{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 0,
  "global_step": 291,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.003436426116838488,
      "grad_norm": 0.4921875,
      "learning_rate": 9.965635738831616e-06,
      "loss": 1.7833,
      "step": 1
    },
    {
      "epoch": 0.006872852233676976,
      "grad_norm": 0.47265625,
      "learning_rate": 9.931271477663231e-06,
      "loss": 1.6757,
      "step": 2
    },
    {
      "epoch": 0.010309278350515464,
      "grad_norm": 0.48828125,
      "learning_rate": 9.896907216494846e-06,
      "loss": 1.7604,
      "step": 3
    },
    {
      "epoch": 0.013745704467353952,
      "grad_norm": 0.43359375,
      "learning_rate": 9.862542955326461e-06,
      "loss": 1.6662,
      "step": 4
    },
    {
      "epoch": 0.01718213058419244,
      "grad_norm": 0.43359375,
      "learning_rate": 9.828178694158076e-06,
      "loss": 1.7293,
      "step": 5
    },
    {
      "epoch": 0.020618556701030927,
      "grad_norm": 0.380859375,
      "learning_rate": 9.793814432989691e-06,
      "loss": 1.6688,
      "step": 6
    },
    {
      "epoch": 0.024054982817869417,
      "grad_norm": 0.34375,
      "learning_rate": 9.759450171821306e-06,
      "loss": 1.6821,
      "step": 7
    },
    {
      "epoch": 0.027491408934707903,
      "grad_norm": 0.314453125,
      "learning_rate": 9.725085910652921e-06,
      "loss": 1.6133,
      "step": 8
    },
    {
      "epoch": 0.030927835051546393,
      "grad_norm": 0.298828125,
      "learning_rate": 9.690721649484536e-06,
      "loss": 1.6739,
      "step": 9
    },
    {
      "epoch": 0.03436426116838488,
      "grad_norm": 0.26171875,
      "learning_rate": 9.656357388316153e-06,
      "loss": 1.5586,
      "step": 10
    },
    {
      "epoch": 0.037800687285223365,
      "grad_norm": 0.2578125,
      "learning_rate": 9.621993127147768e-06,
      "loss": 1.5729,
      "step": 11
    },
    {
      "epoch": 0.041237113402061855,
      "grad_norm": 0.2333984375,
      "learning_rate": 9.587628865979383e-06,
      "loss": 1.5494,
      "step": 12
    },
    {
      "epoch": 0.044673539518900345,
      "grad_norm": 0.244140625,
      "learning_rate": 9.553264604810998e-06,
      "loss": 1.5882,
      "step": 13
    },
    {
      "epoch": 0.048109965635738834,
      "grad_norm": 0.23828125,
      "learning_rate": 9.518900343642611e-06,
      "loss": 1.5447,
      "step": 14
    },
    {
      "epoch": 0.05154639175257732,
      "grad_norm": 0.232421875,
      "learning_rate": 9.484536082474226e-06,
      "loss": 1.5998,
      "step": 15
    },
    {
      "epoch": 0.054982817869415807,
      "grad_norm": 0.2431640625,
      "learning_rate": 9.450171821305843e-06,
      "loss": 1.5377,
      "step": 16
    },
    {
      "epoch": 0.058419243986254296,
      "grad_norm": 0.2314453125,
      "learning_rate": 9.415807560137458e-06,
      "loss": 1.5192,
      "step": 17
    },
    {
      "epoch": 0.061855670103092786,
      "grad_norm": 0.234375,
      "learning_rate": 9.381443298969073e-06,
      "loss": 1.5355,
      "step": 18
    },
    {
      "epoch": 0.06529209621993128,
      "grad_norm": 0.2265625,
      "learning_rate": 9.347079037800688e-06,
      "loss": 1.5234,
      "step": 19
    },
    {
      "epoch": 0.06872852233676977,
      "grad_norm": 0.220703125,
      "learning_rate": 9.312714776632303e-06,
      "loss": 1.499,
      "step": 20
    },
    {
      "epoch": 0.07216494845360824,
      "grad_norm": 0.2060546875,
      "learning_rate": 9.278350515463918e-06,
      "loss": 1.5096,
      "step": 21
    },
    {
      "epoch": 0.07560137457044673,
      "grad_norm": 0.193359375,
      "learning_rate": 9.243986254295533e-06,
      "loss": 1.5011,
      "step": 22
    },
    {
      "epoch": 0.07903780068728522,
      "grad_norm": 0.197265625,
      "learning_rate": 9.209621993127148e-06,
      "loss": 1.4814,
      "step": 23
    },
    {
      "epoch": 0.08247422680412371,
      "grad_norm": 0.18359375,
      "learning_rate": 9.175257731958764e-06,
      "loss": 1.4306,
      "step": 24
    },
    {
      "epoch": 0.0859106529209622,
      "grad_norm": 0.1826171875,
      "learning_rate": 9.140893470790379e-06,
      "loss": 1.4095,
      "step": 25
    },
    {
      "epoch": 0.08934707903780069,
      "grad_norm": 0.1767578125,
      "learning_rate": 9.106529209621994e-06,
      "loss": 1.5332,
      "step": 26
    },
    {
      "epoch": 0.09278350515463918,
      "grad_norm": 0.171875,
      "learning_rate": 9.072164948453609e-06,
      "loss": 1.3951,
      "step": 27
    },
    {
      "epoch": 0.09621993127147767,
      "grad_norm": 0.1650390625,
      "learning_rate": 9.037800687285224e-06,
      "loss": 1.4271,
      "step": 28
    },
    {
      "epoch": 0.09965635738831616,
      "grad_norm": 0.17578125,
      "learning_rate": 9.003436426116839e-06,
      "loss": 1.5286,
      "step": 29
    },
    {
      "epoch": 0.10309278350515463,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.969072164948455e-06,
      "loss": 1.4261,
      "step": 30
    },
    {
      "epoch": 0.10652920962199312,
      "grad_norm": 0.1689453125,
      "learning_rate": 8.93470790378007e-06,
      "loss": 1.4047,
      "step": 31
    },
    {
      "epoch": 0.10996563573883161,
      "grad_norm": 0.1611328125,
      "learning_rate": 8.900343642611684e-06,
      "loss": 1.3345,
      "step": 32
    },
    {
      "epoch": 0.1134020618556701,
      "grad_norm": 0.177734375,
      "learning_rate": 8.865979381443299e-06,
      "loss": 1.3815,
      "step": 33
    },
    {
      "epoch": 0.11683848797250859,
      "grad_norm": 0.1708984375,
      "learning_rate": 8.831615120274914e-06,
      "loss": 1.419,
      "step": 34
    },
    {
      "epoch": 0.12027491408934708,
      "grad_norm": 0.1630859375,
      "learning_rate": 8.797250859106529e-06,
      "loss": 1.4125,
      "step": 35
    },
    {
      "epoch": 0.12371134020618557,
      "grad_norm": 0.16015625,
      "learning_rate": 8.762886597938146e-06,
      "loss": 1.3938,
      "step": 36
    },
    {
      "epoch": 0.12714776632302405,
      "grad_norm": 0.1513671875,
      "learning_rate": 8.72852233676976e-06,
      "loss": 1.3676,
      "step": 37
    },
    {
      "epoch": 0.13058419243986255,
      "grad_norm": 0.1767578125,
      "learning_rate": 8.694158075601376e-06,
      "loss": 1.355,
      "step": 38
    },
    {
      "epoch": 0.13402061855670103,
      "grad_norm": 0.1435546875,
      "learning_rate": 8.65979381443299e-06,
      "loss": 1.2992,
      "step": 39
    },
    {
      "epoch": 0.13745704467353953,
      "grad_norm": 0.1494140625,
      "learning_rate": 8.625429553264606e-06,
      "loss": 1.2822,
      "step": 40
    },
    {
      "epoch": 0.140893470790378,
      "grad_norm": 0.1416015625,
      "learning_rate": 8.591065292096221e-06,
      "loss": 1.3814,
      "step": 41
    },
    {
      "epoch": 0.14432989690721648,
      "grad_norm": 0.1455078125,
      "learning_rate": 8.556701030927836e-06,
      "loss": 1.342,
      "step": 42
    },
    {
      "epoch": 0.14776632302405499,
      "grad_norm": 0.1552734375,
      "learning_rate": 8.522336769759451e-06,
      "loss": 1.3447,
      "step": 43
    },
    {
      "epoch": 0.15120274914089346,
      "grad_norm": 0.146484375,
      "learning_rate": 8.487972508591066e-06,
      "loss": 1.3496,
      "step": 44
    },
    {
      "epoch": 0.15463917525773196,
      "grad_norm": 0.142578125,
      "learning_rate": 8.453608247422681e-06,
      "loss": 1.361,
      "step": 45
    },
    {
      "epoch": 0.15807560137457044,
      "grad_norm": 0.1708984375,
      "learning_rate": 8.419243986254296e-06,
      "loss": 1.3466,
      "step": 46
    },
    {
      "epoch": 0.16151202749140894,
      "grad_norm": 0.1572265625,
      "learning_rate": 8.384879725085911e-06,
      "loss": 1.2897,
      "step": 47
    },
    {
      "epoch": 0.16494845360824742,
      "grad_norm": 0.1318359375,
      "learning_rate": 8.350515463917526e-06,
      "loss": 1.2955,
      "step": 48
    },
    {
      "epoch": 0.16838487972508592,
      "grad_norm": 0.26953125,
      "learning_rate": 8.316151202749141e-06,
      "loss": 1.2537,
      "step": 49
    },
    {
      "epoch": 0.1718213058419244,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.281786941580758e-06,
      "loss": 1.2805,
      "step": 50
    },
    {
      "epoch": 0.17525773195876287,
      "grad_norm": 0.1171875,
      "learning_rate": 8.247422680412371e-06,
      "loss": 1.2535,
      "step": 51
    },
    {
      "epoch": 0.17869415807560138,
      "grad_norm": 0.11962890625,
      "learning_rate": 8.213058419243986e-06,
      "loss": 1.3243,
      "step": 52
    },
    {
      "epoch": 0.18213058419243985,
      "grad_norm": 0.12353515625,
      "learning_rate": 8.178694158075601e-06,
      "loss": 1.262,
      "step": 53
    },
    {
      "epoch": 0.18556701030927836,
      "grad_norm": 0.1259765625,
      "learning_rate": 8.144329896907216e-06,
      "loss": 1.2669,
      "step": 54
    },
    {
      "epoch": 0.18900343642611683,
      "grad_norm": 0.123046875,
      "learning_rate": 8.109965635738832e-06,
      "loss": 1.2936,
      "step": 55
    },
    {
      "epoch": 0.19243986254295534,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.075601374570448e-06,
      "loss": 1.2678,
      "step": 56
    },
    {
      "epoch": 0.1958762886597938,
      "grad_norm": 0.1171875,
      "learning_rate": 8.041237113402063e-06,
      "loss": 1.3035,
      "step": 57
    },
    {
      "epoch": 0.19931271477663232,
      "grad_norm": 0.1357421875,
      "learning_rate": 8.006872852233678e-06,
      "loss": 1.3927,
      "step": 58
    },
    {
      "epoch": 0.2027491408934708,
      "grad_norm": 0.1103515625,
      "learning_rate": 7.972508591065293e-06,
      "loss": 1.2696,
      "step": 59
    },
    {
      "epoch": 0.20618556701030927,
      "grad_norm": 0.1181640625,
      "learning_rate": 7.938144329896907e-06,
      "loss": 1.2684,
      "step": 60
    },
    {
      "epoch": 0.20962199312714777,
      "grad_norm": 0.1474609375,
      "learning_rate": 7.903780068728523e-06,
      "loss": 1.3102,
      "step": 61
    },
    {
      "epoch": 0.21305841924398625,
      "grad_norm": 0.11181640625,
      "learning_rate": 7.869415807560138e-06,
      "loss": 1.2037,
      "step": 62
    },
    {
      "epoch": 0.21649484536082475,
      "grad_norm": 0.11376953125,
      "learning_rate": 7.835051546391754e-06,
      "loss": 1.2694,
      "step": 63
    },
    {
      "epoch": 0.21993127147766323,
      "grad_norm": 0.11474609375,
      "learning_rate": 7.800687285223369e-06,
      "loss": 1.2515,
      "step": 64
    },
    {
      "epoch": 0.22336769759450173,
      "grad_norm": 0.1142578125,
      "learning_rate": 7.766323024054984e-06,
      "loss": 1.2484,
      "step": 65
    },
    {
      "epoch": 0.2268041237113402,
      "grad_norm": 0.11767578125,
      "learning_rate": 7.731958762886599e-06,
      "loss": 1.342,
      "step": 66
    },
    {
      "epoch": 0.23024054982817868,
      "grad_norm": 0.12255859375,
      "learning_rate": 7.697594501718214e-06,
      "loss": 1.2629,
      "step": 67
    },
    {
      "epoch": 0.23367697594501718,
      "grad_norm": 0.1162109375,
      "learning_rate": 7.663230240549829e-06,
      "loss": 1.2718,
      "step": 68
    },
    {
      "epoch": 0.23711340206185566,
      "grad_norm": 0.109375,
      "learning_rate": 7.628865979381444e-06,
      "loss": 1.3001,
      "step": 69
    },
    {
      "epoch": 0.24054982817869416,
      "grad_norm": 0.10986328125,
      "learning_rate": 7.594501718213059e-06,
      "loss": 1.1548,
      "step": 70
    },
    {
      "epoch": 0.24398625429553264,
      "grad_norm": 0.109375,
      "learning_rate": 7.560137457044674e-06,
      "loss": 1.1643,
      "step": 71
    },
    {
      "epoch": 0.24742268041237114,
      "grad_norm": 0.13671875,
      "learning_rate": 7.525773195876289e-06,
      "loss": 1.2197,
      "step": 72
    },
    {
      "epoch": 0.2508591065292096,
      "grad_norm": 0.111328125,
      "learning_rate": 7.491408934707905e-06,
      "loss": 1.2366,
      "step": 73
    },
    {
      "epoch": 0.2542955326460481,
      "grad_norm": 0.11767578125,
      "learning_rate": 7.45704467353952e-06,
      "loss": 1.2623,
      "step": 74
    },
    {
      "epoch": 0.25773195876288657,
      "grad_norm": 0.10693359375,
      "learning_rate": 7.422680412371135e-06,
      "loss": 1.2439,
      "step": 75
    },
    {
      "epoch": 0.2611683848797251,
      "grad_norm": 0.109375,
      "learning_rate": 7.38831615120275e-06,
      "loss": 1.2451,
      "step": 76
    },
    {
      "epoch": 0.2646048109965636,
      "grad_norm": 0.11474609375,
      "learning_rate": 7.353951890034365e-06,
      "loss": 1.1966,
      "step": 77
    },
    {
      "epoch": 0.26804123711340205,
      "grad_norm": 0.11083984375,
      "learning_rate": 7.319587628865979e-06,
      "loss": 1.1983,
      "step": 78
    },
    {
      "epoch": 0.27147766323024053,
      "grad_norm": 0.1044921875,
      "learning_rate": 7.285223367697595e-06,
      "loss": 1.2115,
      "step": 79
    },
    {
      "epoch": 0.27491408934707906,
      "grad_norm": 0.1083984375,
      "learning_rate": 7.25085910652921e-06,
      "loss": 1.1679,
      "step": 80
    },
    {
      "epoch": 0.27835051546391754,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.216494845360825e-06,
      "loss": 1.2571,
      "step": 81
    },
    {
      "epoch": 0.281786941580756,
      "grad_norm": 0.119140625,
      "learning_rate": 7.18213058419244e-06,
      "loss": 1.3151,
      "step": 82
    },
    {
      "epoch": 0.2852233676975945,
      "grad_norm": 0.1357421875,
      "learning_rate": 7.147766323024056e-06,
      "loss": 1.2378,
      "step": 83
    },
    {
      "epoch": 0.28865979381443296,
      "grad_norm": 0.107421875,
      "learning_rate": 7.113402061855671e-06,
      "loss": 1.2514,
      "step": 84
    },
    {
      "epoch": 0.2920962199312715,
      "grad_norm": 0.138671875,
      "learning_rate": 7.079037800687286e-06,
      "loss": 1.2457,
      "step": 85
    },
    {
      "epoch": 0.29553264604810997,
      "grad_norm": 0.1513671875,
      "learning_rate": 7.044673539518901e-06,
      "loss": 1.2397,
      "step": 86
    },
    {
      "epoch": 0.29896907216494845,
      "grad_norm": 0.11181640625,
      "learning_rate": 7.010309278350515e-06,
      "loss": 1.2586,
      "step": 87
    },
    {
      "epoch": 0.3024054982817869,
      "grad_norm": 0.115234375,
      "learning_rate": 6.9759450171821304e-06,
      "loss": 1.2845,
      "step": 88
    },
    {
      "epoch": 0.30584192439862545,
      "grad_norm": 0.1240234375,
      "learning_rate": 6.941580756013746e-06,
      "loss": 1.2553,
      "step": 89
    },
    {
      "epoch": 0.30927835051546393,
      "grad_norm": 0.2353515625,
      "learning_rate": 6.907216494845361e-06,
      "loss": 1.145,
      "step": 90
    },
    {
      "epoch": 0.3127147766323024,
      "grad_norm": 0.1513671875,
      "learning_rate": 6.872852233676976e-06,
      "loss": 1.2295,
      "step": 91
    },
    {
      "epoch": 0.3161512027491409,
      "grad_norm": 0.1435546875,
      "learning_rate": 6.8384879725085914e-06,
      "loss": 1.1771,
      "step": 92
    },
    {
      "epoch": 0.31958762886597936,
      "grad_norm": 0.11962890625,
      "learning_rate": 6.804123711340207e-06,
      "loss": 1.1458,
      "step": 93
    },
    {
      "epoch": 0.3230240549828179,
      "grad_norm": 0.158203125,
      "learning_rate": 6.769759450171822e-06,
      "loss": 1.2342,
      "step": 94
    },
    {
      "epoch": 0.32646048109965636,
      "grad_norm": 0.10888671875,
      "learning_rate": 6.735395189003437e-06,
      "loss": 1.2193,
      "step": 95
    },
    {
      "epoch": 0.32989690721649484,
      "grad_norm": 0.10986328125,
      "learning_rate": 6.701030927835052e-06,
      "loss": 1.2225,
      "step": 96
    },
    {
      "epoch": 0.3333333333333333,
      "grad_norm": 0.10498046875,
      "learning_rate": 6.666666666666667e-06,
      "loss": 1.2053,
      "step": 97
    },
    {
      "epoch": 0.33676975945017185,
      "grad_norm": 0.119140625,
      "learning_rate": 6.632302405498282e-06,
      "loss": 1.1893,
      "step": 98
    },
    {
      "epoch": 0.3402061855670103,
      "grad_norm": 0.10986328125,
      "learning_rate": 6.597938144329898e-06,
      "loss": 1.2301,
      "step": 99
    },
    {
      "epoch": 0.3436426116838488,
      "grad_norm": 0.11328125,
      "learning_rate": 6.563573883161513e-06,
      "loss": 1.1943,
      "step": 100
    },
    {
      "epoch": 0.3470790378006873,
      "grad_norm": 0.1552734375,
      "learning_rate": 6.529209621993128e-06,
      "loss": 1.2589,
      "step": 101
    },
    {
      "epoch": 0.35051546391752575,
      "grad_norm": 0.1025390625,
      "learning_rate": 6.494845360824743e-06,
      "loss": 1.1986,
      "step": 102
    },
    {
      "epoch": 0.3539518900343643,
      "grad_norm": 0.146484375,
      "learning_rate": 6.460481099656359e-06,
      "loss": 1.2073,
      "step": 103
    },
    {
      "epoch": 0.35738831615120276,
      "grad_norm": 0.111328125,
      "learning_rate": 6.426116838487974e-06,
      "loss": 1.2892,
      "step": 104
    },
    {
      "epoch": 0.36082474226804123,
      "grad_norm": 0.1064453125,
      "learning_rate": 6.391752577319588e-06,
      "loss": 1.2071,
      "step": 105
    },
    {
      "epoch": 0.3642611683848797,
      "grad_norm": 0.0986328125,
      "learning_rate": 6.357388316151203e-06,
      "loss": 1.1472,
      "step": 106
    },
    {
      "epoch": 0.36769759450171824,
      "grad_norm": 0.1376953125,
      "learning_rate": 6.323024054982818e-06,
      "loss": 1.1407,
      "step": 107
    },
    {
      "epoch": 0.3711340206185567,
      "grad_norm": 0.1328125,
      "learning_rate": 6.288659793814433e-06,
      "loss": 1.2531,
      "step": 108
    },
    {
      "epoch": 0.3745704467353952,
      "grad_norm": 0.10107421875,
      "learning_rate": 6.254295532646049e-06,
      "loss": 1.1916,
      "step": 109
    },
    {
      "epoch": 0.37800687285223367,
      "grad_norm": 0.109375,
      "learning_rate": 6.219931271477664e-06,
      "loss": 1.1996,
      "step": 110
    },
    {
      "epoch": 0.38144329896907214,
      "grad_norm": 0.10302734375,
      "learning_rate": 6.185567010309279e-06,
      "loss": 1.158,
      "step": 111
    },
    {
      "epoch": 0.3848797250859107,
      "grad_norm": 0.11669921875,
      "learning_rate": 6.151202749140894e-06,
      "loss": 1.2114,
      "step": 112
    },
    {
      "epoch": 0.38831615120274915,
      "grad_norm": 0.09912109375,
      "learning_rate": 6.11683848797251e-06,
      "loss": 1.2039,
      "step": 113
    },
    {
      "epoch": 0.3917525773195876,
      "grad_norm": 0.11572265625,
      "learning_rate": 6.082474226804124e-06,
      "loss": 1.2078,
      "step": 114
    },
    {
      "epoch": 0.3951890034364261,
      "grad_norm": 0.10205078125,
      "learning_rate": 6.048109965635739e-06,
      "loss": 1.18,
      "step": 115
    },
    {
      "epoch": 0.39862542955326463,
      "grad_norm": 0.1171875,
      "learning_rate": 6.013745704467354e-06,
      "loss": 1.2228,
      "step": 116
    },
    {
      "epoch": 0.4020618556701031,
      "grad_norm": 0.103515625,
      "learning_rate": 5.979381443298969e-06,
      "loss": 1.1975,
      "step": 117
    },
    {
      "epoch": 0.4054982817869416,
      "grad_norm": 0.1142578125,
      "learning_rate": 5.945017182130585e-06,
      "loss": 1.14,
      "step": 118
    },
    {
      "epoch": 0.40893470790378006,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.9106529209622e-06,
      "loss": 1.1489,
      "step": 119
    },
    {
      "epoch": 0.41237113402061853,
      "grad_norm": 0.10205078125,
      "learning_rate": 5.876288659793815e-06,
      "loss": 1.22,
      "step": 120
    },
    {
      "epoch": 0.41580756013745707,
      "grad_norm": 0.10205078125,
      "learning_rate": 5.84192439862543e-06,
      "loss": 1.1821,
      "step": 121
    },
    {
      "epoch": 0.41924398625429554,
      "grad_norm": 0.12255859375,
      "learning_rate": 5.807560137457045e-06,
      "loss": 1.1964,
      "step": 122
    },
    {
      "epoch": 0.422680412371134,
      "grad_norm": 0.109375,
      "learning_rate": 5.7731958762886594e-06,
      "loss": 1.1879,
      "step": 123
    },
    {
      "epoch": 0.4261168384879725,
      "grad_norm": 0.1005859375,
      "learning_rate": 5.738831615120275e-06,
      "loss": 1.1497,
      "step": 124
    },
    {
      "epoch": 0.42955326460481097,
      "grad_norm": 0.10107421875,
      "learning_rate": 5.70446735395189e-06,
      "loss": 1.0871,
      "step": 125
    },
    {
      "epoch": 0.4329896907216495,
      "grad_norm": 0.11083984375,
      "learning_rate": 5.670103092783505e-06,
      "loss": 1.2444,
      "step": 126
    },
    {
      "epoch": 0.436426116838488,
      "grad_norm": 0.099609375,
      "learning_rate": 5.6357388316151204e-06,
      "loss": 1.1902,
      "step": 127
    },
    {
      "epoch": 0.43986254295532645,
      "grad_norm": 0.11083984375,
      "learning_rate": 5.601374570446736e-06,
      "loss": 1.2449,
      "step": 128
    },
    {
      "epoch": 0.44329896907216493,
      "grad_norm": 0.111328125,
      "learning_rate": 5.567010309278351e-06,
      "loss": 1.1841,
      "step": 129
    },
    {
      "epoch": 0.44673539518900346,
      "grad_norm": 0.10888671875,
      "learning_rate": 5.532646048109966e-06,
      "loss": 1.2824,
      "step": 130
    },
    {
      "epoch": 0.45017182130584193,
      "grad_norm": 0.1015625,
      "learning_rate": 5.4982817869415815e-06,
      "loss": 1.1912,
      "step": 131
    },
    {
      "epoch": 0.4536082474226804,
      "grad_norm": 0.09521484375,
      "learning_rate": 5.463917525773196e-06,
      "loss": 1.1778,
      "step": 132
    },
    {
      "epoch": 0.4570446735395189,
      "grad_norm": 0.10400390625,
      "learning_rate": 5.429553264604811e-06,
      "loss": 1.1849,
      "step": 133
    },
    {
      "epoch": 0.46048109965635736,
      "grad_norm": 0.10009765625,
      "learning_rate": 5.395189003436427e-06,
      "loss": 1.1319,
      "step": 134
    },
    {
      "epoch": 0.4639175257731959,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.360824742268042e-06,
      "loss": 1.1162,
      "step": 135
    },
    {
      "epoch": 0.46735395189003437,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.326460481099657e-06,
      "loss": 1.1951,
      "step": 136
    },
    {
      "epoch": 0.47079037800687284,
      "grad_norm": 0.0986328125,
      "learning_rate": 5.292096219931272e-06,
      "loss": 1.2168,
      "step": 137
    },
    {
      "epoch": 0.4742268041237113,
      "grad_norm": 0.10986328125,
      "learning_rate": 5.257731958762888e-06,
      "loss": 1.1578,
      "step": 138
    },
    {
      "epoch": 0.47766323024054985,
      "grad_norm": 0.1103515625,
      "learning_rate": 5.223367697594503e-06,
      "loss": 1.3295,
      "step": 139
    },
    {
      "epoch": 0.48109965635738833,
      "grad_norm": 0.10595703125,
      "learning_rate": 5.189003436426118e-06,
      "loss": 1.1521,
      "step": 140
    },
    {
      "epoch": 0.4845360824742268,
      "grad_norm": 0.1044921875,
      "learning_rate": 5.154639175257732e-06,
      "loss": 1.2219,
      "step": 141
    },
    {
      "epoch": 0.4879725085910653,
      "grad_norm": 0.09912109375,
      "learning_rate": 5.120274914089347e-06,
      "loss": 1.1866,
      "step": 142
    },
    {
      "epoch": 0.49140893470790376,
      "grad_norm": 0.10009765625,
      "learning_rate": 5.085910652920962e-06,
      "loss": 1.1922,
      "step": 143
    },
    {
      "epoch": 0.4948453608247423,
      "grad_norm": 0.1005859375,
      "learning_rate": 5.051546391752578e-06,
      "loss": 1.1784,
      "step": 144
    },
    {
      "epoch": 0.49828178694158076,
      "grad_norm": 0.10595703125,
      "learning_rate": 5.017182130584193e-06,
      "loss": 1.1468,
      "step": 145
    },
    {
      "epoch": 0.5017182130584192,
      "grad_norm": 0.103515625,
      "learning_rate": 4.982817869415808e-06,
      "loss": 1.2448,
      "step": 146
    },
    {
      "epoch": 0.5051546391752577,
      "grad_norm": 0.099609375,
      "learning_rate": 4.948453608247423e-06,
      "loss": 1.1422,
      "step": 147
    },
    {
      "epoch": 0.5085910652920962,
      "grad_norm": 0.1123046875,
      "learning_rate": 4.914089347079038e-06,
      "loss": 1.2039,
      "step": 148
    },
    {
      "epoch": 0.5120274914089347,
      "grad_norm": 0.09814453125,
      "learning_rate": 4.879725085910653e-06,
      "loss": 1.2287,
      "step": 149
    },
    {
      "epoch": 0.5154639175257731,
      "grad_norm": 0.10302734375,
      "learning_rate": 4.845360824742268e-06,
      "loss": 1.17,
      "step": 150
    },
    {
      "epoch": 0.5189003436426117,
      "grad_norm": 0.1064453125,
      "learning_rate": 4.810996563573884e-06,
      "loss": 1.0933,
      "step": 151
    },
    {
      "epoch": 0.5223367697594502,
      "grad_norm": 0.1005859375,
      "learning_rate": 4.776632302405499e-06,
      "loss": 1.2443,
      "step": 152
    },
    {
      "epoch": 0.5257731958762887,
      "grad_norm": 0.1220703125,
      "learning_rate": 4.742268041237113e-06,
      "loss": 1.2075,
      "step": 153
    },
    {
      "epoch": 0.5292096219931272,
      "grad_norm": 0.10888671875,
      "learning_rate": 4.707903780068729e-06,
      "loss": 1.1981,
      "step": 154
    },
    {
      "epoch": 0.5326460481099656,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.673539518900344e-06,
      "loss": 1.1483,
      "step": 155
    },
    {
      "epoch": 0.5360824742268041,
      "grad_norm": 0.10205078125,
      "learning_rate": 4.639175257731959e-06,
      "loss": 1.1753,
      "step": 156
    },
    {
      "epoch": 0.5395189003436426,
      "grad_norm": 0.099609375,
      "learning_rate": 4.604810996563574e-06,
      "loss": 1.1578,
      "step": 157
    },
    {
      "epoch": 0.5429553264604811,
      "grad_norm": 0.10498046875,
      "learning_rate": 4.570446735395189e-06,
      "loss": 1.1467,
      "step": 158
    },
    {
      "epoch": 0.5463917525773195,
      "grad_norm": 0.1044921875,
      "learning_rate": 4.536082474226804e-06,
      "loss": 1.188,
      "step": 159
    },
    {
      "epoch": 0.5498281786941581,
      "grad_norm": 0.10595703125,
      "learning_rate": 4.501718213058419e-06,
      "loss": 1.1748,
      "step": 160
    },
    {
      "epoch": 0.5532646048109966,
      "grad_norm": 0.10107421875,
      "learning_rate": 4.467353951890035e-06,
      "loss": 1.1563,
      "step": 161
    },
    {
      "epoch": 0.5567010309278351,
      "grad_norm": 0.103515625,
      "learning_rate": 4.4329896907216494e-06,
      "loss": 1.2162,
      "step": 162
    },
    {
      "epoch": 0.5601374570446735,
      "grad_norm": 0.1708984375,
      "learning_rate": 4.3986254295532645e-06,
      "loss": 1.1612,
      "step": 163
    },
    {
      "epoch": 0.563573883161512,
      "grad_norm": 0.1015625,
      "learning_rate": 4.36426116838488e-06,
      "loss": 1.1738,
      "step": 164
    },
    {
      "epoch": 0.5670103092783505,
      "grad_norm": 0.11669921875,
      "learning_rate": 4.329896907216495e-06,
      "loss": 1.2428,
      "step": 165
    },
    {
      "epoch": 0.570446735395189,
      "grad_norm": 0.1767578125,
      "learning_rate": 4.2955326460481105e-06,
      "loss": 1.2499,
      "step": 166
    },
    {
      "epoch": 0.5738831615120275,
      "grad_norm": 0.15234375,
      "learning_rate": 4.2611683848797255e-06,
      "loss": 1.1381,
      "step": 167
    },
    {
      "epoch": 0.5773195876288659,
      "grad_norm": 0.107421875,
      "learning_rate": 4.2268041237113405e-06,
      "loss": 1.1218,
      "step": 168
    },
    {
      "epoch": 0.5807560137457045,
      "grad_norm": 0.103515625,
      "learning_rate": 4.192439862542956e-06,
      "loss": 1.1917,
      "step": 169
    },
    {
      "epoch": 0.584192439862543,
      "grad_norm": 0.1201171875,
      "learning_rate": 4.158075601374571e-06,
      "loss": 1.2737,
      "step": 170
    },
    {
      "epoch": 0.5876288659793815,
      "grad_norm": 0.1376953125,
      "learning_rate": 4.123711340206186e-06,
      "loss": 1.1914,
      "step": 171
    },
    {
      "epoch": 0.5910652920962199,
      "grad_norm": 0.10546875,
      "learning_rate": 4.089347079037801e-06,
      "loss": 1.1924,
      "step": 172
    },
    {
      "epoch": 0.5945017182130584,
      "grad_norm": 0.10595703125,
      "learning_rate": 4.054982817869416e-06,
      "loss": 1.186,
      "step": 173
    },
    {
      "epoch": 0.5979381443298969,
      "grad_norm": 0.1591796875,
      "learning_rate": 4.020618556701032e-06,
      "loss": 1.2316,
      "step": 174
    },
    {
      "epoch": 0.6013745704467354,
      "grad_norm": 0.1015625,
      "learning_rate": 3.986254295532647e-06,
      "loss": 1.1319,
      "step": 175
    },
    {
      "epoch": 0.6048109965635738,
      "grad_norm": 0.11083984375,
      "learning_rate": 3.951890034364262e-06,
      "loss": 1.1959,
      "step": 176
    },
    {
      "epoch": 0.6082474226804123,
      "grad_norm": 0.1103515625,
      "learning_rate": 3.917525773195877e-06,
      "loss": 1.192,
      "step": 177
    },
    {
      "epoch": 0.6116838487972509,
      "grad_norm": 0.10498046875,
      "learning_rate": 3.883161512027492e-06,
      "loss": 1.1374,
      "step": 178
    },
    {
      "epoch": 0.6151202749140894,
      "grad_norm": 0.11865234375,
      "learning_rate": 3.848797250859107e-06,
      "loss": 1.2014,
      "step": 179
    },
    {
      "epoch": 0.6185567010309279,
      "grad_norm": 0.12890625,
      "learning_rate": 3.814432989690722e-06,
      "loss": 1.2549,
      "step": 180
    },
    {
      "epoch": 0.6219931271477663,
      "grad_norm": 0.10107421875,
      "learning_rate": 3.780068728522337e-06,
      "loss": 1.2026,
      "step": 181
    },
    {
      "epoch": 0.6254295532646048,
      "grad_norm": 0.10693359375,
      "learning_rate": 3.7457044673539524e-06,
      "loss": 1.249,
      "step": 182
    },
    {
      "epoch": 0.6288659793814433,
      "grad_norm": 0.11865234375,
      "learning_rate": 3.7113402061855674e-06,
      "loss": 1.2328,
      "step": 183
    },
    {
      "epoch": 0.6323024054982818,
      "grad_norm": 0.10205078125,
      "learning_rate": 3.6769759450171825e-06,
      "loss": 1.1544,
      "step": 184
    },
    {
      "epoch": 0.6357388316151202,
      "grad_norm": 0.099609375,
      "learning_rate": 3.6426116838487975e-06,
      "loss": 1.175,
      "step": 185
    },
    {
      "epoch": 0.6391752577319587,
      "grad_norm": 0.1552734375,
      "learning_rate": 3.6082474226804126e-06,
      "loss": 1.122,
      "step": 186
    },
    {
      "epoch": 0.6426116838487973,
      "grad_norm": 0.123046875,
      "learning_rate": 3.573883161512028e-06,
      "loss": 1.1623,
      "step": 187
    },
    {
      "epoch": 0.6460481099656358,
      "grad_norm": 0.10302734375,
      "learning_rate": 3.539518900343643e-06,
      "loss": 1.1869,
      "step": 188
    },
    {
      "epoch": 0.6494845360824743,
      "grad_norm": 0.0966796875,
      "learning_rate": 3.5051546391752577e-06,
      "loss": 1.1359,
      "step": 189
    },
    {
      "epoch": 0.6529209621993127,
      "grad_norm": 0.1103515625,
      "learning_rate": 3.470790378006873e-06,
      "loss": 1.1882,
      "step": 190
    },
    {
      "epoch": 0.6563573883161512,
      "grad_norm": 0.109375,
      "learning_rate": 3.436426116838488e-06,
      "loss": 1.245,
      "step": 191
    },
    {
      "epoch": 0.6597938144329897,
      "grad_norm": 0.09912109375,
      "learning_rate": 3.4020618556701037e-06,
      "loss": 1.1544,
      "step": 192
    },
    {
      "epoch": 0.6632302405498282,
      "grad_norm": 0.11474609375,
      "learning_rate": 3.3676975945017187e-06,
      "loss": 1.1664,
      "step": 193
    },
    {
      "epoch": 0.6666666666666666,
      "grad_norm": 0.1123046875,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 1.1409,
      "step": 194
    },
    {
      "epoch": 0.6701030927835051,
      "grad_norm": 0.1044921875,
      "learning_rate": 3.298969072164949e-06,
      "loss": 1.144,
      "step": 195
    },
    {
      "epoch": 0.6735395189003437,
      "grad_norm": 0.111328125,
      "learning_rate": 3.264604810996564e-06,
      "loss": 1.2533,
      "step": 196
    },
    {
      "epoch": 0.6769759450171822,
      "grad_norm": 0.10693359375,
      "learning_rate": 3.2302405498281793e-06,
      "loss": 1.1903,
      "step": 197
    },
    {
      "epoch": 0.6804123711340206,
      "grad_norm": 0.12109375,
      "learning_rate": 3.195876288659794e-06,
      "loss": 1.1853,
      "step": 198
    },
    {
      "epoch": 0.6838487972508591,
      "grad_norm": 0.1083984375,
      "learning_rate": 3.161512027491409e-06,
      "loss": 1.1456,
      "step": 199
    },
    {
      "epoch": 0.6872852233676976,
      "grad_norm": 0.10498046875,
      "learning_rate": 3.1271477663230244e-06,
      "loss": 1.1724,
      "step": 200
    },
    {
      "epoch": 0.6907216494845361,
      "grad_norm": 0.11181640625,
      "learning_rate": 3.0927835051546395e-06,
      "loss": 1.2242,
      "step": 201
    },
    {
      "epoch": 0.6941580756013745,
      "grad_norm": 0.1494140625,
      "learning_rate": 3.058419243986255e-06,
      "loss": 1.2096,
      "step": 202
    },
    {
      "epoch": 0.697594501718213,
      "grad_norm": 0.1171875,
      "learning_rate": 3.0240549828178695e-06,
      "loss": 1.1972,
      "step": 203
    },
    {
      "epoch": 0.7010309278350515,
      "grad_norm": 0.1240234375,
      "learning_rate": 2.9896907216494846e-06,
      "loss": 1.2032,
      "step": 204
    },
    {
      "epoch": 0.7044673539518901,
      "grad_norm": 0.11279296875,
      "learning_rate": 2.9553264604811e-06,
      "loss": 1.1783,
      "step": 205
    },
    {
      "epoch": 0.7079037800687286,
      "grad_norm": 0.10986328125,
      "learning_rate": 2.920962199312715e-06,
      "loss": 1.1432,
      "step": 206
    },
    {
      "epoch": 0.711340206185567,
      "grad_norm": 0.1083984375,
      "learning_rate": 2.8865979381443297e-06,
      "loss": 1.1197,
      "step": 207
    },
    {
      "epoch": 0.7147766323024055,
      "grad_norm": 0.1064453125,
      "learning_rate": 2.852233676975945e-06,
      "loss": 1.1736,
      "step": 208
    },
    {
      "epoch": 0.718213058419244,
      "grad_norm": 0.126953125,
      "learning_rate": 2.8178694158075602e-06,
      "loss": 1.1505,
      "step": 209
    },
    {
      "epoch": 0.7216494845360825,
      "grad_norm": 0.1123046875,
      "learning_rate": 2.7835051546391757e-06,
      "loss": 1.2264,
      "step": 210
    },
    {
      "epoch": 0.7250859106529209,
      "grad_norm": 0.12255859375,
      "learning_rate": 2.7491408934707907e-06,
      "loss": 1.1537,
      "step": 211
    },
    {
      "epoch": 0.7285223367697594,
      "grad_norm": 0.1083984375,
      "learning_rate": 2.7147766323024053e-06,
      "loss": 1.2153,
      "step": 212
    },
    {
      "epoch": 0.7319587628865979,
      "grad_norm": 0.10498046875,
      "learning_rate": 2.680412371134021e-06,
      "loss": 1.1621,
      "step": 213
    },
    {
      "epoch": 0.7353951890034365,
      "grad_norm": 0.109375,
      "learning_rate": 2.646048109965636e-06,
      "loss": 1.1657,
      "step": 214
    },
    {
      "epoch": 0.738831615120275,
      "grad_norm": 0.11376953125,
      "learning_rate": 2.6116838487972513e-06,
      "loss": 1.1345,
      "step": 215
    },
    {
      "epoch": 0.7422680412371134,
      "grad_norm": 0.11328125,
      "learning_rate": 2.577319587628866e-06,
      "loss": 1.1935,
      "step": 216
    },
    {
      "epoch": 0.7457044673539519,
      "grad_norm": 0.10302734375,
      "learning_rate": 2.542955326460481e-06,
      "loss": 1.1348,
      "step": 217
    },
    {
      "epoch": 0.7491408934707904,
      "grad_norm": 0.11083984375,
      "learning_rate": 2.5085910652920964e-06,
      "loss": 1.1279,
      "step": 218
    },
    {
      "epoch": 0.7525773195876289,
      "grad_norm": 0.115234375,
      "learning_rate": 2.4742268041237115e-06,
      "loss": 1.1244,
      "step": 219
    },
    {
      "epoch": 0.7560137457044673,
      "grad_norm": 0.109375,
      "learning_rate": 2.4398625429553265e-06,
      "loss": 1.1416,
      "step": 220
    },
    {
      "epoch": 0.7594501718213058,
      "grad_norm": 0.1396484375,
      "learning_rate": 2.405498281786942e-06,
      "loss": 1.2763,
      "step": 221
    },
    {
      "epoch": 0.7628865979381443,
      "grad_norm": 0.10693359375,
      "learning_rate": 2.3711340206185566e-06,
      "loss": 1.112,
      "step": 222
    },
    {
      "epoch": 0.7663230240549829,
      "grad_norm": 0.11279296875,
      "learning_rate": 2.336769759450172e-06,
      "loss": 1.2058,
      "step": 223
    },
    {
      "epoch": 0.7697594501718213,
      "grad_norm": 0.125,
      "learning_rate": 2.302405498281787e-06,
      "loss": 1.217,
      "step": 224
    },
    {
      "epoch": 0.7731958762886598,
      "grad_norm": 0.11962890625,
      "learning_rate": 2.268041237113402e-06,
      "loss": 1.1224,
      "step": 225
    },
    {
      "epoch": 0.7766323024054983,
      "grad_norm": 0.1181640625,
      "learning_rate": 2.2336769759450176e-06,
      "loss": 1.2149,
      "step": 226
    },
    {
      "epoch": 0.7800687285223368,
      "grad_norm": 0.10693359375,
      "learning_rate": 2.1993127147766322e-06,
      "loss": 1.1129,
      "step": 227
    },
    {
      "epoch": 0.7835051546391752,
      "grad_norm": 0.11767578125,
      "learning_rate": 2.1649484536082477e-06,
      "loss": 1.1593,
      "step": 228
    },
    {
      "epoch": 0.7869415807560137,
      "grad_norm": 0.11279296875,
      "learning_rate": 2.1305841924398628e-06,
      "loss": 1.1298,
      "step": 229
    },
    {
      "epoch": 0.7903780068728522,
      "grad_norm": 0.1220703125,
      "learning_rate": 2.096219931271478e-06,
      "loss": 1.2249,
      "step": 230
    },
    {
      "epoch": 0.7938144329896907,
      "grad_norm": 0.1162109375,
      "learning_rate": 2.061855670103093e-06,
      "loss": 1.1265,
      "step": 231
    },
    {
      "epoch": 0.7972508591065293,
      "grad_norm": 0.10791015625,
      "learning_rate": 2.027491408934708e-06,
      "loss": 1.0844,
      "step": 232
    },
    {
      "epoch": 0.8006872852233677,
      "grad_norm": 0.1083984375,
      "learning_rate": 1.9931271477663233e-06,
      "loss": 1.1828,
      "step": 233
    },
    {
      "epoch": 0.8041237113402062,
      "grad_norm": 0.1572265625,
      "learning_rate": 1.9587628865979384e-06,
      "loss": 1.1879,
      "step": 234
    },
    {
      "epoch": 0.8075601374570447,
      "grad_norm": 0.1220703125,
      "learning_rate": 1.9243986254295534e-06,
      "loss": 1.2128,
      "step": 235
    },
    {
      "epoch": 0.8109965635738832,
      "grad_norm": 0.11474609375,
      "learning_rate": 1.8900343642611685e-06,
      "loss": 1.1451,
      "step": 236
    },
    {
      "epoch": 0.8144329896907216,
      "grad_norm": 0.11279296875,
      "learning_rate": 1.8556701030927837e-06,
      "loss": 1.1279,
      "step": 237
    },
    {
      "epoch": 0.8178694158075601,
      "grad_norm": 0.11083984375,
      "learning_rate": 1.8213058419243988e-06,
      "loss": 1.1636,
      "step": 238
    },
    {
      "epoch": 0.8213058419243986,
      "grad_norm": 0.11083984375,
      "learning_rate": 1.786941580756014e-06,
      "loss": 1.1323,
      "step": 239
    },
    {
      "epoch": 0.8247422680412371,
      "grad_norm": 0.1767578125,
      "learning_rate": 1.7525773195876288e-06,
      "loss": 1.1456,
      "step": 240
    },
    {
      "epoch": 0.8281786941580757,
      "grad_norm": 0.1181640625,
      "learning_rate": 1.718213058419244e-06,
      "loss": 1.1559,
      "step": 241
    },
    {
      "epoch": 0.8316151202749141,
      "grad_norm": 0.1591796875,
      "learning_rate": 1.6838487972508594e-06,
      "loss": 1.1903,
      "step": 242
    },
    {
      "epoch": 0.8350515463917526,
      "grad_norm": 0.1376953125,
      "learning_rate": 1.6494845360824744e-06,
      "loss": 1.1933,
      "step": 243
    },
    {
      "epoch": 0.8384879725085911,
      "grad_norm": 0.1328125,
      "learning_rate": 1.6151202749140896e-06,
      "loss": 1.2029,
      "step": 244
    },
    {
      "epoch": 0.8419243986254296,
      "grad_norm": 0.1201171875,
      "learning_rate": 1.5807560137457045e-06,
      "loss": 1.1551,
      "step": 245
    },
    {
      "epoch": 0.845360824742268,
      "grad_norm": 0.11279296875,
      "learning_rate": 1.5463917525773197e-06,
      "loss": 1.1213,
      "step": 246
    },
    {
      "epoch": 0.8487972508591065,
      "grad_norm": 0.11572265625,
      "learning_rate": 1.5120274914089348e-06,
      "loss": 1.1897,
      "step": 247
    },
    {
      "epoch": 0.852233676975945,
      "grad_norm": 0.1142578125,
      "learning_rate": 1.47766323024055e-06,
      "loss": 1.2245,
      "step": 248
    },
    {
      "epoch": 0.8556701030927835,
      "grad_norm": 0.1162109375,
      "learning_rate": 1.4432989690721649e-06,
      "loss": 1.1553,
      "step": 249
    },
    {
      "epoch": 0.8591065292096219,
      "grad_norm": 0.134765625,
      "learning_rate": 1.4089347079037801e-06,
      "loss": 1.12,
      "step": 250
    },
    {
      "epoch": 0.8625429553264605,
      "grad_norm": 0.10986328125,
      "learning_rate": 1.3745704467353954e-06,
      "loss": 1.0931,
      "step": 251
    },
    {
      "epoch": 0.865979381443299,
      "grad_norm": 0.208984375,
      "learning_rate": 1.3402061855670104e-06,
      "loss": 1.1994,
      "step": 252
    },
    {
      "epoch": 0.8694158075601375,
      "grad_norm": 0.10888671875,
      "learning_rate": 1.3058419243986257e-06,
      "loss": 1.1478,
      "step": 253
    },
    {
      "epoch": 0.872852233676976,
      "grad_norm": 0.1181640625,
      "learning_rate": 1.2714776632302405e-06,
      "loss": 1.1798,
      "step": 254
    },
    {
      "epoch": 0.8762886597938144,
      "grad_norm": 0.1318359375,
      "learning_rate": 1.2371134020618557e-06,
      "loss": 1.1266,
      "step": 255
    },
    {
      "epoch": 0.8797250859106529,
      "grad_norm": 0.11962890625,
      "learning_rate": 1.202749140893471e-06,
      "loss": 1.1522,
      "step": 256
    },
    {
      "epoch": 0.8831615120274914,
      "grad_norm": 0.11083984375,
      "learning_rate": 1.168384879725086e-06,
      "loss": 1.1133,
      "step": 257
    },
    {
      "epoch": 0.8865979381443299,
      "grad_norm": 0.11572265625,
      "learning_rate": 1.134020618556701e-06,
      "loss": 1.2044,
      "step": 258
    },
    {
      "epoch": 0.8900343642611683,
      "grad_norm": 0.1279296875,
      "learning_rate": 1.0996563573883161e-06,
      "loss": 1.2362,
      "step": 259
    },
    {
      "epoch": 0.8934707903780069,
      "grad_norm": 0.1728515625,
      "learning_rate": 1.0652920962199314e-06,
      "loss": 1.1711,
      "step": 260
    },
    {
      "epoch": 0.8969072164948454,
      "grad_norm": 0.125,
      "learning_rate": 1.0309278350515464e-06,
      "loss": 1.1429,
      "step": 261
    },
    {
      "epoch": 0.9003436426116839,
      "grad_norm": 0.14453125,
      "learning_rate": 9.965635738831617e-07,
      "loss": 1.2533,
      "step": 262
    },
    {
      "epoch": 0.9037800687285223,
      "grad_norm": 0.126953125,
      "learning_rate": 9.621993127147767e-07,
      "loss": 1.1734,
      "step": 263
    },
    {
      "epoch": 0.9072164948453608,
      "grad_norm": 0.11572265625,
      "learning_rate": 9.278350515463919e-07,
      "loss": 1.1519,
      "step": 264
    },
    {
      "epoch": 0.9106529209621993,
      "grad_norm": 0.11865234375,
      "learning_rate": 8.93470790378007e-07,
      "loss": 1.1965,
      "step": 265
    },
    {
      "epoch": 0.9140893470790378,
      "grad_norm": 0.123046875,
      "learning_rate": 8.59106529209622e-07,
      "loss": 1.1872,
      "step": 266
    },
    {
      "epoch": 0.9175257731958762,
      "grad_norm": 0.12890625,
      "learning_rate": 8.247422680412372e-07,
      "loss": 1.28,
      "step": 267
    },
    {
      "epoch": 0.9209621993127147,
      "grad_norm": 0.1416015625,
      "learning_rate": 7.903780068728522e-07,
      "loss": 1.0979,
      "step": 268
    },
    {
      "epoch": 0.9243986254295533,
      "grad_norm": 0.1484375,
      "learning_rate": 7.560137457044674e-07,
      "loss": 1.1781,
      "step": 269
    },
    {
      "epoch": 0.9278350515463918,
      "grad_norm": 0.115234375,
      "learning_rate": 7.216494845360824e-07,
      "loss": 1.1292,
      "step": 270
    },
    {
      "epoch": 0.9312714776632303,
      "grad_norm": 0.11376953125,
      "learning_rate": 6.872852233676977e-07,
      "loss": 1.1431,
      "step": 271
    },
    {
      "epoch": 0.9347079037800687,
      "grad_norm": 0.11669921875,
      "learning_rate": 6.529209621993128e-07,
      "loss": 1.1752,
      "step": 272
    },
    {
      "epoch": 0.9381443298969072,
      "grad_norm": 0.11376953125,
      "learning_rate": 6.185567010309279e-07,
      "loss": 1.167,
      "step": 273
    },
    {
      "epoch": 0.9415807560137457,
      "grad_norm": 0.11328125,
      "learning_rate": 5.84192439862543e-07,
      "loss": 1.1292,
      "step": 274
    },
    {
      "epoch": 0.9450171821305842,
      "grad_norm": 0.125,
      "learning_rate": 5.498281786941581e-07,
      "loss": 1.2543,
      "step": 275
    },
    {
      "epoch": 0.9484536082474226,
      "grad_norm": 0.12890625,
      "learning_rate": 5.154639175257732e-07,
      "loss": 1.2113,
      "step": 276
    },
    {
      "epoch": 0.9518900343642611,
      "grad_norm": 0.119140625,
      "learning_rate": 4.810996563573884e-07,
      "loss": 1.1205,
      "step": 277
    },
    {
      "epoch": 0.9553264604810997,
      "grad_norm": 0.12109375,
      "learning_rate": 4.467353951890035e-07,
      "loss": 1.1825,
      "step": 278
    },
    {
      "epoch": 0.9587628865979382,
      "grad_norm": 0.39453125,
      "learning_rate": 4.123711340206186e-07,
      "loss": 1.1801,
      "step": 279
    },
    {
      "epoch": 0.9621993127147767,
      "grad_norm": 0.16015625,
      "learning_rate": 3.780068728522337e-07,
      "loss": 1.1751,
      "step": 280
    },
    {
      "epoch": 0.9656357388316151,
      "grad_norm": 0.1259765625,
      "learning_rate": 3.4364261168384884e-07,
      "loss": 1.1759,
      "step": 281
    },
    {
      "epoch": 0.9690721649484536,
      "grad_norm": 0.1298828125,
      "learning_rate": 3.0927835051546394e-07,
      "loss": 1.1685,
      "step": 282
    },
    {
      "epoch": 0.9725085910652921,
      "grad_norm": 0.1796875,
      "learning_rate": 2.7491408934707903e-07,
      "loss": 1.1825,
      "step": 283
    },
    {
      "epoch": 0.9759450171821306,
      "grad_norm": 0.126953125,
      "learning_rate": 2.405498281786942e-07,
      "loss": 1.1749,
      "step": 284
    },
    {
      "epoch": 0.979381443298969,
      "grad_norm": 0.1337890625,
      "learning_rate": 2.061855670103093e-07,
      "loss": 1.1566,
      "step": 285
    },
    {
      "epoch": 0.9828178694158075,
      "grad_norm": 0.1201171875,
      "learning_rate": 1.7182130584192442e-07,
      "loss": 1.1753,
      "step": 286
    },
    {
      "epoch": 0.9862542955326461,
      "grad_norm": 0.185546875,
      "learning_rate": 1.3745704467353952e-07,
      "loss": 1.149,
      "step": 287
    },
    {
      "epoch": 0.9896907216494846,
      "grad_norm": 0.1171875,
      "learning_rate": 1.0309278350515465e-07,
      "loss": 1.114,
      "step": 288
    },
    {
      "epoch": 0.993127147766323,
      "grad_norm": 0.130859375,
      "learning_rate": 6.872852233676976e-08,
      "loss": 1.1794,
      "step": 289
    },
    {
      "epoch": 0.9965635738831615,
      "grad_norm": 0.14453125,
      "learning_rate": 3.436426116838488e-08,
      "loss": 1.2422,
      "step": 290
    },
    {
      "epoch": 1.0,
      "grad_norm": 0.125,
      "learning_rate": 0.0,
      "loss": 1.257,
      "step": 291
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 291,
  "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": 9.177843676804547e+17,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}