{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.9782608695652173,
  "eval_steps": 115,
  "global_step": 230,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008695652173913044,
      "grad_norm": 27.9770450592041,
      "learning_rate": 2.0000000000000002e-07,
      "loss": 2.2478,
      "step": 1
    },
    {
      "epoch": 0.008695652173913044,
      "eval_loss": 2.349900007247925,
      "eval_runtime": 27.4212,
      "eval_samples_per_second": 1.823,
      "eval_steps_per_second": 0.474,
      "step": 1
    },
    {
      "epoch": 0.017391304347826087,
      "grad_norm": 28.724267959594727,
      "learning_rate": 4.0000000000000003e-07,
      "loss": 2.1994,
      "step": 2
    },
    {
      "epoch": 0.02608695652173913,
      "grad_norm": 28.540616989135742,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.2229,
      "step": 3
    },
    {
      "epoch": 0.034782608695652174,
      "grad_norm": 28.697280883789062,
      "learning_rate": 8.000000000000001e-07,
      "loss": 2.2037,
      "step": 4
    },
    {
      "epoch": 0.043478260869565216,
      "grad_norm": 28.92486000061035,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 2.1981,
      "step": 5
    },
    {
      "epoch": 0.05217391304347826,
      "grad_norm": 29.778339385986328,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 2.1414,
      "step": 6
    },
    {
      "epoch": 0.06086956521739131,
      "grad_norm": 29.21839141845703,
      "learning_rate": 1.4000000000000001e-06,
      "loss": 2.1423,
      "step": 7
    },
    {
      "epoch": 0.06956521739130435,
      "grad_norm": 18.694778442382812,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 1.9974,
      "step": 8
    },
    {
      "epoch": 0.0782608695652174,
      "grad_norm": 17.274688720703125,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.9562,
      "step": 9
    },
    {
      "epoch": 0.08695652173913043,
      "grad_norm": 15.813633918762207,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 1.9486,
      "step": 10
    },
    {
      "epoch": 0.09565217391304348,
      "grad_norm": 6.602424144744873,
      "learning_rate": 2.2e-06,
      "loss": 1.7994,
      "step": 11
    },
    {
      "epoch": 0.10434782608695652,
      "grad_norm": 6.037355422973633,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.766,
      "step": 12
    },
    {
      "epoch": 0.11304347826086956,
      "grad_norm": 4.949301242828369,
      "learning_rate": 2.6e-06,
      "loss": 1.7459,
      "step": 13
    },
    {
      "epoch": 0.12173913043478261,
      "grad_norm": 2.621478796005249,
      "learning_rate": 2.8000000000000003e-06,
      "loss": 1.6899,
      "step": 14
    },
    {
      "epoch": 0.13043478260869565,
      "grad_norm": 3.7320492267608643,
      "learning_rate": 3e-06,
      "loss": 1.7185,
      "step": 15
    },
    {
      "epoch": 0.1391304347826087,
      "grad_norm": 3.5459675788879395,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 1.6988,
      "step": 16
    },
    {
      "epoch": 0.14782608695652175,
      "grad_norm": 3.048675060272217,
      "learning_rate": 3.4000000000000005e-06,
      "loss": 1.6822,
      "step": 17
    },
    {
      "epoch": 0.1565217391304348,
      "grad_norm": 2.540501832962036,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.6814,
      "step": 18
    },
    {
      "epoch": 0.16521739130434782,
      "grad_norm": 2.129035234451294,
      "learning_rate": 3.8000000000000005e-06,
      "loss": 1.659,
      "step": 19
    },
    {
      "epoch": 0.17391304347826086,
      "grad_norm": 1.9875308275222778,
      "learning_rate": 4.000000000000001e-06,
      "loss": 1.6249,
      "step": 20
    },
    {
      "epoch": 0.1826086956521739,
      "grad_norm": 1.7227815389633179,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 1.6261,
      "step": 21
    },
    {
      "epoch": 0.19130434782608696,
      "grad_norm": 1.6309900283813477,
      "learning_rate": 4.4e-06,
      "loss": 1.6043,
      "step": 22
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.4524799585342407,
      "learning_rate": 4.600000000000001e-06,
      "loss": 1.6014,
      "step": 23
    },
    {
      "epoch": 0.20869565217391303,
      "grad_norm": 1.5349876880645752,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.6097,
      "step": 24
    },
    {
      "epoch": 0.21739130434782608,
      "grad_norm": 1.4749583005905151,
      "learning_rate": 5e-06,
      "loss": 1.6108,
      "step": 25
    },
    {
      "epoch": 0.22608695652173913,
      "grad_norm": 1.3951363563537598,
      "learning_rate": 5.2e-06,
      "loss": 1.6179,
      "step": 26
    },
    {
      "epoch": 0.23478260869565218,
      "grad_norm": 1.4121828079223633,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.5842,
      "step": 27
    },
    {
      "epoch": 0.24347826086956523,
      "grad_norm": 1.259391188621521,
      "learning_rate": 5.600000000000001e-06,
      "loss": 1.563,
      "step": 28
    },
    {
      "epoch": 0.25217391304347825,
      "grad_norm": 1.2551765441894531,
      "learning_rate": 5.8e-06,
      "loss": 1.5829,
      "step": 29
    },
    {
      "epoch": 0.2608695652173913,
      "grad_norm": 1.2322245836257935,
      "learning_rate": 6e-06,
      "loss": 1.5877,
      "step": 30
    },
    {
      "epoch": 0.26956521739130435,
      "grad_norm": 1.136198878288269,
      "learning_rate": 6.200000000000001e-06,
      "loss": 1.5464,
      "step": 31
    },
    {
      "epoch": 0.2782608695652174,
      "grad_norm": 1.2440415620803833,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 1.537,
      "step": 32
    },
    {
      "epoch": 0.28695652173913044,
      "grad_norm": 1.1897674798965454,
      "learning_rate": 6.600000000000001e-06,
      "loss": 1.5757,
      "step": 33
    },
    {
      "epoch": 0.2956521739130435,
      "grad_norm": 1.173654556274414,
      "learning_rate": 6.800000000000001e-06,
      "loss": 1.5542,
      "step": 34
    },
    {
      "epoch": 0.30434782608695654,
      "grad_norm": 1.1894692182540894,
      "learning_rate": 7e-06,
      "loss": 1.5822,
      "step": 35
    },
    {
      "epoch": 0.3130434782608696,
      "grad_norm": 1.1219773292541504,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.5596,
      "step": 36
    },
    {
      "epoch": 0.3217391304347826,
      "grad_norm": 1.2288638353347778,
      "learning_rate": 7.4e-06,
      "loss": 1.5512,
      "step": 37
    },
    {
      "epoch": 0.33043478260869563,
      "grad_norm": 1.1236315965652466,
      "learning_rate": 7.600000000000001e-06,
      "loss": 1.5378,
      "step": 38
    },
    {
      "epoch": 0.3391304347826087,
      "grad_norm": 1.086173415184021,
      "learning_rate": 7.800000000000002e-06,
      "loss": 1.5475,
      "step": 39
    },
    {
      "epoch": 0.34782608695652173,
      "grad_norm": 1.13786780834198,
      "learning_rate": 8.000000000000001e-06,
      "loss": 1.5345,
      "step": 40
    },
    {
      "epoch": 0.3565217391304348,
      "grad_norm": 1.078940510749817,
      "learning_rate": 8.2e-06,
      "loss": 1.5736,
      "step": 41
    },
    {
      "epoch": 0.3652173913043478,
      "grad_norm": 1.06036376953125,
      "learning_rate": 8.400000000000001e-06,
      "loss": 1.5402,
      "step": 42
    },
    {
      "epoch": 0.3739130434782609,
      "grad_norm": 1.0889636278152466,
      "learning_rate": 8.6e-06,
      "loss": 1.5594,
      "step": 43
    },
    {
      "epoch": 0.3826086956521739,
      "grad_norm": 1.089091181755066,
      "learning_rate": 8.8e-06,
      "loss": 1.5298,
      "step": 44
    },
    {
      "epoch": 0.391304347826087,
      "grad_norm": 1.1010075807571411,
      "learning_rate": 9e-06,
      "loss": 1.54,
      "step": 45
    },
    {
      "epoch": 0.4,
      "grad_norm": 1.0824369192123413,
      "learning_rate": 9.200000000000002e-06,
      "loss": 1.5344,
      "step": 46
    },
    {
      "epoch": 0.40869565217391307,
      "grad_norm": 1.130077838897705,
      "learning_rate": 9.4e-06,
      "loss": 1.5425,
      "step": 47
    },
    {
      "epoch": 0.41739130434782606,
      "grad_norm": 1.0903602838516235,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.5515,
      "step": 48
    },
    {
      "epoch": 0.4260869565217391,
      "grad_norm": 1.1305510997772217,
      "learning_rate": 9.800000000000001e-06,
      "loss": 1.5288,
      "step": 49
    },
    {
      "epoch": 0.43478260869565216,
      "grad_norm": 1.1045012474060059,
      "learning_rate": 1e-05,
      "loss": 1.5156,
      "step": 50
    },
    {
      "epoch": 0.4434782608695652,
      "grad_norm": 1.132230281829834,
      "learning_rate": 1.02e-05,
      "loss": 1.5193,
      "step": 51
    },
    {
      "epoch": 0.45217391304347826,
      "grad_norm": 1.1561248302459717,
      "learning_rate": 1.04e-05,
      "loss": 1.5198,
      "step": 52
    },
    {
      "epoch": 0.4608695652173913,
      "grad_norm": 1.1394556760787964,
      "learning_rate": 1.0600000000000002e-05,
      "loss": 1.5609,
      "step": 53
    },
    {
      "epoch": 0.46956521739130436,
      "grad_norm": 1.2164558172225952,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 1.5497,
      "step": 54
    },
    {
      "epoch": 0.4782608695652174,
      "grad_norm": 1.1864646673202515,
      "learning_rate": 1.1000000000000001e-05,
      "loss": 1.5278,
      "step": 55
    },
    {
      "epoch": 0.48695652173913045,
      "grad_norm": 1.2279999256134033,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 1.5035,
      "step": 56
    },
    {
      "epoch": 0.4956521739130435,
      "grad_norm": 1.1659705638885498,
      "learning_rate": 1.14e-05,
      "loss": 1.5278,
      "step": 57
    },
    {
      "epoch": 0.5043478260869565,
      "grad_norm": 1.1985366344451904,
      "learning_rate": 1.16e-05,
      "loss": 1.5238,
      "step": 58
    },
    {
      "epoch": 0.5130434782608696,
      "grad_norm": 1.1645257472991943,
      "learning_rate": 1.18e-05,
      "loss": 1.5181,
      "step": 59
    },
    {
      "epoch": 0.5217391304347826,
      "grad_norm": 1.2394540309906006,
      "learning_rate": 1.2e-05,
      "loss": 1.5253,
      "step": 60
    },
    {
      "epoch": 0.5304347826086957,
      "grad_norm": 1.1257425546646118,
      "learning_rate": 1.22e-05,
      "loss": 1.5086,
      "step": 61
    },
    {
      "epoch": 0.5391304347826087,
      "grad_norm": 1.205934762954712,
      "learning_rate": 1.2400000000000002e-05,
      "loss": 1.5302,
      "step": 62
    },
    {
      "epoch": 0.5478260869565217,
      "grad_norm": 1.2120305299758911,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 1.5126,
      "step": 63
    },
    {
      "epoch": 0.5565217391304348,
      "grad_norm": 1.0931782722473145,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 1.4921,
      "step": 64
    },
    {
      "epoch": 0.5652173913043478,
      "grad_norm": 1.0688101053237915,
      "learning_rate": 1.3000000000000001e-05,
      "loss": 1.5567,
      "step": 65
    },
    {
      "epoch": 0.5739130434782609,
      "grad_norm": 1.1482597589492798,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 1.521,
      "step": 66
    },
    {
      "epoch": 0.5826086956521739,
      "grad_norm": 1.11100435256958,
      "learning_rate": 1.3400000000000002e-05,
      "loss": 1.4961,
      "step": 67
    },
    {
      "epoch": 0.591304347826087,
      "grad_norm": 1.1361342668533325,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 1.5143,
      "step": 68
    },
    {
      "epoch": 0.6,
      "grad_norm": 1.0989760160446167,
      "learning_rate": 1.38e-05,
      "loss": 1.5044,
      "step": 69
    },
    {
      "epoch": 0.6086956521739131,
      "grad_norm": 1.1394468545913696,
      "learning_rate": 1.4e-05,
      "loss": 1.5028,
      "step": 70
    },
    {
      "epoch": 0.6173913043478261,
      "grad_norm": 1.1451812982559204,
      "learning_rate": 1.4200000000000001e-05,
      "loss": 1.531,
      "step": 71
    },
    {
      "epoch": 0.6260869565217392,
      "grad_norm": 1.118803858757019,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 1.5319,
      "step": 72
    },
    {
      "epoch": 0.6347826086956522,
      "grad_norm": 1.1544065475463867,
      "learning_rate": 1.46e-05,
      "loss": 1.524,
      "step": 73
    },
    {
      "epoch": 0.6434782608695652,
      "grad_norm": 1.1219969987869263,
      "learning_rate": 1.48e-05,
      "loss": 1.4892,
      "step": 74
    },
    {
      "epoch": 0.6521739130434783,
      "grad_norm": 1.1506638526916504,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.5218,
      "step": 75
    },
    {
      "epoch": 0.6608695652173913,
      "grad_norm": 1.1717455387115479,
      "learning_rate": 1.5200000000000002e-05,
      "loss": 1.5207,
      "step": 76
    },
    {
      "epoch": 0.6695652173913044,
      "grad_norm": 1.110942006111145,
      "learning_rate": 1.54e-05,
      "loss": 1.5658,
      "step": 77
    },
    {
      "epoch": 0.6782608695652174,
      "grad_norm": 1.2768936157226562,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 1.501,
      "step": 78
    },
    {
      "epoch": 0.6869565217391305,
      "grad_norm": 1.26193368434906,
      "learning_rate": 1.58e-05,
      "loss": 1.5385,
      "step": 79
    },
    {
      "epoch": 0.6956521739130435,
      "grad_norm": 1.2869552373886108,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 1.523,
      "step": 80
    },
    {
      "epoch": 0.7043478260869566,
      "grad_norm": 1.2835365533828735,
      "learning_rate": 1.62e-05,
      "loss": 1.5154,
      "step": 81
    },
    {
      "epoch": 0.7130434782608696,
      "grad_norm": 1.2280563116073608,
      "learning_rate": 1.64e-05,
      "loss": 1.5089,
      "step": 82
    },
    {
      "epoch": 0.7217391304347827,
      "grad_norm": 1.2316559553146362,
      "learning_rate": 1.66e-05,
      "loss": 1.5296,
      "step": 83
    },
    {
      "epoch": 0.7304347826086957,
      "grad_norm": 1.2359421253204346,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 1.5187,
      "step": 84
    },
    {
      "epoch": 0.7391304347826086,
      "grad_norm": 1.1773601770401,
      "learning_rate": 1.7e-05,
      "loss": 1.5102,
      "step": 85
    },
    {
      "epoch": 0.7478260869565218,
      "grad_norm": 1.2701928615570068,
      "learning_rate": 1.72e-05,
      "loss": 1.5049,
      "step": 86
    },
    {
      "epoch": 0.7565217391304347,
      "grad_norm": 1.2413982152938843,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 1.4875,
      "step": 87
    },
    {
      "epoch": 0.7652173913043478,
      "grad_norm": 1.1729072332382202,
      "learning_rate": 1.76e-05,
      "loss": 1.508,
      "step": 88
    },
    {
      "epoch": 0.7739130434782608,
      "grad_norm": 1.2796519994735718,
      "learning_rate": 1.7800000000000002e-05,
      "loss": 1.4953,
      "step": 89
    },
    {
      "epoch": 0.782608695652174,
      "grad_norm": 1.1654998064041138,
      "learning_rate": 1.8e-05,
      "loss": 1.5115,
      "step": 90
    },
    {
      "epoch": 0.7913043478260869,
      "grad_norm": 1.2365690469741821,
      "learning_rate": 1.8200000000000002e-05,
      "loss": 1.4861,
      "step": 91
    },
    {
      "epoch": 0.8,
      "grad_norm": 1.2018343210220337,
      "learning_rate": 1.8400000000000003e-05,
      "loss": 1.5328,
      "step": 92
    },
    {
      "epoch": 0.808695652173913,
      "grad_norm": 1.1778483390808105,
      "learning_rate": 1.86e-05,
      "loss": 1.4973,
      "step": 93
    },
    {
      "epoch": 0.8173913043478261,
      "grad_norm": 1.2713650465011597,
      "learning_rate": 1.88e-05,
      "loss": 1.5172,
      "step": 94
    },
    {
      "epoch": 0.8260869565217391,
      "grad_norm": 1.3153022527694702,
      "learning_rate": 1.9e-05,
      "loss": 1.502,
      "step": 95
    },
    {
      "epoch": 0.8347826086956521,
      "grad_norm": 1.1676228046417236,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 1.5095,
      "step": 96
    },
    {
      "epoch": 0.8434782608695652,
      "grad_norm": 1.3076750040054321,
      "learning_rate": 1.94e-05,
      "loss": 1.5199,
      "step": 97
    },
    {
      "epoch": 0.8521739130434782,
      "grad_norm": 1.1498546600341797,
      "learning_rate": 1.9600000000000002e-05,
      "loss": 1.514,
      "step": 98
    },
    {
      "epoch": 0.8608695652173913,
      "grad_norm": 1.2685290575027466,
      "learning_rate": 1.98e-05,
      "loss": 1.503,
      "step": 99
    },
    {
      "epoch": 0.8695652173913043,
      "grad_norm": 1.2121868133544922,
      "learning_rate": 2e-05,
      "loss": 1.5182,
      "step": 100
    },
    {
      "epoch": 0.8782608695652174,
      "grad_norm": 1.269320011138916,
      "learning_rate": 1.9997080140801932e-05,
      "loss": 1.4833,
      "step": 101
    },
    {
      "epoch": 0.8869565217391304,
      "grad_norm": 1.21848726272583,
      "learning_rate": 1.998832226832327e-05,
      "loss": 1.517,
      "step": 102
    },
    {
      "epoch": 0.8956521739130435,
      "grad_norm": 1.2216753959655762,
      "learning_rate": 1.9973731496914914e-05,
      "loss": 1.4846,
      "step": 103
    },
    {
      "epoch": 0.9043478260869565,
      "grad_norm": 1.162424087524414,
      "learning_rate": 1.995331634717649e-05,
      "loss": 1.5128,
      "step": 104
    },
    {
      "epoch": 0.9130434782608695,
      "grad_norm": 1.173030972480774,
      "learning_rate": 1.992708874098054e-05,
      "loss": 1.5261,
      "step": 105
    },
    {
      "epoch": 0.9217391304347826,
      "grad_norm": 1.0961143970489502,
      "learning_rate": 1.9895063994510512e-05,
      "loss": 1.4848,
      "step": 106
    },
    {
      "epoch": 0.9304347826086956,
      "grad_norm": 1.2185754776000977,
      "learning_rate": 1.985726080931651e-05,
      "loss": 1.4958,
      "step": 107
    },
    {
      "epoch": 0.9391304347826087,
      "grad_norm": 1.1143345832824707,
      "learning_rate": 1.9813701261394136e-05,
      "loss": 1.5084,
      "step": 108
    },
    {
      "epoch": 0.9478260869565217,
      "grad_norm": 1.2007545232772827,
      "learning_rate": 1.9764410788292724e-05,
      "loss": 1.5032,
      "step": 109
    },
    {
      "epoch": 0.9565217391304348,
      "grad_norm": 1.1655194759368896,
      "learning_rate": 1.9709418174260523e-05,
      "loss": 1.5104,
      "step": 110
    },
    {
      "epoch": 0.9652173913043478,
      "grad_norm": 1.2042982578277588,
      "learning_rate": 1.9648755533435517e-05,
      "loss": 1.5274,
      "step": 111
    },
    {
      "epoch": 0.9739130434782609,
      "grad_norm": 1.135833740234375,
      "learning_rate": 1.9582458291091664e-05,
      "loss": 1.5159,
      "step": 112
    },
    {
      "epoch": 0.9826086956521739,
      "grad_norm": 1.1194689273834229,
      "learning_rate": 1.9510565162951538e-05,
      "loss": 1.4841,
      "step": 113
    },
    {
      "epoch": 0.991304347826087,
      "grad_norm": 1.1721700429916382,
      "learning_rate": 1.9433118132577432e-05,
      "loss": 1.4958,
      "step": 114
    },
    {
      "epoch": 1.0,
      "grad_norm": 1.1297590732574463,
      "learning_rate": 1.9350162426854152e-05,
      "loss": 1.4974,
      "step": 115
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.658707618713379,
      "eval_runtime": 27.8072,
      "eval_samples_per_second": 1.798,
      "eval_steps_per_second": 0.468,
      "step": 115
    },
    {
      "epoch": 1.008695652173913,
      "grad_norm": 1.189160943031311,
      "learning_rate": 1.9261746489577767e-05,
      "loss": 1.4661,
      "step": 116
    },
    {
      "epoch": 1.017391304347826,
      "grad_norm": 1.2425116300582886,
      "learning_rate": 1.9167921953165827e-05,
      "loss": 1.5087,
      "step": 117
    },
    {
      "epoch": 1.0043478260869565,
      "grad_norm": 1.1860764026641846,
      "learning_rate": 1.9068743608505454e-05,
      "loss": 1.4146,
      "step": 118
    },
    {
      "epoch": 1.0130434782608695,
      "grad_norm": 1.4199373722076416,
      "learning_rate": 1.896426937295704e-05,
      "loss": 1.3588,
      "step": 119
    },
    {
      "epoch": 1.0217391304347827,
      "grad_norm": 1.3657387495040894,
      "learning_rate": 1.8854560256532098e-05,
      "loss": 1.3389,
      "step": 120
    },
    {
      "epoch": 1.0304347826086957,
      "grad_norm": 1.4500665664672852,
      "learning_rate": 1.873968032626518e-05,
      "loss": 1.3416,
      "step": 121
    },
    {
      "epoch": 1.0391304347826087,
      "grad_norm": 1.4582746028900146,
      "learning_rate": 1.8619696668800494e-05,
      "loss": 1.3226,
      "step": 122
    },
    {
      "epoch": 1.0478260869565217,
      "grad_norm": 1.5258561372756958,
      "learning_rate": 1.8494679351215212e-05,
      "loss": 1.3183,
      "step": 123
    },
    {
      "epoch": 1.0565217391304347,
      "grad_norm": 1.42508864402771,
      "learning_rate": 1.8364701380102267e-05,
      "loss": 1.3535,
      "step": 124
    },
    {
      "epoch": 1.065217391304348,
      "grad_norm": 1.4576255083084106,
      "learning_rate": 1.8229838658936566e-05,
      "loss": 1.2911,
      "step": 125
    },
    {
      "epoch": 1.0739130434782609,
      "grad_norm": 1.3179984092712402,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 1.3239,
      "step": 126
    },
    {
      "epoch": 1.0826086956521739,
      "grad_norm": 1.3821628093719482,
      "learning_rate": 1.7945776797137544e-05,
      "loss": 1.3119,
      "step": 127
    },
    {
      "epoch": 1.0913043478260869,
      "grad_norm": 1.3231525421142578,
      "learning_rate": 1.7796743540632226e-05,
      "loss": 1.304,
      "step": 128
    },
    {
      "epoch": 1.1,
      "grad_norm": 1.3649060726165771,
      "learning_rate": 1.7643157205458483e-05,
      "loss": 1.3113,
      "step": 129
    },
    {
      "epoch": 1.108695652173913,
      "grad_norm": 1.2634330987930298,
      "learning_rate": 1.7485107481711014e-05,
      "loss": 1.3116,
      "step": 130
    },
    {
      "epoch": 1.117391304347826,
      "grad_norm": 1.2672367095947266,
      "learning_rate": 1.7322686665977738e-05,
      "loss": 1.338,
      "step": 131
    },
    {
      "epoch": 1.126086956521739,
      "grad_norm": 1.2646681070327759,
      "learning_rate": 1.715598960744121e-05,
      "loss": 1.3083,
      "step": 132
    },
    {
      "epoch": 1.134782608695652,
      "grad_norm": 1.2951526641845703,
      "learning_rate": 1.6985113652489374e-05,
      "loss": 1.3318,
      "step": 133
    },
    {
      "epoch": 1.1434782608695653,
      "grad_norm": 1.2833969593048096,
      "learning_rate": 1.6810158587867973e-05,
      "loss": 1.3544,
      "step": 134
    },
    {
      "epoch": 1.1521739130434783,
      "grad_norm": 1.2921353578567505,
      "learning_rate": 1.6631226582407954e-05,
      "loss": 1.2928,
      "step": 135
    },
    {
      "epoch": 1.1608695652173913,
      "grad_norm": 1.2432658672332764,
      "learning_rate": 1.6448422127361707e-05,
      "loss": 1.3029,
      "step": 136
    },
    {
      "epoch": 1.1695652173913043,
      "grad_norm": 1.2263303995132446,
      "learning_rate": 1.626185197538314e-05,
      "loss": 1.318,
      "step": 137
    },
    {
      "epoch": 1.1782608695652175,
      "grad_norm": 1.3209857940673828,
      "learning_rate": 1.6071625078187113e-05,
      "loss": 1.3315,
      "step": 138
    },
    {
      "epoch": 1.1869565217391305,
      "grad_norm": 1.279836654663086,
      "learning_rate": 1.5877852522924733e-05,
      "loss": 1.3287,
      "step": 139
    },
    {
      "epoch": 1.1956521739130435,
      "grad_norm": 1.2511873245239258,
      "learning_rate": 1.568064746731156e-05,
      "loss": 1.2982,
      "step": 140
    },
    {
      "epoch": 1.2043478260869565,
      "grad_norm": 1.3275624513626099,
      "learning_rate": 1.5480125073546705e-05,
      "loss": 1.2863,
      "step": 141
    },
    {
      "epoch": 1.2130434782608694,
      "grad_norm": 1.2315387725830078,
      "learning_rate": 1.527640244106133e-05,
      "loss": 1.3091,
      "step": 142
    },
    {
      "epoch": 1.2217391304347827,
      "grad_norm": 1.2301815748214722,
      "learning_rate": 1.5069598538135905e-05,
      "loss": 1.3336,
      "step": 143
    },
    {
      "epoch": 1.2304347826086957,
      "grad_norm": 1.2598013877868652,
      "learning_rate": 1.485983413242606e-05,
      "loss": 1.3238,
      "step": 144
    },
    {
      "epoch": 1.2391304347826086,
      "grad_norm": 1.223031759262085,
      "learning_rate": 1.4647231720437687e-05,
      "loss": 1.2943,
      "step": 145
    },
    {
      "epoch": 1.2478260869565219,
      "grad_norm": 1.2285338640213013,
      "learning_rate": 1.4431915455992416e-05,
      "loss": 1.3045,
      "step": 146
    },
    {
      "epoch": 1.2565217391304349,
      "grad_norm": 1.176661729812622,
      "learning_rate": 1.4214011077725293e-05,
      "loss": 1.2958,
      "step": 147
    },
    {
      "epoch": 1.2652173913043478,
      "grad_norm": 1.2064881324768066,
      "learning_rate": 1.3993645835656955e-05,
      "loss": 1.2981,
      "step": 148
    },
    {
      "epoch": 1.2739130434782608,
      "grad_norm": 1.294325828552246,
      "learning_rate": 1.3770948416883205e-05,
      "loss": 1.3422,
      "step": 149
    },
    {
      "epoch": 1.2826086956521738,
      "grad_norm": 1.2420161962509155,
      "learning_rate": 1.3546048870425356e-05,
      "loss": 1.3275,
      "step": 150
    },
    {
      "epoch": 1.2913043478260868,
      "grad_norm": 1.2433290481567383,
      "learning_rate": 1.3319078531285286e-05,
      "loss": 1.2862,
      "step": 151
    },
    {
      "epoch": 1.3,
      "grad_norm": 1.1643811464309692,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 1.2696,
      "step": 152
    },
    {
      "epoch": 1.308695652173913,
      "grad_norm": 1.2214665412902832,
      "learning_rate": 1.2859456783986892e-05,
      "loss": 1.2988,
      "step": 153
    },
    {
      "epoch": 1.317391304347826,
      "grad_norm": 1.2694072723388672,
      "learning_rate": 1.262707378198587e-05,
      "loss": 1.3163,
      "step": 154
    },
    {
      "epoch": 1.3260869565217392,
      "grad_norm": 1.229480504989624,
      "learning_rate": 1.2393156642875579e-05,
      "loss": 1.3025,
      "step": 155
    },
    {
      "epoch": 1.3347826086956522,
      "grad_norm": 1.2135909795761108,
      "learning_rate": 1.2157841967678064e-05,
      "loss": 1.3184,
      "step": 156
    },
    {
      "epoch": 1.3434782608695652,
      "grad_norm": 1.283299446105957,
      "learning_rate": 1.1921267173537085e-05,
      "loss": 1.3269,
      "step": 157
    },
    {
      "epoch": 1.3521739130434782,
      "grad_norm": 1.1701160669326782,
      "learning_rate": 1.1683570413470384e-05,
      "loss": 1.2864,
      "step": 158
    },
    {
      "epoch": 1.3608695652173912,
      "grad_norm": 1.1845210790634155,
      "learning_rate": 1.1444890495692214e-05,
      "loss": 1.3249,
      "step": 159
    },
    {
      "epoch": 1.3695652173913042,
      "grad_norm": 1.2007068395614624,
      "learning_rate": 1.1205366802553231e-05,
      "loss": 1.3045,
      "step": 160
    },
    {
      "epoch": 1.3782608695652174,
      "grad_norm": 1.2322540283203125,
      "learning_rate": 1.0965139209145153e-05,
      "loss": 1.3132,
      "step": 161
    },
    {
      "epoch": 1.3869565217391304,
      "grad_norm": 1.1864914894104004,
      "learning_rate": 1.0724348001617626e-05,
      "loss": 1.3101,
      "step": 162
    },
    {
      "epoch": 1.3956521739130434,
      "grad_norm": 1.1936239004135132,
      "learning_rate": 1.0483133795255072e-05,
      "loss": 1.2983,
      "step": 163
    },
    {
      "epoch": 1.4043478260869566,
      "grad_norm": 1.213137149810791,
      "learning_rate": 1.0241637452361323e-05,
      "loss": 1.336,
      "step": 164
    },
    {
      "epoch": 1.4130434782608696,
      "grad_norm": 1.1825729608535767,
      "learning_rate": 1e-05,
      "loss": 1.3098,
      "step": 165
    },
    {
      "epoch": 1.4217391304347826,
      "grad_norm": 1.1550593376159668,
      "learning_rate": 9.75836254763868e-06,
      "loss": 1.3096,
      "step": 166
    },
    {
      "epoch": 1.4304347826086956,
      "grad_norm": 1.1386010646820068,
      "learning_rate": 9.516866204744932e-06,
      "loss": 1.3167,
      "step": 167
    },
    {
      "epoch": 1.4391304347826086,
      "grad_norm": 1.2719810009002686,
      "learning_rate": 9.275651998382377e-06,
      "loss": 1.2959,
      "step": 168
    },
    {
      "epoch": 1.4478260869565218,
      "grad_norm": 1.2225759029388428,
      "learning_rate": 9.034860790854848e-06,
      "loss": 1.2942,
      "step": 169
    },
    {
      "epoch": 1.4565217391304348,
      "grad_norm": 1.1454390287399292,
      "learning_rate": 8.79463319744677e-06,
      "loss": 1.2971,
      "step": 170
    },
    {
      "epoch": 1.4652173913043478,
      "grad_norm": 1.2099990844726562,
      "learning_rate": 8.55510950430779e-06,
      "loss": 1.2989,
      "step": 171
    },
    {
      "epoch": 1.4739130434782608,
      "grad_norm": 1.2047284841537476,
      "learning_rate": 8.316429586529616e-06,
      "loss": 1.3048,
      "step": 172
    },
    {
      "epoch": 1.482608695652174,
      "grad_norm": 1.1526693105697632,
      "learning_rate": 8.078732826462917e-06,
      "loss": 1.2794,
      "step": 173
    },
    {
      "epoch": 1.491304347826087,
      "grad_norm": 1.188796877861023,
      "learning_rate": 7.84215803232194e-06,
      "loss": 1.305,
      "step": 174
    },
    {
      "epoch": 1.5,
      "grad_norm": 1.1713995933532715,
      "learning_rate": 7.606843357124426e-06,
      "loss": 1.3153,
      "step": 175
    },
    {
      "epoch": 1.508695652173913,
      "grad_norm": 1.2031924724578857,
      "learning_rate": 7.372926218014131e-06,
      "loss": 1.3235,
      "step": 176
    },
    {
      "epoch": 1.517391304347826,
      "grad_norm": 1.1624150276184082,
      "learning_rate": 7.140543216013109e-06,
      "loss": 1.2881,
      "step": 177
    },
    {
      "epoch": 1.526086956521739,
      "grad_norm": 1.1415232419967651,
      "learning_rate": 6.909830056250527e-06,
      "loss": 1.3027,
      "step": 178
    },
    {
      "epoch": 1.5347826086956522,
      "grad_norm": 1.2224611043930054,
      "learning_rate": 6.680921468714718e-06,
      "loss": 1.3007,
      "step": 179
    },
    {
      "epoch": 1.5434782608695652,
      "grad_norm": 1.1667900085449219,
      "learning_rate": 6.453951129574644e-06,
      "loss": 1.2827,
      "step": 180
    },
    {
      "epoch": 1.5521739130434784,
      "grad_norm": 1.1897790431976318,
      "learning_rate": 6.229051583116796e-06,
      "loss": 1.3233,
      "step": 181
    },
    {
      "epoch": 1.5608695652173914,
      "grad_norm": 1.1487510204315186,
      "learning_rate": 6.006354164343047e-06,
      "loss": 1.3236,
      "step": 182
    },
    {
      "epoch": 1.5695652173913044,
      "grad_norm": 1.1928635835647583,
      "learning_rate": 5.785988922274711e-06,
      "loss": 1.3001,
      "step": 183
    },
    {
      "epoch": 1.5782608695652174,
      "grad_norm": 1.1951863765716553,
      "learning_rate": 5.5680845440075885e-06,
      "loss": 1.2964,
      "step": 184
    },
    {
      "epoch": 1.5869565217391304,
      "grad_norm": 1.1351429224014282,
      "learning_rate": 5.352768279562315e-06,
      "loss": 1.3045,
      "step": 185
    },
    {
      "epoch": 1.5956521739130434,
      "grad_norm": 1.117018461227417,
      "learning_rate": 5.14016586757394e-06,
      "loss": 1.2889,
      "step": 186
    },
    {
      "epoch": 1.6043478260869564,
      "grad_norm": 1.1258199214935303,
      "learning_rate": 4.930401461864099e-06,
      "loss": 1.3015,
      "step": 187
    },
    {
      "epoch": 1.6130434782608696,
      "grad_norm": 1.1547445058822632,
      "learning_rate": 4.7235975589386715e-06,
      "loss": 1.2961,
      "step": 188
    },
    {
      "epoch": 1.6217391304347826,
      "grad_norm": 1.1343611478805542,
      "learning_rate": 4.519874926453303e-06,
      "loss": 1.2912,
      "step": 189
    },
    {
      "epoch": 1.6304347826086958,
      "grad_norm": 1.0676140785217285,
      "learning_rate": 4.319352532688444e-06,
      "loss": 1.2999,
      "step": 190
    },
    {
      "epoch": 1.6391304347826088,
      "grad_norm": 1.0918030738830566,
      "learning_rate": 4.12214747707527e-06,
      "loss": 1.2933,
      "step": 191
    },
    {
      "epoch": 1.6478260869565218,
      "grad_norm": 1.1060066223144531,
      "learning_rate": 3.9283749218128885e-06,
      "loss": 1.2918,
      "step": 192
    },
    {
      "epoch": 1.6565217391304348,
      "grad_norm": 1.095795750617981,
      "learning_rate": 3.738148024616863e-06,
      "loss": 1.299,
      "step": 193
    },
    {
      "epoch": 1.6652173913043478,
      "grad_norm": 1.0955934524536133,
      "learning_rate": 3.5515778726382967e-06,
      "loss": 1.2752,
      "step": 194
    },
    {
      "epoch": 1.6739130434782608,
      "grad_norm": 1.0838819742202759,
      "learning_rate": 3.3687734175920505e-06,
      "loss": 1.2836,
      "step": 195
    },
    {
      "epoch": 1.6826086956521737,
      "grad_norm": 1.0857439041137695,
      "learning_rate": 3.1898414121320277e-06,
      "loss": 1.2912,
      "step": 196
    },
    {
      "epoch": 1.691304347826087,
      "grad_norm": 1.1158169507980347,
      "learning_rate": 3.0148863475106315e-06,
      "loss": 1.3102,
      "step": 197
    },
    {
      "epoch": 1.7,
      "grad_norm": 1.0872242450714111,
      "learning_rate": 2.8440103925587904e-06,
      "loss": 1.3049,
      "step": 198
    },
    {
      "epoch": 1.7086956521739132,
      "grad_norm": 1.0633604526519775,
      "learning_rate": 2.6773133340222677e-06,
      "loss": 1.272,
      "step": 199
    },
    {
      "epoch": 1.7173913043478262,
      "grad_norm": 1.0870684385299683,
      "learning_rate": 2.514892518288988e-06,
      "loss": 1.3042,
      "step": 200
    },
    {
      "epoch": 1.7260869565217392,
      "grad_norm": 1.069403052330017,
      "learning_rate": 2.3568427945415163e-06,
      "loss": 1.2767,
      "step": 201
    },
    {
      "epoch": 1.7347826086956522,
      "grad_norm": 1.0961412191390991,
      "learning_rate": 2.2032564593677773e-06,
      "loss": 1.3146,
      "step": 202
    },
    {
      "epoch": 1.7434782608695651,
      "grad_norm": 1.0615010261535645,
      "learning_rate": 2.0542232028624585e-06,
      "loss": 1.3176,
      "step": 203
    },
    {
      "epoch": 1.7521739130434781,
      "grad_norm": 1.081389307975769,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 1.2941,
      "step": 204
    },
    {
      "epoch": 1.7608695652173914,
      "grad_norm": 1.055414080619812,
      "learning_rate": 1.7701613410634367e-06,
      "loss": 1.2926,
      "step": 205
    },
    {
      "epoch": 1.7695652173913043,
      "grad_norm": 1.0605552196502686,
      "learning_rate": 1.6352986198977327e-06,
      "loss": 1.2903,
      "step": 206
    },
    {
      "epoch": 1.7782608695652173,
      "grad_norm": 1.083540916442871,
      "learning_rate": 1.5053206487847916e-06,
      "loss": 1.302,
      "step": 207
    },
    {
      "epoch": 1.7869565217391306,
      "grad_norm": 1.0532197952270508,
      "learning_rate": 1.3803033311995072e-06,
      "loss": 1.3106,
      "step": 208
    },
    {
      "epoch": 1.7956521739130435,
      "grad_norm": 1.0551762580871582,
      "learning_rate": 1.2603196737348211e-06,
      "loss": 1.2788,
      "step": 209
    },
    {
      "epoch": 1.8043478260869565,
      "grad_norm": 1.0607982873916626,
      "learning_rate": 1.1454397434679022e-06,
      "loss": 1.2925,
      "step": 210
    },
    {
      "epoch": 1.8130434782608695,
      "grad_norm": 1.0949007272720337,
      "learning_rate": 1.0357306270429623e-06,
      "loss": 1.3353,
      "step": 211
    },
    {
      "epoch": 1.8217391304347825,
      "grad_norm": 1.0425626039505005,
      "learning_rate": 9.312563914945461e-07,
      "loss": 1.2878,
      "step": 212
    },
    {
      "epoch": 1.8304347826086955,
      "grad_norm": 1.0543763637542725,
      "learning_rate": 8.320780468341761e-07,
      "loss": 1.2827,
      "step": 213
    },
    {
      "epoch": 1.8391304347826087,
      "grad_norm": 1.0291991233825684,
      "learning_rate": 7.382535104222366e-07,
      "loss": 1.2935,
      "step": 214
    },
    {
      "epoch": 1.8478260869565217,
      "grad_norm": 1.0535385608673096,
      "learning_rate": 6.498375731458529e-07,
      "loss": 1.3051,
      "step": 215
    },
    {
      "epoch": 1.856521739130435,
      "grad_norm": 1.0559706687927246,
      "learning_rate": 5.668818674225684e-07,
      "loss": 1.278,
      "step": 216
    },
    {
      "epoch": 1.865217391304348,
      "grad_norm": 1.0581434965133667,
      "learning_rate": 4.894348370484648e-07,
      "loss": 1.3069,
      "step": 217
    },
    {
      "epoch": 1.873913043478261,
      "grad_norm": 1.039284586906433,
      "learning_rate": 4.1754170890833777e-07,
      "loss": 1.2723,
      "step": 218
    },
    {
      "epoch": 1.882608695652174,
      "grad_norm": 1.0665440559387207,
      "learning_rate": 3.5124446656448654e-07,
      "loss": 1.2797,
      "step": 219
    },
    {
      "epoch": 1.891304347826087,
      "grad_norm": 1.045371413230896,
      "learning_rate": 2.905818257394799e-07,
      "loss": 1.307,
      "step": 220
    },
    {
      "epoch": 1.9,
      "grad_norm": 1.035818338394165,
      "learning_rate": 2.355892117072789e-07,
      "loss": 1.3083,
      "step": 221
    },
    {
      "epoch": 1.908695652173913,
      "grad_norm": 1.044579029083252,
      "learning_rate": 1.8629873860586567e-07,
      "loss": 1.2914,
      "step": 222
    },
    {
      "epoch": 1.9173913043478261,
      "grad_norm": 1.0681899785995483,
      "learning_rate": 1.4273919068349184e-07,
      "loss": 1.3293,
      "step": 223
    },
    {
      "epoch": 1.9260869565217391,
      "grad_norm": 1.0404493808746338,
      "learning_rate": 1.0493600548948879e-07,
      "loss": 1.269,
      "step": 224
    },
    {
      "epoch": 1.9347826086956523,
      "grad_norm": 1.0317591428756714,
      "learning_rate": 7.291125901946027e-08,
      "loss": 1.2957,
      "step": 225
    },
    {
      "epoch": 1.9434782608695653,
      "grad_norm": 1.0500375032424927,
      "learning_rate": 4.6683652823513725e-08,
      "loss": 1.2982,
      "step": 226
    },
    {
      "epoch": 1.9521739130434783,
      "grad_norm": 1.0283217430114746,
      "learning_rate": 2.6268503085089547e-08,
      "loss": 1.2891,
      "step": 227
    },
    {
      "epoch": 1.9608695652173913,
      "grad_norm": 1.0270622968673706,
      "learning_rate": 1.1677731676733584e-08,
      "loss": 1.306,
      "step": 228
    },
    {
      "epoch": 1.9695652173913043,
      "grad_norm": 1.0450721979141235,
      "learning_rate": 2.9198591980705847e-09,
      "loss": 1.2695,
      "step": 229
    },
    {
      "epoch": 1.9782608695652173,
      "grad_norm": 1.0649240016937256,
      "learning_rate": 0.0,
      "loss": 1.325,
      "step": 230
    },
    {
      "epoch": 1.9782608695652173,
      "eval_loss": 1.6434073448181152,
      "eval_runtime": 27.6497,
      "eval_samples_per_second": 1.808,
      "eval_steps_per_second": 0.47,
      "step": 230
    }
  ],
  "logging_steps": 1,
  "max_steps": 230,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 39,
  "total_flos": 2.71497194303914e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}