.8218213110396926,
      "learning_rate": 0.00014829032517260488,
      "loss": 3.9487,
      "step": 1649
    },
    {
      "epoch": 0.7567071772529237,
      "grad_norm": 0.8561374963143255,
      "learning_rate": 0.0001477625761340704,
      "loss": 4.0122,
      "step": 1650
    },
    {
      "epoch": 0.7571657876633799,
      "grad_norm": 0.838234017437776,
      "learning_rate": 0.0001472356049980868,
      "loss": 3.82,
      "step": 1651
    },
    {
      "epoch": 0.7576243980738363,
      "grad_norm": 0.8619916378429533,
      "learning_rate": 0.00014670941292844954,
      "loss": 3.7776,
      "step": 1652
    },
    {
      "epoch": 0.7580830084842926,
      "grad_norm": 0.8080172709597696,
      "learning_rate": 0.00014618400108723295,
      "loss": 3.7854,
      "step": 1653
    },
    {
      "epoch": 0.7585416188947489,
      "grad_norm": 0.8316081825641323,
      "learning_rate": 0.00014565937063478862,
      "loss": 3.8453,
      "step": 1654
    },
    {
      "epoch": 0.7590002293052053,
      "grad_norm": 0.8332692375592281,
      "learning_rate": 0.00014513552272974207,
      "loss": 3.8425,
      "step": 1655
    },
    {
      "epoch": 0.7594588397156615,
      "grad_norm": 0.8302135149550943,
      "learning_rate": 0.0001446124585289913,
      "loss": 3.8273,
      "step": 1656
    },
    {
      "epoch": 0.7599174501261179,
      "grad_norm": 0.8477139903764007,
      "learning_rate": 0.00014409017918770266,
      "loss": 3.992,
      "step": 1657
    },
    {
      "epoch": 0.7603760605365741,
      "grad_norm": 0.8869440739297604,
      "learning_rate": 0.00014356868585930994,
      "loss": 3.8177,
      "step": 1658
    },
    {
      "epoch": 0.7608346709470305,
      "grad_norm": 0.8935840135462629,
      "learning_rate": 0.00014304797969551077,
      "loss": 4.2162,
      "step": 1659
    },
    {
      "epoch": 0.7612932813574869,
      "grad_norm": 0.887283718151882,
      "learning_rate": 0.00014252806184626417,
      "loss": 3.9963,
      "step": 1660
    },
    {
      "epoch": 0.7617518917679431,
      "grad_norm": 0.8376316287714409,
      "learning_rate": 0.00014200893345978817,
      "loss": 3.9235,
      "step": 1661
    },
    {
      "epoch": 0.7622105021783995,
      "grad_norm": 0.8799635448422263,
      "learning_rate": 0.00014149059568255778,
      "loss": 3.8793,
      "step": 1662
    },
    {
      "epoch": 0.7626691125888557,
      "grad_norm": 0.8135255398723551,
      "learning_rate": 0.00014097304965930157,
      "loss": 3.9703,
      "step": 1663
    },
    {
      "epoch": 0.7631277229993121,
      "grad_norm": 0.8314251936713516,
      "learning_rate": 0.00014045629653299953,
      "loss": 3.8623,
      "step": 1664
    },
    {
      "epoch": 0.7635863334097684,
      "grad_norm": 0.8387986347777888,
      "learning_rate": 0.00013994033744488076,
      "loss": 3.8174,
      "step": 1665
    },
    {
      "epoch": 0.7640449438202247,
      "grad_norm": 0.8000105908952896,
      "learning_rate": 0.00013942517353442092,
      "loss": 3.8907,
      "step": 1666
    },
    {
      "epoch": 0.7645035542306811,
      "grad_norm": 0.8572646811056924,
      "learning_rate": 0.0001389108059393391,
      "loss": 3.9679,
      "step": 1667
    },
    {
      "epoch": 0.7649621646411373,
      "grad_norm": 0.8433924348923313,
      "learning_rate": 0.00013839723579559581,
      "loss": 3.878,
      "step": 1668
    },
    {
      "epoch": 0.7654207750515937,
      "grad_norm": 0.8182266282459676,
      "learning_rate": 0.00013788446423739103,
      "loss": 3.9194,
      "step": 1669
    },
    {
      "epoch": 0.76587938546205,
      "grad_norm": 0.7890418992445691,
      "learning_rate": 0.00013737249239716042,
      "loss": 3.8839,
      "step": 1670
    },
    {
      "epoch": 0.7663379958725063,
      "grad_norm": 0.8298253342036503,
      "learning_rate": 0.00013686132140557355,
      "loss": 3.891,
      "step": 1671
    },
    {
      "epoch": 0.7667966062829626,
      "grad_norm": 0.8936775706693539,
      "learning_rate": 0.00013635095239153188,
      "loss": 3.8912,
      "step": 1672
    },
    {
      "epoch": 0.7672552166934189,
      "grad_norm": 0.8488535067304734,
      "learning_rate": 0.00013584138648216527,
      "loss": 3.9474,
      "step": 1673
    },
    {
      "epoch": 0.7677138271038753,
      "grad_norm": 0.8882592595246132,
      "learning_rate": 0.0001353326248028298,
      "loss": 3.9844,
      "step": 1674
    },
    {
      "epoch": 0.7681724375143316,
      "grad_norm": 0.8676579413997748,
      "learning_rate": 0.00013482466847710594,
      "loss": 3.991,
      "step": 1675
    },
    {
      "epoch": 0.7686310479247879,
      "grad_norm": 0.8449448281475874,
      "learning_rate": 0.00013431751862679554,
      "loss": 3.7803,
      "step": 1676
    },
    {
      "epoch": 0.7690896583352442,
      "grad_norm": 0.8230013806262261,
      "learning_rate": 0.00013381117637191887,
      "loss": 3.8238,
      "step": 1677
    },
    {
      "epoch": 0.7695482687457005,
      "grad_norm": 0.8254448719292461,
      "learning_rate": 0.00013330564283071293,
      "loss": 3.8477,
      "step": 1678
    },
    {
      "epoch": 0.7700068791561568,
      "grad_norm": 0.8887726928303455,
      "learning_rate": 0.000132800919119629,
      "loss": 3.8711,
      "step": 1679
    },
    {
      "epoch": 0.7704654895666132,
      "grad_norm": 0.8197758687628195,
      "learning_rate": 0.00013229700635332948,
      "loss": 3.878,
      "step": 1680
    },
    {
      "epoch": 0.7709240999770695,
      "grad_norm": 0.8419025121420519,
      "learning_rate": 0.00013179390564468585,
      "loss": 3.893,
      "step": 1681
    },
    {
      "epoch": 0.7713827103875258,
      "grad_norm": 0.8345265499432424,
      "learning_rate": 0.00013129161810477641,
      "loss": 3.9056,
      "step": 1682
    },
    {
      "epoch": 0.7718413207979821,
      "grad_norm": 0.8058721080117095,
      "learning_rate": 0.0001307901448428837,
      "loss": 3.9111,
      "step": 1683
    },
    {
      "epoch": 0.7722999312084384,
      "grad_norm": 0.8318573731981033,
      "learning_rate": 0.0001302894869664916,
      "loss": 3.8214,
      "step": 1684
    },
    {
      "epoch": 0.7727585416188948,
      "grad_norm": 0.7925385111920094,
      "learning_rate": 0.00012978964558128336,
      "loss": 3.8008,
      "step": 1685
    },
    {
      "epoch": 0.773217152029351,
      "grad_norm": 0.8210153520507243,
      "learning_rate": 0.00012929062179113925,
      "loss": 3.8833,
      "step": 1686
    },
    {
      "epoch": 0.7736757624398074,
      "grad_norm": 0.8765438208952743,
      "learning_rate": 0.00012879241669813368,
      "loss": 4.0808,
      "step": 1687
    },
    {
      "epoch": 0.7741343728502637,
      "grad_norm": 0.8044557414165603,
      "learning_rate": 0.00012829503140253295,
      "loss": 3.9188,
      "step": 1688
    },
    {
      "epoch": 0.77459298326072,
      "grad_norm": 0.8434384557841194,
      "learning_rate": 0.0001277984670027933,
      "loss": 3.9608,
      "step": 1689
    },
    {
      "epoch": 0.7750515936711764,
      "grad_norm": 0.8459065588532845,
      "learning_rate": 0.00012730272459555737,
      "loss": 3.8262,
      "step": 1690
    },
    {
      "epoch": 0.7755102040816326,
      "grad_norm": 0.8369521017788505,
      "learning_rate": 0.00012680780527565312,
      "loss": 3.799,
      "step": 1691
    },
    {
      "epoch": 0.775968814492089,
      "grad_norm": 0.8902366935679701,
      "learning_rate": 0.0001263137101360905,
      "loss": 3.8524,
      "step": 1692
    },
    {
      "epoch": 0.7764274249025452,
      "grad_norm": 0.8468439136497087,
      "learning_rate": 0.00012582044026805922,
      "loss": 3.7983,
      "step": 1693
    },
    {
      "epoch": 0.7768860353130016,
      "grad_norm": 0.8251881901021714,
      "learning_rate": 0.00012532799676092627,
      "loss": 3.9906,
      "step": 1694
    },
    {
      "epoch": 0.777344645723458,
      "grad_norm": 0.8738936303887673,
      "learning_rate": 0.00012483638070223414,
      "loss": 3.9058,
      "step": 1695
    },
    {
      "epoch": 0.7778032561339142,
      "grad_norm": 0.8719932056440368,
      "learning_rate": 0.00012434559317769752,
      "loss": 3.9367,
      "step": 1696
    },
    {
      "epoch": 0.7782618665443706,
      "grad_norm": 0.8817683270031609,
      "learning_rate": 0.0001238556352712012,
      "loss": 3.9765,
      "step": 1697
    },
    {
      "epoch": 0.7787204769548268,
      "grad_norm": 0.8822440856217741,
      "learning_rate": 0.00012336650806479827,
      "loss": 3.9281,
      "step": 1698
    },
    {
      "epoch": 0.7791790873652832,
      "grad_norm": 0.866055983819052,
      "learning_rate": 0.00012287821263870708,
      "loss": 4.0236,
      "step": 1699
    },
    {
      "epoch": 0.7796376977757395,
      "grad_norm": 0.8414879622625108,
      "learning_rate": 0.00012239075007130885,
      "loss": 3.8028,
      "step": 1700
    },
    {
      "epoch": 0.7800963081861958,
      "grad_norm": 0.7953764850966831,
      "learning_rate": 0.00012190412143914536,
      "loss": 3.7237,
      "step": 1701
    },
    {
      "epoch": 0.7805549185966522,
      "grad_norm": 0.8130869493379697,
      "learning_rate": 0.0001214183278169172,
      "loss": 4.1512,
      "step": 1702
    },
    {
      "epoch": 0.7810135290071084,
      "grad_norm": 0.7922194266022583,
      "learning_rate": 0.00012093337027748042,
      "loss": 3.8987,
      "step": 1703
    },
    {
      "epoch": 0.7814721394175648,
      "grad_norm": 0.8335002671122194,
      "learning_rate": 0.00012044924989184459,
      "loss": 3.8982,
      "step": 1704
    },
    {
      "epoch": 0.7819307498280211,
      "grad_norm": 0.7994973063028986,
      "learning_rate": 0.0001199659677291709,
      "loss": 4.0349,
      "step": 1705
    },
    {
      "epoch": 0.7823893602384774,
      "grad_norm": 0.8516867149493331,
      "learning_rate": 0.00011948352485676895,
      "loss": 3.9016,
      "step": 1706
    },
    {
      "epoch": 0.7828479706489337,
      "grad_norm": 0.8171214727888845,
      "learning_rate": 0.00011900192234009477,
      "loss": 3.895,
      "step": 1707
    },
    {
      "epoch": 0.78330658105939,
      "grad_norm": 0.8100408068676448,
      "learning_rate": 0.00011852116124274875,
      "loss": 3.9157,
      "step": 1708
    },
    {
      "epoch": 0.7837651914698464,
      "grad_norm": 0.8727392632407555,
      "learning_rate": 0.00011804124262647314,
      "loss": 3.8146,
      "step": 1709
    },
    {
      "epoch": 0.7842238018803027,
      "grad_norm": 0.8284675012458365,
      "learning_rate": 0.00011756216755114929,
      "loss": 4.1124,
      "step": 1710
    },
    {
      "epoch": 0.784682412290759,
      "grad_norm": 0.844695489786009,
      "learning_rate": 0.00011708393707479548,
      "loss": 3.8768,
      "step": 1711
    },
    {
      "epoch": 0.7851410227012153,
      "grad_norm": 0.827597383672807,
      "learning_rate": 0.00011660655225356531,
      "loss": 3.907,
      "step": 1712
    },
    {
      "epoch": 0.7855996331116716,
      "grad_norm": 0.8719806283284544,
      "learning_rate": 0.0001161300141417444,
      "loss": 3.9471,
      "step": 1713
    },
    {
      "epoch": 0.7860582435221279,
      "grad_norm": 0.8200657067422078,
      "learning_rate": 0.00011565432379174823,
      "loss": 3.8966,
      "step": 1714
    },
    {
      "epoch": 0.7865168539325843,
      "grad_norm": 0.8558635032224313,
      "learning_rate": 0.00011517948225412056,
      "loss": 3.9544,
      "step": 1715
    },
    {
      "epoch": 0.7869754643430406,
      "grad_norm": 0.8657795644775393,
      "learning_rate": 0.00011470549057753032,
      "loss": 3.9665,
      "step": 1716
    },
    {
      "epoch": 0.7874340747534969,
      "grad_norm": 0.8154290279225862,
      "learning_rate": 0.00011423234980876957,
      "loss": 3.9197,
      "step": 1717
    },
    {
      "epoch": 0.7878926851639533,
      "grad_norm": 0.8454064664906694,
      "learning_rate": 0.00011376006099275099,
      "loss": 3.9807,
      "step": 1718
    },
    {
      "epoch": 0.7883512955744095,
      "grad_norm": 0.8347746906247907,
      "learning_rate": 0.00011328862517250609,
      "loss": 3.948,
      "step": 1719
    },
    {
      "epoch": 0.7888099059848659,
      "grad_norm": 0.8922285328145984,
      "learning_rate": 0.00011281804338918239,
      "loss": 3.833,
      "step": 1720
    },
    {
      "epoch": 0.7892685163953221,
      "grad_norm": 0.8394554646163404,
      "learning_rate": 0.00011234831668204115,
      "loss": 3.9765,
      "step": 1721
    },
    {
      "epoch": 0.7897271268057785,
      "grad_norm": 0.8305334284232776,
      "learning_rate": 0.00011187944608845569,
      "loss": 3.9569,
      "step": 1722
    },
    {
      "epoch": 0.7901857372162349,
      "grad_norm": 0.8582804691573132,
      "learning_rate": 0.00011141143264390801,
      "loss": 3.9465,
      "step": 1723
    },
    {
      "epoch": 0.7906443476266911,
      "grad_norm": 0.8863205375300387,
      "learning_rate": 0.0001109442773819877,
      "loss": 3.8259,
      "step": 1724
    },
    {
      "epoch": 0.7911029580371475,
      "grad_norm": 0.8466396231560457,
      "learning_rate": 0.0001104779813343889,
      "loss": 3.9425,
      "step": 1725
    },
    {
      "epoch": 0.7915615684476037,
      "grad_norm": 0.851174262497763,
      "learning_rate": 0.00011001254553090812,
      "loss": 3.982,
      "step": 1726
    },
    {
      "epoch": 0.7920201788580601,
      "grad_norm": 0.8369701026217211,
      "learning_rate": 0.00010954797099944186,
      "loss": 3.8355,
      "step": 1727
    },
    {
      "epoch": 0.7924787892685164,
      "grad_norm": 0.8135087560989958,
      "learning_rate": 0.0001090842587659851,
      "loss": 4.0205,
      "step": 1728
    },
    {
      "epoch": 0.7929373996789727,
      "grad_norm": 0.841652868647545,
      "learning_rate": 0.00010862140985462804,
      "loss": 3.7811,
      "step": 1729
    },
    {
      "epoch": 0.7933960100894291,
      "grad_norm": 0.8986128141043016,
      "learning_rate": 0.00010815942528755418,
      "loss": 3.9595,
      "step": 1730
    },
    {
      "epoch": 0.7938546204998853,
      "grad_norm": 0.8618771990874391,
      "learning_rate": 0.00010769830608503844,
      "loss": 3.9472,
      "step": 1731
    },
    {
      "epoch": 0.7943132309103417,
      "grad_norm": 0.8724876437936445,
      "learning_rate": 0.00010723805326544473,
      "loss": 3.9721,
      "step": 1732
    },
    {
      "epoch": 0.794771841320798,
      "grad_norm": 0.8362881193098228,
      "learning_rate": 0.00010677866784522316,
      "loss": 3.821,
      "step": 1733
    },
    {
      "epoch": 0.7952304517312543,
      "grad_norm": 0.8621972650036842,
      "learning_rate": 0.00010632015083890839,
      "loss": 4.1712,
      "step": 1734
    },
    {
      "epoch": 0.7956890621417106,
      "grad_norm": 0.834207874195674,
      "learning_rate": 0.00010586250325911745,
      "loss": 3.9749,
      "step": 1735
    },
    {
      "epoch": 0.7961476725521669,
      "grad_norm": 0.8482579439286547,
      "learning_rate": 0.00010540572611654697,
      "loss": 3.8552,
      "step": 1736
    },
    {
      "epoch": 0.7966062829626233,
      "grad_norm": 0.8421292016870463,
      "learning_rate": 0.00010494982041997126,
      "loss": 3.906,
      "step": 1737
    },
    {
      "epoch": 0.7970648933730796,
      "grad_norm": 0.881060132284894,
      "learning_rate": 0.0001044947871762405,
      "loss": 3.9814,
      "step": 1738
    },
    {
      "epoch": 0.7975235037835359,
      "grad_norm": 0.7970088853280194,
      "learning_rate": 0.00010404062739027753,
      "loss": 4.1063,
      "step": 1739
    },
    {
      "epoch": 0.7979821141939922,
      "grad_norm": 0.8248941800339114,
      "learning_rate": 0.00010358734206507641,
      "loss": 3.8529,
      "step": 1740
    },
    {
      "epoch": 0.7984407246044485,
      "grad_norm": 0.8849692668198678,
      "learning_rate": 0.00010313493220170017,
      "loss": 3.775,
      "step": 1741
    },
    {
      "epoch": 0.7988993350149048,
      "grad_norm": 0.7977066311117996,
      "learning_rate": 0.00010268339879927836,
      "loss": 3.9092,
      "step": 1742
    },
    {
      "epoch": 0.7993579454253612,
      "grad_norm": 0.8418730296960252,
      "learning_rate": 0.00010223274285500466,
      "loss": 4.0561,
      "step": 1743
    },
    {
      "epoch": 0.7998165558358175,
      "grad_norm": 0.8439359818675287,
      "learning_rate": 0.00010178296536413495,
      "loss": 3.8781,
      "step": 1744
    },
    {
      "epoch": 0.8002751662462738,
      "grad_norm": 0.8460980386946729,
      "learning_rate": 0.00010133406731998546,
      "loss": 3.8526,
      "step": 1745
    },
    {
      "epoch": 0.8007337766567301,
      "grad_norm": 0.835275308644109,
      "learning_rate": 0.00010088604971392979,
      "loss": 3.9865,
      "step": 1746
    },
    {
      "epoch": 0.8011923870671864,
      "grad_norm": 0.83331230709501,
      "learning_rate": 0.0001004389135353972,
      "loss": 3.941,
      "step": 1747
    },
    {
      "epoch": 0.8016509974776428,
      "grad_norm": 0.8426246906861796,
      "learning_rate": 9.999265977187049e-05,
      "loss": 3.9494,
      "step": 1748
    },
    {
      "epoch": 0.802109607888099,
      "grad_norm": 0.8082000273732284,
      "learning_rate": 9.95472894088838e-05,
      "loss": 3.8124,
      "step": 1749
    },
    {
      "epoch": 0.8025682182985554,
      "grad_norm": 0.807970401620772,
      "learning_rate": 9.910280343001993e-05,
      "loss": 3.9308,
      "step": 1750
    },
    {
      "epoch": 0.8030268287090117,
      "grad_norm": 0.842310656562353,
      "learning_rate": 9.865920281690866e-05,
      "loss": 3.8806,
      "step": 1751
    },
    {
      "epoch": 0.803485439119468,
      "grad_norm": 0.8477060541231161,
      "learning_rate": 9.821648854922482e-05,
      "loss": 3.9084,
      "step": 1752
    },
    {
      "epoch": 0.8039440495299244,
      "grad_norm": 0.8306231392440592,
      "learning_rate": 9.77746616046854e-05,
      "loss": 3.9358,
      "step": 1753
    },
    {
      "epoch": 0.8044026599403806,
      "grad_norm": 0.8531699899279188,
      "learning_rate": 9.733372295904774e-05,
      "loss": 3.9493,
      "step": 1754
    },
    {
      "epoch": 0.804861270350837,
      "grad_norm": 0.8733028078336182,
      "learning_rate": 9.68936735861079e-05,
      "loss": 4.1274,
      "step": 1755
    },
    {
      "epoch": 0.8053198807612932,
      "grad_norm": 0.9157566887014461,
      "learning_rate": 9.645451445769737e-05,
      "loss": 3.8535,
      "step": 1756
    },
    {
      "epoch": 0.8057784911717496,
      "grad_norm": 0.8319949334685524,
      "learning_rate": 9.601624654368196e-05,
      "loss": 3.8935,
      "step": 1757
    },
    {
      "epoch": 0.806237101582206,
      "grad_norm": 0.8648273331850923,
      "learning_rate": 9.557887081195938e-05,
      "loss": 4.0654,
      "step": 1758
    },
    {
      "epoch": 0.8066957119926622,
      "grad_norm": 0.8967770688740806,
      "learning_rate": 9.514238822845667e-05,
      "loss": 3.7561,
      "step": 1759
    },
    {
      "epoch": 0.8071543224031186,
      "grad_norm": 0.8459313576924138,
      "learning_rate": 9.470679975712837e-05,
      "loss": 3.9122,
      "step": 1760
    },
    {
      "epoch": 0.8076129328135748,
      "grad_norm": 0.8417910309586496,
      "learning_rate": 9.427210635995481e-05,
      "loss": 3.947,
      "step": 1761
    },
    {
      "epoch": 0.8080715432240312,
      "grad_norm": 0.8552258464554909,
      "learning_rate": 9.383830899693923e-05,
      "loss": 3.7783,
      "step": 1762
    },
    {
      "epoch": 0.8085301536344875,
      "grad_norm": 0.8457719222050728,
      "learning_rate": 9.340540862610591e-05,
      "loss": 3.9932,
      "step": 1763
    },
    {
      "epoch": 0.8089887640449438,
      "grad_norm": 0.9230657339086558,
      "learning_rate": 9.297340620349854e-05,
      "loss": 3.8687,
      "step": 1764
    },
    {
      "epoch": 0.8094473744554002,
      "grad_norm": 0.8441957393537736,
      "learning_rate": 9.25423026831777e-05,
      "loss": 3.8851,
      "step": 1765
    },
    {
      "epoch": 0.8099059848658564,
      "grad_norm": 0.8693738394106627,
      "learning_rate": 9.211209901721846e-05,
      "loss": 3.7489,
      "step": 1766
    },
    {
      "epoch": 0.8103645952763128,
      "grad_norm": 0.8289462810183715,
      "learning_rate": 9.168279615570863e-05,
      "loss": 3.9167,
      "step": 1767
    },
    {
      "epoch": 0.8108232056867691,
      "grad_norm": 0.8022043907461626,
      "learning_rate": 9.125439504674699e-05,
      "loss": 3.8559,
      "step": 1768
    },
    {
      "epoch": 0.8112818160972254,
      "grad_norm": 0.8215299413539758,
      "learning_rate": 9.082689663644057e-05,
      "loss": 3.7653,
      "step": 1769
    },
    {
      "epoch": 0.8117404265076817,
      "grad_norm": 0.84344366205893,
      "learning_rate": 9.040030186890264e-05,
      "loss": 3.9718,
      "step": 1770
    },
    {
      "epoch": 0.812199036918138,
      "grad_norm": 0.8641604045350689,
      "learning_rate": 8.997461168625138e-05,
      "loss": 3.9964,
      "step": 1771
    },
    {
      "epoch": 0.8126576473285944,
      "grad_norm": 0.8174436072616433,
      "learning_rate": 8.954982702860664e-05,
      "loss": 3.8093,
      "step": 1772
    },
    {
      "epoch": 0.8131162577390507,
      "grad_norm": 0.8603981559684999,
      "learning_rate": 8.912594883408865e-05,
      "loss": 3.8825,
      "step": 1773
    },
    {
      "epoch": 0.813574868149507,
      "grad_norm": 0.866537502675343,
      "learning_rate": 8.870297803881589e-05,
      "loss": 3.9543,
      "step": 1774
    },
    {
      "epoch": 0.8140334785599633,
      "grad_norm": 0.8780803590908147,
      "learning_rate": 8.828091557690287e-05,
      "loss": 3.88,
      "step": 1775
    },
    {
      "epoch": 0.8144920889704196,
      "grad_norm": 0.7971785296855463,
      "learning_rate": 8.785976238045801e-05,
      "loss": 3.7404,
      "step": 1776
    },
    {
      "epoch": 0.8149506993808759,
      "grad_norm": 0.8309417509471111,
      "learning_rate": 8.743951937958144e-05,
      "loss": 3.8717,
      "step": 1777
    },
    {
      "epoch": 0.8154093097913323,
      "grad_norm": 0.859808105548437,
      "learning_rate": 8.702018750236357e-05,
      "loss": 3.8542,
      "step": 1778
    },
    {
      "epoch": 0.8158679202017886,
      "grad_norm": 0.8400733215368001,
      "learning_rate": 8.660176767488237e-05,
      "loss": 3.9305,
      "step": 1779
    },
    {
      "epoch": 0.8163265306122449,
      "grad_norm": 0.8273572779149242,
      "learning_rate": 8.618426082120146e-05,
      "loss": 3.7909,
      "step": 1780
    },
    {
      "epoch": 0.8167851410227012,
      "grad_norm": 0.8383257681881194,
      "learning_rate": 8.576766786336854e-05,
      "loss": 3.8608,
      "step": 1781
    },
    {
      "epoch": 0.8172437514331575,
      "grad_norm": 0.8555482485598784,
      "learning_rate": 8.535198972141294e-05,
      "loss": 3.9679,
      "step": 1782
    },
    {
      "epoch": 0.8177023618436139,
      "grad_norm": 0.8475032700763724,
      "learning_rate": 8.493722731334347e-05,
      "loss": 4.0694,
      "step": 1783
    },
    {
      "epoch": 0.8181609722540701,
      "grad_norm": 0.8355107385111606,
      "learning_rate": 8.452338155514644e-05,
      "loss": 3.9748,
      "step": 1784
    },
    {
      "epoch": 0.8186195826645265,
      "grad_norm": 0.8240344352196229,
      "learning_rate": 8.411045336078426e-05,
      "loss": 3.9936,
      "step": 1785
    },
    {
      "epoch": 0.8190781930749828,
      "grad_norm": 0.8227450862717396,
      "learning_rate": 8.369844364219264e-05,
      "loss": 3.8226,
      "step": 1786
    },
    {
      "epoch": 0.8195368034854391,
      "grad_norm": 0.8811044893816955,
      "learning_rate": 8.328735330927873e-05,
      "loss": 3.9018,
      "step": 1787
    },
    {
      "epoch": 0.8199954138958955,
      "grad_norm": 0.8619485691626624,
      "learning_rate": 8.287718326991961e-05,
      "loss": 4.074,
      "step": 1788
    },
    {
      "epoch": 0.8204540243063517,
      "grad_norm": 0.8270537351464764,
      "learning_rate": 8.246793442995954e-05,
      "loss": 3.9004,
      "step": 1789
    },
    {
      "epoch": 0.8209126347168081,
      "grad_norm": 0.852160672421944,
      "learning_rate": 8.205960769320875e-05,
      "loss": 3.8249,
      "step": 1790
    },
    {
      "epoch": 0.8213712451272643,
      "grad_norm": 0.8782636794277223,
      "learning_rate": 8.165220396144085e-05,
      "loss": 4.005,
      "step": 1791
    },
    {
      "epoch": 0.8218298555377207,
      "grad_norm": 0.8487186898742015,
      "learning_rate": 8.12457241343909e-05,
      "loss": 3.7834,
      "step": 1792
    },
    {
      "epoch": 0.8222884659481771,
      "grad_norm": 0.84006678092706,
      "learning_rate": 8.084016910975367e-05,
      "loss": 3.8481,
      "step": 1793
    },
    {
      "epoch": 0.8227470763586333,
      "grad_norm": 0.834012657552531,
      "learning_rate": 8.043553978318169e-05,
      "loss": 3.9124,
      "step": 1794
    },
    {
      "epoch": 0.8232056867690897,
      "grad_norm": 0.8763990555514874,
      "learning_rate": 8.003183704828281e-05,
      "loss": 3.942,
      "step": 1795
    },
    {
      "epoch": 0.8236642971795459,
      "grad_norm": 0.8195515134195439,
      "learning_rate": 7.962906179661872e-05,
      "loss": 3.8924,
      "step": 1796
    },
    {
      "epoch": 0.8241229075900023,
      "grad_norm": 0.8143256262597016,
      "learning_rate": 7.922721491770296e-05,
      "loss": 3.9513,
      "step": 1797
    },
    {
      "epoch": 0.8245815180004586,
      "grad_norm": 0.8547146425871314,
      "learning_rate": 7.882629729899832e-05,
      "loss": 3.6645,
      "step": 1798
    },
    {
      "epoch": 0.8250401284109149,
      "grad_norm": 0.824561903862021,
      "learning_rate": 7.842630982591598e-05,
      "loss": 3.8257,
      "step": 1799
    },
    {
      "epoch": 0.8254987388213713,
      "grad_norm": 0.8756605660678058,
      "learning_rate": 7.802725338181232e-05,
      "loss": 3.9646,
      "step": 1800
    },
    {
      "epoch": 0.8259573492318275,
      "grad_norm": 0.8995976998366219,
      "learning_rate": 7.762912884798812e-05,
      "loss": 3.9301,
      "step": 1801
    },
    {
      "epoch": 0.8264159596422839,
      "grad_norm": 0.8372637561748941,
      "learning_rate": 7.723193710368564e-05,
      "loss": 3.8521,
      "step": 1802
    },
    {
      "epoch": 0.8268745700527402,
      "grad_norm": 0.8677276717637402,
      "learning_rate": 7.683567902608729e-05,
      "loss": 3.9825,
      "step": 1803
    },
    {
      "epoch": 0.8273331804631965,
      "grad_norm": 0.85307266800581,
      "learning_rate": 7.644035549031364e-05,
      "loss": 3.7389,
      "step": 1804
    },
    {
      "epoch": 0.8277917908736528,
      "grad_norm": 0.8604915263652924,
      "learning_rate": 7.604596736942115e-05,
      "loss": 3.8544,
      "step": 1805
    },
    {
      "epoch": 0.8282504012841091,
      "grad_norm": 0.8520348503678395,
      "learning_rate": 7.56525155344004e-05,
      "loss": 3.7941,
      "step": 1806
    },
    {
      "epoch": 0.8287090116945655,
      "grad_norm": 0.8574357971199452,
      "learning_rate": 7.52600008541745e-05,
      "loss": 3.9739,
      "step": 1807
    },
    {
      "epoch": 0.8291676221050218,
      "grad_norm": 0.9066844794807549,
      "learning_rate": 7.486842419559681e-05,
      "loss": 3.9616,
      "step": 1808
    },
    {
      "epoch": 0.8296262325154781,
      "grad_norm": 0.8993705996326355,
      "learning_rate": 7.447778642344898e-05,
      "loss": 3.9463,
      "step": 1809
    },
    {
      "epoch": 0.8300848429259344,
      "grad_norm": 0.8373462630419455,
      "learning_rate": 7.408808840043912e-05,
      "loss": 3.8919,
      "step": 1810
    },
    {
      "epoch": 0.8305434533363908,
      "grad_norm": 0.8455405687497736,
      "learning_rate": 7.369933098720021e-05,
      "loss": 3.8471,
      "step": 1811
    },
    {
      "epoch": 0.831002063746847,
      "grad_norm": 0.7760737749379033,
      "learning_rate": 7.331151504228767e-05,
      "loss": 3.902,
      "step": 1812
    },
    {
      "epoch": 0.8314606741573034,
      "grad_norm": 0.7988895920746409,
      "learning_rate": 7.292464142217775e-05,
      "loss": 3.7579,
      "step": 1813
    },
    {
      "epoch": 0.8319192845677597,
      "grad_norm": 0.850902418566754,
      "learning_rate": 7.25387109812658e-05,
      "loss": 3.7864,
      "step": 1814
    },
    {
      "epoch": 0.832377894978216,
      "grad_norm": 0.8586888326595268,
      "learning_rate": 7.215372457186415e-05,
      "loss": 3.8944,
      "step": 1815
    },
    {
      "epoch": 0.8328365053886724,
      "grad_norm": 0.8207283095817377,
      "learning_rate": 7.176968304420007e-05,
      "loss": 3.8461,
      "step": 1816
    },
    {
      "epoch": 0.8332951157991286,
      "grad_norm": 0.8003365791057159,
      "learning_rate": 7.138658724641417e-05,
      "loss": 3.8852,
      "step": 1817
    },
    {
      "epoch": 0.833753726209585,
      "grad_norm": 0.8381840691092338,
      "learning_rate": 7.10044380245587e-05,
      "loss": 3.7899,
      "step": 1818
    },
    {
      "epoch": 0.8342123366200412,
      "grad_norm": 0.861042605217842,
      "learning_rate": 7.062323622259515e-05,
      "loss": 3.811,
      "step": 1819
    },
    {
      "epoch": 0.8346709470304976,
      "grad_norm": 0.8699638646907153,
      "learning_rate": 7.024298268239265e-05,
      "loss": 3.9695,
      "step": 1820
    },
    {
      "epoch": 0.835129557440954,
      "grad_norm": 0.8230350311239117,
      "learning_rate": 6.986367824372647e-05,
      "loss": 3.8723,
      "step": 1821
    },
    {
      "epoch": 0.8355881678514102,
      "grad_norm": 0.8491385750399679,
      "learning_rate": 6.948532374427541e-05,
      "loss": 4.0232,
      "step": 1822
    },
    {
      "epoch": 0.8360467782618666,
      "grad_norm": 0.8170528503565758,
      "learning_rate": 6.910792001962063e-05,
      "loss": 3.8805,
      "step": 1823
    },
    {
      "epoch": 0.8365053886723228,
      "grad_norm": 0.8596238014552037,
      "learning_rate": 6.873146790324358e-05,
      "loss": 3.9325,
      "step": 1824
    },
    {
      "epoch": 0.8369639990827792,
      "grad_norm": 0.8306248065167525,
      "learning_rate": 6.83559682265239e-05,
      "loss": 3.835,
      "step": 1825
    },
    {
      "epoch": 0.8374226094932355,
      "grad_norm": 0.8896693746445915,
      "learning_rate": 6.798142181873784e-05,
      "loss": 3.9551,
      "step": 1826
    },
    {
      "epoch": 0.8378812199036918,
      "grad_norm": 0.8429261704046282,
      "learning_rate": 6.760782950705662e-05,
      "loss": 3.8763,
      "step": 1827
    },
    {
      "epoch": 0.8383398303141482,
      "grad_norm": 0.8171394993231066,
      "learning_rate": 6.723519211654422e-05,
      "loss": 3.915,
      "step": 1828
    },
    {
      "epoch": 0.8387984407246044,
      "grad_norm": 0.8445435728202753,
      "learning_rate": 6.686351047015554e-05,
      "loss": 3.7587,
      "step": 1829
    },
    {
      "epoch": 0.8392570511350608,
      "grad_norm": 0.7925432216198227,
      "learning_rate": 6.649278538873515e-05,
      "loss": 3.897,
      "step": 1830
    },
    {
      "epoch": 0.8397156615455171,
      "grad_norm": 0.8430749454969462,
      "learning_rate": 6.612301769101465e-05,
      "loss": 4.0847,
      "step": 1831
    },
    {
      "epoch": 0.8401742719559734,
      "grad_norm": 0.8936762612399684,
      "learning_rate": 6.575420819361177e-05,
      "loss": 3.8671,
      "step": 1832
    },
    {
      "epoch": 0.8406328823664297,
      "grad_norm": 0.9008498768665099,
      "learning_rate": 6.538635771102757e-05,
      "loss": 4.0745,
      "step": 1833
    },
    {
      "epoch": 0.841091492776886,
      "grad_norm": 0.8153048376911851,
      "learning_rate": 6.501946705564566e-05,
      "loss": 3.7754,
      "step": 1834
    },
    {
      "epoch": 0.8415501031873424,
      "grad_norm": 0.8158218531494716,
      "learning_rate": 6.465353703772959e-05,
      "loss": 3.8613,
      "step": 1835
    },
    {
      "epoch": 0.8420087135977987,
      "grad_norm": 0.8494068370212671,
      "learning_rate": 6.428856846542136e-05,
      "loss": 3.9566,
      "step": 1836
    },
    {
      "epoch": 0.842467324008255,
      "grad_norm": 0.8106142929488781,
      "learning_rate": 6.392456214473996e-05,
      "loss": 3.7995,
      "step": 1837
    },
    {
      "epoch": 0.8429259344187113,
      "grad_norm": 0.812812449368362,
      "learning_rate": 6.3561518879579e-05,
      "loss": 3.7269,
      "step": 1838
    },
    {
      "epoch": 0.8433845448291676,
      "grad_norm": 0.871705370011186,
      "learning_rate": 6.31994394717052e-05,
      "loss": 3.9411,
      "step": 1839
    },
    {
      "epoch": 0.8438431552396239,
      "grad_norm": 0.8589441045508612,
      "learning_rate": 6.283832472075685e-05,
      "loss": 3.9037,
      "step": 1840
    },
    {
      "epoch": 0.8443017656500803,
      "grad_norm": 0.8386467626291731,
      "learning_rate": 6.247817542424178e-05,
      "loss": 3.8657,
      "step": 1841
    },
    {
      "epoch": 0.8447603760605366,
      "grad_norm": 0.8425052899721118,
      "learning_rate": 6.211899237753559e-05,
      "loss": 3.8629,
      "step": 1842
    },
    {
      "epoch": 0.8452189864709929,
      "grad_norm": 0.880587883036184,
      "learning_rate": 6.176077637387984e-05,
      "loss": 3.8204,
      "step": 1843
    },
    {
      "epoch": 0.8456775968814492,
      "grad_norm": 0.8315715846470759,
      "learning_rate": 6.140352820438066e-05,
      "loss": 3.8118,
      "step": 1844
    },
    {
      "epoch": 0.8461362072919055,
      "grad_norm": 0.8439652112726204,
      "learning_rate": 6.104724865800665e-05,
      "loss": 3.9237,
      "step": 1845
    },
    {
      "epoch": 0.8465948177023619,
      "grad_norm": 0.876326829568899,
      "learning_rate": 6.069193852158711e-05,
      "loss": 3.8355,
      "step": 1846
    },
    {
      "epoch": 0.8470534281128181,
      "grad_norm": 0.8855325131036269,
      "learning_rate": 6.0337598579810584e-05,
      "loss": 3.7428,
      "step": 1847
    },
    {
      "epoch": 0.8475120385232745,
      "grad_norm": 0.8488632600945438,
      "learning_rate": 5.9984229615223096e-05,
      "loss": 3.9061,
      "step": 1848
    },
    {
      "epoch": 0.8479706489337308,
      "grad_norm": 0.8435751120573699,
      "learning_rate": 5.963183240822606e-05,
      "loss": 3.8267,
      "step": 1849
    },
    {
      "epoch": 0.8484292593441871,
      "grad_norm": 0.9159999792190291,
      "learning_rate": 5.9280407737074825e-05,
      "loss": 4.0047,
      "step": 1850
    },
    {
      "epoch": 0.8488878697546435,
      "grad_norm": 0.837087128704532,
      "learning_rate": 5.8929956377877125e-05,
      "loss": 3.8373,
      "step": 1851
    },
    {
      "epoch": 0.8493464801650997,
      "grad_norm": 0.8195742529354402,
      "learning_rate": 5.8580479104591075e-05,
      "loss": 3.9239,
      "step": 1852
    },
    {
      "epoch": 0.8498050905755561,
      "grad_norm": 0.8677408293822915,
      "learning_rate": 5.823197668902341e-05,
      "loss": 3.7788,
      "step": 1853
    },
    {
      "epoch": 0.8502637009860123,
      "grad_norm": 0.8445396671374166,
      "learning_rate": 5.78844499008282e-05,
      "loss": 3.8956,
      "step": 1854
    },
    {
      "epoch": 0.8507223113964687,
      "grad_norm": 0.844910622262557,
      "learning_rate": 5.753789950750454e-05,
      "loss": 3.8715,
      "step": 1855
    },
    {
      "epoch": 0.8511809218069251,
      "grad_norm": 0.8413426516320491,
      "learning_rate": 5.719232627439558e-05,
      "loss": 3.6966,
      "step": 1856
    },
    {
      "epoch": 0.8516395322173813,
      "grad_norm": 0.79538232504088,
      "learning_rate": 5.6847730964686315e-05,
      "loss": 4.01,
      "step": 1857
    },
    {
      "epoch": 0.8520981426278377,
      "grad_norm": 0.8332165716993561,
      "learning_rate": 5.650411433940189e-05,
      "loss": 3.7853,
      "step": 1858
    },
    {
      "epoch": 0.8525567530382939,
      "grad_norm": 0.803955606077538,
      "learning_rate": 5.61614771574061e-05,
      "loss": 3.7598,
      "step": 1859
    },
    {
      "epoch": 0.8530153634487503,
      "grad_norm": 0.9150136595001858,
      "learning_rate": 5.581982017539988e-05,
      "loss": 3.8524,
      "step": 1860
    },
    {
      "epoch": 0.8534739738592066,
      "grad_norm": 0.8601215281806416,
      "learning_rate": 5.5479144147919216e-05,
      "loss": 3.7951,
      "step": 1861
    },
    {
      "epoch": 0.8539325842696629,
      "grad_norm": 0.8779376709058448,
      "learning_rate": 5.51394498273336e-05,
      "loss": 3.9927,
      "step": 1862
    },
    {
      "epoch": 0.8543911946801193,
      "grad_norm": 0.837093299859567,
      "learning_rate": 5.480073796384494e-05,
      "loss": 3.9932,
      "step": 1863
    },
    {
      "epoch": 0.8548498050905755,
      "grad_norm": 0.796144369077145,
      "learning_rate": 5.446300930548492e-05,
      "loss": 3.8208,
      "step": 1864
    },
    {
      "epoch": 0.8553084155010319,
      "grad_norm": 0.8172458510577078,
      "learning_rate": 5.412626459811415e-05,
      "loss": 3.8316,
      "step": 1865
    },
    {
      "epoch": 0.8557670259114882,
      "grad_norm": 0.8549762417456219,
      "learning_rate": 5.3790504585419954e-05,
      "loss": 3.6776,
      "step": 1866
    },
    {
      "epoch": 0.8562256363219445,
      "grad_norm": 0.8560181588837545,
      "learning_rate": 5.345573000891541e-05,
      "loss": 3.8316,
      "step": 1867
    },
    {
      "epoch": 0.8566842467324008,
      "grad_norm": 0.847419679113905,
      "learning_rate": 5.312194160793693e-05,
      "loss": 3.9634,
      "step": 1868
    },
    {
      "epoch": 0.8571428571428571,
      "grad_norm": 0.8579339980480146,
      "learning_rate": 5.278914011964303e-05,
      "loss": 3.8472,
      "step": 1869
    },
    {
      "epoch": 0.8576014675533135,
      "grad_norm": 0.8569823555508638,
      "learning_rate": 5.2457326279013006e-05,
      "loss": 4.1016,
      "step": 1870
    },
    {
      "epoch": 0.8580600779637698,
      "grad_norm": 0.819089365641458,
      "learning_rate": 5.2126500818844514e-05,
      "loss": 3.9831,
      "step": 1871
    },
    {
      "epoch": 0.8585186883742261,
      "grad_norm": 0.8233317674881645,
      "learning_rate": 5.1796664469752566e-05,
      "loss": 3.8431,
      "step": 1872
    },
    {
      "epoch": 0.8589772987846824,
      "grad_norm": 0.8249443363810788,
      "learning_rate": 5.1467817960167975e-05,
      "loss": 3.8892,
      "step": 1873
    },
    {
      "epoch": 0.8594359091951387,
      "grad_norm": 0.816707743639336,
      "learning_rate": 5.113996201633536e-05,
      "loss": 3.8067,
      "step": 1874
    },
    {
      "epoch": 0.859894519605595,
      "grad_norm": 0.8005734005281675,
      "learning_rate": 5.0813097362311765e-05,
      "loss": 3.7178,
      "step": 1875
    },
    {
      "epoch": 0.8603531300160514,
      "grad_norm": 0.8250728688448736,
      "learning_rate": 5.048722471996475e-05,
      "loss": 3.9297,
      "step": 1876
    },
    {
      "epoch": 0.8608117404265077,
      "grad_norm": 0.8266816587068168,
      "learning_rate": 5.016234480897158e-05,
      "loss": 3.7404,
      "step": 1877
    },
    {
      "epoch": 0.861270350836964,
      "grad_norm": 0.8263714424745943,
      "learning_rate": 4.9838458346816664e-05,
      "loss": 3.868,
      "step": 1878
    },
    {
      "epoch": 0.8617289612474203,
      "grad_norm": 0.8806647336250433,
      "learning_rate": 4.9515566048790485e-05,
      "loss": 3.8399,
      "step": 1879
    },
    {
      "epoch": 0.8621875716578766,
      "grad_norm": 0.8091174252031158,
      "learning_rate": 4.9193668627988074e-05,
      "loss": 3.9706,
      "step": 1880
    },
    {
      "epoch": 0.862646182068333,
      "grad_norm": 0.8496102489269629,
      "learning_rate": 4.887276679530744e-05,
      "loss": 3.9383,
      "step": 1881
    },
    {
      "epoch": 0.8631047924787892,
      "grad_norm": 0.8467565642393998,
      "learning_rate": 4.855286125944752e-05,
      "loss": 3.9118,
      "step": 1882
    },
    {
      "epoch": 0.8635634028892456,
      "grad_norm": 0.8507539633004192,
      "learning_rate": 4.8233952726907224e-05,
      "loss": 3.8775,
      "step": 1883
    },
    {
      "epoch": 0.8640220132997019,
      "grad_norm": 0.8469815455849562,
      "learning_rate": 4.7916041901983565e-05,
      "loss": 3.8128,
      "step": 1884
    },
    {
      "epoch": 0.8644806237101582,
      "grad_norm": 0.8939498254392466,
      "learning_rate": 4.7599129486770145e-05,
      "loss": 3.8317,
      "step": 1885
    },
    {
      "epoch": 0.8649392341206146,
      "grad_norm": 0.8375374245755477,
      "learning_rate": 4.728321618115555e-05,
      "loss": 3.7945,
      "step": 1886
    },
    {
      "epoch": 0.8653978445310708,
      "grad_norm": 0.8361945308610618,
      "learning_rate": 4.696830268282204e-05,
      "loss": 3.9443,
      "step": 1887
    },
    {
      "epoch": 0.8658564549415272,
      "grad_norm": 0.8477103761562675,
      "learning_rate": 4.665438968724361e-05,
      "loss": 3.8566,
      "step": 1888
    },
    {
      "epoch": 0.8663150653519834,
      "grad_norm": 0.8455716007718013,
      "learning_rate": 4.634147788768489e-05,
      "loss": 3.9781,
      "step": 1889
    },
    {
      "epoch": 0.8667736757624398,
      "grad_norm": 0.8226448077305428,
      "learning_rate": 4.6029567975199414e-05,
      "loss": 3.8851,
      "step": 1890
    },
    {
      "epoch": 0.8672322861728962,
      "grad_norm": 0.840241611853895,
      "learning_rate": 4.571866063862795e-05,
      "loss": 4.0457,
      "step": 1891
    },
    {
      "epoch": 0.8676908965833524,
      "grad_norm": 0.826923039305308,
      "learning_rate": 4.540875656459703e-05,
      "loss": 3.8549,
      "step": 1892
    },
    {
      "epoch": 0.8681495069938088,
      "grad_norm": 0.8443074810267847,
      "learning_rate": 4.509985643751785e-05,
      "loss": 3.8899,
      "step": 1893
    },
    {
      "epoch": 0.868608117404265,
      "grad_norm": 0.8380703897196387,
      "learning_rate": 4.479196093958421e-05,
      "loss": 4.0008,
      "step": 1894
    },
    {
      "epoch": 0.8690667278147214,
      "grad_norm": 0.8303987217725618,
      "learning_rate": 4.4485070750771187e-05,
      "loss": 3.8189,
      "step": 1895
    },
    {
      "epoch": 0.8695253382251777,
      "grad_norm": 0.8740313183306649,
      "learning_rate": 4.417918654883363e-05,
      "loss": 3.9076,
      "step": 1896
    },
    {
      "epoch": 0.869983948635634,
      "grad_norm": 0.8650749696921894,
      "learning_rate": 4.3874309009305e-05,
      "loss": 4.0145,
      "step": 1897
    },
    {
      "epoch": 0.8704425590460904,
      "grad_norm": 0.8470981989666586,
      "learning_rate": 4.357043880549538e-05,
      "loss": 4.0062,
      "step": 1898
    },
    {
      "epoch": 0.8709011694565466,
      "grad_norm": 0.8258317854444175,
      "learning_rate": 4.326757660849012e-05,
      "loss": 3.8168,
      "step": 1899
    },
    {
      "epoch": 0.871359779867003,
      "grad_norm": 0.8606470644619356,
      "learning_rate": 4.2965723087148635e-05,
      "loss": 3.8148,
      "step": 1900
    },
    {
      "epoch": 0.8718183902774593,
      "grad_norm": 0.8499074380190059,
      "learning_rate": 4.266487890810256e-05,
      "loss": 3.7821,
      "step": 1901
    },
    {
      "epoch": 0.8722770006879156,
      "grad_norm": 0.8383673562691168,
      "learning_rate": 4.2365044735754365e-05,
      "loss": 3.7963,
      "step": 1902
    },
    {
      "epoch": 0.8727356110983719,
      "grad_norm": 0.8252105987983893,
      "learning_rate": 4.2066221232276266e-05,
      "loss": 3.7819,
      "step": 1903
    },
    {
      "epoch": 0.8731942215088283,
      "grad_norm": 0.821150675362424,
      "learning_rate": 4.176840905760815e-05,
      "loss": 3.7733,
      "step": 1904
    },
    {
      "epoch": 0.8736528319192846,
      "grad_norm": 0.8859926692875528,
      "learning_rate": 4.1471608869456443e-05,
      "loss": 4.0444,
      "step": 1905
    },
    {
      "epoch": 0.8741114423297409,
      "grad_norm": 0.8351403721195733,
      "learning_rate": 4.117582132329284e-05,
      "loss": 3.9356,
      "step": 1906
    },
    {
      "epoch": 0.8745700527401972,
      "grad_norm": 0.8218163326917144,
      "learning_rate": 4.088104707235263e-05,
      "loss": 3.9284,
      "step": 1907
    },
    {
      "epoch": 0.8750286631506535,
      "grad_norm": 0.8334345113220989,
      "learning_rate": 4.058728676763313e-05,
      "loss": 3.7434,
      "step": 1908
    },
    {
      "epoch": 0.8754872735611099,
      "grad_norm": 0.8494606320852781,
      "learning_rate": 4.0294541057892375e-05,
      "loss": 3.7772,
      "step": 1909
    },
    {
      "epoch": 0.8759458839715661,
      "grad_norm": 0.8965442737190648,
      "learning_rate": 4.000281058964794e-05,
      "loss": 3.772,
      "step": 1910
    },
    {
      "epoch": 0.8764044943820225,
      "grad_norm": 0.8296158861155694,
      "learning_rate": 3.971209600717507e-05,
      "loss": 3.6743,
      "step": 1911
    },
    {
      "epoch": 0.8768631047924788,
      "grad_norm": 0.8254906444703016,
      "learning_rate": 3.9422397952505465e-05,
      "loss": 3.9346,
      "step": 1912
    },
    {
      "epoch": 0.8773217152029351,
      "grad_norm": 0.8125166100399779,
      "learning_rate": 3.913371706542596e-05,
      "loss": 3.7167,
      "step": 1913
    },
    {
      "epoch": 0.8777803256133915,
      "grad_norm": 0.8925100546826128,
      "learning_rate": 3.884605398347707e-05,
      "loss": 4.0733,
      "step": 1914
    },
    {
      "epoch": 0.8782389360238477,
      "grad_norm": 0.8477966647157978,
      "learning_rate": 3.8559409341951456e-05,
      "loss": 3.9208,
      "step": 1915
    },
    {
      "epoch": 0.8786975464343041,
      "grad_norm": 0.8244294184339335,
      "learning_rate": 3.8273783773892404e-05,
      "loss": 3.7912,
      "step": 1916
    },
    {
      "epoch": 0.8791561568447603,
      "grad_norm": 0.8533007197904041,
      "learning_rate": 3.798917791009293e-05,
      "loss": 3.8031,
      "step": 1917
    },
    {
      "epoch": 0.8796147672552167,
      "grad_norm": 0.8280287222182107,
      "learning_rate": 3.770559237909393e-05,
      "loss": 3.7582,
      "step": 1918
    },
    {
      "epoch": 0.8800733776656731,
      "grad_norm": 0.8297559595420562,
      "learning_rate": 3.742302780718288e-05,
      "loss": 3.7662,
      "step": 1919
    },
    {
      "epoch": 0.8805319880761293,
      "grad_norm": 0.7914286497675128,
      "learning_rate": 3.7141484818392635e-05,
      "loss": 3.8004,
      "step": 1920
    },
    {
      "epoch": 0.8809905984865857,
      "grad_norm": 0.8452626258350453,
      "learning_rate": 3.686096403449973e-05,
      "loss": 3.8528,
      "step": 1921
    },
    {
      "epoch": 0.8814492088970419,
      "grad_norm": 0.8019084719170775,
      "learning_rate": 3.658146607502344e-05,
      "loss": 3.9638,
      "step": 1922
    },
    {
      "epoch": 0.8819078193074983,
      "grad_norm": 0.8463714427896485,
      "learning_rate": 3.630299155722411e-05,
      "loss": 3.8718,
      "step": 1923
    },
    {
      "epoch": 0.8823664297179546,
      "grad_norm": 0.8756277842310629,
      "learning_rate": 3.6025541096101676e-05,
      "loss": 3.9785,
      "step": 1924
    },
    {
      "epoch": 0.8828250401284109,
      "grad_norm": 0.8722035839090333,
      "learning_rate": 3.574911530439473e-05,
      "loss": 3.9508,
      "step": 1925
    },
    {
      "epoch": 0.8832836505388673,
      "grad_norm": 0.8862348312081401,
      "learning_rate": 3.5473714792578606e-05,
      "loss": 4.0858,
      "step": 1926
    },
    {
      "epoch": 0.8837422609493235,
      "grad_norm": 0.8260490574755135,
      "learning_rate": 3.519934016886478e-05,
      "loss": 4.0308,
      "step": 1927
    },
    {
      "epoch": 0.8842008713597799,
      "grad_norm": 0.8476017841626508,
      "learning_rate": 3.4925992039198776e-05,
      "loss": 3.9151,
      "step": 1928
    },
    {
      "epoch": 0.8846594817702362,
      "grad_norm": 0.82219210695854,
      "learning_rate": 3.465367100725908e-05,
      "loss": 3.8369,
      "step": 1929
    },
    {
      "epoch": 0.8851180921806925,
      "grad_norm": 0.8087119940901157,
      "learning_rate": 3.438237767445618e-05,
      "loss": 3.9347,
      "step": 1930
    },
    {
      "epoch": 0.8855767025911488,
      "grad_norm": 0.8802403161402887,
      "learning_rate": 3.4112112639930804e-05,
      "loss": 3.8551,
      "step": 1931
    },
    {
      "epoch": 0.8860353130016051,
      "grad_norm": 0.8408076504674384,
      "learning_rate": 3.3842876500552564e-05,
      "loss": 3.9304,
      "step": 1932
    },
    {
      "epoch": 0.8864939234120615,
      "grad_norm": 0.8253297800623369,
      "learning_rate": 3.357466985091906e-05,
      "loss": 3.8517,
      "step": 1933
    },
    {
      "epoch": 0.8869525338225178,
      "grad_norm": 0.8637982218685483,
      "learning_rate": 3.330749328335414e-05,
      "loss": 3.7874,
      "step": 1934
    },
    {
      "epoch": 0.8874111442329741,
      "grad_norm": 0.8674668203746955,
      "learning_rate": 3.304134738790659e-05,
      "loss": 3.9118,
      "step": 1935
    },
    {
      "epoch": 0.8878697546434304,
      "grad_norm": 0.8384396998114987,
      "learning_rate": 3.277623275234953e-05,
      "loss": 3.9602,
      "step": 1936
    },
    {
      "epoch": 0.8883283650538867,
      "grad_norm": 0.8298233903577324,
      "learning_rate": 3.2512149962177994e-05,
      "loss": 3.7413,
      "step": 1937
    },
    {
      "epoch": 0.888786975464343,
      "grad_norm": 0.8881192936410269,
      "learning_rate": 3.224909960060851e-05,
      "loss": 3.9305,
      "step": 1938
    },
    {
      "epoch": 0.8892455858747994,
      "grad_norm": 0.8697561926434358,
      "learning_rate": 3.198708224857755e-05,
      "loss": 3.8033,
      "step": 1939
    },
    {
      "epoch": 0.8897041962852557,
      "grad_norm": 0.8328976969257236,
      "learning_rate": 3.172609848474023e-05,
      "loss": 3.8703,
      "step": 1940
    },
    {
      "epoch": 0.890162806695712,
      "grad_norm": 0.8355667821903328,
      "learning_rate": 3.1466148885468895e-05,
      "loss": 3.9877,
      "step": 1941
    },
    {
      "epoch": 0.8906214171061683,
      "grad_norm": 0.8808758801776873,
      "learning_rate": 3.120723402485198e-05,
      "loss": 3.7663,
      "step": 1942
    },
    {
      "epoch": 0.8910800275166246,
      "grad_norm": 0.849072999562872,
      "learning_rate": 3.094935447469294e-05,
      "loss": 3.9897,
      "step": 1943
    },
    {
      "epoch": 0.891538637927081,
      "grad_norm": 0.822827478226575,
      "learning_rate": 3.069251080450863e-05,
      "loss": 3.8816,
      "step": 1944
    },
    {
      "epoch": 0.8919972483375372,
      "grad_norm": 0.8446352440751402,
      "learning_rate": 3.0436703581528113e-05,
      "loss": 3.9551,
      "step": 1945
    },
    {
      "epoch": 0.8924558587479936,
      "grad_norm": 0.7949071052125789,
      "learning_rate": 3.0181933370691694e-05,
      "loss": 3.7742,
      "step": 1946
    },
    {
      "epoch": 0.8929144691584499,
      "grad_norm": 0.8291955713167838,
      "learning_rate": 2.9928200734649523e-05,
      "loss": 3.8611,
      "step": 1947
    },
    {
      "epoch": 0.8933730795689062,
      "grad_norm": 0.8726989700418658,
      "learning_rate": 2.9675506233760142e-05,
      "loss": 3.8754,
      "step": 1948
    },
    {
      "epoch": 0.8938316899793626,
      "grad_norm": 0.8247517542008261,
      "learning_rate": 2.942385042608925e-05,
      "loss": 3.9028,
      "step": 1949
    },
    {
      "epoch": 0.8942903003898188,
      "grad_norm": 0.8447981258353934,
      "learning_rate": 2.9173233867409054e-05,
      "loss": 3.9187,
      "step": 1950
    },
    {
      "epoch": 0.8947489108002752,
      "grad_norm": 0.8453575646807977,
      "learning_rate": 2.892365711119638e-05,
      "loss": 3.8742,
      "step": 1951
    },
    {
      "epoch": 0.8952075212107314,
      "grad_norm": 0.8079795410829534,
      "learning_rate": 2.8675120708631596e-05,
      "loss": 3.8528,
      "step": 1952
    },
    {
      "epoch": 0.8956661316211878,
      "grad_norm": 0.8419849673514438,
      "learning_rate": 2.8427625208597764e-05,
      "loss": 3.946,
      "step": 1953
    },
    {
      "epoch": 0.8961247420316442,
      "grad_norm": 0.8473312385499908,
      "learning_rate": 2.8181171157678874e-05,
      "loss": 3.9689,
      "step": 1954
    },
    {
      "epoch": 0.8965833524421004,
      "grad_norm": 0.8501795053381869,
      "learning_rate": 2.7935759100159053e-05,
      "loss": 3.85,
      "step": 1955
    },
    {
      "epoch": 0.8970419628525568,
      "grad_norm": 0.840966497178792,
      "learning_rate": 2.7691389578021365e-05,
      "loss": 3.8811,
      "step": 1956
    },
    {
      "epoch": 0.897500573263013,
      "grad_norm": 0.840894514724081,
      "learning_rate": 2.7448063130946223e-05,
      "loss": 3.8973,
      "step": 1957
    },
    {
      "epoch": 0.8979591836734694,
      "grad_norm": 0.8209373192061823,
      "learning_rate": 2.7205780296310544e-05,
      "loss": 3.8518,
      "step": 1958
    },
    {
      "epoch": 0.8984177940839257,
      "grad_norm": 0.8785141957912973,
      "learning_rate": 2.6964541609186378e-05,
      "loss": 3.865,
      "step": 1959
    },
    {
      "epoch": 0.898876404494382,
      "grad_norm": 0.828952655742634,
      "learning_rate": 2.6724347602340104e-05,
      "loss": 3.8993,
      "step": 1960
    },
    {
      "epoch": 0.8993350149048384,
      "grad_norm": 0.8528656710794904,
      "learning_rate": 2.6485198806230682e-05,
      "loss": 4.0181,
      "step": 1961
    },
    {
      "epoch": 0.8997936253152946,
      "grad_norm": 0.8501449530172904,
      "learning_rate": 2.6247095749008797e-05,
      "loss": 3.9853,
      "step": 1962
    },
    {
      "epoch": 0.900252235725751,
      "grad_norm": 0.8543744992545613,
      "learning_rate": 2.6010038956515826e-05,
      "loss": 3.9149,
      "step": 1963
    },
    {
      "epoch": 0.9007108461362073,
      "grad_norm": 0.8326406072015868,
      "learning_rate": 2.5774028952282423e-05,
      "loss": 3.8049,
      "step": 1964
    },
    {
      "epoch": 0.9011694565466636,
      "grad_norm": 0.820753076858836,
      "learning_rate": 2.5539066257527277e-05,
      "loss": 3.9403,
      "step": 1965
    },
    {
      "epoch": 0.90162806695712,
      "grad_norm": 0.8483764126118326,
      "learning_rate": 2.530515139115652e-05,
      "loss": 3.9236,
      "step": 1966
    },
    {
      "epoch": 0.9020866773675762,
      "grad_norm": 0.8325375608130782,
      "learning_rate": 2.5072284869761874e-05,
      "loss": 3.8096,
      "step": 1967
    },
    {
      "epoch": 0.9025452877780326,
      "grad_norm": 0.8271304224860582,
      "learning_rate": 2.4840467207619786e-05,
      "loss": 3.9324,
      "step": 1968
    },
    {
      "epoch": 0.9030038981884889,
      "grad_norm": 0.8373713835172847,
      "learning_rate": 2.460969891669068e-05,
      "loss": 3.9643,
      "step": 1969
    },
    {
      "epoch": 0.9034625085989452,
      "grad_norm": 0.8271048043417418,
      "learning_rate": 2.4379980506617272e-05,
      "loss": 3.9486,
      "step": 1970
    },
    {
      "epoch": 0.9039211190094015,
      "grad_norm": 0.8628503435484537,
      "learning_rate": 2.4151312484723464e-05,
      "loss": 3.8424,
      "step": 1971
    },
    {
      "epoch": 0.9043797294198578,
      "grad_norm": 0.7872814364301522,
      "learning_rate": 2.3923695356013798e-05,
      "loss": 3.8261,
      "step": 1972
    },
    {
      "epoch": 0.9048383398303141,
      "grad_norm": 0.8559090339186114,
      "learning_rate": 2.3697129623171833e-05,
      "loss": 4.0225,
      "step": 1973
    },
    {
      "epoch": 0.9052969502407705,
      "grad_norm": 0.8398429454755256,
      "learning_rate": 2.3471615786559042e-05,
      "loss": 3.7971,
      "step": 1974
    },
    {
      "epoch": 0.9057555606512268,
      "grad_norm": 0.8390016603434238,
      "learning_rate": 2.3247154344213818e-05,
      "loss": 3.7982,
      "step": 1975
    },
    {
      "epoch": 0.9062141710616831,
      "grad_norm": 0.817700228991357,
      "learning_rate": 2.3023745791850625e-05,
      "loss": 3.9776,
      "step": 1976
    },
    {
      "epoch": 0.9066727814721394,
      "grad_norm": 0.8493850088380667,
      "learning_rate": 2.2801390622858354e-05,
      "loss": 3.8105,
      "step": 1977
    },
    {
      "epoch": 0.9071313918825957,
      "grad_norm": 0.8545227909057439,
      "learning_rate": 2.2580089328299746e-05,
      "loss": 3.9154,
      "step": 1978
    },
    {
      "epoch": 0.9075900022930521,
      "grad_norm": 0.8513676437933291,
      "learning_rate": 2.235984239690997e-05,
      "loss": 3.8411,
      "step": 1979
    },
    {
      "epoch": 0.9080486127035083,
      "grad_norm": 0.8346570490909112,
      "learning_rate": 2.2140650315095934e-05,
      "loss": 3.9886,
      "step": 1980
    },
    {
      "epoch": 0.9085072231139647,
      "grad_norm": 0.8662982327384878,
      "learning_rate": 2.192251356693459e-05,
      "loss": 4.0172,
      "step": 1981
    },
    {
      "epoch": 0.908965833524421,
      "grad_norm": 0.8672868383382905,
      "learning_rate": 2.170543263417246e-05,
      "loss": 3.9597,
      "step": 1982
    },
    {
      "epoch": 0.9094244439348773,
      "grad_norm": 0.8407095615759864,
      "learning_rate": 2.1489407996224286e-05,
      "loss": 3.9495,
      "step": 1983
    },
    {
      "epoch": 0.9098830543453337,
      "grad_norm": 0.8000117829715154,
      "learning_rate": 2.127444013017199e-05,
      "loss": 3.8045,
      "step": 1984
    },
    {
      "epoch": 0.9103416647557899,
      "grad_norm": 0.8537838994892079,
      "learning_rate": 2.1060529510763648e-05,
      "loss": 3.9533,
      "step": 1985
    },
    {
      "epoch": 0.9108002751662463,
      "grad_norm": 0.8661065812883498,
      "learning_rate": 2.084767661041259e-05,
      "loss": 3.9162,
      "step": 1986
    },
    {
      "epoch": 0.9112588855767025,
      "grad_norm": 0.8620142461074801,
      "learning_rate": 2.063588189919596e-05,
      "loss": 3.8445,
      "step": 1987
    },
    {
      "epoch": 0.9117174959871589,
      "grad_norm": 0.88133984850228,
      "learning_rate": 2.0425145844854275e-05,
      "loss": 3.9034,
      "step": 1988
    },
    {
      "epoch": 0.9121761063976153,
      "grad_norm": 0.8514348129032506,
      "learning_rate": 2.0215468912789693e-05,
      "loss": 3.8436,
      "step": 1989
    },
    {
      "epoch": 0.9126347168080715,
      "grad_norm": 0.8674772654805663,
      "learning_rate": 2.0006851566065575e-05,
      "loss": 3.7999,
      "step": 1990
    },
    {
      "epoch": 0.9130933272185279,
      "grad_norm": 0.8309050973364643,
      "learning_rate": 1.9799294265405166e-05,
      "loss": 3.95,
      "step": 1991
    },
    {
      "epoch": 0.9135519376289841,
      "grad_norm": 0.8230229764097498,
      "learning_rate": 1.9592797469190572e-05,
      "loss": 3.8842,
      "step": 1992
    },
    {
      "epoch": 0.9140105480394405,
      "grad_norm": 0.8740163890724908,
      "learning_rate": 1.938736163346194e-05,
      "loss": 3.7656,
      "step": 1993
    },
    {
      "epoch": 0.9144691584498968,
      "grad_norm": 0.861097459008977,
      "learning_rate": 1.9182987211916246e-05,
      "loss": 3.8877,
      "step": 1994
    },
    {
      "epoch": 0.9149277688603531,
      "grad_norm": 0.8663929051428375,
      "learning_rate": 1.8979674655906332e-05,
      "loss": 3.6803,
      "step": 1995
    },
    {
      "epoch": 0.9153863792708095,
      "grad_norm": 0.8588063558659643,
      "learning_rate": 1.8777424414440024e-05,
      "loss": 3.9762,
      "step": 1996
    },
    {
      "epoch": 0.9158449896812658,
      "grad_norm": 0.8474644590103753,
      "learning_rate": 1.8576236934179202e-05,
      "loss": 4.0526,
      "step": 1997
    },
    {
      "epoch": 0.9163036000917221,
      "grad_norm": 0.8158827716442782,
      "learning_rate": 1.8376112659438393e-05,
      "loss": 3.8505,
      "step": 1998
    },
    {
      "epoch": 0.9167622105021784,
      "grad_norm": 0.8643175796813047,
      "learning_rate": 1.8177052032184282e-05,
      "loss": 4.1238,
      "step": 1999
    },
    {
      "epoch": 0.9172208209126347,
      "grad_norm": 0.867861512119339,
      "learning_rate": 1.7979055492034435e-05,
      "loss": 3.8445,
      "step": 2000
    },
    {
      "epoch": 0.917679431323091,
      "grad_norm": 0.8599921367694354,
      "learning_rate": 1.7782123476256407e-05,
      "loss": 3.9807,
      "step": 2001
    },
    {
      "epoch": 0.9181380417335474,
      "grad_norm": 0.8101063875765772,
      "learning_rate": 1.7586256419766965e-05,
      "loss": 3.8282,
      "step": 2002
    },
    {
      "epoch": 0.9185966521440037,
      "grad_norm": 0.8378972765185031,
      "learning_rate": 1.7391454755130766e-05,
      "loss": 3.7908,
      "step": 2003
    },
    {
      "epoch": 0.91905526255446,
      "grad_norm": 0.8688994362318205,
      "learning_rate": 1.7197718912559557e-05,
      "loss": 3.9536,
      "step": 2004
    },
    {
      "epoch": 0.9195138729649163,
      "grad_norm": 0.8371079202560124,
      "learning_rate": 1.700504931991148e-05,
      "loss": 3.8356,
      "step": 2005
    },
    {
      "epoch": 0.9199724833753726,
      "grad_norm": 0.8386078456961309,
      "learning_rate": 1.681344640268978e-05,
      "loss": 4.0408,
      "step": 2006
    },
    {
      "epoch": 0.920431093785829,
      "grad_norm": 0.8742005696166191,
      "learning_rate": 1.6622910584041974e-05,
      "loss": 3.8202,
      "step": 2007
    },
    {
      "epoch": 0.9208897041962852,
      "grad_norm": 0.8686301768381685,
      "learning_rate": 1.6433442284758903e-05,
      "loss": 3.8642,
      "step": 2008
    },
    {
      "epoch": 0.9213483146067416,
      "grad_norm": 0.8474627223900476,
      "learning_rate": 1.624504192327392e-05,
      "loss": 3.8409,
      "step": 2009
    },
    {
      "epoch": 0.9218069250171979,
      "grad_norm": 0.831629774901717,
      "learning_rate": 1.6057709915661856e-05,
      "loss": 4.0369,
      "step": 2010
    },
    {
      "epoch": 0.9222655354276542,
      "grad_norm": 0.8882341207987431,
      "learning_rate": 1.5871446675638057e-05,
      "loss": 3.8323,
      "step": 2011
    },
    {
      "epoch": 0.9227241458381106,
      "grad_norm": 0.8442889907658873,
      "learning_rate": 1.5686252614557638e-05,
      "loss": 3.9805,
      "step": 2012
    },
    {
      "epoch": 0.9231827562485668,
      "grad_norm": 0.8512581157673809,
      "learning_rate": 1.5502128141414497e-05,
      "loss": 3.7806,
      "step": 2013
    },
    {
      "epoch": 0.9236413666590232,
      "grad_norm": 0.8627277245973806,
      "learning_rate": 1.5319073662840188e-05,
      "loss": 4.0332,
      "step": 2014
    },
    {
      "epoch": 0.9240999770694794,
      "grad_norm": 0.8703741554181751,
      "learning_rate": 1.5137089583103391e-05,
      "loss": 3.9774,
      "step": 2015
    },
    {
      "epoch": 0.9245585874799358,
      "grad_norm": 0.9365924813732052,
      "learning_rate": 1.4956176304108893e-05,
      "loss": 3.9755,
      "step": 2016
    },
    {
      "epoch": 0.9250171978903922,
      "grad_norm": 0.8654169550363484,
      "learning_rate": 1.4776334225396481e-05,
      "loss": 3.7319,
      "step": 2017
    },
    {
      "epoch": 0.9254758083008484,
      "grad_norm": 0.8561034496244034,
      "learning_rate": 1.4597563744140397e-05,
      "loss": 3.9291,
      "step": 2018
    },
    {
      "epoch": 0.9259344187113048,
      "grad_norm": 0.8658208681070794,
      "learning_rate": 1.4419865255148269e-05,
      "loss": 3.6516,
      "step": 2019
    },
    {
      "epoch": 0.926393029121761,
      "grad_norm": 0.8286871169108122,
      "learning_rate": 1.4243239150860122e-05,
      "loss": 3.7881,
      "step": 2020
    },
    {
      "epoch": 0.9268516395322174,
      "grad_norm": 0.8148342708404772,
      "learning_rate": 1.4067685821347932e-05,
      "loss": 3.9682,
      "step": 2021
    },
    {
      "epoch": 0.9273102499426737,
      "grad_norm": 0.847582016168636,
      "learning_rate": 1.389320565431429e-05,
      "loss": 3.7945,
      "step": 2022
    },
    {
      "epoch": 0.92776886035313,
      "grad_norm": 0.8009112909609668,
      "learning_rate": 1.3719799035091851e-05,
      "loss": 3.8859,
      "step": 2023
    },
    {
      "epoch": 0.9282274707635864,
      "grad_norm": 0.8528612332120525,
      "learning_rate": 1.3547466346642278e-05,
      "loss": 3.7644,
      "step": 2024
    },
    {
      "epoch": 0.9286860811740426,
      "grad_norm": 0.8501777468776505,
      "learning_rate": 1.3376207969555577e-05,
      "loss": 3.7813,
      "step": 2025
    },
    {
      "epoch": 0.929144691584499,
      "grad_norm": 0.8098712808624919,
      "learning_rate": 1.32060242820492e-05,
      "loss": 3.8149,
      "step": 2026
    },
    {
      "epoch": 0.9296033019949553,
      "grad_norm": 0.8196946283650222,
      "learning_rate": 1.3036915659967118e-05,
      "loss": 3.8842,
      "step": 2027
    },
    {
      "epoch": 0.9300619124054116,
      "grad_norm": 0.8459372914814807,
      "learning_rate": 1.2868882476779087e-05,
      "loss": 3.8111,
      "step": 2028
    },
    {
      "epoch": 0.930520522815868,
      "grad_norm": 0.8202437096657841,
      "learning_rate": 1.2701925103579815e-05,
      "loss": 3.9147,
      "step": 2029
    },
    {
      "epoch": 0.9309791332263242,
      "grad_norm": 0.867413250056903,
      "learning_rate": 1.2536043909088191e-05,
      "loss": 3.963,
      "step": 2030
    },
    {
      "epoch": 0.9314377436367806,
      "grad_norm": 0.8385223361515965,
      "learning_rate": 1.2371239259646228e-05,
      "loss": 3.9964,
      "step": 2031
    },
    {
      "epoch": 0.9318963540472369,
      "grad_norm": 0.9346062083809694,
      "learning_rate": 1.2207511519218672e-05,
      "loss": 3.9191,
      "step": 2032
    },
    {
      "epoch": 0.9323549644576932,
      "grad_norm": 0.8613123762546415,
      "learning_rate": 1.2044861049391676e-05,
      "loss": 3.7522,
      "step": 2033
    },
    {
      "epoch": 0.9328135748681495,
      "grad_norm": 0.8371314088700297,
      "learning_rate": 1.1883288209372512e-05,
      "loss": 3.9878,
      "step": 2034
    },
    {
      "epoch": 0.9332721852786058,
      "grad_norm": 0.8033209326083413,
      "learning_rate": 1.1722793355988471e-05,
      "loss": 3.896,
      "step": 2035
    },
    {
      "epoch": 0.9337307956890621,
      "grad_norm": 0.8423748930521902,
      "learning_rate": 1.1563376843686135e-05,
      "loss": 3.7302,
      "step": 2036
    },
    {
      "epoch": 0.9341894060995185,
      "grad_norm": 0.8303239968743652,
      "learning_rate": 1.140503902453055e-05,
      "loss": 3.8029,
      "step": 2037
    },
    {
      "epoch": 0.9346480165099748,
      "grad_norm": 0.8339072575414753,
      "learning_rate": 1.1247780248204665e-05,
      "loss": 3.9935,
      "step": 2038
    },
    {
      "epoch": 0.9351066269204311,
      "grad_norm": 0.8442262981837397,
      "learning_rate": 1.1091600862008333e-05,
      "loss": 3.8296,
      "step": 2039
    },
    {
      "epoch": 0.9355652373308874,
      "grad_norm": 0.8449071410001567,
      "learning_rate": 1.0936501210857652e-05,
      "loss": 3.9556,
      "step": 2040
    },
    {
      "epoch": 0.9360238477413437,
      "grad_norm": 0.826945321310232,
      "learning_rate": 1.0782481637284013e-05,
      "loss": 3.9166,
      "step": 2041
    },
    {
      "epoch": 0.9364824581518001,
      "grad_norm": 0.8264214483107898,
      "learning_rate": 1.0629542481433663e-05,
      "loss": 3.936,
      "step": 2042
    },
    {
      "epoch": 0.9369410685622563,
      "grad_norm": 0.9082940038209367,
      "learning_rate": 1.0477684081066751e-05,
      "loss": 3.983,
      "step": 2043
    },
    {
      "epoch": 0.9373996789727127,
      "grad_norm": 0.809605826023425,
      "learning_rate": 1.0326906771556566e-05,
      "loss": 3.8669,
      "step": 2044
    },
    {
      "epoch": 0.937858289383169,
      "grad_norm": 0.8178205195490351,
      "learning_rate": 1.017721088588891e-05,
      "loss": 3.738,
      "step": 2045
    },
    {
      "epoch": 0.9383168997936253,
      "grad_norm": 0.8361251695039015,
      "learning_rate": 1.0028596754661334e-05,
      "loss": 3.8437,
      "step": 2046
    },
    {
      "epoch": 0.9387755102040817,
      "grad_norm": 0.8562456208638298,
      "learning_rate": 9.881064706082298e-06,
      "loss": 3.8345,
      "step": 2047
    },
    {
      "epoch": 0.9392341206145379,
      "grad_norm": 0.8305406316943807,
      "learning_rate": 9.734615065970454e-06,
      "loss": 3.9437,
      "step": 2048
    },
    {
      "epoch": 0.9396927310249943,
      "grad_norm": 0.8741568889710702,
      "learning_rate": 9.58924815775425e-06,
      "loss": 3.8612,
      "step": 2049
    },
    {
      "epoch": 0.9401513414354505,
      "grad_norm": 0.8079847153397736,
      "learning_rate": 9.444964302470715e-06,
      "loss": 3.9002,
      "step": 2050
    },
    {
      "epoch": 0.9406099518459069,
      "grad_norm": 0.9153020436588157,
      "learning_rate": 9.301763818765018e-06,
      "loss": 3.9662,
      "step": 2051
    },
    {
      "epoch": 0.9410685622563633,
      "grad_norm": 0.8477617049322664,
      "learning_rate": 9.15964702288996e-06,
      "loss": 3.7232,
      "step": 2052
    },
    {
      "epoch": 0.9415271726668195,
      "grad_norm": 0.8145117184033174,
      "learning_rate": 9.018614228704925e-06,
      "loss": 3.7929,
      "step": 2053
    },
    {
      "epoch": 0.9419857830772759,
      "grad_norm": 0.8484136750632856,
      "learning_rate": 8.878665747675152e-06,
      "loss": 3.8646,
      "step": 2054
    },
    {
      "epoch": 0.9424443934877321,
      "grad_norm": 0.8379671899897309,
      "learning_rate": 8.739801888871469e-06,
      "loss": 4.0031,
      "step": 2055
    },
    {
      "epoch": 0.9429030038981885,
      "grad_norm": 0.8135366409047714,
      "learning_rate": 8.602022958969336e-06,
      "loss": 3.8307,
      "step": 2056
    },
    {
      "epoch": 0.9433616143086448,
      "grad_norm": 0.8332705424414986,
      "learning_rate": 8.465329262248078e-06,
      "loss": 3.7887,
      "step": 2057
    },
    {
      "epoch": 0.9438202247191011,
      "grad_norm": 0.8340838591293532,
      "learning_rate": 8.32972110059027e-06,
      "loss": 3.9817,
      "step": 2058
    },
    {
      "epoch": 0.9442788351295575,
      "grad_norm": 0.8514020382009008,
      "learning_rate": 8.195198773481406e-06,
      "loss": 3.8839,
      "step": 2059
    },
    {
      "epoch": 0.9447374455400137,
      "grad_norm": 0.8180510769842605,
      "learning_rate": 8.061762578008613e-06,
      "loss": 3.8627,
      "step": 2060
    },
    {
      "epoch": 0.9451960559504701,
      "grad_norm": 0.8810942634293348,
      "learning_rate": 7.929412808860559e-06,
      "loss": 3.9834,
      "step": 2061
    },
    {
      "epoch": 0.9456546663609264,
      "grad_norm": 0.9099296914078776,
      "learning_rate": 7.79814975832649e-06,
      "loss": 3.864,
      "step": 2062
    },
    {
      "epoch": 0.9461132767713827,
      "grad_norm": 0.8214441072577684,
      "learning_rate": 7.667973716295851e-06,
      "loss": 3.8642,
      "step": 2063
    },
    {
      "epoch": 0.946571887181839,
      "grad_norm": 0.8577244386067902,
      "learning_rate": 7.5388849702571205e-06,
      "loss": 3.8447,
      "step": 2064
    },
    {
      "epoch": 0.9470304975922953,
      "grad_norm": 0.8444743124337305,
      "learning_rate": 7.4108838052979185e-06,
      "loss": 3.7855,
      "step": 2065
    },
    {
      "epoch": 0.9474891080027517,
      "grad_norm": 0.8554710718563221,
      "learning_rate": 7.283970504103732e-06,
      "loss": 3.8719,
      "step": 2066
    },
    {
      "epoch": 0.947947718413208,
      "grad_norm": 0.8372022776266841,
      "learning_rate": 7.1581453469575785e-06,
      "loss": 3.8396,
      "step": 2067
    },
    {
      "epoch": 0.9484063288236643,
      "grad_norm": 0.8838036694030891,
      "learning_rate": 7.033408611739456e-06,
      "loss": 3.9597,
      "step": 2068
    },
    {
      "epoch": 0.9488649392341206,
      "grad_norm": 0.8266569828686807,
      "learning_rate": 6.909760573925561e-06,
      "loss": 3.8057,
      "step": 2069
    },
    {
      "epoch": 0.9493235496445769,
      "grad_norm": 0.868736352551762,
      "learning_rate": 6.787201506587626e-06,
      "loss": 3.965,
      "step": 2070
    },
    {
      "epoch": 0.9497821600550332,
      "grad_norm": 0.8306329359648865,
      "learning_rate": 6.66573168039264e-06,
      "loss": 4.0009,
      "step": 2071
    },
    {
      "epoch": 0.9502407704654896,
      "grad_norm": 0.8809657748915067,
      "learning_rate": 6.545351363601959e-06,
      "loss": 3.8078,
      "step": 2072
    },
    {
      "epoch": 0.9506993808759459,
      "grad_norm": 0.7920426838860011,
      "learning_rate": 6.426060822070812e-06,
      "loss": 3.9438,
      "step": 2073
    },
    {
      "epoch": 0.9511579912864022,
      "grad_norm": 1.0223228927047494,
      "learning_rate": 6.3078603192475716e-06,
      "loss": 3.9883,
      "step": 2074
    },
    {
      "epoch": 0.9516166016968585,
      "grad_norm": 0.8624401493747123,
      "learning_rate": 6.1907501161735934e-06,
      "loss": 3.7507,
      "step": 2075
    },
    {
      "epoch": 0.9520752121073148,
      "grad_norm": 0.8385366450837136,
      "learning_rate": 6.074730471482049e-06,
      "loss": 3.8189,
      "step": 2076
    },
    {
      "epoch": 0.9525338225177712,
      "grad_norm": 0.8082462482047105,
      "learning_rate": 5.959801641397755e-06,
      "loss": 3.7655,
      "step": 2077
    },
    {
      "epoch": 0.9529924329282274,
      "grad_norm": 0.8290034520834061,
      "learning_rate": 5.845963879736627e-06,
      "loss": 3.823,
      "step": 2078
    },
    {
      "epoch": 0.9534510433386838,
      "grad_norm": 0.8241047822771208,
      "learning_rate": 5.733217437904892e-06,
      "loss": 3.9027,
      "step": 2079
    },
    {
      "epoch": 0.95390965374914,
      "grad_norm": 0.83981116974645,
      "learning_rate": 5.621562564898597e-06,
      "loss": 3.9019,
      "step": 2080
    },
    {
      "epoch": 0.9543682641595964,
      "grad_norm": 0.8164022040038718,
      "learning_rate": 5.51099950730316e-06,
      "loss": 4.0005,
      "step": 2081
    },
    {
      "epoch": 0.9548268745700528,
      "grad_norm": 0.8583683067215391,
      "learning_rate": 5.401528509292763e-06,
      "loss": 3.9584,
      "step": 2082
    },
    {
      "epoch": 0.955285484980509,
      "grad_norm": 0.841285688372956,
      "learning_rate": 5.2931498126298495e-06,
      "loss": 3.8573,
      "step": 2083
    },
    {
      "epoch": 0.9557440953909654,
      "grad_norm": 0.8631098264736263,
      "learning_rate": 5.1858636566645135e-06,
      "loss": 3.6911,
      "step": 2084
    },
    {
      "epoch": 0.9562027058014216,
      "grad_norm": 0.8466894642552518,
      "learning_rate": 5.0796702783340035e-06,
      "loss": 3.7852,
      "step": 2085
    },
    {
      "epoch": 0.956661316211878,
      "grad_norm": 0.8423353904450127,
      "learning_rate": 4.97456991216233e-06,
      "loss": 3.972,
      "step": 2086
    },
    {
      "epoch": 0.9571199266223344,
      "grad_norm": 0.8575073799129915,
      "learning_rate": 4.870562790259325e-06,
      "loss": 3.7915,
      "step": 2087
    },
    {
      "epoch": 0.9575785370327906,
      "grad_norm": 0.8425766081281263,
      "learning_rate": 4.7676491423208625e-06,
      "loss": 3.9245,
      "step": 2088
    },
    {
      "epoch": 0.958037147443247,
      "grad_norm": 0.8524922486962105,
      "learning_rate": 4.66582919562758e-06,
      "loss": 3.936,
      "step": 2089
    },
    {
      "epoch": 0.9584957578537033,
      "grad_norm": 0.8370686614175017,
      "learning_rate": 4.5651031750448825e-06,
      "loss": 3.8184,
      "step": 2090
    },
    {
      "epoch": 0.9589543682641596,
      "grad_norm": 0.8801589991731269,
      "learning_rate": 4.465471303022217e-06,
      "loss": 3.9745,
      "step": 2091
    },
    {
      "epoch": 0.959412978674616,
      "grad_norm": 0.8829890727062999,
      "learning_rate": 4.366933799592743e-06,
      "loss": 3.7803,
      "step": 2092
    },
    {
      "epoch": 0.9598715890850722,
      "grad_norm": 0.8330892766761371,
      "learning_rate": 4.269490882372551e-06,
      "loss": 3.883,
      "step": 2093
    },
    {
      "epoch": 0.9603301994955286,
      "grad_norm": 0.8666276904933145,
      "learning_rate": 4.1731427665606115e-06,
      "loss": 3.8281,
      "step": 2094
    },
    {
      "epoch": 0.9607888099059849,
      "grad_norm": 0.8497904411937507,
      "learning_rate": 4.077889664937884e-06,
      "loss": 3.7654,
      "step": 2095
    },
    {
      "epoch": 0.9612474203164412,
      "grad_norm": 0.855942757025609,
      "learning_rate": 3.983731787867207e-06,
      "loss": 3.9469,
      "step": 2096
    },
    {
      "epoch": 0.9617060307268975,
      "grad_norm": 0.8784613913999026,
      "learning_rate": 3.890669343292464e-06,
      "loss": 3.9233,
      "step": 2097
    },
    {
      "epoch": 0.9621646411373538,
      "grad_norm": 0.8401857250951897,
      "learning_rate": 3.7987025367384743e-06,
      "loss": 3.8933,
      "step": 2098
    },
    {
      "epoch": 0.9626232515478101,
      "grad_norm": 0.831504671656338,
      "learning_rate": 3.707831571310327e-06,
      "loss": 3.8555,
      "step": 2099
    },
    {
      "epoch": 0.9630818619582665,
      "grad_norm": 0.8748149428668217,
      "learning_rate": 3.6180566476929912e-06,
      "loss": 3.8311,
      "step": 2100
    },
    {
      "epoch": 0.9635404723687228,
      "grad_norm": 0.8517084942574455,
      "learning_rate": 3.529377964150815e-06,
      "loss": 3.9928,
      "step": 2101
    },
    {
      "epoch": 0.9639990827791791,
      "grad_norm": 0.8287256034899216,
      "learning_rate": 3.441795716527307e-06,
      "loss": 3.8743,
      "step": 2102
    },
    {
      "epoch": 0.9644576931896354,
      "grad_norm": 0.8545255234891419,
      "learning_rate": 3.355310098244302e-06,
      "loss": 3.7821,
      "step": 2103
    },
    {
      "epoch": 0.9649163036000917,
      "grad_norm": 0.8827601418328855,
      "learning_rate": 3.269921300301959e-06,
      "loss": 3.8856,
      "step": 2104
    },
    {
      "epoch": 0.9653749140105481,
      "grad_norm": 0.8583649581590631,
      "learning_rate": 3.1856295112780988e-06,
      "loss": 3.8873,
      "step": 2105
    },
    {
      "epoch": 0.9658335244210043,
      "grad_norm": 0.8586582044666037,
      "learning_rate": 3.102434917327812e-06,
      "loss": 4.0188,
      "step": 2106
    },
    {
      "epoch": 0.9662921348314607,
      "grad_norm": 0.8247603415587091,
      "learning_rate": 3.0203377021831292e-06,
      "loss": 3.7413,
      "step": 2107
    },
    {
      "epoch": 0.966750745241917,
      "grad_norm": 0.842672819682094,
      "learning_rate": 2.939338047152573e-06,
      "loss": 3.7777,
      "step": 2108
    },
    {
      "epoch": 0.9672093556523733,
      "grad_norm": 0.8246654112654029,
      "learning_rate": 2.8594361311206073e-06,
      "loss": 3.8927,
      "step": 2109
    },
    {
      "epoch": 0.9676679660628297,
      "grad_norm": 0.8490739683726153,
      "learning_rate": 2.7806321305475225e-06,
      "loss": 3.8482,
      "step": 2110
    },
    {
      "epoch": 0.9681265764732859,
      "grad_norm": 0.8358164960415104,
      "learning_rate": 2.7029262194688818e-06,
      "loss": 3.9114,
      "step": 2111
    },
    {
      "epoch": 0.9685851868837423,
      "grad_norm": 0.8430828284103903,
      "learning_rate": 2.626318569495134e-06,
      "loss": 3.8531,
      "step": 2112
    },
    {
      "epoch": 0.9690437972941985,
      "grad_norm": 0.8524547843307241,
      "learning_rate": 2.550809349811334e-06,
      "loss": 3.9045,
      "step": 2113
    },
    {
      "epoch": 0.9695024077046549,
      "grad_norm": 0.8836921243562457,
      "learning_rate": 2.476398727176532e-06,
      "loss": 3.9509,
      "step": 2114
    },
    {
      "epoch": 0.9699610181151113,
      "grad_norm": 0.8377025174684698,
      "learning_rate": 2.4030868659237204e-06,
      "loss": 3.9013,
      "step": 2115
    },
    {
      "epoch": 0.9704196285255675,
      "grad_norm": 0.8148751506191039,
      "learning_rate": 2.3308739279593317e-06,
      "loss": 3.7915,
      "step": 2116
    },
    {
      "epoch": 0.9708782389360239,
      "grad_norm": 0.8585410466172051,
      "learning_rate": 2.2597600727626845e-06,
      "loss": 3.885,
      "step": 2117
    },
    {
      "epoch": 0.9713368493464801,
      "grad_norm": 0.8456035185338863,
      "learning_rate": 2.1897454573860387e-06,
      "loss": 3.805,
      "step": 2118
    },
    {
      "epoch": 0.9717954597569365,
      "grad_norm": 0.8272685323519661,
      "learning_rate": 2.1208302364538746e-06,
      "loss": 3.8536,
      "step": 2119
    },
    {
      "epoch": 0.9722540701673928,
      "grad_norm": 0.8603501373997986,
      "learning_rate": 2.0530145621627804e-06,
      "loss": 3.9297,
      "step": 2120
    },
    {
      "epoch": 0.9727126805778491,
      "grad_norm": 0.9429249061645142,
      "learning_rate": 1.9862985842810653e-06,
      "loss": 3.8396,
      "step": 2121
    },
    {
      "epoch": 0.9731712909883055,
      "grad_norm": 0.8456223941575688,
      "learning_rate": 1.920682450148259e-06,
      "loss": 4.0025,
      "step": 2122
    },
    {
      "epoch": 0.9736299013987617,
      "grad_norm": 0.8499384242870541,
      "learning_rate": 1.856166304675111e-06,
      "loss": 3.9548,
      "step": 2123
    },
    {
      "epoch": 0.9740885118092181,
      "grad_norm": 0.9245937700046619,
      "learning_rate": 1.792750290342926e-06,
      "loss": 3.9681,
      "step": 2124
    },
    {
      "epoch": 0.9745471222196744,
      "grad_norm": 0.8490082304926304,
      "learning_rate": 1.7304345472035632e-06,
      "loss": 3.9159,
      "step": 2125
    },
    {
      "epoch": 0.9750057326301307,
      "grad_norm": 0.880193890844829,
      "learning_rate": 1.6692192128788253e-06,
      "loss": 3.923,
      "step": 2126
    },
    {
      "epoch": 0.975464343040587,
      "grad_norm": 0.8168073184081396,
      "learning_rate": 1.6091044225604035e-06,
      "loss": 3.7085,
      "step": 2127
    },
    {
      "epoch": 0.9759229534510433,
      "grad_norm": 0.8081302970345136,
      "learning_rate": 1.5500903090094888e-06,
      "loss": 3.7687,
      "step": 2128
    },
    {
      "epoch": 0.9763815638614997,
      "grad_norm": 0.8390136974942781,
      "learning_rate": 1.492177002556383e-06,
      "loss": 3.7101,
      "step": 2129
    },
    {
      "epoch": 0.976840174271956,
      "grad_norm": 0.8447715251179975,
      "learning_rate": 1.4353646311004443e-06,
      "loss": 3.8803,
      "step": 2130
    },
    {
      "epoch": 0.9772987846824123,
      "grad_norm": 0.8957939062745698,
      "learning_rate": 1.3796533201094752e-06,
      "loss": 4.004,
      "step": 2131
    },
    {
      "epoch": 0.9777573950928686,
      "grad_norm": 0.8530251880175247,
      "learning_rate": 1.3250431926197793e-06,
      "loss": 3.9913,
      "step": 2132
    },
    {
      "epoch": 0.9782160055033249,
      "grad_norm": 0.8234832377959644,
      "learning_rate": 1.2715343692356607e-06,
      "loss": 3.9346,
      "step": 2133
    },
    {
      "epoch": 0.9786746159137812,
      "grad_norm": 0.8490251261512456,
      "learning_rate": 1.2191269681292582e-06,
      "loss": 3.9683,
      "step": 2134
    },
    {
      "epoch": 0.9791332263242376,
      "grad_norm": 0.8594555410660307,
      "learning_rate": 1.1678211050402676e-06,
      "loss": 3.9399,
      "step": 2135
    },
    {
      "epoch": 0.9795918367346939,
      "grad_norm": 0.8358827393635083,
      "learning_rate": 1.117616893275719e-06,
      "loss": 3.8784,
      "step": 2136
    },
    {
      "epoch": 0.9800504471451502,
      "grad_norm": 0.8494916722004088,
      "learning_rate": 1.068514443709534e-06,
      "loss": 3.9019,
      "step": 2137
    },
    {
      "epoch": 0.9805090575556065,
      "grad_norm": 0.8563772547484699,
      "learning_rate": 1.0205138647826905e-06,
      "loss": 4.0,
      "step": 2138
    },
    {
      "epoch": 0.9809676679660628,
      "grad_norm": 0.8652585230206649,
      "learning_rate": 9.73615262502503e-07,
      "loss": 3.9501,
      "step": 2139
    },
    {
      "epoch": 0.9814262783765192,
      "grad_norm": 0.8720537290352637,
      "learning_rate": 9.278187404426763e-07,
      "loss": 4.012,
      "step": 2140
    },
    {
      "epoch": 0.9818848887869754,
      "grad_norm": 0.8738055133324034,
      "learning_rate": 8.831243997431404e-07,
      "loss": 3.8936,
      "step": 2141
    },
    {
      "epoch": 0.9823434991974318,
      "grad_norm": 0.8279339003607601,
      "learning_rate": 8.395323391094944e-07,
      "loss": 3.7996,
      "step": 2142
    },
    {
      "epoch": 0.982802109607888,
      "grad_norm": 0.8732625515687198,
      "learning_rate": 7.970426548131183e-07,
      "loss": 3.7147,
      "step": 2143
    },
    {
      "epoch": 0.9832607200183444,
      "grad_norm": 0.8923938545880581,
      "learning_rate": 7.556554406908389e-07,
      "loss": 4.0501,
      "step": 2144
    },
    {
      "epoch": 0.9837193304288008,
      "grad_norm": 0.8810402402727103,
      "learning_rate": 7.153707881446536e-07,
      "loss": 3.7853,
      "step": 2145
    },
    {
      "epoch": 0.984177940839257,
      "grad_norm": 0.8774799318299431,
      "learning_rate": 6.761887861417293e-07,
      "loss": 3.7989,
      "step": 2146
    },
    {
      "epoch": 0.9846365512497134,
      "grad_norm": 0.8543518534273976,
      "learning_rate": 6.381095212139032e-07,
      "loss": 3.9939,
      "step": 2147
    },
    {
      "epoch": 0.9850951616601696,
      "grad_norm": 0.8681844844604579,
      "learning_rate": 6.011330774577384e-07,
      "loss": 3.8993,
      "step": 2148
    },
    {
      "epoch": 0.985553772070626,
      "grad_norm": 0.8617170668345188,
      "learning_rate": 5.652595365343016e-07,
      "loss": 3.9876,
      "step": 2149
    },
    {
      "epoch": 0.9860123824810824,
      "grad_norm": 0.8431471513232357,
      "learning_rate": 5.304889776688859e-07,
      "loss": 3.8941,
      "step": 2150
    },
    {
      "epoch": 0.9864709928915386,
      "grad_norm": 0.8378679663144202,
      "learning_rate": 4.968214776508994e-07,
      "loss": 3.7839,
      "step": 2151
    },
    {
      "epoch": 0.986929603301995,
      "grad_norm": 0.835175997103173,
      "learning_rate": 4.6425711083375454e-07,
      "loss": 3.8013,
      "step": 2152
    },
    {
      "epoch": 0.9873882137124512,
      "grad_norm": 0.8549666523429778,
      "learning_rate": 4.3279594913447906e-07,
      "loss": 3.7675,
      "step": 2153
    },
    {
      "epoch": 0.9878468241229076,
      "grad_norm": 0.8631633752508621,
      "learning_rate": 4.02438062033883e-07,
      "loss": 3.797,
      "step": 2154
    },
    {
      "epoch": 0.988305434533364,
      "grad_norm": 0.841692691250541,
      "learning_rate": 3.7318351657616987e-07,
      "loss": 3.944,
      "step": 2155
    },
    {
      "epoch": 0.9887640449438202,
      "grad_norm": 0.843999669545464,
      "learning_rate": 3.4503237736882573e-07,
      "loss": 3.8504,
      "step": 2156
    },
    {
      "epoch": 0.9892226553542766,
      "grad_norm": 0.8520113864852823,
      "learning_rate": 3.179847065825081e-07,
      "loss": 3.888,
      "step": 2157
    },
    {
      "epoch": 0.9896812657647328,
      "grad_norm": 0.8192944817404859,
      "learning_rate": 2.9204056395104594e-07,
      "loss": 3.6924,
      "step": 2158
    },
    {
      "epoch": 0.9901398761751892,
      "grad_norm": 0.8691277106536485,
      "learning_rate": 2.672000067709956e-07,
      "loss": 4.0074,
      "step": 2159
    },
    {
      "epoch": 0.9905984865856455,
      "grad_norm": 0.88603385262786,
      "learning_rate": 2.4346308990175204e-07,
      "loss": 3.9212,
      "step": 2160
    },
    {
      "epoch": 0.9910570969961018,
      "grad_norm": 0.8248551110421741,
      "learning_rate": 2.208298657653818e-07,
      "loss": 3.8716,
      "step": 2161
    },
    {
      "epoch": 0.9915157074065581,
      "grad_norm": 0.8691477660089372,
      "learning_rate": 1.9930038434645692e-07,
      "loss": 3.8655,
      "step": 2162
    },
    {
      "epoch": 0.9919743178170144,
      "grad_norm": 0.8265969454487336,
      "learning_rate": 1.7887469319205484e-07,
      "loss": 3.9797,
      "step": 2163
    },
    {
      "epoch": 0.9924329282274708,
      "grad_norm": 0.8491536773629087,
      "learning_rate": 1.5955283741142523e-07,
      "loss": 3.9033,
      "step": 2164
    },
    {
      "epoch": 0.9928915386379271,
      "grad_norm": 0.88530382151361,
      "learning_rate": 1.4133485967615655e-07,
      "loss": 4.0326,
      "step": 2165
    },
    {
      "epoch": 0.9933501490483834,
      "grad_norm": 0.8500286070965702,
      "learning_rate": 1.2422080021995407e-07,
      "loss": 3.7687,
      "step": 2166
    },
    {
      "epoch": 0.9938087594588397,
      "grad_norm": 0.8448340259397703,
      "learning_rate": 1.082106968385288e-07,
      "loss": 4.0198,
      "step": 2167
    },
    {
      "epoch": 0.994267369869296,
      "grad_norm": 0.838664810384553,
      "learning_rate": 9.330458488959748e-08,
      "loss": 3.82,
      "step": 2168
    },
    {
      "epoch": 0.9947259802797523,
      "grad_norm": 0.8550643280662361,
      "learning_rate": 7.950249729271608e-08,
      "loss": 3.8989,
      "step": 2169
    },
    {
      "epoch": 0.9951845906902087,
      "grad_norm": 0.8277100454610066,
      "learning_rate": 6.680446452922429e-08,
      "loss": 3.8186,
      "step": 2170
    },
    {
      "epoch": 0.995643201100665,
      "grad_norm": 0.8657962170226962,
      "learning_rate": 5.521051464230098e-08,
      "loss": 4.0412,
      "step": 2171
    },
    {
      "epoch": 0.9961018115111213,
      "grad_norm": 0.8285271957291828,
      "learning_rate": 4.4720673236631206e-08,
      "loss": 3.8734,
      "step": 2172
    },
    {
      "epoch": 0.9965604219215776,
      "grad_norm": 0.8228909123729956,
      "learning_rate": 3.53349634786837e-08,
      "loss": 4.0032,
      "step": 2173
    },
    {
      "epoch": 0.9970190323320339,
      "grad_norm": 0.8332227019421334,
      "learning_rate": 2.7053406096433365e-08,
      "loss": 3.8867,
      "step": 2174
    },
    {
      "epoch": 0.9974776427424903,
      "grad_norm": 0.8451164781727469,
      "learning_rate": 1.987601937930572e-08,
      "loss": 3.8538,
      "step": 2175
    },
    {
      "epoch": 0.9979362531529465,
      "grad_norm": 0.8431323896012494,
      "learning_rate": 1.3802819178398984e-08,
      "loss": 3.7884,
      "step": 2176
    },
    {
      "epoch": 0.9983948635634029,
      "grad_norm": 0.8168501109366243,
      "learning_rate": 8.833818906039959e-09,
      "loss": 3.9006,
      "step": 2177
    },
    {
      "epoch": 0.9988534739738592,
      "grad_norm": 0.8148284125804769,
      "learning_rate": 4.969029536061598e-09,
      "loss": 3.9347,
      "step": 2178
    },
    {
      "epoch": 0.9993120843843155,
      "grad_norm": 0.8422059107556905,
      "learning_rate": 2.2084596038030037e-09,
      "loss": 3.9078,
      "step": 2179
    },
    {
      "epoch": 0.9997706947947719,
      "grad_norm": 0.8498093758780108,
      "learning_rate": 5.521152057763601e-10,
      "loss": 3.7544,
      "step": 2180
    },
    {
      "epoch": 0.9997706947947719,
      "step": 2180,
      "total_flos": 4.23221669785292e+18,
      "train_loss": 4.302980575867749,
      "train_runtime": 25712.4659,
      "train_samples_per_second": 21.707,
      "train_steps_per_second": 0.085
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 2180,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 24000,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 4.23221669785292e+18,
  "train_batch_size": 32,
  "trial_name": null,
  "trial_params": null
}