{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.8158640226628895,
  "global_step": 300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 4e-05,
      "loss": 1.9088,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 8e-05,
      "loss": 1.9287,
      "step": 4
    },
    {
      "epoch": 0.04,
      "learning_rate": 9.99989723479183e-05,
      "loss": 1.7736,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 9.999075138471951e-05,
      "loss": 1.9095,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 9.99743108100344e-05,
      "loss": 1.918,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.994965332706573e-05,
      "loss": 1.8706,
      "step": 12
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.991678299006205e-05,
      "loss": 1.9351,
      "step": 14
    },
    {
      "epoch": 0.1,
      "learning_rate": 9.987570520365104e-05,
      "loss": 1.8793,
      "step": 16
    },
    {
      "epoch": 0.11,
      "learning_rate": 9.982642672195092e-05,
      "loss": 1.8827,
      "step": 18
    },
    {
      "epoch": 0.12,
      "learning_rate": 9.976895564745991e-05,
      "loss": 1.8303,
      "step": 20
    },
    {
      "epoch": 0.13,
      "learning_rate": 9.970330142972401e-05,
      "loss": 1.8379,
      "step": 22
    },
    {
      "epoch": 0.15,
      "learning_rate": 9.962947486378326e-05,
      "loss": 1.8987,
      "step": 24
    },
    {
      "epoch": 0.16,
      "learning_rate": 9.954748808839674e-05,
      "loss": 1.9518,
      "step": 26
    },
    {
      "epoch": 0.17,
      "learning_rate": 9.945735458404681e-05,
      "loss": 1.8511,
      "step": 28
    },
    {
      "epoch": 0.18,
      "learning_rate": 9.935908917072252e-05,
      "loss": 1.9064,
      "step": 30
    },
    {
      "epoch": 0.19,
      "learning_rate": 9.925270800548285e-05,
      "loss": 1.8882,
      "step": 32
    },
    {
      "epoch": 0.21,
      "learning_rate": 9.91382285798002e-05,
      "loss": 1.9368,
      "step": 34
    },
    {
      "epoch": 0.22,
      "learning_rate": 9.901566971668437e-05,
      "loss": 1.8747,
      "step": 36
    },
    {
      "epoch": 0.23,
      "learning_rate": 9.888505156758759e-05,
      "loss": 1.9426,
      "step": 38
    },
    {
      "epoch": 0.24,
      "learning_rate": 9.874639560909117e-05,
      "loss": 1.9316,
      "step": 40
    },
    {
      "epoch": 0.25,
      "learning_rate": 9.859972463937441e-05,
      "loss": 1.867,
      "step": 42
    },
    {
      "epoch": 0.27,
      "learning_rate": 9.844506277446577e-05,
      "loss": 1.8927,
      "step": 44
    },
    {
      "epoch": 0.28,
      "learning_rate": 9.828243544427796e-05,
      "loss": 1.9449,
      "step": 46
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.811186938842645e-05,
      "loss": 1.8571,
      "step": 48
    },
    {
      "epoch": 0.3,
      "learning_rate": 9.793339265183303e-05,
      "loss": 1.8346,
      "step": 50
    },
    {
      "epoch": 0.31,
      "learning_rate": 9.774703458011453e-05,
      "loss": 1.9036,
      "step": 52
    },
    {
      "epoch": 0.33,
      "learning_rate": 9.755282581475769e-05,
      "loss": 1.8838,
      "step": 54
    },
    {
      "epoch": 0.34,
      "learning_rate": 9.735079828808107e-05,
      "loss": 1.959,
      "step": 56
    },
    {
      "epoch": 0.35,
      "learning_rate": 9.714098521798465e-05,
      "loss": 1.9376,
      "step": 58
    },
    {
      "epoch": 0.36,
      "learning_rate": 9.692342110248802e-05,
      "loss": 1.9326,
      "step": 60
    },
    {
      "epoch": 0.37,
      "learning_rate": 9.669814171405816e-05,
      "loss": 1.9344,
      "step": 62
    },
    {
      "epoch": 0.39,
      "learning_rate": 9.64651840937276e-05,
      "loss": 1.9038,
      "step": 64
    },
    {
      "epoch": 0.4,
      "learning_rate": 9.622458654500409e-05,
      "loss": 1.9294,
      "step": 66
    },
    {
      "epoch": 0.41,
      "learning_rate": 9.597638862757255e-05,
      "loss": 1.847,
      "step": 68
    },
    {
      "epoch": 0.42,
      "learning_rate": 9.572063115079063e-05,
      "loss": 1.9167,
      "step": 70
    },
    {
      "epoch": 0.44,
      "learning_rate": 9.545735616697875e-05,
      "loss": 1.8309,
      "step": 72
    },
    {
      "epoch": 0.45,
      "learning_rate": 9.518660696450568e-05,
      "loss": 1.8982,
      "step": 74
    },
    {
      "epoch": 0.46,
      "learning_rate": 9.490842806067095e-05,
      "loss": 1.8734,
      "step": 76
    },
    {
      "epoch": 0.47,
      "learning_rate": 9.46228651943853e-05,
      "loss": 1.8738,
      "step": 78
    },
    {
      "epoch": 0.48,
      "learning_rate": 9.432996531865002e-05,
      "loss": 1.9731,
      "step": 80
    },
    {
      "epoch": 0.5,
      "learning_rate": 9.40297765928369e-05,
      "loss": 1.9105,
      "step": 82
    },
    {
      "epoch": 0.51,
      "learning_rate": 9.372234837476978e-05,
      "loss": 1.8629,
      "step": 84
    },
    {
      "epoch": 0.52,
      "learning_rate": 9.340773121260893e-05,
      "loss": 1.8565,
      "step": 86
    },
    {
      "epoch": 0.53,
      "learning_rate": 9.308597683653975e-05,
      "loss": 1.9962,
      "step": 88
    },
    {
      "epoch": 0.54,
      "learning_rate": 9.275713815026731e-05,
      "loss": 1.8981,
      "step": 90
    },
    {
      "epoch": 0.56,
      "learning_rate": 9.242126922231763e-05,
      "loss": 1.9132,
      "step": 92
    },
    {
      "epoch": 0.57,
      "learning_rate": 9.207842527714767e-05,
      "loss": 1.9159,
      "step": 94
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.172866268606513e-05,
      "loss": 1.9617,
      "step": 96
    },
    {
      "epoch": 0.59,
      "learning_rate": 9.137203895795983e-05,
      "loss": 1.8794,
      "step": 98
    },
    {
      "epoch": 0.6,
      "learning_rate": 9.10086127298478e-05,
      "loss": 2.0122,
      "step": 100
    },
    {
      "epoch": 0.62,
      "learning_rate": 9.063844375723014e-05,
      "loss": 1.8813,
      "step": 102
    },
    {
      "epoch": 0.63,
      "learning_rate": 9.02615929042678e-05,
      "loss": 1.9717,
      "step": 104
    },
    {
      "epoch": 0.64,
      "learning_rate": 8.987812213377424e-05,
      "loss": 1.9022,
      "step": 106
    },
    {
      "epoch": 0.65,
      "learning_rate": 8.948809449702711e-05,
      "loss": 1.9205,
      "step": 108
    },
    {
      "epoch": 0.66,
      "learning_rate": 8.90915741234015e-05,
      "loss": 1.8173,
      "step": 110
    },
    {
      "epoch": 0.68,
      "learning_rate": 8.868862620982534e-05,
      "loss": 2.0031,
      "step": 112
    },
    {
      "epoch": 0.69,
      "learning_rate": 8.827931701005974e-05,
      "loss": 1.9448,
      "step": 114
    },
    {
      "epoch": 0.7,
      "learning_rate": 8.786371382380528e-05,
      "loss": 1.9093,
      "step": 116
    },
    {
      "epoch": 0.71,
      "learning_rate": 8.744188498563641e-05,
      "loss": 1.9093,
      "step": 118
    },
    {
      "epoch": 0.73,
      "learning_rate": 8.701389985376578e-05,
      "loss": 1.8793,
      "step": 120
    },
    {
      "epoch": 0.74,
      "learning_rate": 8.657982879864007e-05,
      "loss": 1.9477,
      "step": 122
    },
    {
      "epoch": 0.75,
      "learning_rate": 8.613974319136958e-05,
      "loss": 1.8808,
      "step": 124
    },
    {
      "epoch": 0.76,
      "learning_rate": 8.569371539199316e-05,
      "loss": 1.9425,
      "step": 126
    },
    {
      "epoch": 0.77,
      "learning_rate": 8.524181873758059e-05,
      "loss": 1.8945,
      "step": 128
    },
    {
      "epoch": 0.79,
      "learning_rate": 8.478412753017433e-05,
      "loss": 1.9579,
      "step": 130
    },
    {
      "epoch": 0.8,
      "learning_rate": 8.432071702457252e-05,
      "loss": 1.9224,
      "step": 132
    },
    {
      "epoch": 0.81,
      "learning_rate": 8.385166341595548e-05,
      "loss": 1.9117,
      "step": 134
    },
    {
      "epoch": 0.82,
      "learning_rate": 8.33770438273574e-05,
      "loss": 1.9455,
      "step": 136
    },
    {
      "epoch": 0.83,
      "learning_rate": 8.289693629698564e-05,
      "loss": 1.8856,
      "step": 138
    },
    {
      "epoch": 0.85,
      "learning_rate": 8.241141976538943e-05,
      "loss": 1.915,
      "step": 140
    },
    {
      "epoch": 0.86,
      "learning_rate": 8.192057406248028e-05,
      "loss": 1.9302,
      "step": 142
    },
    {
      "epoch": 0.87,
      "learning_rate": 8.142447989440618e-05,
      "loss": 1.8647,
      "step": 144
    },
    {
      "epoch": 0.88,
      "learning_rate": 8.092321883028158e-05,
      "loss": 1.8653,
      "step": 146
    },
    {
      "epoch": 0.89,
      "learning_rate": 8.041687328877567e-05,
      "loss": 1.9183,
      "step": 148
    },
    {
      "epoch": 0.91,
      "learning_rate": 7.990552652456081e-05,
      "loss": 1.852,
      "step": 150
    },
    {
      "epoch": 0.92,
      "learning_rate": 7.938926261462366e-05,
      "loss": 1.9445,
      "step": 152
    },
    {
      "epoch": 0.93,
      "learning_rate": 7.886816644444098e-05,
      "loss": 1.8727,
      "step": 154
    },
    {
      "epoch": 0.94,
      "learning_rate": 7.83423236940225e-05,
      "loss": 1.9182,
      "step": 156
    },
    {
      "epoch": 0.95,
      "learning_rate": 7.781182082382325e-05,
      "loss": 1.9708,
      "step": 158
    },
    {
      "epoch": 0.97,
      "learning_rate": 7.727674506052743e-05,
      "loss": 1.8826,
      "step": 160
    },
    {
      "epoch": 0.98,
      "learning_rate": 7.673718438270648e-05,
      "loss": 1.9755,
      "step": 162
    },
    {
      "epoch": 0.99,
      "learning_rate": 7.619322750635327e-05,
      "loss": 1.947,
      "step": 164
    },
    {
      "epoch": 1.01,
      "learning_rate": 7.564496387029532e-05,
      "loss": 2.3347,
      "step": 166
    },
    {
      "epoch": 1.02,
      "learning_rate": 7.509248362148889e-05,
      "loss": 1.9177,
      "step": 168
    },
    {
      "epoch": 1.03,
      "learning_rate": 7.45358776001969e-05,
      "loss": 1.8889,
      "step": 170
    },
    {
      "epoch": 1.04,
      "learning_rate": 7.39752373250527e-05,
      "loss": 1.8581,
      "step": 172
    },
    {
      "epoch": 1.05,
      "learning_rate": 7.34106549780123e-05,
      "loss": 1.8564,
      "step": 174
    },
    {
      "epoch": 1.07,
      "learning_rate": 7.284222338919758e-05,
      "loss": 1.8354,
      "step": 176
    },
    {
      "epoch": 1.08,
      "learning_rate": 7.227003602163295e-05,
      "loss": 1.89,
      "step": 178
    },
    {
      "epoch": 1.09,
      "learning_rate": 7.169418695587791e-05,
      "loss": 1.9877,
      "step": 180
    },
    {
      "epoch": 1.1,
      "learning_rate": 7.1114770874558e-05,
      "loss": 1.7948,
      "step": 182
    },
    {
      "epoch": 1.11,
      "learning_rate": 7.05318830467969e-05,
      "loss": 1.9217,
      "step": 184
    },
    {
      "epoch": 1.13,
      "learning_rate": 6.99456193125521e-05,
      "loss": 1.7795,
      "step": 186
    },
    {
      "epoch": 1.14,
      "learning_rate": 6.935607606685642e-05,
      "loss": 1.9189,
      "step": 188
    },
    {
      "epoch": 1.15,
      "learning_rate": 6.876335024396872e-05,
      "loss": 1.8445,
      "step": 190
    },
    {
      "epoch": 1.16,
      "learning_rate": 6.816753930143558e-05,
      "loss": 1.852,
      "step": 192
    },
    {
      "epoch": 1.18,
      "learning_rate": 6.756874120406714e-05,
      "loss": 1.8622,
      "step": 194
    },
    {
      "epoch": 1.19,
      "learning_rate": 6.696705440782938e-05,
      "loss": 1.9092,
      "step": 196
    },
    {
      "epoch": 1.2,
      "learning_rate": 6.636257784365584e-05,
      "loss": 1.8074,
      "step": 198
    },
    {
      "epoch": 1.21,
      "learning_rate": 6.575541090118105e-05,
      "loss": 1.8295,
      "step": 200
    },
    {
      "epoch": 1.22,
      "learning_rate": 6.514565341239861e-05,
      "loss": 1.9595,
      "step": 202
    },
    {
      "epoch": 1.24,
      "learning_rate": 6.453340563524669e-05,
      "loss": 1.8733,
      "step": 204
    },
    {
      "epoch": 1.25,
      "learning_rate": 6.391876823712317e-05,
      "loss": 1.8277,
      "step": 206
    },
    {
      "epoch": 1.26,
      "learning_rate": 6.330184227833376e-05,
      "loss": 1.914,
      "step": 208
    },
    {
      "epoch": 1.27,
      "learning_rate": 6.268272919547537e-05,
      "loss": 1.8684,
      "step": 210
    },
    {
      "epoch": 1.28,
      "learning_rate": 6.206153078475763e-05,
      "loss": 1.9092,
      "step": 212
    },
    {
      "epoch": 1.3,
      "learning_rate": 6.143834918526527e-05,
      "loss": 1.7996,
      "step": 214
    },
    {
      "epoch": 1.31,
      "learning_rate": 6.081328686216418e-05,
      "loss": 1.8958,
      "step": 216
    },
    {
      "epoch": 1.32,
      "learning_rate": 6.0186446589853784e-05,
      "loss": 1.9597,
      "step": 218
    },
    {
      "epoch": 1.33,
      "learning_rate": 5.955793143506863e-05,
      "loss": 1.9054,
      "step": 220
    },
    {
      "epoch": 1.34,
      "learning_rate": 5.8927844739931834e-05,
      "loss": 1.8064,
      "step": 222
    },
    {
      "epoch": 1.36,
      "learning_rate": 5.82962901049634e-05,
      "loss": 1.9105,
      "step": 224
    },
    {
      "epoch": 1.37,
      "learning_rate": 5.766337137204579e-05,
      "loss": 1.8535,
      "step": 226
    },
    {
      "epoch": 1.38,
      "learning_rate": 5.7029192607350146e-05,
      "loss": 1.8961,
      "step": 228
    },
    {
      "epoch": 1.39,
      "learning_rate": 5.6393858084225305e-05,
      "loss": 1.8679,
      "step": 230
    },
    {
      "epoch": 1.4,
      "learning_rate": 5.575747226605298e-05,
      "loss": 1.8808,
      "step": 232
    },
    {
      "epoch": 1.42,
      "learning_rate": 5.512013978907157e-05,
      "loss": 1.8165,
      "step": 234
    },
    {
      "epoch": 1.43,
      "learning_rate": 5.448196544517168e-05,
      "loss": 1.849,
      "step": 236
    },
    {
      "epoch": 1.44,
      "learning_rate": 5.384305416466584e-05,
      "loss": 1.8585,
      "step": 238
    },
    {
      "epoch": 1.45,
      "learning_rate": 5.320351099903565e-05,
      "loss": 1.8477,
      "step": 240
    },
    {
      "epoch": 1.47,
      "learning_rate": 5.256344110365896e-05,
      "loss": 1.84,
      "step": 242
    },
    {
      "epoch": 1.48,
      "learning_rate": 5.192294972051992e-05,
      "loss": 1.847,
      "step": 244
    },
    {
      "epoch": 1.49,
      "learning_rate": 5.128214216090478e-05,
      "loss": 1.9149,
      "step": 246
    },
    {
      "epoch": 1.5,
      "learning_rate": 5.064112378808637e-05,
      "loss": 1.8937,
      "step": 248
    },
    {
      "epoch": 1.51,
      "learning_rate": 5e-05,
      "loss": 1.8805,
      "step": 250
    },
    {
      "epoch": 1.53,
      "learning_rate": 4.935887621191364e-05,
      "loss": 1.862,
      "step": 252
    },
    {
      "epoch": 1.54,
      "learning_rate": 4.871785783909523e-05,
      "loss": 1.8958,
      "step": 254
    },
    {
      "epoch": 1.55,
      "learning_rate": 4.807705027948008e-05,
      "loss": 1.9492,
      "step": 256
    },
    {
      "epoch": 1.56,
      "learning_rate": 4.743655889634105e-05,
      "loss": 1.918,
      "step": 258
    },
    {
      "epoch": 1.57,
      "learning_rate": 4.679648900096436e-05,
      "loss": 1.9169,
      "step": 260
    },
    {
      "epoch": 1.59,
      "learning_rate": 4.6156945835334184e-05,
      "loss": 1.8927,
      "step": 262
    },
    {
      "epoch": 1.6,
      "learning_rate": 4.551803455482833e-05,
      "loss": 1.9368,
      "step": 264
    },
    {
      "epoch": 1.61,
      "learning_rate": 4.487986021092844e-05,
      "loss": 1.9573,
      "step": 266
    },
    {
      "epoch": 1.62,
      "learning_rate": 4.424252773394704e-05,
      "loss": 1.9401,
      "step": 268
    },
    {
      "epoch": 1.63,
      "learning_rate": 4.3606141915774693e-05,
      "loss": 1.8759,
      "step": 270
    },
    {
      "epoch": 1.65,
      "learning_rate": 4.297080739264987e-05,
      "loss": 1.9353,
      "step": 272
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.23366286279542e-05,
      "loss": 1.8375,
      "step": 274
    },
    {
      "epoch": 1.67,
      "learning_rate": 4.1703709895036625e-05,
      "loss": 1.9202,
      "step": 276
    },
    {
      "epoch": 1.68,
      "learning_rate": 4.107215526006817e-05,
      "loss": 1.9248,
      "step": 278
    },
    {
      "epoch": 1.69,
      "learning_rate": 4.04420685649314e-05,
      "loss": 1.7924,
      "step": 280
    },
    {
      "epoch": 1.71,
      "learning_rate": 3.981355341014623e-05,
      "loss": 1.7682,
      "step": 282
    },
    {
      "epoch": 1.72,
      "learning_rate": 3.9186713137835826e-05,
      "loss": 1.8612,
      "step": 284
    },
    {
      "epoch": 1.73,
      "learning_rate": 3.856165081473474e-05,
      "loss": 1.8759,
      "step": 286
    },
    {
      "epoch": 1.74,
      "learning_rate": 3.793846921524237e-05,
      "loss": 1.875,
      "step": 288
    },
    {
      "epoch": 1.76,
      "learning_rate": 3.731727080452464e-05,
      "loss": 1.8429,
      "step": 290
    },
    {
      "epoch": 1.77,
      "learning_rate": 3.6698157721666246e-05,
      "loss": 1.907,
      "step": 292
    },
    {
      "epoch": 1.78,
      "learning_rate": 3.608123176287685e-05,
      "loss": 1.8561,
      "step": 294
    },
    {
      "epoch": 1.79,
      "learning_rate": 3.5466594364753326e-05,
      "loss": 1.9138,
      "step": 296
    },
    {
      "epoch": 1.8,
      "learning_rate": 3.48543465876014e-05,
      "loss": 1.893,
      "step": 298
    },
    {
      "epoch": 1.82,
      "learning_rate": 3.424458909881897e-05,
      "loss": 1.8505,
      "step": 300
    }
  ],
  "max_steps": 495,
  "num_train_epochs": 3,
  "total_flos": 1.4182542633473147e+18,
  "trial_name": null,
  "trial_params": null
}