{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 116,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008620689655172414,
      "grad_norm": 0.490234375,
      "learning_rate": 9.913793103448277e-06,
      "loss": 1.8897,
      "step": 1
    },
    {
      "epoch": 0.017241379310344827,
      "grad_norm": 0.44921875,
      "learning_rate": 9.827586206896553e-06,
      "loss": 1.7847,
      "step": 2
    },
    {
      "epoch": 0.02586206896551724,
      "grad_norm": 0.470703125,
      "learning_rate": 9.741379310344829e-06,
      "loss": 1.8435,
      "step": 3
    },
    {
      "epoch": 0.034482758620689655,
      "grad_norm": 0.435546875,
      "learning_rate": 9.655172413793105e-06,
      "loss": 1.8,
      "step": 4
    },
    {
      "epoch": 0.04310344827586207,
      "grad_norm": 0.41796875,
      "learning_rate": 9.56896551724138e-06,
      "loss": 1.8393,
      "step": 5
    },
    {
      "epoch": 0.05172413793103448,
      "grad_norm": 0.466796875,
      "learning_rate": 9.482758620689655e-06,
      "loss": 1.8275,
      "step": 6
    },
    {
      "epoch": 0.0603448275862069,
      "grad_norm": 0.365234375,
      "learning_rate": 9.396551724137931e-06,
      "loss": 1.7559,
      "step": 7
    },
    {
      "epoch": 0.06896551724137931,
      "grad_norm": 0.3203125,
      "learning_rate": 9.310344827586207e-06,
      "loss": 1.7784,
      "step": 8
    },
    {
      "epoch": 0.07758620689655173,
      "grad_norm": 0.3046875,
      "learning_rate": 9.224137931034484e-06,
      "loss": 1.7637,
      "step": 9
    },
    {
      "epoch": 0.08620689655172414,
      "grad_norm": 0.28515625,
      "learning_rate": 9.13793103448276e-06,
      "loss": 1.7164,
      "step": 10
    },
    {
      "epoch": 0.09482758620689655,
      "grad_norm": 0.294921875,
      "learning_rate": 9.051724137931036e-06,
      "loss": 1.7796,
      "step": 11
    },
    {
      "epoch": 0.10344827586206896,
      "grad_norm": 0.265625,
      "learning_rate": 8.965517241379312e-06,
      "loss": 1.698,
      "step": 12
    },
    {
      "epoch": 0.11206896551724138,
      "grad_norm": 0.2412109375,
      "learning_rate": 8.879310344827588e-06,
      "loss": 1.7032,
      "step": 13
    },
    {
      "epoch": 0.1206896551724138,
      "grad_norm": 0.2392578125,
      "learning_rate": 8.793103448275862e-06,
      "loss": 1.7151,
      "step": 14
    },
    {
      "epoch": 0.12931034482758622,
      "grad_norm": 0.265625,
      "learning_rate": 8.706896551724138e-06,
      "loss": 1.6807,
      "step": 15
    },
    {
      "epoch": 0.13793103448275862,
      "grad_norm": 0.2373046875,
      "learning_rate": 8.620689655172414e-06,
      "loss": 1.7027,
      "step": 16
    },
    {
      "epoch": 0.14655172413793102,
      "grad_norm": 0.244140625,
      "learning_rate": 8.53448275862069e-06,
      "loss": 1.6944,
      "step": 17
    },
    {
      "epoch": 0.15517241379310345,
      "grad_norm": 0.2392578125,
      "learning_rate": 8.448275862068966e-06,
      "loss": 1.7346,
      "step": 18
    },
    {
      "epoch": 0.16379310344827586,
      "grad_norm": 0.22265625,
      "learning_rate": 8.362068965517242e-06,
      "loss": 1.5981,
      "step": 19
    },
    {
      "epoch": 0.1724137931034483,
      "grad_norm": 0.2451171875,
      "learning_rate": 8.275862068965518e-06,
      "loss": 1.6794,
      "step": 20
    },
    {
      "epoch": 0.1810344827586207,
      "grad_norm": 0.212890625,
      "learning_rate": 8.189655172413794e-06,
      "loss": 1.6385,
      "step": 21
    },
    {
      "epoch": 0.1896551724137931,
      "grad_norm": 0.205078125,
      "learning_rate": 8.103448275862069e-06,
      "loss": 1.6853,
      "step": 22
    },
    {
      "epoch": 0.19827586206896552,
      "grad_norm": 0.2021484375,
      "learning_rate": 8.017241379310345e-06,
      "loss": 1.6833,
      "step": 23
    },
    {
      "epoch": 0.20689655172413793,
      "grad_norm": 0.1875,
      "learning_rate": 7.93103448275862e-06,
      "loss": 1.5735,
      "step": 24
    },
    {
      "epoch": 0.21551724137931033,
      "grad_norm": 0.1806640625,
      "learning_rate": 7.844827586206897e-06,
      "loss": 1.6481,
      "step": 25
    },
    {
      "epoch": 0.22413793103448276,
      "grad_norm": 0.1796875,
      "learning_rate": 7.758620689655173e-06,
      "loss": 1.6415,
      "step": 26
    },
    {
      "epoch": 0.23275862068965517,
      "grad_norm": 0.2109375,
      "learning_rate": 7.672413793103449e-06,
      "loss": 1.697,
      "step": 27
    },
    {
      "epoch": 0.2413793103448276,
      "grad_norm": 0.16796875,
      "learning_rate": 7.586206896551724e-06,
      "loss": 1.6084,
      "step": 28
    },
    {
      "epoch": 0.25,
      "grad_norm": 0.1884765625,
      "learning_rate": 7.500000000000001e-06,
      "loss": 1.6449,
      "step": 29
    },
    {
      "epoch": 0.25862068965517243,
      "grad_norm": 0.216796875,
      "learning_rate": 7.413793103448277e-06,
      "loss": 1.6571,
      "step": 30
    },
    {
      "epoch": 0.2672413793103448,
      "grad_norm": 0.193359375,
      "learning_rate": 7.327586206896552e-06,
      "loss": 1.5427,
      "step": 31
    },
    {
      "epoch": 0.27586206896551724,
      "grad_norm": 0.1708984375,
      "learning_rate": 7.241379310344828e-06,
      "loss": 1.498,
      "step": 32
    },
    {
      "epoch": 0.28448275862068967,
      "grad_norm": 0.1865234375,
      "learning_rate": 7.155172413793104e-06,
      "loss": 1.6642,
      "step": 33
    },
    {
      "epoch": 0.29310344827586204,
      "grad_norm": 0.212890625,
      "learning_rate": 7.0689655172413796e-06,
      "loss": 1.6553,
      "step": 34
    },
    {
      "epoch": 0.3017241379310345,
      "grad_norm": 0.1796875,
      "learning_rate": 6.982758620689656e-06,
      "loss": 1.5789,
      "step": 35
    },
    {
      "epoch": 0.3103448275862069,
      "grad_norm": 0.2041015625,
      "learning_rate": 6.896551724137932e-06,
      "loss": 1.568,
      "step": 36
    },
    {
      "epoch": 0.31896551724137934,
      "grad_norm": 0.189453125,
      "learning_rate": 6.810344827586207e-06,
      "loss": 1.5269,
      "step": 37
    },
    {
      "epoch": 0.3275862068965517,
      "grad_norm": 0.2001953125,
      "learning_rate": 6.724137931034484e-06,
      "loss": 1.6063,
      "step": 38
    },
    {
      "epoch": 0.33620689655172414,
      "grad_norm": 0.181640625,
      "learning_rate": 6.63793103448276e-06,
      "loss": 1.5043,
      "step": 39
    },
    {
      "epoch": 0.3448275862068966,
      "grad_norm": 0.1904296875,
      "learning_rate": 6.551724137931035e-06,
      "loss": 1.6623,
      "step": 40
    },
    {
      "epoch": 0.35344827586206895,
      "grad_norm": 0.1875,
      "learning_rate": 6.465517241379311e-06,
      "loss": 1.5848,
      "step": 41
    },
    {
      "epoch": 0.3620689655172414,
      "grad_norm": 0.17578125,
      "learning_rate": 6.379310344827587e-06,
      "loss": 1.5198,
      "step": 42
    },
    {
      "epoch": 0.3706896551724138,
      "grad_norm": 0.1982421875,
      "learning_rate": 6.293103448275862e-06,
      "loss": 1.5934,
      "step": 43
    },
    {
      "epoch": 0.3793103448275862,
      "grad_norm": 0.23828125,
      "learning_rate": 6.206896551724138e-06,
      "loss": 1.5823,
      "step": 44
    },
    {
      "epoch": 0.3879310344827586,
      "grad_norm": 0.1533203125,
      "learning_rate": 6.1206896551724135e-06,
      "loss": 1.5819,
      "step": 45
    },
    {
      "epoch": 0.39655172413793105,
      "grad_norm": 0.162109375,
      "learning_rate": 6.03448275862069e-06,
      "loss": 1.5034,
      "step": 46
    },
    {
      "epoch": 0.4051724137931034,
      "grad_norm": 0.1943359375,
      "learning_rate": 5.9482758620689665e-06,
      "loss": 1.5815,
      "step": 47
    },
    {
      "epoch": 0.41379310344827586,
      "grad_norm": 0.1796875,
      "learning_rate": 5.862068965517242e-06,
      "loss": 1.5191,
      "step": 48
    },
    {
      "epoch": 0.4224137931034483,
      "grad_norm": 0.1708984375,
      "learning_rate": 5.775862068965518e-06,
      "loss": 1.5469,
      "step": 49
    },
    {
      "epoch": 0.43103448275862066,
      "grad_norm": 0.158203125,
      "learning_rate": 5.689655172413794e-06,
      "loss": 1.5112,
      "step": 50
    },
    {
      "epoch": 0.4396551724137931,
      "grad_norm": 0.1748046875,
      "learning_rate": 5.603448275862069e-06,
      "loss": 1.5272,
      "step": 51
    },
    {
      "epoch": 0.4482758620689655,
      "grad_norm": 0.14453125,
      "learning_rate": 5.517241379310345e-06,
      "loss": 1.4435,
      "step": 52
    },
    {
      "epoch": 0.45689655172413796,
      "grad_norm": 0.185546875,
      "learning_rate": 5.431034482758621e-06,
      "loss": 1.5291,
      "step": 53
    },
    {
      "epoch": 0.46551724137931033,
      "grad_norm": 0.1923828125,
      "learning_rate": 5.344827586206896e-06,
      "loss": 1.4499,
      "step": 54
    },
    {
      "epoch": 0.47413793103448276,
      "grad_norm": 0.2294921875,
      "learning_rate": 5.258620689655173e-06,
      "loss": 1.5481,
      "step": 55
    },
    {
      "epoch": 0.4827586206896552,
      "grad_norm": 0.16796875,
      "learning_rate": 5.172413793103449e-06,
      "loss": 1.4589,
      "step": 56
    },
    {
      "epoch": 0.49137931034482757,
      "grad_norm": 0.166015625,
      "learning_rate": 5.086206896551724e-06,
      "loss": 1.5175,
      "step": 57
    },
    {
      "epoch": 0.5,
      "grad_norm": 0.1630859375,
      "learning_rate": 5e-06,
      "loss": 1.5369,
      "step": 58
    },
    {
      "epoch": 0.5086206896551724,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.9137931034482765e-06,
      "loss": 1.5117,
      "step": 59
    },
    {
      "epoch": 0.5172413793103449,
      "grad_norm": 0.1376953125,
      "learning_rate": 4.8275862068965525e-06,
      "loss": 1.5455,
      "step": 60
    },
    {
      "epoch": 0.5258620689655172,
      "grad_norm": 0.1650390625,
      "learning_rate": 4.741379310344828e-06,
      "loss": 1.5353,
      "step": 61
    },
    {
      "epoch": 0.5344827586206896,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.655172413793104e-06,
      "loss": 1.4975,
      "step": 62
    },
    {
      "epoch": 0.5431034482758621,
      "grad_norm": 0.15234375,
      "learning_rate": 4.56896551724138e-06,
      "loss": 1.5309,
      "step": 63
    },
    {
      "epoch": 0.5517241379310345,
      "grad_norm": 0.14453125,
      "learning_rate": 4.482758620689656e-06,
      "loss": 1.5189,
      "step": 64
    },
    {
      "epoch": 0.5603448275862069,
      "grad_norm": 0.134765625,
      "learning_rate": 4.396551724137931e-06,
      "loss": 1.4481,
      "step": 65
    },
    {
      "epoch": 0.5689655172413793,
      "grad_norm": 0.1484375,
      "learning_rate": 4.310344827586207e-06,
      "loss": 1.5394,
      "step": 66
    },
    {
      "epoch": 0.5775862068965517,
      "grad_norm": 0.1572265625,
      "learning_rate": 4.224137931034483e-06,
      "loss": 1.5147,
      "step": 67
    },
    {
      "epoch": 0.5862068965517241,
      "grad_norm": 0.14453125,
      "learning_rate": 4.137931034482759e-06,
      "loss": 1.5097,
      "step": 68
    },
    {
      "epoch": 0.5948275862068966,
      "grad_norm": 0.1826171875,
      "learning_rate": 4.051724137931034e-06,
      "loss": 1.5006,
      "step": 69
    },
    {
      "epoch": 0.603448275862069,
      "grad_norm": 0.166015625,
      "learning_rate": 3.96551724137931e-06,
      "loss": 1.5042,
      "step": 70
    },
    {
      "epoch": 0.6120689655172413,
      "grad_norm": 0.1591796875,
      "learning_rate": 3.8793103448275865e-06,
      "loss": 1.5101,
      "step": 71
    },
    {
      "epoch": 0.6206896551724138,
      "grad_norm": 0.16796875,
      "learning_rate": 3.793103448275862e-06,
      "loss": 1.4706,
      "step": 72
    },
    {
      "epoch": 0.6293103448275862,
      "grad_norm": 0.1669921875,
      "learning_rate": 3.7068965517241385e-06,
      "loss": 1.513,
      "step": 73
    },
    {
      "epoch": 0.6379310344827587,
      "grad_norm": 0.1650390625,
      "learning_rate": 3.620689655172414e-06,
      "loss": 1.4574,
      "step": 74
    },
    {
      "epoch": 0.646551724137931,
      "grad_norm": 0.18359375,
      "learning_rate": 3.5344827586206898e-06,
      "loss": 1.4928,
      "step": 75
    },
    {
      "epoch": 0.6551724137931034,
      "grad_norm": 0.1572265625,
      "learning_rate": 3.448275862068966e-06,
      "loss": 1.4709,
      "step": 76
    },
    {
      "epoch": 0.6637931034482759,
      "grad_norm": 0.12890625,
      "learning_rate": 3.362068965517242e-06,
      "loss": 1.428,
      "step": 77
    },
    {
      "epoch": 0.6724137931034483,
      "grad_norm": 0.1669921875,
      "learning_rate": 3.2758620689655175e-06,
      "loss": 1.5212,
      "step": 78
    },
    {
      "epoch": 0.6810344827586207,
      "grad_norm": 0.1416015625,
      "learning_rate": 3.1896551724137935e-06,
      "loss": 1.492,
      "step": 79
    },
    {
      "epoch": 0.6896551724137931,
      "grad_norm": 0.1572265625,
      "learning_rate": 3.103448275862069e-06,
      "loss": 1.3807,
      "step": 80
    },
    {
      "epoch": 0.6982758620689655,
      "grad_norm": 0.1416015625,
      "learning_rate": 3.017241379310345e-06,
      "loss": 1.4559,
      "step": 81
    },
    {
      "epoch": 0.7068965517241379,
      "grad_norm": 0.1484375,
      "learning_rate": 2.931034482758621e-06,
      "loss": 1.5057,
      "step": 82
    },
    {
      "epoch": 0.7155172413793104,
      "grad_norm": 0.185546875,
      "learning_rate": 2.844827586206897e-06,
      "loss": 1.4579,
      "step": 83
    },
    {
      "epoch": 0.7241379310344828,
      "grad_norm": 0.1455078125,
      "learning_rate": 2.7586206896551725e-06,
      "loss": 1.4931,
      "step": 84
    },
    {
      "epoch": 0.7327586206896551,
      "grad_norm": 0.1552734375,
      "learning_rate": 2.672413793103448e-06,
      "loss": 1.4055,
      "step": 85
    },
    {
      "epoch": 0.7413793103448276,
      "grad_norm": 0.1484375,
      "learning_rate": 2.5862068965517246e-06,
      "loss": 1.4233,
      "step": 86
    },
    {
      "epoch": 0.75,
      "grad_norm": 0.2041015625,
      "learning_rate": 2.5e-06,
      "loss": 1.4571,
      "step": 87
    },
    {
      "epoch": 0.7586206896551724,
      "grad_norm": 0.1748046875,
      "learning_rate": 2.4137931034482762e-06,
      "loss": 1.503,
      "step": 88
    },
    {
      "epoch": 0.7672413793103449,
      "grad_norm": 0.18359375,
      "learning_rate": 2.327586206896552e-06,
      "loss": 1.4796,
      "step": 89
    },
    {
      "epoch": 0.7758620689655172,
      "grad_norm": 0.1591796875,
      "learning_rate": 2.241379310344828e-06,
      "loss": 1.4198,
      "step": 90
    },
    {
      "epoch": 0.7844827586206896,
      "grad_norm": 0.1845703125,
      "learning_rate": 2.1551724137931035e-06,
      "loss": 1.4005,
      "step": 91
    },
    {
      "epoch": 0.7931034482758621,
      "grad_norm": 0.1328125,
      "learning_rate": 2.0689655172413796e-06,
      "loss": 1.4767,
      "step": 92
    },
    {
      "epoch": 0.8017241379310345,
      "grad_norm": 0.185546875,
      "learning_rate": 1.982758620689655e-06,
      "loss": 1.5157,
      "step": 93
    },
    {
      "epoch": 0.8103448275862069,
      "grad_norm": 0.2138671875,
      "learning_rate": 1.896551724137931e-06,
      "loss": 1.4623,
      "step": 94
    },
    {
      "epoch": 0.8189655172413793,
      "grad_norm": 0.1826171875,
      "learning_rate": 1.810344827586207e-06,
      "loss": 1.4235,
      "step": 95
    },
    {
      "epoch": 0.8275862068965517,
      "grad_norm": 0.173828125,
      "learning_rate": 1.724137931034483e-06,
      "loss": 1.5049,
      "step": 96
    },
    {
      "epoch": 0.8362068965517241,
      "grad_norm": 0.13671875,
      "learning_rate": 1.6379310344827587e-06,
      "loss": 1.4492,
      "step": 97
    },
    {
      "epoch": 0.8448275862068966,
      "grad_norm": 0.1943359375,
      "learning_rate": 1.5517241379310346e-06,
      "loss": 1.4447,
      "step": 98
    },
    {
      "epoch": 0.853448275862069,
      "grad_norm": 0.1455078125,
      "learning_rate": 1.4655172413793104e-06,
      "loss": 1.4843,
      "step": 99
    },
    {
      "epoch": 0.8620689655172413,
      "grad_norm": 0.1767578125,
      "learning_rate": 1.3793103448275862e-06,
      "loss": 1.3772,
      "step": 100
    },
    {
      "epoch": 0.8706896551724138,
      "grad_norm": 0.2109375,
      "learning_rate": 1.2931034482758623e-06,
      "loss": 1.5097,
      "step": 101
    },
    {
      "epoch": 0.8793103448275862,
      "grad_norm": 0.1689453125,
      "learning_rate": 1.2068965517241381e-06,
      "loss": 1.4115,
      "step": 102
    },
    {
      "epoch": 0.8879310344827587,
      "grad_norm": 0.13671875,
      "learning_rate": 1.120689655172414e-06,
      "loss": 1.4794,
      "step": 103
    },
    {
      "epoch": 0.896551724137931,
      "grad_norm": 0.17578125,
      "learning_rate": 1.0344827586206898e-06,
      "loss": 1.4197,
      "step": 104
    },
    {
      "epoch": 0.9051724137931034,
      "grad_norm": 0.2080078125,
      "learning_rate": 9.482758620689655e-07,
      "loss": 1.4999,
      "step": 105
    },
    {
      "epoch": 0.9137931034482759,
      "grad_norm": 0.1337890625,
      "learning_rate": 8.620689655172415e-07,
      "loss": 1.4207,
      "step": 106
    },
    {
      "epoch": 0.9224137931034483,
      "grad_norm": 0.1416015625,
      "learning_rate": 7.758620689655173e-07,
      "loss": 1.396,
      "step": 107
    },
    {
      "epoch": 0.9310344827586207,
      "grad_norm": 0.162109375,
      "learning_rate": 6.896551724137931e-07,
      "loss": 1.4244,
      "step": 108
    },
    {
      "epoch": 0.9396551724137931,
      "grad_norm": 0.185546875,
      "learning_rate": 6.034482758620691e-07,
      "loss": 1.4731,
      "step": 109
    },
    {
      "epoch": 0.9482758620689655,
      "grad_norm": 0.1396484375,
      "learning_rate": 5.172413793103449e-07,
      "loss": 1.5113,
      "step": 110
    },
    {
      "epoch": 0.9568965517241379,
      "grad_norm": 0.1630859375,
      "learning_rate": 4.3103448275862073e-07,
      "loss": 1.468,
      "step": 111
    },
    {
      "epoch": 0.9655172413793104,
      "grad_norm": 0.162109375,
      "learning_rate": 3.4482758620689656e-07,
      "loss": 1.4651,
      "step": 112
    },
    {
      "epoch": 0.9741379310344828,
      "grad_norm": 0.2001953125,
      "learning_rate": 2.5862068965517245e-07,
      "loss": 1.4632,
      "step": 113
    },
    {
      "epoch": 0.9827586206896551,
      "grad_norm": 0.1337890625,
      "learning_rate": 1.7241379310344828e-07,
      "loss": 1.4311,
      "step": 114
    },
    {
      "epoch": 0.9913793103448276,
      "grad_norm": 0.134765625,
      "learning_rate": 8.620689655172414e-08,
      "loss": 1.423,
      "step": 115
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.1494140625,
      "learning_rate": 0.0,
      "loss": 1.4911,
      "step": 116
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 116,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 0,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 3.1862449256372634e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}