{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 2.0,
  "global_step": 250,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.25e-05,
      "loss": 35.6005,
      "step": 1
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000125,
      "loss": 29.7436,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001875,
      "loss": 53.5485,
      "step": 3
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00025,
      "loss": 25.7467,
      "step": 4
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0003125,
      "loss": 33.2158,
      "step": 5
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000375,
      "loss": 19.5161,
      "step": 6
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0004375,
      "loss": 13.8304,
      "step": 7
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0005,
      "loss": 22.178,
      "step": 8
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0004999789344439543,
      "loss": 34.5256,
      "step": 9
    },
    {
      "epoch": 0.08,
      "learning_rate": 0.0004999157413258782,
      "loss": 10.8449,
      "step": 10
    },
    {
      "epoch": 0.09,
      "learning_rate": 0.000499810431295357,
      "loss": 6.3006,
      "step": 11
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0004996630220997058,
      "loss": 7.4694,
      "step": 12
    },
    {
      "epoch": 0.1,
      "learning_rate": 0.0004994735385809777,
      "loss": 14.6974,
      "step": 13
    },
    {
      "epoch": 0.11,
      "learning_rate": 0.0004992420126717783,
      "loss": 12.4919,
      "step": 14
    },
    {
      "epoch": 0.12,
      "learning_rate": 0.0004989684833898838,
      "loss": 5.8078,
      "step": 15
    },
    {
      "epoch": 0.13,
      "learning_rate": 0.0004986529968316653,
      "loss": 14.5466,
      "step": 16
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0004982956061643213,
      "loss": 8.1059,
      "step": 17
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0004978963716169166,
      "loss": 35.4656,
      "step": 18
    },
    {
      "epoch": 0.15,
      "learning_rate": 0.0004974553604702333,
      "loss": 39.1298,
      "step": 19
    },
    {
      "epoch": 0.16,
      "learning_rate": 0.0004969726470454314,
      "loss": 32.5985,
      "step": 20
    },
    {
      "epoch": 0.17,
      "learning_rate": 0.0004964483126915245,
      "loss": 29.7271,
      "step": 21
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0004958824457716707,
      "loss": 55.4317,
      "step": 22
    },
    {
      "epoch": 0.18,
      "learning_rate": 0.0004952751416482801,
      "loss": 29.8617,
      "step": 23
    },
    {
      "epoch": 0.19,
      "learning_rate": 0.0004946265026669455,
      "loss": 43.5113,
      "step": 24
    },
    {
      "epoch": 0.2,
      "learning_rate": 0.000493936638139193,
      "loss": 12.1452,
      "step": 25
    },
    {
      "epoch": 0.21,
      "learning_rate": 0.0004932056643240618,
      "loss": 24.7202,
      "step": 26
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.000492433704408511,
      "loss": 12.7435,
      "step": 27
    },
    {
      "epoch": 0.22,
      "learning_rate": 0.0004916208884866593,
      "loss": 32.6779,
      "step": 28
    },
    {
      "epoch": 0.23,
      "learning_rate": 0.0004907673535378617,
      "loss": 14.9898,
      "step": 29
    },
    {
      "epoch": 0.24,
      "learning_rate": 0.0004898732434036243,
      "loss": 21.4031,
      "step": 30
    },
    {
      "epoch": 0.25,
      "learning_rate": 0.0004889387087633647,
      "loss": 16.9956,
      "step": 31
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.0004879639071090174,
      "loss": 17.35,
      "step": 32
    },
    {
      "epoch": 0.26,
      "learning_rate": 0.00048694900271849354,
      "loss": 17.5599,
      "step": 33
    },
    {
      "epoch": 0.27,
      "learning_rate": 0.00048589416662799553,
      "loss": 16.571,
      "step": 34
    },
    {
      "epoch": 0.28,
      "learning_rate": 0.00048479957660319374,
      "loss": 5.7298,
      "step": 35
    },
    {
      "epoch": 0.29,
      "learning_rate": 0.0004836654171092683,
      "loss": 12.8386,
      "step": 36
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.0004824918792798222,
      "loss": 14.2135,
      "step": 37
    },
    {
      "epoch": 0.3,
      "learning_rate": 0.00048127916088467097,
      "loss": 11.3054,
      "step": 38
    },
    {
      "epoch": 0.31,
      "learning_rate": 0.00048002746629651296,
      "loss": 16.8367,
      "step": 39
    },
    {
      "epoch": 0.32,
      "learning_rate": 0.0004787370064564883,
      "loss": 5.4942,
      "step": 40
    },
    {
      "epoch": 0.33,
      "learning_rate": 0.0004774079988386296,
      "loss": 4.8653,
      "step": 41
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.00047604066741321254,
      "loss": 24.3939,
      "step": 42
    },
    {
      "epoch": 0.34,
      "learning_rate": 0.0004746352426090114,
      "loss": 9.7793,
      "step": 43
    },
    {
      "epoch": 0.35,
      "learning_rate": 0.00047319196127446593,
      "loss": 5.1413,
      "step": 44
    },
    {
      "epoch": 0.36,
      "learning_rate": 0.0004717110666377669,
      "loss": 8.2737,
      "step": 45
    },
    {
      "epoch": 0.37,
      "learning_rate": 0.0004701928082658661,
      "loss": 15.5838,
      "step": 46
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00046863744202241806,
      "loss": 2.9668,
      "step": 47
    },
    {
      "epoch": 0.38,
      "learning_rate": 0.00046704523002466097,
      "loss": 4.941,
      "step": 48
    },
    {
      "epoch": 0.39,
      "learning_rate": 0.00046541644059924347,
      "loss": 5.507,
      "step": 49
    },
    {
      "epoch": 0.4,
      "learning_rate": 0.000463751348237005,
      "loss": 3.6334,
      "step": 50
    },
    {
      "epoch": 0.41,
      "learning_rate": 0.0004620502335467174,
      "loss": 5.5532,
      "step": 51
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.0004603133832077953,
      "loss": 4.0241,
      "step": 52
    },
    {
      "epoch": 0.42,
      "learning_rate": 0.00045854108992198417,
      "loss": 4.7032,
      "step": 53
    },
    {
      "epoch": 0.43,
      "learning_rate": 0.00045673365236403217,
      "loss": 5.4451,
      "step": 54
    },
    {
      "epoch": 0.44,
      "learning_rate": 0.0004548913751313568,
      "loss": 25.8355,
      "step": 55
    },
    {
      "epoch": 0.45,
      "learning_rate": 0.0004530145686927125,
      "loss": 3.9343,
      "step": 56
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.00045110354933586896,
      "loss": 8.62,
      "step": 57
    },
    {
      "epoch": 0.46,
      "learning_rate": 0.000449158639114309,
      "loss": 4.9604,
      "step": 58
    },
    {
      "epoch": 0.47,
      "learning_rate": 0.0004471801657929551,
      "loss": 8.0785,
      "step": 59
    },
    {
      "epoch": 0.48,
      "learning_rate": 0.000445168462792932,
      "loss": 4.1231,
      "step": 60
    },
    {
      "epoch": 0.49,
      "learning_rate": 0.0004431238691353784,
      "loss": 6.2251,
      "step": 61
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0004410467293843123,
      "loss": 5.098,
      "step": 62
    },
    {
      "epoch": 0.5,
      "learning_rate": 0.0004389373935885646,
      "loss": 3.343,
      "step": 63
    },
    {
      "epoch": 0.51,
      "learning_rate": 0.0004367962172227866,
      "loss": 3.7448,
      "step": 64
    },
    {
      "epoch": 0.52,
      "learning_rate": 0.0004346235611275443,
      "loss": 3.0778,
      "step": 65
    },
    {
      "epoch": 0.53,
      "learning_rate": 0.0004324197914485075,
      "loss": 3.2933,
      "step": 66
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.0004301852795747458,
      "loss": 2.6587,
      "step": 67
    },
    {
      "epoch": 0.54,
      "learning_rate": 0.00042792040207614005,
      "loss": 47.8005,
      "step": 68
    },
    {
      "epoch": 0.55,
      "learning_rate": 0.0004256255406399213,
      "loss": 4.3677,
      "step": 69
    },
    {
      "epoch": 0.56,
      "learning_rate": 0.00042330108200634725,
      "loss": 3.8955,
      "step": 70
    },
    {
      "epoch": 0.57,
      "learning_rate": 0.00042094741790352675,
      "loss": 2.8307,
      "step": 71
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.0004185649449814045,
      "loss": 3.4969,
      "step": 72
    },
    {
      "epoch": 0.58,
      "learning_rate": 0.00041615406474491533,
      "loss": 4.0794,
      "step": 73
    },
    {
      "epoch": 0.59,
      "learning_rate": 0.0004137151834863213,
      "loss": 3.8236,
      "step": 74
    },
    {
      "epoch": 0.6,
      "learning_rate": 0.000411248712216741,
      "loss": 4.3189,
      "step": 75
    },
    {
      "epoch": 0.61,
      "learning_rate": 0.0004087550665968846,
      "loss": 58.4084,
      "step": 76
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0004062346668670046,
      "loss": 3.7195,
      "step": 77
    },
    {
      "epoch": 0.62,
      "learning_rate": 0.0004036879377760753,
      "loss": 3.752,
      "step": 78
    },
    {
      "epoch": 0.63,
      "learning_rate": 0.0004011153085102116,
      "loss": 3.5386,
      "step": 79
    },
    {
      "epoch": 0.64,
      "learning_rate": 0.00039851721262034156,
      "loss": 4.0915,
      "step": 80
    },
    {
      "epoch": 0.65,
      "learning_rate": 0.0003958940879491418,
      "loss": 3.6754,
      "step": 81
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00039324637655725053,
      "loss": 3.4652,
      "step": 82
    },
    {
      "epoch": 0.66,
      "learning_rate": 0.00039057452464876945,
      "loss": 3.5402,
      "step": 83
    },
    {
      "epoch": 0.67,
      "learning_rate": 0.00038787898249606766,
      "loss": 35.8811,
      "step": 84
    },
    {
      "epoch": 0.68,
      "learning_rate": 0.0003851602043638994,
      "loss": 4.1382,
      "step": 85
    },
    {
      "epoch": 0.69,
      "learning_rate": 0.00038241864843284966,
      "loss": 3.0867,
      "step": 86
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.00037965477672211934,
      "loss": 3.657,
      "step": 87
    },
    {
      "epoch": 0.7,
      "learning_rate": 0.0003768690550116639,
      "loss": 3.2442,
      "step": 88
    },
    {
      "epoch": 0.71,
      "learning_rate": 0.00037406195276369793,
      "loss": 2.8701,
      "step": 89
    },
    {
      "epoch": 0.72,
      "learning_rate": 0.0003712339430435792,
      "loss": 2.6621,
      "step": 90
    },
    {
      "epoch": 0.73,
      "learning_rate": 0.00036838550244008573,
      "loss": 3.0335,
      "step": 91
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.00036551711098509903,
      "loss": 2.8238,
      "step": 92
    },
    {
      "epoch": 0.74,
      "learning_rate": 0.0003626292520727067,
      "loss": 3.2807,
      "step": 93
    },
    {
      "epoch": 0.75,
      "learning_rate": 0.0003597224123777389,
      "loss": 3.054,
      "step": 94
    },
    {
      "epoch": 0.76,
      "learning_rate": 0.0003567970817737518,
      "loss": 2.7397,
      "step": 95
    },
    {
      "epoch": 0.77,
      "learning_rate": 0.00035385375325047166,
      "loss": 4.0169,
      "step": 96
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0003508929228307142,
      "loss": 5.2793,
      "step": 97
    },
    {
      "epoch": 0.78,
      "learning_rate": 0.0003479150894867926,
      "loss": 14.8092,
      "step": 98
    },
    {
      "epoch": 0.79,
      "learning_rate": 0.0003449207550564285,
      "loss": 2.8473,
      "step": 99
    },
    {
      "epoch": 0.8,
      "learning_rate": 0.00034191042415818,
      "loss": 5.8621,
      "step": 100
    },
    {
      "epoch": 0.81,
      "learning_rate": 0.0003388846041064012,
      "loss": 8.8111,
      "step": 101
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0003358438048257472,
      "loss": 3.2919,
      "step": 102
    },
    {
      "epoch": 0.82,
      "learning_rate": 0.0003327885387652391,
      "loss": 3.0,
      "step": 103
    },
    {
      "epoch": 0.83,
      "learning_rate": 0.0003297193208119047,
      "loss": 3.848,
      "step": 104
    },
    {
      "epoch": 0.84,
      "learning_rate": 0.0003266366682040063,
      "loss": 2.577,
      "step": 105
    },
    {
      "epoch": 0.85,
      "learning_rate": 0.0003235411004438741,
      "loss": 4.1369,
      "step": 106
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00032043313921035745,
      "loss": 3.0288,
      "step": 107
    },
    {
      "epoch": 0.86,
      "learning_rate": 0.00031731330827090864,
      "loss": 2.8884,
      "step": 108
    },
    {
      "epoch": 0.87,
      "learning_rate": 0.0003141821333933158,
      "loss": 3.185,
      "step": 109
    },
    {
      "epoch": 0.88,
      "learning_rate": 0.00031104014225709787,
      "loss": 3.4899,
      "step": 110
    },
    {
      "epoch": 0.89,
      "learning_rate": 0.0003078878643645778,
      "loss": 2.6116,
      "step": 111
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00030472583095164874,
      "loss": 3.3813,
      "step": 112
    },
    {
      "epoch": 0.9,
      "learning_rate": 0.00030155457489824706,
      "loss": 2.9228,
      "step": 113
    },
    {
      "epoch": 0.91,
      "learning_rate": 0.0002983746306385499,
      "loss": 5.7183,
      "step": 114
    },
    {
      "epoch": 0.92,
      "learning_rate": 0.00029518653407090953,
      "loss": 2.9872,
      "step": 115
    },
    {
      "epoch": 0.93,
      "learning_rate": 0.00029199082246754123,
      "loss": 2.6678,
      "step": 116
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00028878803438398015,
      "loss": 3.0527,
      "step": 117
    },
    {
      "epoch": 0.94,
      "learning_rate": 0.00028557870956832135,
      "loss": 3.4959,
      "step": 118
    },
    {
      "epoch": 0.95,
      "learning_rate": 0.00028236338887025885,
      "loss": 21.3875,
      "step": 119
    },
    {
      "epoch": 0.96,
      "learning_rate": 0.0002791426141499398,
      "loss": 3.4579,
      "step": 120
    },
    {
      "epoch": 0.97,
      "learning_rate": 0.0002759169281866472,
      "loss": 3.7853,
      "step": 121
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.0002726868745873286,
      "loss": 3.8194,
      "step": 122
    },
    {
      "epoch": 0.98,
      "learning_rate": 0.00026945299769498494,
      "loss": 3.8317,
      "step": 123
    },
    {
      "epoch": 0.99,
      "learning_rate": 0.00026621584249693575,
      "loss": 3.7197,
      "step": 124
    },
    {
      "epoch": 1.0,
      "learning_rate": 0.00026297595453297494,
      "loss": 3.6754,
      "step": 125
    },
    {
      "epoch": 1.01,
      "learning_rate": 0.0002597338798034344,
      "loss": 2.6789,
      "step": 126
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.0002564901646771696,
      "loss": 2.7223,
      "step": 127
    },
    {
      "epoch": 1.02,
      "learning_rate": 0.00025324535579948274,
      "loss": 3.8515,
      "step": 128
    },
    {
      "epoch": 1.03,
      "learning_rate": 0.00025,
      "loss": 4.6488,
      "step": 129
    },
    {
      "epoch": 1.04,
      "learning_rate": 0.00024675464420051733,
      "loss": 40.6186,
      "step": 130
    },
    {
      "epoch": 1.05,
      "learning_rate": 0.00024350983532283048,
      "loss": 2.1917,
      "step": 131
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002402661201965656,
      "loss": 11.4722,
      "step": 132
    },
    {
      "epoch": 1.06,
      "learning_rate": 0.0002370240454670251,
      "loss": 3.3231,
      "step": 133
    },
    {
      "epoch": 1.07,
      "learning_rate": 0.00023378415750306423,
      "loss": 11.9216,
      "step": 134
    },
    {
      "epoch": 1.08,
      "learning_rate": 0.000230547002305015,
      "loss": 2.9565,
      "step": 135
    },
    {
      "epoch": 1.09,
      "learning_rate": 0.00022731312541267144,
      "loss": 2.6065,
      "step": 136
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00022408307181335285,
      "loss": 3.0483,
      "step": 137
    },
    {
      "epoch": 1.1,
      "learning_rate": 0.00022085738585006024,
      "loss": 3.5948,
      "step": 138
    },
    {
      "epoch": 1.11,
      "learning_rate": 0.00021763661112974113,
      "loss": 2.7134,
      "step": 139
    },
    {
      "epoch": 1.12,
      "learning_rate": 0.00021442129043167875,
      "loss": 3.4743,
      "step": 140
    },
    {
      "epoch": 1.13,
      "learning_rate": 0.00021121196561601994,
      "loss": 3.4124,
      "step": 141
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.00020800917753245875,
      "loss": 2.7959,
      "step": 142
    },
    {
      "epoch": 1.14,
      "learning_rate": 0.0002048134659290905,
      "loss": 3.1839,
      "step": 143
    },
    {
      "epoch": 1.15,
      "learning_rate": 0.0002016253693614501,
      "loss": 4.1872,
      "step": 144
    },
    {
      "epoch": 1.16,
      "learning_rate": 0.000198445425101753,
      "loss": 2.611,
      "step": 145
    },
    {
      "epoch": 1.17,
      "learning_rate": 0.00019527416904835133,
      "loss": 3.0293,
      "step": 146
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.00019211213563542217,
      "loss": 2.8123,
      "step": 147
    },
    {
      "epoch": 1.18,
      "learning_rate": 0.0001889598577429022,
      "loss": 2.5599,
      "step": 148
    },
    {
      "epoch": 1.19,
      "learning_rate": 0.00018581786660668432,
      "loss": 3.3894,
      "step": 149
    },
    {
      "epoch": 1.2,
      "learning_rate": 0.00018268669172909137,
      "loss": 8.1448,
      "step": 150
    },
    {
      "epoch": 1.21,
      "learning_rate": 0.00017956686078964256,
      "loss": 3.2351,
      "step": 151
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00017645889955612593,
      "loss": 3.1908,
      "step": 152
    },
    {
      "epoch": 1.22,
      "learning_rate": 0.00017336333179599378,
      "loss": 3.2858,
      "step": 153
    },
    {
      "epoch": 1.23,
      "learning_rate": 0.00017028067918809536,
      "loss": 2.2132,
      "step": 154
    },
    {
      "epoch": 1.24,
      "learning_rate": 0.00016721146123476095,
      "loss": 2.6308,
      "step": 155
    },
    {
      "epoch": 1.25,
      "learning_rate": 0.00016415619517425294,
      "loss": 2.7597,
      "step": 156
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00016111539589359886,
      "loss": 35.8525,
      "step": 157
    },
    {
      "epoch": 1.26,
      "learning_rate": 0.00015808957584181996,
      "loss": 3.0954,
      "step": 158
    },
    {
      "epoch": 1.27,
      "learning_rate": 0.00015507924494357153,
      "loss": 3.1205,
      "step": 159
    },
    {
      "epoch": 1.28,
      "learning_rate": 0.00015208491051320745,
      "loss": 2.669,
      "step": 160
    },
    {
      "epoch": 1.29,
      "learning_rate": 0.00014910707716928588,
      "loss": 3.0896,
      "step": 161
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001461462467495284,
      "loss": 4.3308,
      "step": 162
    },
    {
      "epoch": 1.3,
      "learning_rate": 0.0001432029182262482,
      "loss": 2.1798,
      "step": 163
    },
    {
      "epoch": 1.31,
      "learning_rate": 0.00014027758762226108,
      "loss": 2.8421,
      "step": 164
    },
    {
      "epoch": 1.32,
      "learning_rate": 0.0001373707479272933,
      "loss": 2.0236,
      "step": 165
    },
    {
      "epoch": 1.33,
      "learning_rate": 0.00013448288901490095,
      "loss": 2.7477,
      "step": 166
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00013161449755991425,
      "loss": 2.5402,
      "step": 167
    },
    {
      "epoch": 1.34,
      "learning_rate": 0.00012876605695642086,
      "loss": 2.2462,
      "step": 168
    },
    {
      "epoch": 1.35,
      "learning_rate": 0.00012593804723630208,
      "loss": 2.118,
      "step": 169
    },
    {
      "epoch": 1.36,
      "learning_rate": 0.0001231309449883361,
      "loss": 2.9153,
      "step": 170
    },
    {
      "epoch": 1.37,
      "learning_rate": 0.0001203452232778807,
      "loss": 2.4183,
      "step": 171
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.00011758135156715042,
      "loss": 2.525,
      "step": 172
    },
    {
      "epoch": 1.38,
      "learning_rate": 0.0001148397956361007,
      "loss": 1.7479,
      "step": 173
    },
    {
      "epoch": 1.39,
      "learning_rate": 0.00011212101750393236,
      "loss": 5.24,
      "step": 174
    },
    {
      "epoch": 1.4,
      "learning_rate": 0.00010942547535123056,
      "loss": 3.7628,
      "step": 175
    },
    {
      "epoch": 1.41,
      "learning_rate": 0.00010675362344274952,
      "loss": 3.0845,
      "step": 176
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00010410591205085817,
      "loss": 3.484,
      "step": 177
    },
    {
      "epoch": 1.42,
      "learning_rate": 0.00010148278737965844,
      "loss": 2.6217,
      "step": 178
    },
    {
      "epoch": 1.43,
      "learning_rate": 9.888469148978843e-05,
      "loss": 2.7539,
      "step": 179
    },
    {
      "epoch": 1.44,
      "learning_rate": 9.63120622239248e-05,
      "loss": 3.0892,
      "step": 180
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.376533313299543e-05,
      "loss": 2.8202,
      "step": 181
    },
    {
      "epoch": 1.46,
      "learning_rate": 9.124493340311537e-05,
      "loss": 31.165,
      "step": 182
    },
    {
      "epoch": 1.46,
      "learning_rate": 8.875128778325903e-05,
      "loss": 3.4765,
      "step": 183
    },
    {
      "epoch": 1.47,
      "learning_rate": 8.628481651367875e-05,
      "loss": 3.3355,
      "step": 184
    },
    {
      "epoch": 1.48,
      "learning_rate": 8.384593525508466e-05,
      "loss": 2.2746,
      "step": 185
    },
    {
      "epoch": 1.49,
      "learning_rate": 8.143505501859552e-05,
      "loss": 2.425,
      "step": 186
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.905258209647325e-05,
      "loss": 4.0229,
      "step": 187
    },
    {
      "epoch": 1.5,
      "learning_rate": 7.669891799365283e-05,
      "loss": 3.619,
      "step": 188
    },
    {
      "epoch": 1.51,
      "learning_rate": 7.437445936007866e-05,
      "loss": 3.0008,
      "step": 189
    },
    {
      "epoch": 1.52,
      "learning_rate": 7.207959792385999e-05,
      "loss": 3.3409,
      "step": 190
    },
    {
      "epoch": 1.53,
      "learning_rate": 6.981472042525417e-05,
      "loss": 2.4765,
      "step": 191
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.75802085514925e-05,
      "loss": 3.2739,
      "step": 192
    },
    {
      "epoch": 1.54,
      "learning_rate": 6.537643887245573e-05,
      "loss": 2.5563,
      "step": 193
    },
    {
      "epoch": 1.55,
      "learning_rate": 6.320378277721342e-05,
      "loss": 2.7439,
      "step": 194
    },
    {
      "epoch": 1.56,
      "learning_rate": 6.106260641143546e-05,
      "loss": 2.0306,
      "step": 195
    },
    {
      "epoch": 1.57,
      "learning_rate": 5.895327061568775e-05,
      "loss": 1.9613,
      "step": 196
    },
    {
      "epoch": 1.58,
      "learning_rate": 5.6876130864621706e-05,
      "loss": 3.8642,
      "step": 197
    },
    {
      "epoch": 1.58,
      "learning_rate": 5.483153720706799e-05,
      "loss": 5.3053,
      "step": 198
    },
    {
      "epoch": 1.59,
      "learning_rate": 5.281983420704498e-05,
      "loss": 2.8855,
      "step": 199
    },
    {
      "epoch": 1.6,
      "learning_rate": 5.0841360885691e-05,
      "loss": 3.592,
      "step": 200
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.889645066413112e-05,
      "loss": 3.0089,
      "step": 201
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.698543130728755e-05,
      "loss": 2.8367,
      "step": 202
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.510862486864317e-05,
      "loss": 2.4621,
      "step": 203
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3266347635967846e-05,
      "loss": 3.6,
      "step": 204
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.1458910078015886e-05,
      "loss": 3.218,
      "step": 205
    },
    {
      "epoch": 1.65,
      "learning_rate": 3.968661679220467e-05,
      "loss": 3.1191,
      "step": 206
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.7949766453282656e-05,
      "loss": 3.4187,
      "step": 207
    },
    {
      "epoch": 1.66,
      "learning_rate": 3.6248651762995e-05,
      "loss": 3.1861,
      "step": 208
    },
    {
      "epoch": 1.67,
      "learning_rate": 3.458355940075653e-05,
      "loss": 27.1506,
      "step": 209
    },
    {
      "epoch": 1.68,
      "learning_rate": 3.295476997533905e-05,
      "loss": 3.8461,
      "step": 210
    },
    {
      "epoch": 1.69,
      "learning_rate": 3.1362557977582e-05,
      "loss": 2.7402,
      "step": 211
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.980719173413396e-05,
      "loss": 3.333,
      "step": 212
    },
    {
      "epoch": 1.7,
      "learning_rate": 2.828893336223315e-05,
      "loss": 2.0692,
      "step": 213
    },
    {
      "epoch": 1.71,
      "learning_rate": 2.6808038725534083e-05,
      "loss": 3.3265,
      "step": 214
    },
    {
      "epoch": 1.72,
      "learning_rate": 2.536475739098862e-05,
      "loss": 6.6827,
      "step": 215
    },
    {
      "epoch": 1.73,
      "learning_rate": 2.395933258678745e-05,
      "loss": 2.6437,
      "step": 216
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.259200116137039e-05,
      "loss": 3.0813,
      "step": 217
    },
    {
      "epoch": 1.74,
      "learning_rate": 2.1262993543511715e-05,
      "loss": 2.4426,
      "step": 218
    },
    {
      "epoch": 1.75,
      "learning_rate": 1.9972533703487035e-05,
      "loss": 3.2463,
      "step": 219
    },
    {
      "epoch": 1.76,
      "learning_rate": 1.872083911532907e-05,
      "loss": 3.2592,
      "step": 220
    },
    {
      "epoch": 1.77,
      "learning_rate": 1.7508120720177794e-05,
      "loss": 2.7844,
      "step": 221
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.6334582890731697e-05,
      "loss": 1.9245,
      "step": 222
    },
    {
      "epoch": 1.78,
      "learning_rate": 1.5200423396806273e-05,
      "loss": 3.0307,
      "step": 223
    },
    {
      "epoch": 1.79,
      "learning_rate": 1.4105833372004522e-05,
      "loss": 2.2147,
      "step": 224
    },
    {
      "epoch": 1.8,
      "learning_rate": 1.305099728150655e-05,
      "loss": 3.1483,
      "step": 225
    },
    {
      "epoch": 1.81,
      "learning_rate": 1.203609289098262e-05,
      "loss": 7.4484,
      "step": 226
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.1061291236635295e-05,
      "loss": 3.8132,
      "step": 227
    },
    {
      "epoch": 1.82,
      "learning_rate": 1.0126756596375685e-05,
      "loss": 2.7503,
      "step": 228
    },
    {
      "epoch": 1.83,
      "learning_rate": 9.232646462138423e-06,
      "loss": 3.3353,
      "step": 229
    },
    {
      "epoch": 1.84,
      "learning_rate": 8.379111513340754e-06,
      "loss": 6.692,
      "step": 230
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.566295591489053e-06,
      "loss": 2.8828,
      "step": 231
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.794335675938179e-06,
      "loss": 2.8349,
      "step": 232
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.063361860807026e-06,
      "loss": 3.9368,
      "step": 233
    },
    {
      "epoch": 1.87,
      "learning_rate": 5.3734973330546165e-06,
      "loss": 2.1829,
      "step": 234
    },
    {
      "epoch": 1.88,
      "learning_rate": 4.7248583517199014e-06,
      "loss": 2.3841,
      "step": 235
    },
    {
      "epoch": 1.89,
      "learning_rate": 4.117554228329407e-06,
      "loss": 2.8279,
      "step": 236
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.551687308475521e-06,
      "loss": 3.5517,
      "step": 237
    },
    {
      "epoch": 1.9,
      "learning_rate": 3.027352954568713e-06,
      "loss": 2.1687,
      "step": 238
    },
    {
      "epoch": 1.91,
      "learning_rate": 2.544639529766829e-06,
      "loss": 3.3842,
      "step": 239
    },
    {
      "epoch": 1.92,
      "learning_rate": 2.1036283830834224e-06,
      "loss": 2.2917,
      "step": 240
    },
    {
      "epoch": 1.93,
      "learning_rate": 1.704393835678747e-06,
      "loss": 1.9778,
      "step": 241
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.347003168334665e-06,
      "loss": 2.9791,
      "step": 242
    },
    {
      "epoch": 1.94,
      "learning_rate": 1.0315166101162454e-06,
      "loss": 3.732,
      "step": 243
    },
    {
      "epoch": 1.95,
      "learning_rate": 7.579873282216598e-07,
      "loss": 6.2757,
      "step": 244
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.264614190222949e-07,
      "loss": 4.7044,
      "step": 245
    },
    {
      "epoch": 1.97,
      "learning_rate": 3.3697790029424415e-07,
      "loss": 2.3531,
      "step": 246
    },
    {
      "epoch": 1.98,
      "learning_rate": 1.8956870464298393e-07,
      "loss": 2.4217,
      "step": 247
    },
    {
      "epoch": 1.98,
      "learning_rate": 8.425867412190091e-08,
      "loss": 2.2583,
      "step": 248
    },
    {
      "epoch": 1.99,
      "learning_rate": 2.1065556045779532e-08,
      "loss": 3.9272,
      "step": 249
    },
    {
      "epoch": 2.0,
      "learning_rate": 0.0,
      "loss": 2.4315,
      "step": 250
    },
    {
      "epoch": 2.0,
      "step": 250,
      "total_flos": 4091138433294336.0,
      "train_loss": 7.845541751384735,
      "train_runtime": 5854.3349,
      "train_samples_per_second": 0.171,
      "train_steps_per_second": 0.043
    }
  ],
  "max_steps": 250,
  "num_train_epochs": 2,
  "total_flos": 4091138433294336.0,
  "trial_name": null,
  "trial_params": null
}