{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.9724310776942353,
  "eval_steps": 500,
  "global_step": 147,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.020050125313283207,
      "grad_norm": 6.100090649437154,
      "learning_rate": 5.333333333333334e-06,
      "loss": 1.0088,
      "step": 1
    },
    {
      "epoch": 0.040100250626566414,
      "grad_norm": 6.165069480911681,
      "learning_rate": 1.0666666666666667e-05,
      "loss": 1.0202,
      "step": 2
    },
    {
      "epoch": 0.06015037593984962,
      "grad_norm": 4.472412116195948,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.9496,
      "step": 3
    },
    {
      "epoch": 0.08020050125313283,
      "grad_norm": 4.963164130948997,
      "learning_rate": 2.1333333333333335e-05,
      "loss": 0.9406,
      "step": 4
    },
    {
      "epoch": 0.10025062656641603,
      "grad_norm": 4.041819143127636,
      "learning_rate": 2.6666666666666667e-05,
      "loss": 0.8794,
      "step": 5
    },
    {
      "epoch": 0.12030075187969924,
      "grad_norm": 3.719380540370943,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.9034,
      "step": 6
    },
    {
      "epoch": 0.14035087719298245,
      "grad_norm": 2.3309985890149765,
      "learning_rate": 3.733333333333334e-05,
      "loss": 0.8102,
      "step": 7
    },
    {
      "epoch": 0.16040100250626566,
      "grad_norm": 2.2467101719324454,
      "learning_rate": 4.266666666666667e-05,
      "loss": 0.786,
      "step": 8
    },
    {
      "epoch": 0.18045112781954886,
      "grad_norm": 2.8539093369835227,
      "learning_rate": 4.8e-05,
      "loss": 0.7531,
      "step": 9
    },
    {
      "epoch": 0.20050125313283207,
      "grad_norm": 2.0083048527276404,
      "learning_rate": 5.333333333333333e-05,
      "loss": 0.7468,
      "step": 10
    },
    {
      "epoch": 0.22055137844611528,
      "grad_norm": 2.165125600420533,
      "learning_rate": 5.8666666666666665e-05,
      "loss": 0.7391,
      "step": 11
    },
    {
      "epoch": 0.24060150375939848,
      "grad_norm": 2.122858303676939,
      "learning_rate": 6.400000000000001e-05,
      "loss": 0.7233,
      "step": 12
    },
    {
      "epoch": 0.2606516290726817,
      "grad_norm": 2.2789276233273843,
      "learning_rate": 6.933333333333334e-05,
      "loss": 0.7286,
      "step": 13
    },
    {
      "epoch": 0.2807017543859649,
      "grad_norm": 2.1616002933600393,
      "learning_rate": 7.466666666666667e-05,
      "loss": 0.7199,
      "step": 14
    },
    {
      "epoch": 0.3007518796992481,
      "grad_norm": 3.2259560404881986,
      "learning_rate": 8e-05,
      "loss": 0.7223,
      "step": 15
    },
    {
      "epoch": 0.3208020050125313,
      "grad_norm": 1.5369895434247212,
      "learning_rate": 7.998867178772517e-05,
      "loss": 0.7063,
      "step": 16
    },
    {
      "epoch": 0.3408521303258145,
      "grad_norm": 3.28272311228189,
      "learning_rate": 7.995469356732033e-05,
      "loss": 0.7174,
      "step": 17
    },
    {
      "epoch": 0.3609022556390977,
      "grad_norm": 2.1611599200246974,
      "learning_rate": 7.989808458441014e-05,
      "loss": 0.6981,
      "step": 18
    },
    {
      "epoch": 0.38095238095238093,
      "grad_norm": 1.9324451852274291,
      "learning_rate": 7.981887690292339e-05,
      "loss": 0.6943,
      "step": 19
    },
    {
      "epoch": 0.40100250626566414,
      "grad_norm": 18.248869277300834,
      "learning_rate": 7.971711538693153e-05,
      "loss": 0.6998,
      "step": 20
    },
    {
      "epoch": 0.42105263157894735,
      "grad_norm": 2.181882821404686,
      "learning_rate": 7.959285767523732e-05,
      "loss": 0.7193,
      "step": 21
    },
    {
      "epoch": 0.44110275689223055,
      "grad_norm": 1.2480300021039636,
      "learning_rate": 7.944617414872747e-05,
      "loss": 0.6843,
      "step": 22
    },
    {
      "epoch": 0.46115288220551376,
      "grad_norm": 2.3412336103636693,
      "learning_rate": 7.927714789050826e-05,
      "loss": 0.7089,
      "step": 23
    },
    {
      "epoch": 0.48120300751879697,
      "grad_norm": 1.9642689883848843,
      "learning_rate": 7.908587463884638e-05,
      "loss": 0.6787,
      "step": 24
    },
    {
      "epoch": 0.5012531328320802,
      "grad_norm": 1.3434579333848315,
      "learning_rate": 7.887246273294167e-05,
      "loss": 0.6773,
      "step": 25
    },
    {
      "epoch": 0.5213032581453634,
      "grad_norm": 1.440602627293297,
      "learning_rate": 7.863703305156273e-05,
      "loss": 0.6792,
      "step": 26
    },
    {
      "epoch": 0.5413533834586466,
      "grad_norm": 1.3175913095028617,
      "learning_rate": 7.837971894457991e-05,
      "loss": 0.6654,
      "step": 27
    },
    {
      "epoch": 0.5614035087719298,
      "grad_norm": 1.3198703652450279,
      "learning_rate": 7.810066615743443e-05,
      "loss": 0.6524,
      "step": 28
    },
    {
      "epoch": 0.581453634085213,
      "grad_norm": 0.781557000061513,
      "learning_rate": 7.780003274858674e-05,
      "loss": 0.6573,
      "step": 29
    },
    {
      "epoch": 0.6015037593984962,
      "grad_norm": 1.1674794127969026,
      "learning_rate": 7.747798899999048e-05,
      "loss": 0.6664,
      "step": 30
    },
    {
      "epoch": 0.6215538847117794,
      "grad_norm": 1.623871556077205,
      "learning_rate": 7.71347173206429e-05,
      "loss": 0.6722,
      "step": 31
    },
    {
      "epoch": 0.6416040100250626,
      "grad_norm": 1.1713900589893373,
      "learning_rate": 7.677041214326663e-05,
      "loss": 0.6403,
      "step": 32
    },
    {
      "epoch": 0.6616541353383458,
      "grad_norm": 1.1207108511375592,
      "learning_rate": 7.638527981418075e-05,
      "loss": 0.6427,
      "step": 33
    },
    {
      "epoch": 0.681704260651629,
      "grad_norm": 1.6390101502728163,
      "learning_rate": 7.597953847642413e-05,
      "loss": 0.6451,
      "step": 34
    },
    {
      "epoch": 0.7017543859649122,
      "grad_norm": 0.9703477640239104,
      "learning_rate": 7.555341794619695e-05,
      "loss": 0.6371,
      "step": 35
    },
    {
      "epoch": 0.7218045112781954,
      "grad_norm": 1.888557130236824,
      "learning_rate": 7.510715958269023e-05,
      "loss": 0.6385,
      "step": 36
    },
    {
      "epoch": 0.7418546365914787,
      "grad_norm": 1.4709265925596289,
      "learning_rate": 7.464101615137756e-05,
      "loss": 0.6468,
      "step": 37
    },
    {
      "epoch": 0.7619047619047619,
      "grad_norm": 1.3340842858015245,
      "learning_rate": 7.415525168084593e-05,
      "loss": 0.636,
      "step": 38
    },
    {
      "epoch": 0.7819548872180451,
      "grad_norm": 1.2331491459930695,
      "learning_rate": 7.365014131324725e-05,
      "loss": 0.6423,
      "step": 39
    },
    {
      "epoch": 0.8020050125313283,
      "grad_norm": 1.0459942770103978,
      "learning_rate": 7.312597114845483e-05,
      "loss": 0.6405,
      "step": 40
    },
    {
      "epoch": 0.8220551378446115,
      "grad_norm": 0.9742677496779505,
      "learning_rate": 7.258303808201343e-05,
      "loss": 0.619,
      "step": 41
    },
    {
      "epoch": 0.8421052631578947,
      "grad_norm": 0.9126761857775636,
      "learning_rate": 7.202164963697442e-05,
      "loss": 0.6237,
      "step": 42
    },
    {
      "epoch": 0.8621553884711779,
      "grad_norm": 1.1034224988906396,
      "learning_rate": 7.144212378971151e-05,
      "loss": 0.6126,
      "step": 43
    },
    {
      "epoch": 0.8822055137844611,
      "grad_norm": 0.9050299139188852,
      "learning_rate": 7.084478878981552e-05,
      "loss": 0.6199,
      "step": 44
    },
    {
      "epoch": 0.9022556390977443,
      "grad_norm": 0.9484346415934313,
      "learning_rate": 7.022998297417034e-05,
      "loss": 0.6242,
      "step": 45
    },
    {
      "epoch": 0.9223057644110275,
      "grad_norm": 0.6761984153684619,
      "learning_rate": 6.959805457531536e-05,
      "loss": 0.6271,
      "step": 46
    },
    {
      "epoch": 0.9423558897243107,
      "grad_norm": 0.7734163089859195,
      "learning_rate": 6.89493615242028e-05,
      "loss": 0.6057,
      "step": 47
    },
    {
      "epoch": 0.9624060150375939,
      "grad_norm": 0.7126127374462361,
      "learning_rate": 6.828427124746191e-05,
      "loss": 0.6143,
      "step": 48
    },
    {
      "epoch": 0.9824561403508771,
      "grad_norm": 0.7040138292374798,
      "learning_rate": 6.760316045928449e-05,
      "loss": 0.5971,
      "step": 49
    },
    {
      "epoch": 1.0150375939849625,
      "grad_norm": 1.0358340840041869,
      "learning_rate": 6.690641494805011e-05,
      "loss": 1.0623,
      "step": 50
    },
    {
      "epoch": 1.0350877192982457,
      "grad_norm": 0.975942175036343,
      "learning_rate": 6.619442935781141e-05,
      "loss": 0.5949,
      "step": 51
    },
    {
      "epoch": 1.055137844611529,
      "grad_norm": 1.3817333278949258,
      "learning_rate": 6.546760696476354e-05,
      "loss": 0.5965,
      "step": 52
    },
    {
      "epoch": 1.0751879699248121,
      "grad_norm": 0.5594198623207998,
      "learning_rate": 6.472635944882421e-05,
      "loss": 0.5817,
      "step": 53
    },
    {
      "epoch": 1.0952380952380953,
      "grad_norm": 1.3502604906380835,
      "learning_rate": 6.397110666045388e-05,
      "loss": 0.5936,
      "step": 54
    },
    {
      "epoch": 1.1152882205513786,
      "grad_norm": 0.662382371248644,
      "learning_rate": 6.320227638284793e-05,
      "loss": 0.597,
      "step": 55
    },
    {
      "epoch": 1.1353383458646618,
      "grad_norm": 1.0092828128239282,
      "learning_rate": 6.242030408963576e-05,
      "loss": 0.5895,
      "step": 56
    },
    {
      "epoch": 1.155388471177945,
      "grad_norm": 0.8148251028605993,
      "learning_rate": 6.162563269822391e-05,
      "loss": 0.5796,
      "step": 57
    },
    {
      "epoch": 1.1754385964912282,
      "grad_norm": 0.7504641683568708,
      "learning_rate": 6.0818712318922894e-05,
      "loss": 0.5756,
      "step": 58
    },
    {
      "epoch": 1.1954887218045114,
      "grad_norm": 0.7075215261142646,
      "learning_rate": 6.000000000000001e-05,
      "loss": 0.5899,
      "step": 59
    },
    {
      "epoch": 1.2155388471177946,
      "grad_norm": 0.616627011160271,
      "learning_rate": 5.916995946880228e-05,
      "loss": 0.5756,
      "step": 60
    },
    {
      "epoch": 1.2355889724310778,
      "grad_norm": 0.6224665456566449,
      "learning_rate": 5.832906086909642e-05,
      "loss": 0.5717,
      "step": 61
    },
    {
      "epoch": 1.255639097744361,
      "grad_norm": 0.5410026994188678,
      "learning_rate": 5.747778049477438e-05,
      "loss": 0.5719,
      "step": 62
    },
    {
      "epoch": 1.2756892230576442,
      "grad_norm": 0.5138816389771951,
      "learning_rate": 5.661660052007547e-05,
      "loss": 0.5767,
      "step": 63
    },
    {
      "epoch": 1.2957393483709274,
      "grad_norm": 0.538651466379072,
      "learning_rate": 5.574600872647766e-05,
      "loss": 0.5754,
      "step": 64
    },
    {
      "epoch": 1.3157894736842106,
      "grad_norm": 0.39832101327162095,
      "learning_rate": 5.48664982264131e-05,
      "loss": 0.5806,
      "step": 65
    },
    {
      "epoch": 1.3358395989974938,
      "grad_norm": 0.46665463498566045,
      "learning_rate": 5.397856718396394e-05,
      "loss": 0.5622,
      "step": 66
    },
    {
      "epoch": 1.355889724310777,
      "grad_norm": 0.33596753910393834,
      "learning_rate": 5.3082718532696874e-05,
      "loss": 0.5635,
      "step": 67
    },
    {
      "epoch": 1.3759398496240602,
      "grad_norm": 0.36330721064185245,
      "learning_rate": 5.217945969079629e-05,
      "loss": 0.5728,
      "step": 68
    },
    {
      "epoch": 1.3959899749373434,
      "grad_norm": 0.31668248946183414,
      "learning_rate": 5.1269302273657195e-05,
      "loss": 0.5829,
      "step": 69
    },
    {
      "epoch": 1.4160401002506267,
      "grad_norm": 0.3108858835369522,
      "learning_rate": 5.0352761804100835e-05,
      "loss": 0.5797,
      "step": 70
    },
    {
      "epoch": 1.4360902255639099,
      "grad_norm": 0.37076038809913947,
      "learning_rate": 4.94303574203771e-05,
      "loss": 0.5678,
      "step": 71
    },
    {
      "epoch": 1.456140350877193,
      "grad_norm": 0.27669921803024705,
      "learning_rate": 4.8502611582119065e-05,
      "loss": 0.5644,
      "step": 72
    },
    {
      "epoch": 1.4761904761904763,
      "grad_norm": 0.3206960394723747,
      "learning_rate": 4.7570049774416414e-05,
      "loss": 0.5696,
      "step": 73
    },
    {
      "epoch": 1.4962406015037595,
      "grad_norm": 0.35836009606291436,
      "learning_rate": 4.663320021017497e-05,
      "loss": 0.5574,
      "step": 74
    },
    {
      "epoch": 1.5162907268170427,
      "grad_norm": 0.26025491223433994,
      "learning_rate": 4.5692593530931416e-05,
      "loss": 0.5683,
      "step": 75
    },
    {
      "epoch": 1.536340852130326,
      "grad_norm": 0.25457026883587025,
      "learning_rate": 4.474876250629221e-05,
      "loss": 0.565,
      "step": 76
    },
    {
      "epoch": 1.556390977443609,
      "grad_norm": 0.32769122481932667,
      "learning_rate": 4.38022417321673e-05,
      "loss": 0.5641,
      "step": 77
    },
    {
      "epoch": 1.5764411027568923,
      "grad_norm": 0.22460593199796938,
      "learning_rate": 4.2853567327969296e-05,
      "loss": 0.557,
      "step": 78
    },
    {
      "epoch": 1.5964912280701755,
      "grad_norm": 0.21753199007833202,
      "learning_rate": 4.19032766329497e-05,
      "loss": 0.5578,
      "step": 79
    },
    {
      "epoch": 1.6165413533834587,
      "grad_norm": 0.23083331300054297,
      "learning_rate": 4.0951907901844296e-05,
      "loss": 0.5622,
      "step": 80
    },
    {
      "epoch": 1.636591478696742,
      "grad_norm": 0.2564157643733938,
      "learning_rate": 4e-05,
      "loss": 0.5657,
      "step": 81
    },
    {
      "epoch": 1.6566416040100251,
      "grad_norm": 0.27813133557097336,
      "learning_rate": 3.904809209815571e-05,
      "loss": 0.5603,
      "step": 82
    },
    {
      "epoch": 1.6766917293233083,
      "grad_norm": 0.2038258932695245,
      "learning_rate": 3.809672336705031e-05,
      "loss": 0.5572,
      "step": 83
    },
    {
      "epoch": 1.6967418546365916,
      "grad_norm": 0.29155489774909854,
      "learning_rate": 3.714643267203071e-05,
      "loss": 0.5544,
      "step": 84
    },
    {
      "epoch": 1.7167919799498748,
      "grad_norm": 0.16516627488221486,
      "learning_rate": 3.6197758267832705e-05,
      "loss": 0.5584,
      "step": 85
    },
    {
      "epoch": 1.736842105263158,
      "grad_norm": 0.29249607068570543,
      "learning_rate": 3.5251237493707804e-05,
      "loss": 0.5677,
      "step": 86
    },
    {
      "epoch": 1.7568922305764412,
      "grad_norm": 0.19265408478071472,
      "learning_rate": 3.4307406469068604e-05,
      "loss": 0.5632,
      "step": 87
    },
    {
      "epoch": 1.7769423558897244,
      "grad_norm": 0.21254433897953356,
      "learning_rate": 3.3366799789825044e-05,
      "loss": 0.5512,
      "step": 88
    },
    {
      "epoch": 1.7969924812030076,
      "grad_norm": 0.2170294921675785,
      "learning_rate": 3.2429950225583606e-05,
      "loss": 0.5493,
      "step": 89
    },
    {
      "epoch": 1.8170426065162908,
      "grad_norm": 0.17517457106296594,
      "learning_rate": 3.1497388417880935e-05,
      "loss": 0.5522,
      "step": 90
    },
    {
      "epoch": 1.837092731829574,
      "grad_norm": 0.21184436677480925,
      "learning_rate": 3.0569642579622905e-05,
      "loss": 0.5533,
      "step": 91
    },
    {
      "epoch": 1.8571428571428572,
      "grad_norm": 0.1566054507779003,
      "learning_rate": 2.9647238195899168e-05,
      "loss": 0.538,
      "step": 92
    },
    {
      "epoch": 1.8771929824561404,
      "grad_norm": 0.19291518917014827,
      "learning_rate": 2.873069772634281e-05,
      "loss": 0.5613,
      "step": 93
    },
    {
      "epoch": 1.8972431077694236,
      "grad_norm": 0.14403155629811726,
      "learning_rate": 2.7820540309203728e-05,
      "loss": 0.5561,
      "step": 94
    },
    {
      "epoch": 1.9172932330827068,
      "grad_norm": 0.18577775134211605,
      "learning_rate": 2.691728146730314e-05,
      "loss": 0.5619,
      "step": 95
    },
    {
      "epoch": 1.93734335839599,
      "grad_norm": 0.17655395819783837,
      "learning_rate": 2.6021432816036073e-05,
      "loss": 0.557,
      "step": 96
    },
    {
      "epoch": 1.9573934837092732,
      "grad_norm": 0.15678742149650277,
      "learning_rate": 2.5133501773586905e-05,
      "loss": 0.55,
      "step": 97
    },
    {
      "epoch": 1.9774436090225564,
      "grad_norm": 0.147011080540491,
      "learning_rate": 2.425399127352235e-05,
      "loss": 0.5615,
      "step": 98
    },
    {
      "epoch": 2.0100250626566414,
      "grad_norm": 0.2896327588272941,
      "learning_rate": 2.338339947992455e-05,
      "loss": 0.984,
      "step": 99
    },
    {
      "epoch": 2.030075187969925,
      "grad_norm": 0.17749955956373084,
      "learning_rate": 2.2522219505225627e-05,
      "loss": 0.5472,
      "step": 100
    },
    {
      "epoch": 2.050125313283208,
      "grad_norm": 0.18960883016778624,
      "learning_rate": 2.1670939130903585e-05,
      "loss": 0.5246,
      "step": 101
    },
    {
      "epoch": 2.0701754385964914,
      "grad_norm": 0.19580310466277323,
      "learning_rate": 2.0830040531197744e-05,
      "loss": 0.5333,
      "step": 102
    },
    {
      "epoch": 2.090225563909774,
      "grad_norm": 0.17002438913970971,
      "learning_rate": 2.0000000000000012e-05,
      "loss": 0.5232,
      "step": 103
    },
    {
      "epoch": 2.110275689223058,
      "grad_norm": 0.23635746246402964,
      "learning_rate": 1.9181287681077116e-05,
      "loss": 0.5299,
      "step": 104
    },
    {
      "epoch": 2.1303258145363406,
      "grad_norm": 0.15533319903966208,
      "learning_rate": 1.8374367301776112e-05,
      "loss": 0.5193,
      "step": 105
    },
    {
      "epoch": 2.1503759398496243,
      "grad_norm": 0.2082307249817901,
      "learning_rate": 1.7579695910364235e-05,
      "loss": 0.5342,
      "step": 106
    },
    {
      "epoch": 2.170426065162907,
      "grad_norm": 0.14435259557657618,
      "learning_rate": 1.679772361715208e-05,
      "loss": 0.5361,
      "step": 107
    },
    {
      "epoch": 2.1904761904761907,
      "grad_norm": 0.20712238757176837,
      "learning_rate": 1.6028893339546122e-05,
      "loss": 0.5331,
      "step": 108
    },
    {
      "epoch": 2.2105263157894735,
      "grad_norm": 0.13522587328433971,
      "learning_rate": 1.527364055117579e-05,
      "loss": 0.5329,
      "step": 109
    },
    {
      "epoch": 2.230576441102757,
      "grad_norm": 0.1593070945975799,
      "learning_rate": 1.4532393035236477e-05,
      "loss": 0.5323,
      "step": 110
    },
    {
      "epoch": 2.25062656641604,
      "grad_norm": 0.1461967123983933,
      "learning_rate": 1.3805570642188602e-05,
      "loss": 0.5162,
      "step": 111
    },
    {
      "epoch": 2.2706766917293235,
      "grad_norm": 0.11139864548024211,
      "learning_rate": 1.30935850519499e-05,
      "loss": 0.5258,
      "step": 112
    },
    {
      "epoch": 2.2907268170426063,
      "grad_norm": 0.1459949648258527,
      "learning_rate": 1.2396839540715528e-05,
      "loss": 0.5249,
      "step": 113
    },
    {
      "epoch": 2.31077694235589,
      "grad_norm": 0.1134072731915256,
      "learning_rate": 1.1715728752538103e-05,
      "loss": 0.5283,
      "step": 114
    },
    {
      "epoch": 2.3308270676691727,
      "grad_norm": 0.10061797227153219,
      "learning_rate": 1.1050638475797193e-05,
      "loss": 0.5264,
      "step": 115
    },
    {
      "epoch": 2.3508771929824563,
      "grad_norm": 0.11967704006720019,
      "learning_rate": 1.0401945424684653e-05,
      "loss": 0.5258,
      "step": 116
    },
    {
      "epoch": 2.370927318295739,
      "grad_norm": 0.10081258865949869,
      "learning_rate": 9.770017025829675e-06,
      "loss": 0.5125,
      "step": 117
    },
    {
      "epoch": 2.3909774436090228,
      "grad_norm": 0.0910736424605034,
      "learning_rate": 9.155211210184495e-06,
      "loss": 0.5215,
      "step": 118
    },
    {
      "epoch": 2.4110275689223055,
      "grad_norm": 0.10320752546874508,
      "learning_rate": 8.55787621028851e-06,
      "loss": 0.5162,
      "step": 119
    },
    {
      "epoch": 2.431077694235589,
      "grad_norm": 0.09217800293365436,
      "learning_rate": 7.978350363025588e-06,
      "loss": 0.5343,
      "step": 120
    },
    {
      "epoch": 2.451127819548872,
      "grad_norm": 0.08490382451555073,
      "learning_rate": 7.416961917986572e-06,
      "loss": 0.5219,
      "step": 121
    },
    {
      "epoch": 2.4711779448621556,
      "grad_norm": 0.08656662969137564,
      "learning_rate": 6.874028851545174e-06,
      "loss": 0.5212,
      "step": 122
    },
    {
      "epoch": 2.4912280701754383,
      "grad_norm": 0.09050314726798651,
      "learning_rate": 6.349858686752748e-06,
      "loss": 0.5328,
      "step": 123
    },
    {
      "epoch": 2.511278195488722,
      "grad_norm": 0.0824254065764698,
      "learning_rate": 5.8447483191540784e-06,
      "loss": 0.5282,
      "step": 124
    },
    {
      "epoch": 2.5313283208020048,
      "grad_norm": 0.09004753737634522,
      "learning_rate": 5.358983848622452e-06,
      "loss": 0.5291,
      "step": 125
    },
    {
      "epoch": 2.5513784461152884,
      "grad_norm": 0.08491851685766005,
      "learning_rate": 4.892840417309775e-06,
      "loss": 0.5174,
      "step": 126
    },
    {
      "epoch": 2.571428571428571,
      "grad_norm": 0.08667374099676277,
      "learning_rate": 4.446582053803066e-06,
      "loss": 0.5269,
      "step": 127
    },
    {
      "epoch": 2.591478696741855,
      "grad_norm": 0.08634185611814611,
      "learning_rate": 4.020461523575873e-06,
      "loss": 0.5404,
      "step": 128
    },
    {
      "epoch": 2.6115288220551376,
      "grad_norm": 0.08142061930820212,
      "learning_rate": 3.6147201858192627e-06,
      "loss": 0.5297,
      "step": 129
    },
    {
      "epoch": 2.6315789473684212,
      "grad_norm": 0.08234802409772482,
      "learning_rate": 3.2295878567333784e-06,
      "loss": 0.5347,
      "step": 130
    },
    {
      "epoch": 2.651629072681704,
      "grad_norm": 0.0850700303547024,
      "learning_rate": 2.8652826793570975e-06,
      "loss": 0.5309,
      "step": 131
    },
    {
      "epoch": 2.6716791979949877,
      "grad_norm": 0.08576766617100663,
      "learning_rate": 2.5220110000095366e-06,
      "loss": 0.529,
      "step": 132
    },
    {
      "epoch": 2.6917293233082704,
      "grad_norm": 0.08621611962612956,
      "learning_rate": 2.199967251413262e-06,
      "loss": 0.526,
      "step": 133
    },
    {
      "epoch": 2.711779448621554,
      "grad_norm": 0.07973648331165978,
      "learning_rate": 1.8993338425655805e-06,
      "loss": 0.5291,
      "step": 134
    },
    {
      "epoch": 2.731829573934837,
      "grad_norm": 0.07607751613838545,
      "learning_rate": 1.6202810554201099e-06,
      "loss": 0.5287,
      "step": 135
    },
    {
      "epoch": 2.7518796992481205,
      "grad_norm": 0.0764140719381241,
      "learning_rate": 1.3629669484372722e-06,
      "loss": 0.519,
      "step": 136
    },
    {
      "epoch": 2.7719298245614032,
      "grad_norm": 0.07473144353146483,
      "learning_rate": 1.127537267058334e-06,
      "loss": 0.5299,
      "step": 137
    },
    {
      "epoch": 2.791979949874687,
      "grad_norm": 0.07476164790510335,
      "learning_rate": 9.141253611536238e-07,
      "loss": 0.5335,
      "step": 138
    },
    {
      "epoch": 2.8120300751879697,
      "grad_norm": 0.07472989997236351,
      "learning_rate": 7.228521094917318e-07,
      "loss": 0.5156,
      "step": 139
    },
    {
      "epoch": 2.8320802005012533,
      "grad_norm": 0.07763611917758945,
      "learning_rate": 5.538258512725403e-07,
      "loss": 0.528,
      "step": 140
    },
    {
      "epoch": 2.852130325814536,
      "grad_norm": 0.07864532159225761,
      "learning_rate": 4.0714232476269265e-07,
      "loss": 0.5205,
      "step": 141
    },
    {
      "epoch": 2.8721804511278197,
      "grad_norm": 0.07157416039875991,
      "learning_rate": 2.8288461306846817e-07,
      "loss": 0.5251,
      "step": 142
    },
    {
      "epoch": 2.8922305764411025,
      "grad_norm": 0.07874362826484563,
      "learning_rate": 1.8112309707661647e-07,
      "loss": 0.5326,
      "step": 143
    },
    {
      "epoch": 2.912280701754386,
      "grad_norm": 0.07678251046065998,
      "learning_rate": 1.019154155898594e-07,
      "loss": 0.5325,
      "step": 144
    },
    {
      "epoch": 2.932330827067669,
      "grad_norm": 0.07928853233638587,
      "learning_rate": 4.530643267968149e-08,
      "loss": 0.5283,
      "step": 145
    },
    {
      "epoch": 2.9523809523809526,
      "grad_norm": 0.07049261679058248,
      "learning_rate": 1.1328212274839267e-08,
      "loss": 0.5307,
      "step": 146
    },
    {
      "epoch": 2.9724310776942353,
      "grad_norm": 0.07518275904349948,
      "learning_rate": 0.0,
      "loss": 0.5333,
      "step": 147
    },
    {
      "epoch": 2.9724310776942353,
      "step": 147,
      "total_flos": 3.782746824809382e+18,
      "train_loss": 0.6079183743924511,
      "train_runtime": 22677.177,
      "train_samples_per_second": 3.374,
      "train_steps_per_second": 0.006
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 147,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.782746824809382e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}