{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "eval_steps": 500,
  "global_step": 338,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 1.8181818181818183e-06,
      "loss": 1.144,
      "step": 1
    },
    {
      "epoch": 0.01,
      "learning_rate": 3.6363636363636366e-06,
      "loss": 1.1406,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 5.4545454545454545e-06,
      "loss": 1.1239,
      "step": 3
    },
    {
      "epoch": 0.02,
      "learning_rate": 7.272727272727273e-06,
      "loss": 0.8465,
      "step": 4
    },
    {
      "epoch": 0.03,
      "learning_rate": 9.090909090909091e-06,
      "loss": 0.4127,
      "step": 5
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.0909090909090909e-05,
      "loss": 0.2055,
      "step": 6
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.2727272727272728e-05,
      "loss": 0.1968,
      "step": 7
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.4545454545454546e-05,
      "loss": 0.3186,
      "step": 8
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.6363636363636366e-05,
      "loss": 0.2235,
      "step": 9
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.8181818181818182e-05,
      "loss": 0.2093,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 2e-05,
      "loss": 0.1955,
      "step": 11
    },
    {
      "epoch": 0.07,
      "learning_rate": 1.9999538500851633e-05,
      "loss": 0.22,
      "step": 12
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9998154046002822e-05,
      "loss": 0.1701,
      "step": 13
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.9995846763238514e-05,
      "loss": 0.1713,
      "step": 14
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9992616865520515e-05,
      "loss": 0.1572,
      "step": 15
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.9988464650967834e-05,
      "loss": 0.1561,
      "step": 16
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.9983390502829168e-05,
      "loss": 0.1571,
      "step": 17
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9977394889447526e-05,
      "loss": 0.1357,
      "step": 18
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.9970478364217e-05,
      "loss": 0.1586,
      "step": 19
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9962641565531694e-05,
      "loss": 0.1416,
      "step": 20
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.9953885216726788e-05,
      "loss": 0.1712,
      "step": 21
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.994421012601179e-05,
      "loss": 0.147,
      "step": 22
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.9933617186395917e-05,
      "loss": 0.1501,
      "step": 23
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.99221073756057e-05,
      "loss": 0.1286,
      "step": 24
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.990968175599471e-05,
      "loss": 0.1238,
      "step": 25
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.9896341474445526e-05,
      "loss": 0.144,
      "step": 26
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9882087762263857e-05,
      "loss": 0.131,
      "step": 27
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9866921935064907e-05,
      "loss": 0.1301,
      "step": 28
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.985084539265195e-05,
      "loss": 0.1158,
      "step": 29
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.983385961888711e-05,
      "loss": 0.1269,
      "step": 30
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9815966181554412e-05,
      "loss": 0.1138,
      "step": 31
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9797166732215078e-05,
      "loss": 0.1185,
      "step": 32
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.977746300605507e-05,
      "loss": 0.1287,
      "step": 33
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.975685682172497e-05,
      "loss": 0.1131,
      "step": 34
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.973535008117207e-05,
      "loss": 0.103,
      "step": 35
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9712944769464864e-05,
      "loss": 0.117,
      "step": 36
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9689642954609808e-05,
      "loss": 0.1209,
      "step": 37
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9665446787360444e-05,
      "loss": 0.1057,
      "step": 38
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9640358501018885e-05,
      "loss": 0.1293,
      "step": 39
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9614380411229693e-05,
      "loss": 0.123,
      "step": 40
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9587514915766124e-05,
      "loss": 0.1011,
      "step": 41
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9559764494308838e-05,
      "loss": 0.1025,
      "step": 42
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9531131708217005e-05,
      "loss": 0.1164,
      "step": 43
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.950161920029191e-05,
      "loss": 0.1286,
      "step": 44
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9471229694533003e-05,
      "loss": 0.1237,
      "step": 45
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.943996599588649e-05,
      "loss": 0.1082,
      "step": 46
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.940783098998643e-05,
      "loss": 0.1072,
      "step": 47
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.93748276428884e-05,
      "loss": 0.0998,
      "step": 48
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9340959000795707e-05,
      "loss": 0.12,
      "step": 49
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9306228189778255e-05,
      "loss": 0.1007,
      "step": 50
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.927063841548398e-05,
      "loss": 0.0954,
      "step": 51
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9234192962842996e-05,
      "loss": 0.106,
      "step": 52
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9196895195764363e-05,
      "loss": 0.1151,
      "step": 53
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9158748556825637e-05,
      "loss": 0.1041,
      "step": 54
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9119756566955092e-05,
      "loss": 0.1177,
      "step": 55
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.907992282510675e-05,
      "loss": 0.081,
      "step": 56
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.90392510079282e-05,
      "loss": 0.1008,
      "step": 57
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.8997744869421248e-05,
      "loss": 0.1025,
      "step": 58
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.8955408240595396e-05,
      "loss": 0.1213,
      "step": 59
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.891224502911428e-05,
      "loss": 0.1097,
      "step": 60
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.886825921893497e-05,
      "loss": 0.0903,
      "step": 61
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.8823454869940243e-05,
      "loss": 0.0842,
      "step": 62
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.8777836117563894e-05,
      "loss": 0.1116,
      "step": 63
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.873140717240899e-05,
      "loss": 0.1087,
      "step": 64
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.8684172319859258e-05,
      "loss": 0.0894,
      "step": 65
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.863613591968355e-05,
      "loss": 0.1136,
      "step": 66
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.858730240563342e-05,
      "loss": 0.1016,
      "step": 67
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.8537676285033886e-05,
      "loss": 0.099,
      "step": 68
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.848726213836744e-05,
      "loss": 0.0976,
      "step": 69
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.8436064618851225e-05,
      "loss": 0.0952,
      "step": 70
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.838408845200758e-05,
      "loss": 0.0901,
      "step": 71
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.8331338435227838e-05,
      "loss": 0.0824,
      "step": 72
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.8277819437329577e-05,
      "loss": 0.0988,
      "step": 73
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.8223536398107177e-05,
      "loss": 0.0961,
      "step": 74
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.8168494327875918e-05,
      "loss": 0.096,
      "step": 75
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.8112698307009506e-05,
      "loss": 0.089,
      "step": 76
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.8056153485471167e-05,
      "loss": 0.1083,
      "step": 77
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.799886508233829e-05,
      "loss": 0.0913,
      "step": 78
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.7940838385320732e-05,
      "loss": 0.1032,
      "step": 79
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.788207875027274e-05,
      "loss": 0.0825,
      "step": 80
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.7822591600698632e-05,
      "loss": 0.09,
      "step": 81
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.776238242725217e-05,
      "loss": 0.0951,
      "step": 82
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.7701456787229805e-05,
      "loss": 0.0804,
      "step": 83
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.7639820304057745e-05,
      "loss": 0.0846,
      "step": 84
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.7577478666772886e-05,
      "loss": 0.125,
      "step": 85
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.751443762949772e-05,
      "loss": 0.0926,
      "step": 86
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.7450703010909263e-05,
      "loss": 0.1083,
      "step": 87
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.738628069370195e-05,
      "loss": 0.0887,
      "step": 88
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.732117662404469e-05,
      "loss": 0.105,
      "step": 89
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.7255396811032014e-05,
      "loss": 0.1011,
      "step": 90
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.718894732612947e-05,
      "loss": 0.1104,
      "step": 91
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.712183430261319e-05,
      "loss": 0.0963,
      "step": 92
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.7054063935003813e-05,
      "loss": 0.0901,
      "step": 93
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.698564247849473e-05,
      "loss": 0.0903,
      "step": 94
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.691657624837472e-05,
      "loss": 0.0849,
      "step": 95
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.684687161944506e-05,
      "loss": 0.0955,
      "step": 96
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.677653502543113e-05,
      "loss": 0.0969,
      "step": 97
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.6705572958388576e-05,
      "loss": 0.0777,
      "step": 98
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6633991968104095e-05,
      "loss": 0.0783,
      "step": 99
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.6561798661490904e-05,
      "loss": 0.0864,
      "step": 100
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.6488999701978905e-05,
      "loss": 0.0887,
      "step": 101
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.6415601808899658e-05,
      "loss": 0.0776,
      "step": 102
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.63416117568662e-05,
      "loss": 0.0829,
      "step": 103
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.6267036375147728e-05,
      "loss": 0.0693,
      "step": 104
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.619188254703927e-05,
      "loss": 0.0837,
      "step": 105
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.6116157209226356e-05,
      "loss": 0.0977,
      "step": 106
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.6039867351144778e-05,
      "loss": 0.083,
      "step": 107
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.5963020014335437e-05,
      "loss": 0.0781,
      "step": 108
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.588562229179443e-05,
      "loss": 0.0697,
      "step": 109
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.5807681327318372e-05,
      "loss": 0.0794,
      "step": 110
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.5729204314845002e-05,
      "loss": 0.0984,
      "step": 111
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.56501984977892e-05,
      "loss": 0.0914,
      "step": 112
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.557067116837444e-05,
      "loss": 0.076,
      "step": 113
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.5490629666959668e-05,
      "loss": 0.0884,
      "step": 114
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.541008138136183e-05,
      "loss": 0.0852,
      "step": 115
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.5329033746173975e-05,
      "loss": 0.0822,
      "step": 116
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.5247494242079024e-05,
      "loss": 0.0888,
      "step": 117
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5165470395159314e-05,
      "loss": 0.076,
      "step": 118
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.5082969776201948e-05,
      "loss": 0.0895,
      "step": 119
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 0.085,
      "step": 120
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.4916568724649688e-05,
      "loss": 0.0934,
      "step": 121
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.483268365084351e-05,
      "loss": 0.0818,
      "step": 122
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.4748352521159492e-05,
      "loss": 0.0862,
      "step": 123
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.466358311934654e-05,
      "loss": 0.0947,
      "step": 124
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.4578383269606004e-05,
      "loss": 0.0886,
      "step": 125
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.4492760835869504e-05,
      "loss": 0.0813,
      "step": 126
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.4406723721073088e-05,
      "loss": 0.0699,
      "step": 127
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.4320279866427798e-05,
      "loss": 0.0701,
      "step": 128
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.4233437250686695e-05,
      "loss": 0.0822,
      "step": 129
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.4146203889408418e-05,
      "loss": 0.0725,
      "step": 130
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.4058587834217356e-05,
      "loss": 0.0811,
      "step": 131
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.3970597172060482e-05,
      "loss": 0.0693,
      "step": 132
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.3882240024460928e-05,
      "loss": 0.0656,
      "step": 133
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.3793524546768358e-05,
      "loss": 0.0754,
      "step": 134
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.3704458927406261e-05,
      "loss": 0.075,
      "step": 135
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.3615051387116131e-05,
      "loss": 0.0601,
      "step": 136
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.3525310178198707e-05,
      "loss": 0.087,
      "step": 137
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.3435243583752294e-05,
      "loss": 0.0762,
      "step": 138
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.3344859916908206e-05,
      "loss": 0.0751,
      "step": 139
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.325416752006351e-05,
      "loss": 0.0677,
      "step": 140
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.3163174764110985e-05,
      "loss": 0.0707,
      "step": 141
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.3071890047666498e-05,
      "loss": 0.0749,
      "step": 142
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.2980321796293838e-05,
      "loss": 0.1004,
      "step": 143
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.288847846172701e-05,
      "loss": 0.0725,
      "step": 144
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.2796368521090143e-05,
      "loss": 0.0796,
      "step": 145
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.2704000476115079e-05,
      "loss": 0.0638,
      "step": 146
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.2611382852356632e-05,
      "loss": 0.0692,
      "step": 147
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.2518524198405699e-05,
      "loss": 0.0756,
      "step": 148
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.2425433085100224e-05,
      "loss": 0.0829,
      "step": 149
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.233211810473411e-05,
      "loss": 0.0718,
      "step": 150
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.2238587870264152e-05,
      "loss": 0.0769,
      "step": 151
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.2144851014515055e-05,
      "loss": 0.0862,
      "step": 152
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.2050916189382646e-05,
      "loss": 0.077,
      "step": 153
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.1956792065035281e-05,
      "loss": 0.0728,
      "step": 154
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.1862487329113606e-05,
      "loss": 0.078,
      "step": 155
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.1768010685928686e-05,
      "loss": 0.0738,
      "step": 156
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1673370855658592e-05,
      "loss": 0.0729,
      "step": 157
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.1578576573543541e-05,
      "loss": 0.0584,
      "step": 158
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.1483636589079627e-05,
      "loss": 0.0784,
      "step": 159
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1388559665211241e-05,
      "loss": 0.0856,
      "step": 160
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.1293354577522264e-05,
      "loss": 0.0849,
      "step": 161
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.1198030113426074e-05,
      "loss": 0.0707,
      "step": 162
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.1102595071354471e-05,
      "loss": 0.0974,
      "step": 163
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.1007058259945584e-05,
      "loss": 0.0773,
      "step": 164
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.0911428497230834e-05,
      "loss": 0.0682,
      "step": 165
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.0815714609821027e-05,
      "loss": 0.0772,
      "step": 166
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0719925432091671e-05,
      "loss": 0.0852,
      "step": 167
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.0624069805367558e-05,
      "loss": 0.0724,
      "step": 168
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.0528156577106703e-05,
      "loss": 0.0674,
      "step": 169
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.043219460008374e-05,
      "loss": 0.0773,
      "step": 170
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.0336192731572805e-05,
      "loss": 0.0627,
      "step": 171
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.0240159832530007e-05,
      "loss": 0.069,
      "step": 172
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.0144104766775574e-05,
      "loss": 0.0609,
      "step": 173
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.004803640017571e-05,
      "loss": 0.0526,
      "step": 174
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.951963599824294e-06,
      "loss": 0.0666,
      "step": 175
    },
    {
      "epoch": 1.04,
      "learning_rate": 9.855895233224431e-06,
      "loss": 0.0767,
      "step": 176
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.759840167469995e-06,
      "loss": 0.0647,
      "step": 177
    },
    {
      "epoch": 1.05,
      "learning_rate": 9.663807268427197e-06,
      "loss": 0.0529,
      "step": 178
    },
    {
      "epoch": 1.06,
      "learning_rate": 9.56780539991626e-06,
      "loss": 0.0542,
      "step": 179
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.471843422893299e-06,
      "loss": 0.0598,
      "step": 180
    },
    {
      "epoch": 1.07,
      "learning_rate": 9.375930194632447e-06,
      "loss": 0.0537,
      "step": 181
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.28007456790833e-06,
      "loss": 0.0653,
      "step": 182
    },
    {
      "epoch": 1.08,
      "learning_rate": 9.184285390178978e-06,
      "loss": 0.0544,
      "step": 183
    },
    {
      "epoch": 1.09,
      "learning_rate": 9.08857150276917e-06,
      "loss": 0.0579,
      "step": 184
    },
    {
      "epoch": 1.09,
      "learning_rate": 8.992941740054418e-06,
      "loss": 0.0522,
      "step": 185
    },
    {
      "epoch": 1.1,
      "learning_rate": 8.897404928645529e-06,
      "loss": 0.0526,
      "step": 186
    },
    {
      "epoch": 1.11,
      "learning_rate": 8.80196988657393e-06,
      "loss": 0.0563,
      "step": 187
    },
    {
      "epoch": 1.11,
      "learning_rate": 8.706645422477739e-06,
      "loss": 0.0505,
      "step": 188
    },
    {
      "epoch": 1.12,
      "learning_rate": 8.611440334788762e-06,
      "loss": 0.0562,
      "step": 189
    },
    {
      "epoch": 1.12,
      "learning_rate": 8.516363410920376e-06,
      "loss": 0.0468,
      "step": 190
    },
    {
      "epoch": 1.13,
      "learning_rate": 8.42142342645646e-06,
      "loss": 0.0553,
      "step": 191
    },
    {
      "epoch": 1.14,
      "learning_rate": 8.326629144341408e-06,
      "loss": 0.0523,
      "step": 192
    },
    {
      "epoch": 1.14,
      "learning_rate": 8.231989314071318e-06,
      "loss": 0.0523,
      "step": 193
    },
    {
      "epoch": 1.15,
      "learning_rate": 8.137512670886397e-06,
      "loss": 0.0594,
      "step": 194
    },
    {
      "epoch": 1.15,
      "learning_rate": 8.043207934964722e-06,
      "loss": 0.0574,
      "step": 195
    },
    {
      "epoch": 1.16,
      "learning_rate": 7.949083810617358e-06,
      "loss": 0.0615,
      "step": 196
    },
    {
      "epoch": 1.17,
      "learning_rate": 7.855148985484946e-06,
      "loss": 0.0645,
      "step": 197
    },
    {
      "epoch": 1.17,
      "learning_rate": 7.761412129735853e-06,
      "loss": 0.0656,
      "step": 198
    },
    {
      "epoch": 1.18,
      "learning_rate": 7.667881895265895e-06,
      "loss": 0.0623,
      "step": 199
    },
    {
      "epoch": 1.18,
      "learning_rate": 7.574566914899779e-06,
      "loss": 0.0635,
      "step": 200
    },
    {
      "epoch": 1.19,
      "learning_rate": 7.481475801594302e-06,
      "loss": 0.0464,
      "step": 201
    },
    {
      "epoch": 1.2,
      "learning_rate": 7.388617147643371e-06,
      "loss": 0.0387,
      "step": 202
    },
    {
      "epoch": 1.2,
      "learning_rate": 7.295999523884921e-06,
      "loss": 0.0404,
      "step": 203
    },
    {
      "epoch": 1.21,
      "learning_rate": 7.203631478909857e-06,
      "loss": 0.051,
      "step": 204
    },
    {
      "epoch": 1.21,
      "learning_rate": 7.111521538272997e-06,
      "loss": 0.0633,
      "step": 205
    },
    {
      "epoch": 1.22,
      "learning_rate": 7.019678203706164e-06,
      "loss": 0.0585,
      "step": 206
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.928109952333506e-06,
      "loss": 0.0579,
      "step": 207
    },
    {
      "epoch": 1.23,
      "learning_rate": 6.83682523588902e-06,
      "loss": 0.0635,
      "step": 208
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.745832479936492e-06,
      "loss": 0.0859,
      "step": 209
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.655140083091794e-06,
      "loss": 0.0492,
      "step": 210
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.564756416247712e-06,
      "loss": 0.0497,
      "step": 211
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.474689821801295e-06,
      "loss": 0.0575,
      "step": 212
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.384948612883872e-06,
      "loss": 0.0465,
      "step": 213
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.2955410725937405e-06,
      "loss": 0.0533,
      "step": 214
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.206475453231644e-06,
      "loss": 0.0421,
      "step": 215
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.117759975539075e-06,
      "loss": 0.0543,
      "step": 216
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.029402827939519e-06,
      "loss": 0.0706,
      "step": 217
    },
    {
      "epoch": 1.29,
      "learning_rate": 5.941412165782645e-06,
      "loss": 0.0543,
      "step": 218
    },
    {
      "epoch": 1.3,
      "learning_rate": 5.853796110591583e-06,
      "loss": 0.045,
      "step": 219
    },
    {
      "epoch": 1.3,
      "learning_rate": 5.766562749313309e-06,
      "loss": 0.0587,
      "step": 220
    },
    {
      "epoch": 1.31,
      "learning_rate": 5.6797201335722064e-06,
      "loss": 0.0605,
      "step": 221
    },
    {
      "epoch": 1.31,
      "learning_rate": 5.593276278926912e-06,
      "loss": 0.061,
      "step": 222
    },
    {
      "epoch": 1.32,
      "learning_rate": 5.507239164130501e-06,
      "loss": 0.0594,
      "step": 223
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.421616730394e-06,
      "loss": 0.0511,
      "step": 224
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.336416880653461e-06,
      "loss": 0.0626,
      "step": 225
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.251647478840511e-06,
      "loss": 0.048,
      "step": 226
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.167316349156495e-06,
      "loss": 0.0581,
      "step": 227
    },
    {
      "epoch": 1.35,
      "learning_rate": 5.083431275350312e-06,
      "loss": 0.0653,
      "step": 228
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.000000000000003e-06,
      "loss": 0.0541,
      "step": 229
    },
    {
      "epoch": 1.36,
      "learning_rate": 4.917030223798057e-06,
      "loss": 0.0596,
      "step": 230
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.834529604840686e-06,
      "loss": 0.0441,
      "step": 231
    },
    {
      "epoch": 1.37,
      "learning_rate": 4.7525057579209775e-06,
      "loss": 0.065,
      "step": 232
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.670966253826027e-06,
      "loss": 0.0498,
      "step": 233
    },
    {
      "epoch": 1.38,
      "learning_rate": 4.589918618638173e-06,
      "loss": 0.0591,
      "step": 234
    },
    {
      "epoch": 1.39,
      "learning_rate": 4.5093703330403385e-06,
      "loss": 0.0554,
      "step": 235
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.429328831625565e-06,
      "loss": 0.0432,
      "step": 236
    },
    {
      "epoch": 1.4,
      "learning_rate": 4.349801502210801e-06,
      "loss": 0.0526,
      "step": 237
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.270795685155001e-06,
      "loss": 0.0485,
      "step": 238
    },
    {
      "epoch": 1.41,
      "learning_rate": 4.192318672681631e-06,
      "loss": 0.0538,
      "step": 239
    },
    {
      "epoch": 1.42,
      "learning_rate": 4.1143777082055715e-06,
      "loss": 0.0476,
      "step": 240
    },
    {
      "epoch": 1.43,
      "learning_rate": 4.036979985664566e-06,
      "loss": 0.0514,
      "step": 241
    },
    {
      "epoch": 1.43,
      "learning_rate": 3.960132648855226e-06,
      "loss": 0.0615,
      "step": 242
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.883842790773647e-06,
      "loss": 0.0584,
      "step": 243
    },
    {
      "epoch": 1.44,
      "learning_rate": 3.8081174529607346e-06,
      "loss": 0.0443,
      "step": 244
    },
    {
      "epoch": 1.45,
      "learning_rate": 3.732963624852275e-06,
      "loss": 0.0623,
      "step": 245
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.6583882431338047e-06,
      "loss": 0.0575,
      "step": 246
    },
    {
      "epoch": 1.46,
      "learning_rate": 3.584398191100341e-06,
      "loss": 0.0464,
      "step": 247
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.511000298021098e-06,
      "loss": 0.0572,
      "step": 248
    },
    {
      "epoch": 1.47,
      "learning_rate": 3.4382013385090985e-06,
      "loss": 0.0492,
      "step": 249
    },
    {
      "epoch": 1.48,
      "learning_rate": 3.3660080318959043e-06,
      "loss": 0.0528,
      "step": 250
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.2944270416114256e-06,
      "loss": 0.0433,
      "step": 251
    },
    {
      "epoch": 1.49,
      "learning_rate": 3.223464974568874e-06,
      "loss": 0.0768,
      "step": 252
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.153128380554941e-06,
      "loss": 0.0523,
      "step": 253
    },
    {
      "epoch": 1.5,
      "learning_rate": 3.0834237516252817e-06,
      "loss": 0.0575,
      "step": 254
    },
    {
      "epoch": 1.51,
      "learning_rate": 3.0143575215052732e-06,
      "loss": 0.0471,
      "step": 255
    },
    {
      "epoch": 1.51,
      "learning_rate": 2.94593606499619e-06,
      "loss": 0.0572,
      "step": 256
    },
    {
      "epoch": 1.52,
      "learning_rate": 2.878165697386812e-06,
      "loss": 0.0414,
      "step": 257
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.8110526738705345e-06,
      "loss": 0.0537,
      "step": 258
    },
    {
      "epoch": 1.53,
      "learning_rate": 2.7446031889679893e-06,
      "loss": 0.0548,
      "step": 259
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.678823375955314e-06,
      "loss": 0.0488,
      "step": 260
    },
    {
      "epoch": 1.54,
      "learning_rate": 2.6137193062980506e-06,
      "loss": 0.0612,
      "step": 261
    },
    {
      "epoch": 1.55,
      "learning_rate": 2.5492969890907383e-06,
      "loss": 0.0533,
      "step": 262
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.485562370502279e-06,
      "loss": 0.0556,
      "step": 263
    },
    {
      "epoch": 1.56,
      "learning_rate": 2.4225213332271203e-06,
      "loss": 0.0654,
      "step": 264
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.3601796959422585e-06,
      "loss": 0.0382,
      "step": 265
    },
    {
      "epoch": 1.57,
      "learning_rate": 2.2985432127701945e-06,
      "loss": 0.0559,
      "step": 266
    },
    {
      "epoch": 1.58,
      "learning_rate": 2.2376175727478346e-06,
      "loss": 0.0525,
      "step": 267
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.1774083993013715e-06,
      "loss": 0.0491,
      "step": 268
    },
    {
      "epoch": 1.59,
      "learning_rate": 2.1179212497272582e-06,
      "loss": 0.0542,
      "step": 269
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.0591616146792705e-06,
      "loss": 0.05,
      "step": 270
    },
    {
      "epoch": 1.6,
      "learning_rate": 2.0011349176617133e-06,
      "loss": 0.0554,
      "step": 271
    },
    {
      "epoch": 1.61,
      "learning_rate": 1.9438465145288377e-06,
      "loss": 0.0506,
      "step": 272
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.8873016929904942e-06,
      "loss": 0.0505,
      "step": 273
    },
    {
      "epoch": 1.62,
      "learning_rate": 1.8315056721240831e-06,
      "loss": 0.06,
      "step": 274
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.7764636018928249e-06,
      "loss": 0.0557,
      "step": 275
    },
    {
      "epoch": 1.63,
      "learning_rate": 1.722180562670428e-06,
      "loss": 0.0423,
      "step": 276
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.6686615647721638e-06,
      "loss": 0.0521,
      "step": 277
    },
    {
      "epoch": 1.64,
      "learning_rate": 1.6159115479924259e-06,
      "loss": 0.0487,
      "step": 278
    },
    {
      "epoch": 1.65,
      "learning_rate": 1.5639353811487744e-06,
      "loss": 0.0613,
      "step": 279
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.5127378616325606e-06,
      "loss": 0.0592,
      "step": 280
    },
    {
      "epoch": 1.66,
      "learning_rate": 1.462323714966114e-06,
      "loss": 0.0501,
      "step": 281
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.4126975943665844e-06,
      "loss": 0.0523,
      "step": 282
    },
    {
      "epoch": 1.67,
      "learning_rate": 1.3638640803164516e-06,
      "loss": 0.0396,
      "step": 283
    },
    {
      "epoch": 1.68,
      "learning_rate": 1.3158276801407432e-06,
      "loss": 0.0541,
      "step": 284
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.2685928275910142e-06,
      "loss": 0.062,
      "step": 285
    },
    {
      "epoch": 1.69,
      "learning_rate": 1.222163882436107e-06,
      "loss": 0.0471,
      "step": 286
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.1765451300597574e-06,
      "loss": 0.06,
      "step": 287
    },
    {
      "epoch": 1.7,
      "learning_rate": 1.1317407810650372e-06,
      "loss": 0.051,
      "step": 288
    },
    {
      "epoch": 1.71,
      "learning_rate": 1.0877549708857228e-06,
      "loss": 0.0465,
      "step": 289
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.0445917594046073e-06,
      "loss": 0.0503,
      "step": 290
    },
    {
      "epoch": 1.72,
      "learning_rate": 1.0022551305787564e-06,
      "loss": 0.0548,
      "step": 291
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.607489920717983e-07,
      "loss": 0.0457,
      "step": 292
    },
    {
      "epoch": 1.73,
      "learning_rate": 9.200771748932513e-07,
      "loss": 0.0558,
      "step": 293
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.802434330449128e-07,
      "loss": 0.0459,
      "step": 294
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.412514431743657e-07,
      "loss": 0.0595,
      "step": 295
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.031048042356393e-07,
      "loss": 0.0587,
      "step": 296
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.65807037157007e-07,
      "loss": 0.0528,
      "step": 297
    },
    {
      "epoch": 1.76,
      "learning_rate": 7.293615845160196e-07,
      "loss": 0.0528,
      "step": 298
    },
    {
      "epoch": 1.77,
      "learning_rate": 6.937718102217461e-07,
      "loss": 0.0553,
      "step": 299
    },
    {
      "epoch": 1.78,
      "learning_rate": 6.590409992042957e-07,
      "loss": 0.0427,
      "step": 300
    },
    {
      "epoch": 1.78,
      "learning_rate": 6.251723571116031e-07,
      "loss": 0.0533,
      "step": 301
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.921690100135713e-07,
      "loss": 0.0431,
      "step": 302
    },
    {
      "epoch": 1.79,
      "learning_rate": 5.600340041135133e-07,
      "loss": 0.0429,
      "step": 303
    },
    {
      "epoch": 1.8,
      "learning_rate": 5.287703054670012e-07,
      "loss": 0.0376,
      "step": 304
    },
    {
      "epoch": 1.8,
      "learning_rate": 4.983807997080925e-07,
      "loss": 0.0553,
      "step": 305
    },
    {
      "epoch": 1.81,
      "learning_rate": 4.6886829178299676e-07,
      "loss": 0.0522,
      "step": 306
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.402355056911656e-07,
      "loss": 0.0463,
      "step": 307
    },
    {
      "epoch": 1.82,
      "learning_rate": 4.124850842338779e-07,
      "loss": 0.0461,
      "step": 308
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.8561958877030957e-07,
      "loss": 0.047,
      "step": 309
    },
    {
      "epoch": 1.83,
      "learning_rate": 3.5964149898111587e-07,
      "loss": 0.0594,
      "step": 310
    },
    {
      "epoch": 1.84,
      "learning_rate": 3.345532126395579e-07,
      "loss": 0.0475,
      "step": 311
    },
    {
      "epoch": 1.85,
      "learning_rate": 3.1035704539019384e-07,
      "loss": 0.0431,
      "step": 312
    },
    {
      "epoch": 1.85,
      "learning_rate": 2.870552305351382e-07,
      "loss": 0.0471,
      "step": 313
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.646499188279328e-07,
      "loss": 0.0577,
      "step": 314
    },
    {
      "epoch": 1.86,
      "learning_rate": 2.4314317827503375e-07,
      "loss": 0.0387,
      "step": 315
    },
    {
      "epoch": 1.87,
      "learning_rate": 2.2253699394493066e-07,
      "loss": 0.047,
      "step": 316
    },
    {
      "epoch": 1.88,
      "learning_rate": 2.028332677849254e-07,
      "loss": 0.0526,
      "step": 317
    },
    {
      "epoch": 1.88,
      "learning_rate": 1.840338184455881e-07,
      "loss": 0.0459,
      "step": 318
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.6614038111289034e-07,
      "loss": 0.0438,
      "step": 319
    },
    {
      "epoch": 1.89,
      "learning_rate": 1.49154607348051e-07,
      "loss": 0.0442,
      "step": 320
    },
    {
      "epoch": 1.9,
      "learning_rate": 1.330780649350938e-07,
      "loss": 0.0468,
      "step": 321
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.1791223773614635e-07,
      "loss": 0.0577,
      "step": 322
    },
    {
      "epoch": 1.91,
      "learning_rate": 1.0365852555447642e-07,
      "loss": 0.0416,
      "step": 323
    },
    {
      "epoch": 1.92,
      "learning_rate": 9.031824400528854e-08,
      "loss": 0.0551,
      "step": 324
    },
    {
      "epoch": 1.92,
      "learning_rate": 7.789262439430012e-08,
      "loss": 0.0335,
      "step": 325
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.638281360408339e-08,
      "loss": 0.0382,
      "step": 326
    },
    {
      "epoch": 1.93,
      "learning_rate": 5.578987398821345e-08,
      "loss": 0.0406,
      "step": 327
    },
    {
      "epoch": 1.94,
      "learning_rate": 4.6114783273213395e-08,
      "loss": 0.0492,
      "step": 328
    },
    {
      "epoch": 1.95,
      "learning_rate": 3.735843446830867e-08,
      "loss": 0.0495,
      "step": 329
    },
    {
      "epoch": 1.95,
      "learning_rate": 2.9521635783001932e-08,
      "loss": 0.0464,
      "step": 330
    },
    {
      "epoch": 1.96,
      "learning_rate": 2.2605110552477162e-08,
      "loss": 0.0428,
      "step": 331
    },
    {
      "epoch": 1.96,
      "learning_rate": 1.6609497170834154e-08,
      "loss": 0.0326,
      "step": 332
    },
    {
      "epoch": 1.97,
      "learning_rate": 1.1535349032167908e-08,
      "loss": 0.0516,
      "step": 333
    },
    {
      "epoch": 1.98,
      "learning_rate": 7.3831344794872415e-09,
      "loss": 0.0468,
      "step": 334
    },
    {
      "epoch": 1.98,
      "learning_rate": 4.153236761488266e-09,
      "loss": 0.0444,
      "step": 335
    },
    {
      "epoch": 1.99,
      "learning_rate": 1.8459539971804608e-09,
      "loss": 0.0427,
      "step": 336
    },
    {
      "epoch": 1.99,
      "learning_rate": 4.614991483686826e-10,
      "loss": 0.0394,
      "step": 337
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 0.0317,
      "step": 338
    },
    {
      "epoch": 2.0,
      "step": 338,
      "total_flos": 14257366695936.0,
      "train_loss": 0.08990215752604445,
      "train_runtime": 3909.568,
      "train_samples_per_second": 9.664,
      "train_steps_per_second": 0.086
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 338,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 2718,
  "total_flos": 14257366695936.0,
  "train_batch_size": 16,
  "trial_name": null,
  "trial_params": null
}