{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9954545454545456,
  "eval_steps": 500,
  "global_step": 292,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.006818181818181818,
      "grad_norm": 1.8943578004837036,
      "learning_rate": 5.0000000000000004e-08,
      "loss": 0.8676,
      "step": 1
    },
    {
      "epoch": 0.013636363636363636,
      "grad_norm": 7.775813102722168,
      "learning_rate": 1.0000000000000001e-07,
      "loss": 0.8513,
      "step": 2
    },
    {
      "epoch": 0.020454545454545454,
      "grad_norm": 1.768803596496582,
      "learning_rate": 1.5000000000000002e-07,
      "loss": 0.8339,
      "step": 3
    },
    {
      "epoch": 0.02727272727272727,
      "grad_norm": 1.7032263278961182,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 0.8348,
      "step": 4
    },
    {
      "epoch": 0.03409090909090909,
      "grad_norm": 1.8244796991348267,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.8687,
      "step": 5
    },
    {
      "epoch": 0.04090909090909091,
      "grad_norm": 1.7780181169509888,
      "learning_rate": 3.0000000000000004e-07,
      "loss": 0.8113,
      "step": 6
    },
    {
      "epoch": 0.04772727272727273,
      "grad_norm": 1.8362191915512085,
      "learning_rate": 3.5000000000000004e-07,
      "loss": 0.8463,
      "step": 7
    },
    {
      "epoch": 0.05454545454545454,
      "grad_norm": 1.6640979051589966,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 0.8336,
      "step": 8
    },
    {
      "epoch": 0.06136363636363636,
      "grad_norm": 1.8013464212417603,
      "learning_rate": 4.5000000000000003e-07,
      "loss": 0.8572,
      "step": 9
    },
    {
      "epoch": 0.06818181818181818,
      "grad_norm": 1.7237999439239502,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.8501,
      "step": 10
    },
    {
      "epoch": 0.075,
      "grad_norm": 1.989346981048584,
      "learning_rate": 5.5e-07,
      "loss": 0.8615,
      "step": 11
    },
    {
      "epoch": 0.08181818181818182,
      "grad_norm": 2.0312082767486572,
      "learning_rate": 6.000000000000001e-07,
      "loss": 0.858,
      "step": 12
    },
    {
      "epoch": 0.08863636363636364,
      "grad_norm": 1.6921299695968628,
      "learning_rate": 6.5e-07,
      "loss": 0.8577,
      "step": 13
    },
    {
      "epoch": 0.09545454545454546,
      "grad_norm": 2.4207160472869873,
      "learning_rate": 7.000000000000001e-07,
      "loss": 0.8564,
      "step": 14
    },
    {
      "epoch": 0.10227272727272728,
      "grad_norm": 1.7572585344314575,
      "learning_rate": 7.5e-07,
      "loss": 0.8409,
      "step": 15
    },
    {
      "epoch": 0.10909090909090909,
      "grad_norm": 1.6135950088500977,
      "learning_rate": 8.000000000000001e-07,
      "loss": 0.8086,
      "step": 16
    },
    {
      "epoch": 0.1159090909090909,
      "grad_norm": 2.226036787033081,
      "learning_rate": 8.500000000000001e-07,
      "loss": 0.8106,
      "step": 17
    },
    {
      "epoch": 0.12272727272727273,
      "grad_norm": 1.7164231538772583,
      "learning_rate": 9.000000000000001e-07,
      "loss": 0.8494,
      "step": 18
    },
    {
      "epoch": 0.12954545454545455,
      "grad_norm": 1.6572000980377197,
      "learning_rate": 9.500000000000001e-07,
      "loss": 0.8023,
      "step": 19
    },
    {
      "epoch": 0.13636363636363635,
      "grad_norm": 1.5672118663787842,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.787,
      "step": 20
    },
    {
      "epoch": 0.1431818181818182,
      "grad_norm": 1.5080257654190063,
      "learning_rate": 1.0500000000000001e-06,
      "loss": 0.8105,
      "step": 21
    },
    {
      "epoch": 0.15,
      "grad_norm": 1.484892725944519,
      "learning_rate": 1.1e-06,
      "loss": 0.8258,
      "step": 22
    },
    {
      "epoch": 0.15681818181818183,
      "grad_norm": 1.4591134786605835,
      "learning_rate": 1.1500000000000002e-06,
      "loss": 0.8032,
      "step": 23
    },
    {
      "epoch": 0.16363636363636364,
      "grad_norm": 1.5506278276443481,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 0.8236,
      "step": 24
    },
    {
      "epoch": 0.17045454545454544,
      "grad_norm": 1.639350414276123,
      "learning_rate": 1.25e-06,
      "loss": 0.8185,
      "step": 25
    },
    {
      "epoch": 0.17727272727272728,
      "grad_norm": 1.592640995979309,
      "learning_rate": 1.3e-06,
      "loss": 0.7898,
      "step": 26
    },
    {
      "epoch": 0.18409090909090908,
      "grad_norm": 1.5280641317367554,
      "learning_rate": 1.3500000000000002e-06,
      "loss": 0.7731,
      "step": 27
    },
    {
      "epoch": 0.19090909090909092,
      "grad_norm": 1.4015443325042725,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 0.7956,
      "step": 28
    },
    {
      "epoch": 0.19772727272727272,
      "grad_norm": 1.4890056848526,
      "learning_rate": 1.45e-06,
      "loss": 0.7838,
      "step": 29
    },
    {
      "epoch": 0.20454545454545456,
      "grad_norm": 2.4770750999450684,
      "learning_rate": 1.5e-06,
      "loss": 0.7766,
      "step": 30
    },
    {
      "epoch": 0.21136363636363636,
      "grad_norm": 1.5863685607910156,
      "learning_rate": 1.5500000000000002e-06,
      "loss": 0.7353,
      "step": 31
    },
    {
      "epoch": 0.21818181818181817,
      "grad_norm": 1.058114767074585,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.7237,
      "step": 32
    },
    {
      "epoch": 0.225,
      "grad_norm": 1.5869101285934448,
      "learning_rate": 1.6500000000000003e-06,
      "loss": 0.7597,
      "step": 33
    },
    {
      "epoch": 0.2318181818181818,
      "grad_norm": 1.0012413263320923,
      "learning_rate": 1.7000000000000002e-06,
      "loss": 0.7388,
      "step": 34
    },
    {
      "epoch": 0.23863636363636365,
      "grad_norm": 1.118056058883667,
      "learning_rate": 1.75e-06,
      "loss": 0.7786,
      "step": 35
    },
    {
      "epoch": 0.24545454545454545,
      "grad_norm": 1.0648820400238037,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 0.7502,
      "step": 36
    },
    {
      "epoch": 0.25227272727272726,
      "grad_norm": 0.9440382719039917,
      "learning_rate": 1.85e-06,
      "loss": 0.6942,
      "step": 37
    },
    {
      "epoch": 0.2590909090909091,
      "grad_norm": 0.8300187587738037,
      "learning_rate": 1.9000000000000002e-06,
      "loss": 0.7312,
      "step": 38
    },
    {
      "epoch": 0.26590909090909093,
      "grad_norm": 0.8296191692352295,
      "learning_rate": 1.9500000000000004e-06,
      "loss": 0.7461,
      "step": 39
    },
    {
      "epoch": 0.2727272727272727,
      "grad_norm": 0.7797508239746094,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.7234,
      "step": 40
    },
    {
      "epoch": 0.27954545454545454,
      "grad_norm": 0.7335019707679749,
      "learning_rate": 2.05e-06,
      "loss": 0.6927,
      "step": 41
    },
    {
      "epoch": 0.2863636363636364,
      "grad_norm": 0.7062106728553772,
      "learning_rate": 2.1000000000000002e-06,
      "loss": 0.6552,
      "step": 42
    },
    {
      "epoch": 0.29318181818181815,
      "grad_norm": 1.2342087030410767,
      "learning_rate": 2.15e-06,
      "loss": 0.6959,
      "step": 43
    },
    {
      "epoch": 0.3,
      "grad_norm": 0.8967597484588623,
      "learning_rate": 2.2e-06,
      "loss": 0.704,
      "step": 44
    },
    {
      "epoch": 0.3068181818181818,
      "grad_norm": 0.7228404879570007,
      "learning_rate": 2.25e-06,
      "loss": 0.6777,
      "step": 45
    },
    {
      "epoch": 0.31363636363636366,
      "grad_norm": 0.6731488704681396,
      "learning_rate": 2.3000000000000004e-06,
      "loss": 0.6652,
      "step": 46
    },
    {
      "epoch": 0.32045454545454544,
      "grad_norm": 0.71966952085495,
      "learning_rate": 2.35e-06,
      "loss": 0.7281,
      "step": 47
    },
    {
      "epoch": 0.32727272727272727,
      "grad_norm": 0.7057356238365173,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.6476,
      "step": 48
    },
    {
      "epoch": 0.3340909090909091,
      "grad_norm": 0.6346054077148438,
      "learning_rate": 2.4500000000000003e-06,
      "loss": 0.6865,
      "step": 49
    },
    {
      "epoch": 0.3409090909090909,
      "grad_norm": 0.6938223838806152,
      "learning_rate": 2.5e-06,
      "loss": 0.6808,
      "step": 50
    },
    {
      "epoch": 0.3477272727272727,
      "grad_norm": 0.7425184845924377,
      "learning_rate": 2.55e-06,
      "loss": 0.7074,
      "step": 51
    },
    {
      "epoch": 0.35454545454545455,
      "grad_norm": 0.6743818521499634,
      "learning_rate": 2.6e-06,
      "loss": 0.657,
      "step": 52
    },
    {
      "epoch": 0.3613636363636364,
      "grad_norm": 0.6785942316055298,
      "learning_rate": 2.6500000000000005e-06,
      "loss": 0.6768,
      "step": 53
    },
    {
      "epoch": 0.36818181818181817,
      "grad_norm": 0.586974024772644,
      "learning_rate": 2.7000000000000004e-06,
      "loss": 0.651,
      "step": 54
    },
    {
      "epoch": 0.375,
      "grad_norm": 0.5727331042289734,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.669,
      "step": 55
    },
    {
      "epoch": 0.38181818181818183,
      "grad_norm": 0.7333543300628662,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 0.6571,
      "step": 56
    },
    {
      "epoch": 0.3886363636363636,
      "grad_norm": 0.5710961818695068,
      "learning_rate": 2.85e-06,
      "loss": 0.6475,
      "step": 57
    },
    {
      "epoch": 0.39545454545454545,
      "grad_norm": 0.5614489912986755,
      "learning_rate": 2.9e-06,
      "loss": 0.6314,
      "step": 58
    },
    {
      "epoch": 0.4022727272727273,
      "grad_norm": 0.5007341504096985,
      "learning_rate": 2.95e-06,
      "loss": 0.6346,
      "step": 59
    },
    {
      "epoch": 0.4090909090909091,
      "grad_norm": 0.54221510887146,
      "learning_rate": 3e-06,
      "loss": 0.6715,
      "step": 60
    },
    {
      "epoch": 0.4159090909090909,
      "grad_norm": 0.5177039504051208,
      "learning_rate": 3.05e-06,
      "loss": 0.6459,
      "step": 61
    },
    {
      "epoch": 0.42272727272727273,
      "grad_norm": 0.4741189181804657,
      "learning_rate": 3.1000000000000004e-06,
      "loss": 0.592,
      "step": 62
    },
    {
      "epoch": 0.42954545454545456,
      "grad_norm": 0.5637328028678894,
      "learning_rate": 3.1500000000000003e-06,
      "loss": 0.6459,
      "step": 63
    },
    {
      "epoch": 0.43636363636363634,
      "grad_norm": 0.5478869676589966,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.6447,
      "step": 64
    },
    {
      "epoch": 0.4431818181818182,
      "grad_norm": 0.5483130216598511,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.6319,
      "step": 65
    },
    {
      "epoch": 0.45,
      "grad_norm": 0.478081613779068,
      "learning_rate": 3.3000000000000006e-06,
      "loss": 0.6102,
      "step": 66
    },
    {
      "epoch": 0.45681818181818185,
      "grad_norm": 0.4777645766735077,
      "learning_rate": 3.3500000000000005e-06,
      "loss": 0.5986,
      "step": 67
    },
    {
      "epoch": 0.4636363636363636,
      "grad_norm": 0.5291482210159302,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 0.6314,
      "step": 68
    },
    {
      "epoch": 0.47045454545454546,
      "grad_norm": 0.5224051475524902,
      "learning_rate": 3.45e-06,
      "loss": 0.6278,
      "step": 69
    },
    {
      "epoch": 0.4772727272727273,
      "grad_norm": 0.6474127173423767,
      "learning_rate": 3.5e-06,
      "loss": 0.6398,
      "step": 70
    },
    {
      "epoch": 0.48409090909090907,
      "grad_norm": 0.5060178637504578,
      "learning_rate": 3.5500000000000003e-06,
      "loss": 0.5936,
      "step": 71
    },
    {
      "epoch": 0.4909090909090909,
      "grad_norm": 3.0769243240356445,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 0.5975,
      "step": 72
    },
    {
      "epoch": 0.49772727272727274,
      "grad_norm": 0.5319749116897583,
      "learning_rate": 3.65e-06,
      "loss": 0.6119,
      "step": 73
    },
    {
      "epoch": 0.5045454545454545,
      "grad_norm": 0.6745399236679077,
      "learning_rate": 3.7e-06,
      "loss": 0.6213,
      "step": 74
    },
    {
      "epoch": 0.5113636363636364,
      "grad_norm": 0.5156731009483337,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.6262,
      "step": 75
    },
    {
      "epoch": 0.5181818181818182,
      "grad_norm": 0.9138725399971008,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 0.6244,
      "step": 76
    },
    {
      "epoch": 0.525,
      "grad_norm": 0.4362037777900696,
      "learning_rate": 3.85e-06,
      "loss": 0.5874,
      "step": 77
    },
    {
      "epoch": 0.5318181818181819,
      "grad_norm": 0.4674088954925537,
      "learning_rate": 3.900000000000001e-06,
      "loss": 0.615,
      "step": 78
    },
    {
      "epoch": 0.5386363636363637,
      "grad_norm": 0.4590819180011749,
      "learning_rate": 3.95e-06,
      "loss": 0.6003,
      "step": 79
    },
    {
      "epoch": 0.5454545454545454,
      "grad_norm": 0.5145031809806824,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.6079,
      "step": 80
    },
    {
      "epoch": 0.5522727272727272,
      "grad_norm": 0.46660348773002625,
      "learning_rate": 4.05e-06,
      "loss": 0.6174,
      "step": 81
    },
    {
      "epoch": 0.5590909090909091,
      "grad_norm": 0.48053789138793945,
      "learning_rate": 4.1e-06,
      "loss": 0.6232,
      "step": 82
    },
    {
      "epoch": 0.5659090909090909,
      "grad_norm": 0.540793776512146,
      "learning_rate": 4.15e-06,
      "loss": 0.619,
      "step": 83
    },
    {
      "epoch": 0.5727272727272728,
      "grad_norm": 0.4925222396850586,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 0.5838,
      "step": 84
    },
    {
      "epoch": 0.5795454545454546,
      "grad_norm": 0.44989290833473206,
      "learning_rate": 4.25e-06,
      "loss": 0.5972,
      "step": 85
    },
    {
      "epoch": 0.5863636363636363,
      "grad_norm": 0.42691388726234436,
      "learning_rate": 4.3e-06,
      "loss": 0.6046,
      "step": 86
    },
    {
      "epoch": 0.5931818181818181,
      "grad_norm": 0.4512398838996887,
      "learning_rate": 4.350000000000001e-06,
      "loss": 0.5828,
      "step": 87
    },
    {
      "epoch": 0.6,
      "grad_norm": 0.44399499893188477,
      "learning_rate": 4.4e-06,
      "loss": 0.6143,
      "step": 88
    },
    {
      "epoch": 0.6068181818181818,
      "grad_norm": 0.46058326959609985,
      "learning_rate": 4.450000000000001e-06,
      "loss": 0.6117,
      "step": 89
    },
    {
      "epoch": 0.6136363636363636,
      "grad_norm": 0.8795785903930664,
      "learning_rate": 4.5e-06,
      "loss": 0.6,
      "step": 90
    },
    {
      "epoch": 0.6204545454545455,
      "grad_norm": 0.5148798227310181,
      "learning_rate": 4.5500000000000005e-06,
      "loss": 0.6041,
      "step": 91
    },
    {
      "epoch": 0.6272727272727273,
      "grad_norm": 0.48829612135887146,
      "learning_rate": 4.600000000000001e-06,
      "loss": 0.5492,
      "step": 92
    },
    {
      "epoch": 0.634090909090909,
      "grad_norm": 0.5211894512176514,
      "learning_rate": 4.65e-06,
      "loss": 0.6044,
      "step": 93
    },
    {
      "epoch": 0.6409090909090909,
      "grad_norm": 0.44403275847435,
      "learning_rate": 4.7e-06,
      "loss": 0.5929,
      "step": 94
    },
    {
      "epoch": 0.6477272727272727,
      "grad_norm": 0.6037693619728088,
      "learning_rate": 4.75e-06,
      "loss": 0.6132,
      "step": 95
    },
    {
      "epoch": 0.6545454545454545,
      "grad_norm": 0.4381515085697174,
      "learning_rate": 4.800000000000001e-06,
      "loss": 0.5822,
      "step": 96
    },
    {
      "epoch": 0.6613636363636364,
      "grad_norm": 0.4997427761554718,
      "learning_rate": 4.85e-06,
      "loss": 0.593,
      "step": 97
    },
    {
      "epoch": 0.6681818181818182,
      "grad_norm": 0.440571665763855,
      "learning_rate": 4.9000000000000005e-06,
      "loss": 0.6072,
      "step": 98
    },
    {
      "epoch": 0.675,
      "grad_norm": 0.5020624995231628,
      "learning_rate": 4.95e-06,
      "loss": 0.5861,
      "step": 99
    },
    {
      "epoch": 0.6818181818181818,
      "grad_norm": 0.4280160963535309,
      "learning_rate": 5e-06,
      "loss": 0.6163,
      "step": 100
    },
    {
      "epoch": 0.6886363636363636,
      "grad_norm": 0.4584537148475647,
      "learning_rate": 4.9999795126530275e-06,
      "loss": 0.6177,
      "step": 101
    },
    {
      "epoch": 0.6954545454545454,
      "grad_norm": 0.6835049390792847,
      "learning_rate": 4.999918050947891e-06,
      "loss": 0.579,
      "step": 102
    },
    {
      "epoch": 0.7022727272727273,
      "grad_norm": 0.4551607072353363,
      "learning_rate": 4.999815615891943e-06,
      "loss": 0.5927,
      "step": 103
    },
    {
      "epoch": 0.7090909090909091,
      "grad_norm": 0.5893972516059875,
      "learning_rate": 4.9996722091640805e-06,
      "loss": 0.5775,
      "step": 104
    },
    {
      "epoch": 0.7159090909090909,
      "grad_norm": 0.4574092924594879,
      "learning_rate": 4.9994878331147225e-06,
      "loss": 0.5863,
      "step": 105
    },
    {
      "epoch": 0.7227272727272728,
      "grad_norm": 0.5354658365249634,
      "learning_rate": 4.99926249076577e-06,
      "loss": 0.5453,
      "step": 106
    },
    {
      "epoch": 0.7295454545454545,
      "grad_norm": 0.4347354471683502,
      "learning_rate": 4.998996185810557e-06,
      "loss": 0.5913,
      "step": 107
    },
    {
      "epoch": 0.7363636363636363,
      "grad_norm": 0.4487966299057007,
      "learning_rate": 4.998688922613788e-06,
      "loss": 0.5749,
      "step": 108
    },
    {
      "epoch": 0.7431818181818182,
      "grad_norm": 0.42477577924728394,
      "learning_rate": 4.9983407062114695e-06,
      "loss": 0.5761,
      "step": 109
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.4801446497440338,
      "learning_rate": 4.9979515423108255e-06,
      "loss": 0.5977,
      "step": 110
    },
    {
      "epoch": 0.7568181818181818,
      "grad_norm": 0.4876883029937744,
      "learning_rate": 4.997521437290205e-06,
      "loss": 0.5849,
      "step": 111
    },
    {
      "epoch": 0.7636363636363637,
      "grad_norm": 0.4328872263431549,
      "learning_rate": 4.997050398198977e-06,
      "loss": 0.5988,
      "step": 112
    },
    {
      "epoch": 0.7704545454545455,
      "grad_norm": 0.4332719147205353,
      "learning_rate": 4.996538432757414e-06,
      "loss": 0.6059,
      "step": 113
    },
    {
      "epoch": 0.7772727272727272,
      "grad_norm": 0.4455336630344391,
      "learning_rate": 4.995985549356568e-06,
      "loss": 0.5725,
      "step": 114
    },
    {
      "epoch": 0.7840909090909091,
      "grad_norm": 0.4197766184806824,
      "learning_rate": 4.995391757058129e-06,
      "loss": 0.5715,
      "step": 115
    },
    {
      "epoch": 0.7909090909090909,
      "grad_norm": 0.41945499181747437,
      "learning_rate": 4.99475706559428e-06,
      "loss": 0.5845,
      "step": 116
    },
    {
      "epoch": 0.7977272727272727,
      "grad_norm": 0.42414823174476624,
      "learning_rate": 4.994081485367537e-06,
      "loss": 0.5595,
      "step": 117
    },
    {
      "epoch": 0.8045454545454546,
      "grad_norm": 0.4056423008441925,
      "learning_rate": 4.993365027450576e-06,
      "loss": 0.582,
      "step": 118
    },
    {
      "epoch": 0.8113636363636364,
      "grad_norm": 0.4554205536842346,
      "learning_rate": 4.992607703586058e-06,
      "loss": 0.5591,
      "step": 119
    },
    {
      "epoch": 0.8181818181818182,
      "grad_norm": 0.4307089149951935,
      "learning_rate": 4.991809526186424e-06,
      "loss": 0.5927,
      "step": 120
    },
    {
      "epoch": 0.825,
      "grad_norm": 0.48133528232574463,
      "learning_rate": 4.990970508333707e-06,
      "loss": 0.5703,
      "step": 121
    },
    {
      "epoch": 0.8318181818181818,
      "grad_norm": 0.48582738637924194,
      "learning_rate": 4.990090663779305e-06,
      "loss": 0.5491,
      "step": 122
    },
    {
      "epoch": 0.8386363636363636,
      "grad_norm": 0.44926196336746216,
      "learning_rate": 4.9891700069437635e-06,
      "loss": 0.5803,
      "step": 123
    },
    {
      "epoch": 0.8454545454545455,
      "grad_norm": 0.4562165141105652,
      "learning_rate": 4.988208552916535e-06,
      "loss": 0.5945,
      "step": 124
    },
    {
      "epoch": 0.8522727272727273,
      "grad_norm": 0.5885360836982727,
      "learning_rate": 4.987206317455734e-06,
      "loss": 0.5632,
      "step": 125
    },
    {
      "epoch": 0.8590909090909091,
      "grad_norm": 0.4525550603866577,
      "learning_rate": 4.986163316987877e-06,
      "loss": 0.5221,
      "step": 126
    },
    {
      "epoch": 0.865909090909091,
      "grad_norm": 0.4545478820800781,
      "learning_rate": 4.985079568607613e-06,
      "loss": 0.587,
      "step": 127
    },
    {
      "epoch": 0.8727272727272727,
      "grad_norm": 0.4364396333694458,
      "learning_rate": 4.983955090077445e-06,
      "loss": 0.5626,
      "step": 128
    },
    {
      "epoch": 0.8795454545454545,
      "grad_norm": 0.4354369640350342,
      "learning_rate": 4.982789899827439e-06,
      "loss": 0.5674,
      "step": 129
    },
    {
      "epoch": 0.8863636363636364,
      "grad_norm": 0.4822773337364197,
      "learning_rate": 4.9815840169549216e-06,
      "loss": 0.5685,
      "step": 130
    },
    {
      "epoch": 0.8931818181818182,
      "grad_norm": 0.44673025608062744,
      "learning_rate": 4.980337461224164e-06,
      "loss": 0.5733,
      "step": 131
    },
    {
      "epoch": 0.9,
      "grad_norm": 0.6294501423835754,
      "learning_rate": 4.979050253066064e-06,
      "loss": 0.5396,
      "step": 132
    },
    {
      "epoch": 0.9068181818181819,
      "grad_norm": 0.42890027165412903,
      "learning_rate": 4.977722413577802e-06,
      "loss": 0.5718,
      "step": 133
    },
    {
      "epoch": 0.9136363636363637,
      "grad_norm": 0.470587819814682,
      "learning_rate": 4.976353964522509e-06,
      "loss": 0.5705,
      "step": 134
    },
    {
      "epoch": 0.9204545454545454,
      "grad_norm": 0.4632768929004669,
      "learning_rate": 4.974944928328894e-06,
      "loss": 0.5394,
      "step": 135
    },
    {
      "epoch": 0.9272727272727272,
      "grad_norm": 0.5304691195487976,
      "learning_rate": 4.973495328090891e-06,
      "loss": 0.5762,
      "step": 136
    },
    {
      "epoch": 0.9340909090909091,
      "grad_norm": 0.42036354541778564,
      "learning_rate": 4.972005187567267e-06,
      "loss": 0.5649,
      "step": 137
    },
    {
      "epoch": 0.9409090909090909,
      "grad_norm": 0.45189663767814636,
      "learning_rate": 4.970474531181245e-06,
      "loss": 0.5554,
      "step": 138
    },
    {
      "epoch": 0.9477272727272728,
      "grad_norm": 0.47352468967437744,
      "learning_rate": 4.968903384020095e-06,
      "loss": 0.5511,
      "step": 139
    },
    {
      "epoch": 0.9545454545454546,
      "grad_norm": 0.525702953338623,
      "learning_rate": 4.967291771834727e-06,
      "loss": 0.5777,
      "step": 140
    },
    {
      "epoch": 0.9613636363636363,
      "grad_norm": 0.46146854758262634,
      "learning_rate": 4.965639721039267e-06,
      "loss": 0.5537,
      "step": 141
    },
    {
      "epoch": 0.9681818181818181,
      "grad_norm": 0.5200941562652588,
      "learning_rate": 4.963947258710626e-06,
      "loss": 0.5522,
      "step": 142
    },
    {
      "epoch": 0.975,
      "grad_norm": 0.5260375142097473,
      "learning_rate": 4.962214412588053e-06,
      "loss": 0.5747,
      "step": 143
    },
    {
      "epoch": 0.9818181818181818,
      "grad_norm": 0.518645703792572,
      "learning_rate": 4.960441211072686e-06,
      "loss": 0.5257,
      "step": 144
    },
    {
      "epoch": 0.9886363636363636,
      "grad_norm": 0.4528891444206238,
      "learning_rate": 4.9586276832270785e-06,
      "loss": 0.5548,
      "step": 145
    },
    {
      "epoch": 0.9954545454545455,
      "grad_norm": 0.5137557983398438,
      "learning_rate": 4.9567738587747314e-06,
      "loss": 0.5526,
      "step": 146
    },
    {
      "epoch": 1.0068181818181818,
      "grad_norm": 0.8773607611656189,
      "learning_rate": 4.954879768099599e-06,
      "loss": 1.1,
      "step": 147
    },
    {
      "epoch": 1.0136363636363637,
      "grad_norm": 0.39079979062080383,
      "learning_rate": 4.952945442245598e-06,
      "loss": 0.568,
      "step": 148
    },
    {
      "epoch": 1.0204545454545455,
      "grad_norm": 0.45004600286483765,
      "learning_rate": 4.95097091291609e-06,
      "loss": 0.5383,
      "step": 149
    },
    {
      "epoch": 1.0272727272727273,
      "grad_norm": 0.41872867941856384,
      "learning_rate": 4.948956212473371e-06,
      "loss": 0.5436,
      "step": 150
    },
    {
      "epoch": 1.0340909090909092,
      "grad_norm": 0.46829020977020264,
      "learning_rate": 4.946901373938132e-06,
      "loss": 0.5336,
      "step": 151
    },
    {
      "epoch": 1.040909090909091,
      "grad_norm": 0.45038115978240967,
      "learning_rate": 4.944806430988927e-06,
      "loss": 0.538,
      "step": 152
    },
    {
      "epoch": 1.0477272727272728,
      "grad_norm": 0.6496132016181946,
      "learning_rate": 4.942671417961615e-06,
      "loss": 0.5459,
      "step": 153
    },
    {
      "epoch": 1.0545454545454545,
      "grad_norm": 0.4842703640460968,
      "learning_rate": 4.940496369848795e-06,
      "loss": 0.553,
      "step": 154
    },
    {
      "epoch": 1.0613636363636363,
      "grad_norm": 0.4338674545288086,
      "learning_rate": 4.938281322299243e-06,
      "loss": 0.5366,
      "step": 155
    },
    {
      "epoch": 1.0681818181818181,
      "grad_norm": 0.44630634784698486,
      "learning_rate": 4.936026311617316e-06,
      "loss": 0.5517,
      "step": 156
    },
    {
      "epoch": 1.075,
      "grad_norm": 0.4606929421424866,
      "learning_rate": 4.933731374762361e-06,
      "loss": 0.5405,
      "step": 157
    },
    {
      "epoch": 1.0818181818181818,
      "grad_norm": 0.48667654395103455,
      "learning_rate": 4.931396549348115e-06,
      "loss": 0.5376,
      "step": 158
    },
    {
      "epoch": 1.0886363636363636,
      "grad_norm": 0.4516271650791168,
      "learning_rate": 4.9290218736420795e-06,
      "loss": 0.5206,
      "step": 159
    },
    {
      "epoch": 1.0954545454545455,
      "grad_norm": 0.44719284772872925,
      "learning_rate": 4.926607386564898e-06,
      "loss": 0.5397,
      "step": 160
    },
    {
      "epoch": 1.1022727272727273,
      "grad_norm": 0.40752628445625305,
      "learning_rate": 4.9241531276897196e-06,
      "loss": 0.5466,
      "step": 161
    },
    {
      "epoch": 1.1090909090909091,
      "grad_norm": 0.45423609018325806,
      "learning_rate": 4.921659137241544e-06,
      "loss": 0.5227,
      "step": 162
    },
    {
      "epoch": 1.115909090909091,
      "grad_norm": 0.4185822308063507,
      "learning_rate": 4.919125456096574e-06,
      "loss": 0.5144,
      "step": 163
    },
    {
      "epoch": 1.1227272727272728,
      "grad_norm": 0.7753382921218872,
      "learning_rate": 4.916552125781529e-06,
      "loss": 0.5352,
      "step": 164
    },
    {
      "epoch": 1.1295454545454546,
      "grad_norm": 0.47211477160453796,
      "learning_rate": 4.913939188472979e-06,
      "loss": 0.5505,
      "step": 165
    },
    {
      "epoch": 1.1363636363636362,
      "grad_norm": 0.42742082476615906,
      "learning_rate": 4.911286686996648e-06,
      "loss": 0.529,
      "step": 166
    },
    {
      "epoch": 1.143181818181818,
      "grad_norm": 0.41340839862823486,
      "learning_rate": 4.908594664826708e-06,
      "loss": 0.5292,
      "step": 167
    },
    {
      "epoch": 1.15,
      "grad_norm": 0.44648951292037964,
      "learning_rate": 4.905863166085076e-06,
      "loss": 0.5757,
      "step": 168
    },
    {
      "epoch": 1.1568181818181817,
      "grad_norm": 0.4285544753074646,
      "learning_rate": 4.903092235540679e-06,
      "loss": 0.5102,
      "step": 169
    },
    {
      "epoch": 1.1636363636363636,
      "grad_norm": 0.42675015330314636,
      "learning_rate": 4.900281918608732e-06,
      "loss": 0.5397,
      "step": 170
    },
    {
      "epoch": 1.1704545454545454,
      "grad_norm": 0.47636961936950684,
      "learning_rate": 4.897432261349984e-06,
      "loss": 0.5307,
      "step": 171
    },
    {
      "epoch": 1.1772727272727272,
      "grad_norm": 0.4213427007198334,
      "learning_rate": 4.894543310469968e-06,
      "loss": 0.54,
      "step": 172
    },
    {
      "epoch": 1.184090909090909,
      "grad_norm": 0.41153451800346375,
      "learning_rate": 4.891615113318236e-06,
      "loss": 0.5423,
      "step": 173
    },
    {
      "epoch": 1.190909090909091,
      "grad_norm": 0.459185391664505,
      "learning_rate": 4.888647717887582e-06,
      "loss": 0.5239,
      "step": 174
    },
    {
      "epoch": 1.1977272727272728,
      "grad_norm": 0.48279181122779846,
      "learning_rate": 4.8856411728132526e-06,
      "loss": 0.5336,
      "step": 175
    },
    {
      "epoch": 1.2045454545454546,
      "grad_norm": 0.4186796247959137,
      "learning_rate": 4.8825955273721524e-06,
      "loss": 0.5509,
      "step": 176
    },
    {
      "epoch": 1.2113636363636364,
      "grad_norm": 0.4345344007015228,
      "learning_rate": 4.879510831482039e-06,
      "loss": 0.5415,
      "step": 177
    },
    {
      "epoch": 1.2181818181818183,
      "grad_norm": 0.461592435836792,
      "learning_rate": 4.876387135700701e-06,
      "loss": 0.5256,
      "step": 178
    },
    {
      "epoch": 1.225,
      "grad_norm": 0.5960240960121155,
      "learning_rate": 4.873224491225128e-06,
      "loss": 0.5599,
      "step": 179
    },
    {
      "epoch": 1.231818181818182,
      "grad_norm": 0.4662570357322693,
      "learning_rate": 4.870022949890676e-06,
      "loss": 0.5647,
      "step": 180
    },
    {
      "epoch": 1.2386363636363638,
      "grad_norm": 0.4355057179927826,
      "learning_rate": 4.866782564170217e-06,
      "loss": 0.5056,
      "step": 181
    },
    {
      "epoch": 1.2454545454545454,
      "grad_norm": 0.48699110746383667,
      "learning_rate": 4.863503387173276e-06,
      "loss": 0.5394,
      "step": 182
    },
    {
      "epoch": 1.2522727272727272,
      "grad_norm": 0.48679423332214355,
      "learning_rate": 4.860185472645161e-06,
      "loss": 0.5394,
      "step": 183
    },
    {
      "epoch": 1.259090909090909,
      "grad_norm": 0.6564393043518066,
      "learning_rate": 4.856828874966086e-06,
      "loss": 0.5259,
      "step": 184
    },
    {
      "epoch": 1.2659090909090909,
      "grad_norm": 0.44762882590293884,
      "learning_rate": 4.853433649150276e-06,
      "loss": 0.5563,
      "step": 185
    },
    {
      "epoch": 1.2727272727272727,
      "grad_norm": 0.5391108989715576,
      "learning_rate": 4.849999850845066e-06,
      "loss": 0.5326,
      "step": 186
    },
    {
      "epoch": 1.2795454545454545,
      "grad_norm": 0.4201100170612335,
      "learning_rate": 4.8465275363299905e-06,
      "loss": 0.526,
      "step": 187
    },
    {
      "epoch": 1.2863636363636364,
      "grad_norm": 0.46211543679237366,
      "learning_rate": 4.84301676251586e-06,
      "loss": 0.5498,
      "step": 188
    },
    {
      "epoch": 1.2931818181818182,
      "grad_norm": 0.4325430989265442,
      "learning_rate": 4.839467586943825e-06,
      "loss": 0.556,
      "step": 189
    },
    {
      "epoch": 1.3,
      "grad_norm": 0.46057143807411194,
      "learning_rate": 4.835880067784441e-06,
      "loss": 0.5079,
      "step": 190
    },
    {
      "epoch": 1.3068181818181819,
      "grad_norm": 0.4920806288719177,
      "learning_rate": 4.832254263836708e-06,
      "loss": 0.533,
      "step": 191
    },
    {
      "epoch": 1.3136363636363637,
      "grad_norm": 0.46693211793899536,
      "learning_rate": 4.828590234527107e-06,
      "loss": 0.5542,
      "step": 192
    },
    {
      "epoch": 1.3204545454545453,
      "grad_norm": 0.4227873682975769,
      "learning_rate": 4.82488803990863e-06,
      "loss": 0.544,
      "step": 193
    },
    {
      "epoch": 1.3272727272727272,
      "grad_norm": 0.44745922088623047,
      "learning_rate": 4.821147740659795e-06,
      "loss": 0.4962,
      "step": 194
    },
    {
      "epoch": 1.334090909090909,
      "grad_norm": 0.46137240529060364,
      "learning_rate": 4.817369398083648e-06,
      "loss": 0.5359,
      "step": 195
    },
    {
      "epoch": 1.3409090909090908,
      "grad_norm": 0.6483879685401917,
      "learning_rate": 4.813553074106761e-06,
      "loss": 0.5323,
      "step": 196
    },
    {
      "epoch": 1.3477272727272727,
      "grad_norm": 0.6602574586868286,
      "learning_rate": 4.809698831278217e-06,
      "loss": 0.5161,
      "step": 197
    },
    {
      "epoch": 1.3545454545454545,
      "grad_norm": 0.47758543491363525,
      "learning_rate": 4.805806732768585e-06,
      "loss": 0.5214,
      "step": 198
    },
    {
      "epoch": 1.3613636363636363,
      "grad_norm": 0.4587438106536865,
      "learning_rate": 4.801876842368882e-06,
      "loss": 0.5208,
      "step": 199
    },
    {
      "epoch": 1.3681818181818182,
      "grad_norm": 0.46055835485458374,
      "learning_rate": 4.797909224489531e-06,
      "loss": 0.5374,
      "step": 200
    },
    {
      "epoch": 1.375,
      "grad_norm": 0.43727558851242065,
      "learning_rate": 4.793903944159303e-06,
      "loss": 0.5351,
      "step": 201
    },
    {
      "epoch": 1.3818181818181818,
      "grad_norm": 0.4573572874069214,
      "learning_rate": 4.789861067024253e-06,
      "loss": 0.4997,
      "step": 202
    },
    {
      "epoch": 1.3886363636363637,
      "grad_norm": 0.5921999216079712,
      "learning_rate": 4.785780659346642e-06,
      "loss": 0.5168,
      "step": 203
    },
    {
      "epoch": 1.3954545454545455,
      "grad_norm": 0.47943904995918274,
      "learning_rate": 4.781662788003851e-06,
      "loss": 0.5513,
      "step": 204
    },
    {
      "epoch": 1.4022727272727273,
      "grad_norm": 0.43733686208724976,
      "learning_rate": 4.777507520487289e-06,
      "loss": 0.5281,
      "step": 205
    },
    {
      "epoch": 1.4090909090909092,
      "grad_norm": 0.4289107620716095,
      "learning_rate": 4.773314924901281e-06,
      "loss": 0.524,
      "step": 206
    },
    {
      "epoch": 1.415909090909091,
      "grad_norm": 0.4203416109085083,
      "learning_rate": 4.769085069961955e-06,
      "loss": 0.5563,
      "step": 207
    },
    {
      "epoch": 1.4227272727272728,
      "grad_norm": 0.4205201268196106,
      "learning_rate": 4.764818024996117e-06,
      "loss": 0.5202,
      "step": 208
    },
    {
      "epoch": 1.4295454545454547,
      "grad_norm": 0.4296267032623291,
      "learning_rate": 4.760513859940112e-06,
      "loss": 0.5113,
      "step": 209
    },
    {
      "epoch": 1.4363636363636363,
      "grad_norm": 0.3979087471961975,
      "learning_rate": 4.756172645338675e-06,
      "loss": 0.5313,
      "step": 210
    },
    {
      "epoch": 1.4431818181818181,
      "grad_norm": 0.45232149958610535,
      "learning_rate": 4.751794452343785e-06,
      "loss": 0.5204,
      "step": 211
    },
    {
      "epoch": 1.45,
      "grad_norm": 0.4309398829936981,
      "learning_rate": 4.747379352713489e-06,
      "loss": 0.5216,
      "step": 212
    },
    {
      "epoch": 1.4568181818181818,
      "grad_norm": 0.45721709728240967,
      "learning_rate": 4.7429274188107275e-06,
      "loss": 0.5279,
      "step": 213
    },
    {
      "epoch": 1.4636363636363636,
      "grad_norm": 0.46138322353363037,
      "learning_rate": 4.738438723602154e-06,
      "loss": 0.5287,
      "step": 214
    },
    {
      "epoch": 1.4704545454545455,
      "grad_norm": 0.41752108931541443,
      "learning_rate": 4.733913340656933e-06,
      "loss": 0.5399,
      "step": 215
    },
    {
      "epoch": 1.4772727272727273,
      "grad_norm": 0.45447197556495667,
      "learning_rate": 4.729351344145536e-06,
      "loss": 0.5277,
      "step": 216
    },
    {
      "epoch": 1.4840909090909091,
      "grad_norm": 0.42524856328964233,
      "learning_rate": 4.7247528088385296e-06,
      "loss": 0.5103,
      "step": 217
    },
    {
      "epoch": 1.490909090909091,
      "grad_norm": 0.46024221181869507,
      "learning_rate": 4.720117810105341e-06,
      "loss": 0.5172,
      "step": 218
    },
    {
      "epoch": 1.4977272727272728,
      "grad_norm": 0.5868630409240723,
      "learning_rate": 4.715446423913036e-06,
      "loss": 0.5409,
      "step": 219
    },
    {
      "epoch": 1.5045454545454544,
      "grad_norm": 0.424813836812973,
      "learning_rate": 4.710738726825059e-06,
      "loss": 0.5495,
      "step": 220
    },
    {
      "epoch": 1.5113636363636362,
      "grad_norm": 0.49356943368911743,
      "learning_rate": 4.705994795999991e-06,
      "loss": 0.519,
      "step": 221
    },
    {
      "epoch": 1.518181818181818,
      "grad_norm": 0.44291776418685913,
      "learning_rate": 4.701214709190277e-06,
      "loss": 0.5232,
      "step": 222
    },
    {
      "epoch": 1.525,
      "grad_norm": 0.48171764612197876,
      "learning_rate": 4.696398544740955e-06,
      "loss": 0.5501,
      "step": 223
    },
    {
      "epoch": 1.5318181818181817,
      "grad_norm": 0.4296175539493561,
      "learning_rate": 4.69154638158837e-06,
      "loss": 0.5087,
      "step": 224
    },
    {
      "epoch": 1.5386363636363636,
      "grad_norm": 0.4715246260166168,
      "learning_rate": 4.686658299258881e-06,
      "loss": 0.5363,
      "step": 225
    },
    {
      "epoch": 1.5454545454545454,
      "grad_norm": 0.44422832131385803,
      "learning_rate": 4.681734377867562e-06,
      "loss": 0.5371,
      "step": 226
    },
    {
      "epoch": 1.5522727272727272,
      "grad_norm": 4.221654891967773,
      "learning_rate": 4.67677469811688e-06,
      "loss": 0.5191,
      "step": 227
    },
    {
      "epoch": 1.559090909090909,
      "grad_norm": 0.41857993602752686,
      "learning_rate": 4.671779341295378e-06,
      "loss": 0.5191,
      "step": 228
    },
    {
      "epoch": 1.565909090909091,
      "grad_norm": 0.4950161278247833,
      "learning_rate": 4.666748389276344e-06,
      "loss": 0.5311,
      "step": 229
    },
    {
      "epoch": 1.5727272727272728,
      "grad_norm": 0.44641605019569397,
      "learning_rate": 4.661681924516466e-06,
      "loss": 0.5174,
      "step": 230
    },
    {
      "epoch": 1.5795454545454546,
      "grad_norm": 0.39064037799835205,
      "learning_rate": 4.6565800300544805e-06,
      "loss": 0.515,
      "step": 231
    },
    {
      "epoch": 1.5863636363636364,
      "grad_norm": 0.7764399647712708,
      "learning_rate": 4.651442789509813e-06,
      "loss": 0.5282,
      "step": 232
    },
    {
      "epoch": 1.5931818181818183,
      "grad_norm": 0.44421035051345825,
      "learning_rate": 4.646270287081208e-06,
      "loss": 0.5324,
      "step": 233
    },
    {
      "epoch": 1.6,
      "grad_norm": 0.5149928331375122,
      "learning_rate": 4.641062607545347e-06,
      "loss": 0.5264,
      "step": 234
    },
    {
      "epoch": 1.606818181818182,
      "grad_norm": 0.4784093201160431,
      "learning_rate": 4.6358198362554585e-06,
      "loss": 0.552,
      "step": 235
    },
    {
      "epoch": 1.6136363636363638,
      "grad_norm": 0.4138723313808441,
      "learning_rate": 4.630542059139923e-06,
      "loss": 0.5202,
      "step": 236
    },
    {
      "epoch": 1.6204545454545456,
      "grad_norm": 0.4509594440460205,
      "learning_rate": 4.625229362700863e-06,
      "loss": 0.5318,
      "step": 237
    },
    {
      "epoch": 1.6272727272727274,
      "grad_norm": 0.44979560375213623,
      "learning_rate": 4.61988183401272e-06,
      "loss": 0.521,
      "step": 238
    },
    {
      "epoch": 1.634090909090909,
      "grad_norm": 0.4525575041770935,
      "learning_rate": 4.614499560720837e-06,
      "loss": 0.5073,
      "step": 239
    },
    {
      "epoch": 1.6409090909090909,
      "grad_norm": 0.44126543402671814,
      "learning_rate": 4.609082631040012e-06,
      "loss": 0.5058,
      "step": 240
    },
    {
      "epoch": 1.6477272727272727,
      "grad_norm": 0.887867271900177,
      "learning_rate": 4.603631133753061e-06,
      "loss": 0.5471,
      "step": 241
    },
    {
      "epoch": 1.6545454545454545,
      "grad_norm": 0.4497375786304474,
      "learning_rate": 4.598145158209356e-06,
      "loss": 0.501,
      "step": 242
    },
    {
      "epoch": 1.6613636363636364,
      "grad_norm": 0.45681032538414,
      "learning_rate": 4.592624794323366e-06,
      "loss": 0.5306,
      "step": 243
    },
    {
      "epoch": 1.6681818181818182,
      "grad_norm": 0.5186678171157837,
      "learning_rate": 4.587070132573178e-06,
      "loss": 0.5097,
      "step": 244
    },
    {
      "epoch": 1.675,
      "grad_norm": 0.44289669394493103,
      "learning_rate": 4.581481263999019e-06,
      "loss": 0.5085,
      "step": 245
    },
    {
      "epoch": 1.6818181818181817,
      "grad_norm": 0.4571002125740051,
      "learning_rate": 4.575858280201761e-06,
      "loss": 0.5327,
      "step": 246
    },
    {
      "epoch": 1.6886363636363635,
      "grad_norm": 0.43753257393836975,
      "learning_rate": 4.570201273341418e-06,
      "loss": 0.5167,
      "step": 247
    },
    {
      "epoch": 1.6954545454545453,
      "grad_norm": 0.44994574785232544,
      "learning_rate": 4.564510336135642e-06,
      "loss": 0.5554,
      "step": 248
    },
    {
      "epoch": 1.7022727272727272,
      "grad_norm": 0.6185874938964844,
      "learning_rate": 4.558785561858196e-06,
      "loss": 0.5115,
      "step": 249
    },
    {
      "epoch": 1.709090909090909,
      "grad_norm": 0.4883944094181061,
      "learning_rate": 4.5530270443374305e-06,
      "loss": 0.5312,
      "step": 250
    },
    {
      "epoch": 1.7159090909090908,
      "grad_norm": 0.46433499455451965,
      "learning_rate": 4.547234877954741e-06,
      "loss": 0.5368,
      "step": 251
    },
    {
      "epoch": 1.7227272727272727,
      "grad_norm": 0.4270826578140259,
      "learning_rate": 4.541409157643027e-06,
      "loss": 0.5258,
      "step": 252
    },
    {
      "epoch": 1.7295454545454545,
      "grad_norm": 0.4390241205692291,
      "learning_rate": 4.535549978885132e-06,
      "loss": 0.5259,
      "step": 253
    },
    {
      "epoch": 1.7363636363636363,
      "grad_norm": 0.42756086587905884,
      "learning_rate": 4.5296574377122765e-06,
      "loss": 0.5217,
      "step": 254
    },
    {
      "epoch": 1.7431818181818182,
      "grad_norm": 0.48124927282333374,
      "learning_rate": 4.5237316307024895e-06,
      "loss": 0.5095,
      "step": 255
    },
    {
      "epoch": 1.75,
      "grad_norm": 0.4394672214984894,
      "learning_rate": 4.517772654979024e-06,
      "loss": 0.5116,
      "step": 256
    },
    {
      "epoch": 1.7568181818181818,
      "grad_norm": 0.5805886387825012,
      "learning_rate": 4.51178060820876e-06,
      "loss": 0.5145,
      "step": 257
    },
    {
      "epoch": 1.7636363636363637,
      "grad_norm": 0.4845556616783142,
      "learning_rate": 4.505755588600613e-06,
      "loss": 0.5064,
      "step": 258
    },
    {
      "epoch": 1.7704545454545455,
      "grad_norm": 0.45300963521003723,
      "learning_rate": 4.499697694903915e-06,
      "loss": 0.5362,
      "step": 259
    },
    {
      "epoch": 1.7772727272727273,
      "grad_norm": 0.45417436957359314,
      "learning_rate": 4.493607026406802e-06,
      "loss": 0.5234,
      "step": 260
    },
    {
      "epoch": 1.7840909090909092,
      "grad_norm": 0.4577132761478424,
      "learning_rate": 4.487483682934587e-06,
      "loss": 0.5271,
      "step": 261
    },
    {
      "epoch": 1.790909090909091,
      "grad_norm": 0.4518176019191742,
      "learning_rate": 4.481327764848118e-06,
      "loss": 0.4853,
      "step": 262
    },
    {
      "epoch": 1.7977272727272728,
      "grad_norm": 0.43804875016212463,
      "learning_rate": 4.47513937304214e-06,
      "loss": 0.5188,
      "step": 263
    },
    {
      "epoch": 1.8045454545454547,
      "grad_norm": 0.4401354193687439,
      "learning_rate": 4.4689186089436365e-06,
      "loss": 0.5079,
      "step": 264
    },
    {
      "epoch": 1.8113636363636365,
      "grad_norm": 0.4051227569580078,
      "learning_rate": 4.462665574510169e-06,
      "loss": 0.5084,
      "step": 265
    },
    {
      "epoch": 1.8181818181818183,
      "grad_norm": 0.5269004106521606,
      "learning_rate": 4.456380372228208e-06,
      "loss": 0.528,
      "step": 266
    },
    {
      "epoch": 1.825,
      "grad_norm": 0.44371986389160156,
      "learning_rate": 4.450063105111447e-06,
      "loss": 0.5388,
      "step": 267
    },
    {
      "epoch": 1.8318181818181818,
      "grad_norm": 0.41871729493141174,
      "learning_rate": 4.443713876699124e-06,
      "loss": 0.502,
      "step": 268
    },
    {
      "epoch": 1.8386363636363636,
      "grad_norm": 0.464652419090271,
      "learning_rate": 4.4373327910543125e-06,
      "loss": 0.5045,
      "step": 269
    },
    {
      "epoch": 1.8454545454545455,
      "grad_norm": 0.43198078870773315,
      "learning_rate": 4.430919952762226e-06,
      "loss": 0.5149,
      "step": 270
    },
    {
      "epoch": 1.8522727272727273,
      "grad_norm": 0.42869722843170166,
      "learning_rate": 4.424475466928499e-06,
      "loss": 0.5329,
      "step": 271
    },
    {
      "epoch": 1.8590909090909091,
      "grad_norm": 0.529680609703064,
      "learning_rate": 4.417999439177465e-06,
      "loss": 0.4939,
      "step": 272
    },
    {
      "epoch": 1.865909090909091,
      "grad_norm": 0.4562613368034363,
      "learning_rate": 4.4114919756504275e-06,
      "loss": 0.5389,
      "step": 273
    },
    {
      "epoch": 1.8727272727272726,
      "grad_norm": 0.4480995237827301,
      "learning_rate": 4.404953183003916e-06,
      "loss": 0.51,
      "step": 274
    },
    {
      "epoch": 1.8795454545454544,
      "grad_norm": 0.41215088963508606,
      "learning_rate": 4.398383168407941e-06,
      "loss": 0.5386,
      "step": 275
    },
    {
      "epoch": 1.8863636363636362,
      "grad_norm": 0.8426409363746643,
      "learning_rate": 4.391782039544239e-06,
      "loss": 0.498,
      "step": 276
    },
    {
      "epoch": 1.893181818181818,
      "grad_norm": 0.5359936356544495,
      "learning_rate": 4.385149904604502e-06,
      "loss": 0.51,
      "step": 277
    },
    {
      "epoch": 1.9,
      "grad_norm": 0.4462437033653259,
      "learning_rate": 4.378486872288611e-06,
      "loss": 0.5082,
      "step": 278
    },
    {
      "epoch": 1.9068181818181817,
      "grad_norm": 0.445359468460083,
      "learning_rate": 4.371793051802849e-06,
      "loss": 0.5171,
      "step": 279
    },
    {
      "epoch": 1.9136363636363636,
      "grad_norm": 0.4223516583442688,
      "learning_rate": 4.365068552858116e-06,
      "loss": 0.5218,
      "step": 280
    },
    {
      "epoch": 1.9204545454545454,
      "grad_norm": 0.43929681181907654,
      "learning_rate": 4.358313485668124e-06,
      "loss": 0.5156,
      "step": 281
    },
    {
      "epoch": 1.9272727272727272,
      "grad_norm": 0.42535650730133057,
      "learning_rate": 4.3515279609476e-06,
      "loss": 0.4964,
      "step": 282
    },
    {
      "epoch": 1.934090909090909,
      "grad_norm": 0.46448391675949097,
      "learning_rate": 4.3447120899104615e-06,
      "loss": 0.5182,
      "step": 283
    },
    {
      "epoch": 1.940909090909091,
      "grad_norm": 0.4082206189632416,
      "learning_rate": 4.337865984268002e-06,
      "loss": 0.5175,
      "step": 284
    },
    {
      "epoch": 1.9477272727272728,
      "grad_norm": 0.42423564195632935,
      "learning_rate": 4.3309897562270525e-06,
      "loss": 0.5352,
      "step": 285
    },
    {
      "epoch": 1.9545454545454546,
      "grad_norm": 0.5240079760551453,
      "learning_rate": 4.324083518488151e-06,
      "loss": 0.5157,
      "step": 286
    },
    {
      "epoch": 1.9613636363636364,
      "grad_norm": 0.4517645835876465,
      "learning_rate": 4.317147384243688e-06,
      "loss": 0.5399,
      "step": 287
    },
    {
      "epoch": 1.9681818181818183,
      "grad_norm": 0.44054290652275085,
      "learning_rate": 4.3101814671760546e-06,
      "loss": 0.5293,
      "step": 288
    },
    {
      "epoch": 1.975,
      "grad_norm": 0.4256063401699066,
      "learning_rate": 4.303185881455778e-06,
      "loss": 0.5323,
      "step": 289
    },
    {
      "epoch": 1.981818181818182,
      "grad_norm": 0.43106094002723694,
      "learning_rate": 4.296160741739652e-06,
      "loss": 0.5259,
      "step": 290
    },
    {
      "epoch": 1.9886363636363638,
      "grad_norm": 0.43816477060317993,
      "learning_rate": 4.289106163168858e-06,
      "loss": 0.5098,
      "step": 291
    },
    {
      "epoch": 1.9954545454545456,
      "grad_norm": 1.1177959442138672,
      "learning_rate": 4.282022261367074e-06,
      "loss": 0.513,
      "step": 292
    }
  ],
  "logging_steps": 1,
  "max_steps": 876,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 6,
  "save_steps": 146,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 1.880633908528572e+19,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}