{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.7166085946573753,
  "eval_steps": 500,
  "global_step": 600,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "learning_rate": 6.020599913279623e-06,
      "loss": 0.8986,
      "step": 2
    },
    {
      "epoch": 0.02,
      "learning_rate": 1.2041199826559246e-05,
      "loss": 0.7523,
      "step": 4
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.5563025007672873e-05,
      "loss": 0.6042,
      "step": 6
    },
    {
      "epoch": 0.05,
      "learning_rate": 1.806179973983887e-05,
      "loss": 0.511,
      "step": 8
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.9999999999999998e-05,
      "loss": 0.457,
      "step": 10
    },
    {
      "epoch": 0.07,
      "learning_rate": 2e-05,
      "loss": 0.4316,
      "step": 12
    },
    {
      "epoch": 0.09,
      "learning_rate": 2e-05,
      "loss": 0.4123,
      "step": 14
    },
    {
      "epoch": 0.1,
      "learning_rate": 2e-05,
      "loss": 0.3952,
      "step": 16
    },
    {
      "epoch": 0.11,
      "learning_rate": 2e-05,
      "loss": 0.3816,
      "step": 18
    },
    {
      "epoch": 0.12,
      "learning_rate": 2e-05,
      "loss": 0.3786,
      "step": 20
    },
    {
      "epoch": 0.14,
      "learning_rate": 2e-05,
      "loss": 0.3675,
      "step": 22
    },
    {
      "epoch": 0.15,
      "learning_rate": 2e-05,
      "loss": 0.3584,
      "step": 24
    },
    {
      "epoch": 0.16,
      "learning_rate": 2e-05,
      "loss": 0.3562,
      "step": 26
    },
    {
      "epoch": 0.17,
      "learning_rate": 2e-05,
      "loss": 0.3543,
      "step": 28
    },
    {
      "epoch": 0.19,
      "learning_rate": 2e-05,
      "loss": 0.3517,
      "step": 30
    },
    {
      "epoch": 0.2,
      "learning_rate": 2e-05,
      "loss": 0.3405,
      "step": 32
    },
    {
      "epoch": 0.21,
      "learning_rate": 2e-05,
      "loss": 0.337,
      "step": 34
    },
    {
      "epoch": 0.22,
      "learning_rate": 2e-05,
      "loss": 0.3362,
      "step": 36
    },
    {
      "epoch": 0.24,
      "learning_rate": 2e-05,
      "loss": 0.3315,
      "step": 38
    },
    {
      "epoch": 0.25,
      "learning_rate": 2e-05,
      "loss": 0.3255,
      "step": 40
    },
    {
      "epoch": 0.26,
      "learning_rate": 2e-05,
      "loss": 0.3223,
      "step": 42
    },
    {
      "epoch": 0.27,
      "learning_rate": 2e-05,
      "loss": 0.3184,
      "step": 44
    },
    {
      "epoch": 0.28,
      "learning_rate": 2e-05,
      "loss": 0.3211,
      "step": 46
    },
    {
      "epoch": 0.3,
      "learning_rate": 2e-05,
      "loss": 0.3139,
      "step": 48
    },
    {
      "epoch": 0.31,
      "learning_rate": 2e-05,
      "loss": 0.3112,
      "step": 50
    },
    {
      "epoch": 0.32,
      "learning_rate": 2e-05,
      "loss": 0.3143,
      "step": 52
    },
    {
      "epoch": 0.33,
      "learning_rate": 2e-05,
      "loss": 0.3086,
      "step": 54
    },
    {
      "epoch": 0.35,
      "learning_rate": 2e-05,
      "loss": 0.3025,
      "step": 56
    },
    {
      "epoch": 0.36,
      "learning_rate": 2e-05,
      "loss": 0.3055,
      "step": 58
    },
    {
      "epoch": 0.37,
      "learning_rate": 2e-05,
      "loss": 0.3046,
      "step": 60
    },
    {
      "epoch": 0.38,
      "learning_rate": 2e-05,
      "loss": 0.2948,
      "step": 62
    },
    {
      "epoch": 0.4,
      "learning_rate": 2e-05,
      "loss": 0.2951,
      "step": 64
    },
    {
      "epoch": 0.41,
      "learning_rate": 2e-05,
      "loss": 0.2957,
      "step": 66
    },
    {
      "epoch": 0.42,
      "learning_rate": 2e-05,
      "loss": 0.2899,
      "step": 68
    },
    {
      "epoch": 0.43,
      "learning_rate": 2e-05,
      "loss": 0.2927,
      "step": 70
    },
    {
      "epoch": 0.45,
      "learning_rate": 2e-05,
      "loss": 0.2863,
      "step": 72
    },
    {
      "epoch": 0.46,
      "learning_rate": 2e-05,
      "loss": 0.2812,
      "step": 74
    },
    {
      "epoch": 0.47,
      "learning_rate": 2e-05,
      "loss": 0.2818,
      "step": 76
    },
    {
      "epoch": 0.48,
      "learning_rate": 2e-05,
      "loss": 0.2795,
      "step": 78
    },
    {
      "epoch": 0.5,
      "learning_rate": 2e-05,
      "loss": 0.2782,
      "step": 80
    },
    {
      "epoch": 0.51,
      "learning_rate": 2e-05,
      "loss": 0.2773,
      "step": 82
    },
    {
      "epoch": 0.52,
      "learning_rate": 2e-05,
      "loss": 0.2717,
      "step": 84
    },
    {
      "epoch": 0.53,
      "learning_rate": 2e-05,
      "loss": 0.2716,
      "step": 86
    },
    {
      "epoch": 0.55,
      "learning_rate": 2e-05,
      "loss": 0.2699,
      "step": 88
    },
    {
      "epoch": 0.56,
      "learning_rate": 2e-05,
      "loss": 0.2692,
      "step": 90
    },
    {
      "epoch": 0.57,
      "learning_rate": 2e-05,
      "loss": 0.2683,
      "step": 92
    },
    {
      "epoch": 0.58,
      "learning_rate": 2e-05,
      "loss": 0.266,
      "step": 94
    },
    {
      "epoch": 0.59,
      "learning_rate": 2e-05,
      "loss": 0.2653,
      "step": 96
    },
    {
      "epoch": 0.61,
      "learning_rate": 2e-05,
      "loss": 0.2627,
      "step": 98
    },
    {
      "epoch": 0.62,
      "learning_rate": 2e-05,
      "loss": 0.2599,
      "step": 100
    },
    {
      "epoch": 0.63,
      "learning_rate": 2e-05,
      "loss": 0.262,
      "step": 102
    },
    {
      "epoch": 0.64,
      "learning_rate": 2e-05,
      "loss": 0.2601,
      "step": 104
    },
    {
      "epoch": 0.66,
      "learning_rate": 2e-05,
      "loss": 0.2582,
      "step": 106
    },
    {
      "epoch": 0.67,
      "learning_rate": 2e-05,
      "loss": 0.2565,
      "step": 108
    },
    {
      "epoch": 0.68,
      "learning_rate": 2e-05,
      "loss": 0.2528,
      "step": 110
    },
    {
      "epoch": 0.69,
      "learning_rate": 2e-05,
      "loss": 0.2513,
      "step": 112
    },
    {
      "epoch": 0.71,
      "learning_rate": 2e-05,
      "loss": 0.2506,
      "step": 114
    },
    {
      "epoch": 0.72,
      "learning_rate": 2e-05,
      "loss": 0.2501,
      "step": 116
    },
    {
      "epoch": 0.73,
      "learning_rate": 2e-05,
      "loss": 0.2495,
      "step": 118
    },
    {
      "epoch": 0.74,
      "learning_rate": 2e-05,
      "loss": 0.248,
      "step": 120
    },
    {
      "epoch": 0.76,
      "learning_rate": 2e-05,
      "loss": 0.2454,
      "step": 122
    },
    {
      "epoch": 0.77,
      "learning_rate": 2e-05,
      "loss": 0.2434,
      "step": 124
    },
    {
      "epoch": 0.78,
      "learning_rate": 2e-05,
      "loss": 0.2409,
      "step": 126
    },
    {
      "epoch": 0.79,
      "learning_rate": 2e-05,
      "loss": 0.2389,
      "step": 128
    },
    {
      "epoch": 0.81,
      "learning_rate": 2e-05,
      "loss": 0.2412,
      "step": 130
    },
    {
      "epoch": 0.82,
      "learning_rate": 2e-05,
      "loss": 0.238,
      "step": 132
    },
    {
      "epoch": 0.83,
      "learning_rate": 2e-05,
      "loss": 0.2405,
      "step": 134
    },
    {
      "epoch": 0.84,
      "learning_rate": 2e-05,
      "loss": 0.234,
      "step": 136
    },
    {
      "epoch": 0.85,
      "learning_rate": 2e-05,
      "loss": 0.2349,
      "step": 138
    },
    {
      "epoch": 0.87,
      "learning_rate": 2e-05,
      "loss": 0.233,
      "step": 140
    },
    {
      "epoch": 0.88,
      "learning_rate": 2e-05,
      "loss": 0.2303,
      "step": 142
    },
    {
      "epoch": 0.89,
      "learning_rate": 2e-05,
      "loss": 0.2352,
      "step": 144
    },
    {
      "epoch": 0.9,
      "learning_rate": 2e-05,
      "loss": 0.2322,
      "step": 146
    },
    {
      "epoch": 0.92,
      "learning_rate": 2e-05,
      "loss": 0.2287,
      "step": 148
    },
    {
      "epoch": 0.93,
      "learning_rate": 2e-05,
      "loss": 0.2281,
      "step": 150
    },
    {
      "epoch": 0.94,
      "learning_rate": 2e-05,
      "loss": 0.2282,
      "step": 152
    },
    {
      "epoch": 0.95,
      "learning_rate": 2e-05,
      "loss": 0.2241,
      "step": 154
    },
    {
      "epoch": 0.97,
      "learning_rate": 2e-05,
      "loss": 0.2211,
      "step": 156
    },
    {
      "epoch": 0.98,
      "learning_rate": 2e-05,
      "loss": 0.2238,
      "step": 158
    },
    {
      "epoch": 0.99,
      "learning_rate": 2e-05,
      "loss": 0.2184,
      "step": 160
    },
    {
      "epoch": 1.0,
      "learning_rate": 2e-05,
      "loss": 0.2157,
      "step": 162
    },
    {
      "epoch": 1.02,
      "learning_rate": 2e-05,
      "loss": 0.2071,
      "step": 164
    },
    {
      "epoch": 1.03,
      "learning_rate": 2e-05,
      "loss": 0.2022,
      "step": 166
    },
    {
      "epoch": 1.04,
      "learning_rate": 2e-05,
      "loss": 0.2012,
      "step": 168
    },
    {
      "epoch": 1.05,
      "learning_rate": 2e-05,
      "loss": 0.199,
      "step": 170
    },
    {
      "epoch": 1.07,
      "learning_rate": 2e-05,
      "loss": 0.2018,
      "step": 172
    },
    {
      "epoch": 1.08,
      "learning_rate": 2e-05,
      "loss": 0.2009,
      "step": 174
    },
    {
      "epoch": 1.09,
      "learning_rate": 2e-05,
      "loss": 0.2018,
      "step": 176
    },
    {
      "epoch": 1.1,
      "learning_rate": 2e-05,
      "loss": 0.1956,
      "step": 178
    },
    {
      "epoch": 1.11,
      "learning_rate": 2e-05,
      "loss": 0.2,
      "step": 180
    },
    {
      "epoch": 1.13,
      "learning_rate": 2e-05,
      "loss": 0.1955,
      "step": 182
    },
    {
      "epoch": 1.14,
      "learning_rate": 2e-05,
      "loss": 0.195,
      "step": 184
    },
    {
      "epoch": 1.15,
      "learning_rate": 2e-05,
      "loss": 0.1951,
      "step": 186
    },
    {
      "epoch": 1.16,
      "learning_rate": 2e-05,
      "loss": 0.1931,
      "step": 188
    },
    {
      "epoch": 1.18,
      "learning_rate": 2e-05,
      "loss": 0.1925,
      "step": 190
    },
    {
      "epoch": 1.19,
      "learning_rate": 2e-05,
      "loss": 0.1942,
      "step": 192
    },
    {
      "epoch": 1.2,
      "learning_rate": 2e-05,
      "loss": 0.1906,
      "step": 194
    },
    {
      "epoch": 1.21,
      "learning_rate": 2e-05,
      "loss": 0.191,
      "step": 196
    },
    {
      "epoch": 1.23,
      "learning_rate": 2e-05,
      "loss": 0.192,
      "step": 198
    },
    {
      "epoch": 1.24,
      "learning_rate": 2e-05,
      "loss": 0.1907,
      "step": 200
    },
    {
      "epoch": 1.25,
      "learning_rate": 2e-05,
      "loss": 0.1905,
      "step": 202
    },
    {
      "epoch": 1.26,
      "learning_rate": 2e-05,
      "loss": 0.1887,
      "step": 204
    },
    {
      "epoch": 1.28,
      "learning_rate": 2e-05,
      "loss": 0.1861,
      "step": 206
    },
    {
      "epoch": 1.29,
      "learning_rate": 2e-05,
      "loss": 0.1878,
      "step": 208
    },
    {
      "epoch": 1.3,
      "learning_rate": 2e-05,
      "loss": 0.1847,
      "step": 210
    },
    {
      "epoch": 1.31,
      "learning_rate": 2e-05,
      "loss": 0.1832,
      "step": 212
    },
    {
      "epoch": 1.33,
      "learning_rate": 2e-05,
      "loss": 0.1826,
      "step": 214
    },
    {
      "epoch": 1.34,
      "learning_rate": 2e-05,
      "loss": 0.1796,
      "step": 216
    },
    {
      "epoch": 1.35,
      "learning_rate": 2e-05,
      "loss": 0.1827,
      "step": 218
    },
    {
      "epoch": 1.36,
      "learning_rate": 2e-05,
      "loss": 0.1816,
      "step": 220
    },
    {
      "epoch": 1.38,
      "learning_rate": 2e-05,
      "loss": 0.1786,
      "step": 222
    },
    {
      "epoch": 1.39,
      "learning_rate": 2e-05,
      "loss": 0.1788,
      "step": 224
    },
    {
      "epoch": 1.4,
      "learning_rate": 2e-05,
      "loss": 0.1754,
      "step": 226
    },
    {
      "epoch": 1.41,
      "learning_rate": 2e-05,
      "loss": 0.1752,
      "step": 228
    },
    {
      "epoch": 1.42,
      "learning_rate": 2e-05,
      "loss": 0.1774,
      "step": 230
    },
    {
      "epoch": 1.44,
      "learning_rate": 2e-05,
      "loss": 0.1774,
      "step": 232
    },
    {
      "epoch": 1.45,
      "learning_rate": 2e-05,
      "loss": 0.1727,
      "step": 234
    },
    {
      "epoch": 1.46,
      "learning_rate": 2e-05,
      "loss": 0.1742,
      "step": 236
    },
    {
      "epoch": 1.47,
      "learning_rate": 2e-05,
      "loss": 0.172,
      "step": 238
    },
    {
      "epoch": 1.49,
      "learning_rate": 2e-05,
      "loss": 0.1736,
      "step": 240
    },
    {
      "epoch": 1.5,
      "learning_rate": 2e-05,
      "loss": 0.1768,
      "step": 242
    },
    {
      "epoch": 1.51,
      "learning_rate": 2e-05,
      "loss": 0.1732,
      "step": 244
    },
    {
      "epoch": 1.52,
      "learning_rate": 2e-05,
      "loss": 0.1681,
      "step": 246
    },
    {
      "epoch": 1.54,
      "learning_rate": 2e-05,
      "loss": 0.1686,
      "step": 248
    },
    {
      "epoch": 1.55,
      "learning_rate": 2e-05,
      "loss": 0.167,
      "step": 250
    },
    {
      "epoch": 1.56,
      "learning_rate": 2e-05,
      "loss": 0.1684,
      "step": 252
    },
    {
      "epoch": 1.57,
      "learning_rate": 2e-05,
      "loss": 0.1697,
      "step": 254
    },
    {
      "epoch": 1.59,
      "learning_rate": 2e-05,
      "loss": 0.1669,
      "step": 256
    },
    {
      "epoch": 1.6,
      "learning_rate": 2e-05,
      "loss": 0.1661,
      "step": 258
    },
    {
      "epoch": 1.61,
      "learning_rate": 2e-05,
      "loss": 0.167,
      "step": 260
    },
    {
      "epoch": 1.62,
      "learning_rate": 2e-05,
      "loss": 0.1635,
      "step": 262
    },
    {
      "epoch": 1.64,
      "learning_rate": 2e-05,
      "loss": 0.1658,
      "step": 264
    },
    {
      "epoch": 1.65,
      "learning_rate": 2e-05,
      "loss": 0.1654,
      "step": 266
    },
    {
      "epoch": 1.66,
      "learning_rate": 2e-05,
      "loss": 0.1628,
      "step": 268
    },
    {
      "epoch": 1.67,
      "learning_rate": 2e-05,
      "loss": 0.1629,
      "step": 270
    },
    {
      "epoch": 1.68,
      "learning_rate": 2e-05,
      "loss": 0.1627,
      "step": 272
    },
    {
      "epoch": 1.7,
      "learning_rate": 2e-05,
      "loss": 0.1619,
      "step": 274
    },
    {
      "epoch": 1.71,
      "learning_rate": 2e-05,
      "loss": 0.1631,
      "step": 276
    },
    {
      "epoch": 1.72,
      "learning_rate": 2e-05,
      "loss": 0.1602,
      "step": 278
    },
    {
      "epoch": 1.73,
      "learning_rate": 2e-05,
      "loss": 0.1621,
      "step": 280
    },
    {
      "epoch": 1.75,
      "learning_rate": 2e-05,
      "loss": 0.1586,
      "step": 282
    },
    {
      "epoch": 1.76,
      "learning_rate": 2e-05,
      "loss": 0.1567,
      "step": 284
    },
    {
      "epoch": 1.77,
      "learning_rate": 2e-05,
      "loss": 0.1585,
      "step": 286
    },
    {
      "epoch": 1.78,
      "learning_rate": 2e-05,
      "loss": 0.1572,
      "step": 288
    },
    {
      "epoch": 1.8,
      "learning_rate": 2e-05,
      "loss": 0.1546,
      "step": 290
    },
    {
      "epoch": 1.81,
      "learning_rate": 2e-05,
      "loss": 0.1555,
      "step": 292
    },
    {
      "epoch": 1.82,
      "learning_rate": 2e-05,
      "loss": 0.1546,
      "step": 294
    },
    {
      "epoch": 1.83,
      "learning_rate": 2e-05,
      "loss": 0.1549,
      "step": 296
    },
    {
      "epoch": 1.85,
      "learning_rate": 2e-05,
      "loss": 0.1539,
      "step": 298
    },
    {
      "epoch": 1.86,
      "learning_rate": 2e-05,
      "loss": 0.1553,
      "step": 300
    },
    {
      "epoch": 1.87,
      "learning_rate": 2e-05,
      "loss": 0.1554,
      "step": 302
    },
    {
      "epoch": 1.88,
      "learning_rate": 2e-05,
      "loss": 0.1512,
      "step": 304
    },
    {
      "epoch": 1.9,
      "learning_rate": 2e-05,
      "loss": 0.1514,
      "step": 306
    },
    {
      "epoch": 1.91,
      "learning_rate": 2e-05,
      "loss": 0.1511,
      "step": 308
    },
    {
      "epoch": 1.92,
      "learning_rate": 2e-05,
      "loss": 0.1497,
      "step": 310
    },
    {
      "epoch": 1.93,
      "learning_rate": 2e-05,
      "loss": 0.1504,
      "step": 312
    },
    {
      "epoch": 1.95,
      "learning_rate": 2e-05,
      "loss": 0.1495,
      "step": 314
    },
    {
      "epoch": 1.96,
      "learning_rate": 2e-05,
      "loss": 0.147,
      "step": 316
    },
    {
      "epoch": 1.97,
      "learning_rate": 2e-05,
      "loss": 0.1479,
      "step": 318
    },
    {
      "epoch": 1.98,
      "learning_rate": 2e-05,
      "loss": 0.146,
      "step": 320
    },
    {
      "epoch": 1.99,
      "learning_rate": 2e-05,
      "loss": 0.1436,
      "step": 322
    },
    {
      "epoch": 2.01,
      "learning_rate": 2e-05,
      "loss": 0.1313,
      "step": 324
    },
    {
      "epoch": 2.02,
      "learning_rate": 2e-05,
      "loss": 0.1249,
      "step": 326
    },
    {
      "epoch": 2.03,
      "learning_rate": 2e-05,
      "loss": 0.1251,
      "step": 328
    },
    {
      "epoch": 2.04,
      "learning_rate": 2e-05,
      "loss": 0.1236,
      "step": 330
    },
    {
      "epoch": 2.06,
      "learning_rate": 2e-05,
      "loss": 0.1199,
      "step": 332
    },
    {
      "epoch": 2.07,
      "learning_rate": 2e-05,
      "loss": 0.1186,
      "step": 334
    },
    {
      "epoch": 2.08,
      "learning_rate": 2e-05,
      "loss": 0.1184,
      "step": 336
    },
    {
      "epoch": 2.09,
      "learning_rate": 2e-05,
      "loss": 0.12,
      "step": 338
    },
    {
      "epoch": 2.11,
      "learning_rate": 2e-05,
      "loss": 0.1163,
      "step": 340
    },
    {
      "epoch": 2.12,
      "learning_rate": 2e-05,
      "loss": 0.1191,
      "step": 342
    },
    {
      "epoch": 2.13,
      "learning_rate": 2e-05,
      "loss": 0.1181,
      "step": 344
    },
    {
      "epoch": 2.14,
      "learning_rate": 2e-05,
      "loss": 0.1171,
      "step": 346
    },
    {
      "epoch": 2.16,
      "learning_rate": 2e-05,
      "loss": 0.1141,
      "step": 348
    },
    {
      "epoch": 2.17,
      "learning_rate": 2e-05,
      "loss": 0.1146,
      "step": 350
    },
    {
      "epoch": 2.18,
      "learning_rate": 2e-05,
      "loss": 0.1178,
      "step": 352
    },
    {
      "epoch": 2.19,
      "learning_rate": 2e-05,
      "loss": 0.1134,
      "step": 354
    },
    {
      "epoch": 2.21,
      "learning_rate": 2e-05,
      "loss": 0.1164,
      "step": 356
    },
    {
      "epoch": 2.22,
      "learning_rate": 2e-05,
      "loss": 0.1164,
      "step": 358
    },
    {
      "epoch": 2.23,
      "learning_rate": 2e-05,
      "loss": 0.1163,
      "step": 360
    },
    {
      "epoch": 2.24,
      "learning_rate": 2e-05,
      "loss": 0.1149,
      "step": 362
    },
    {
      "epoch": 2.25,
      "learning_rate": 2e-05,
      "loss": 0.1132,
      "step": 364
    },
    {
      "epoch": 2.27,
      "learning_rate": 2e-05,
      "loss": 0.1118,
      "step": 366
    },
    {
      "epoch": 2.28,
      "learning_rate": 2e-05,
      "loss": 0.1117,
      "step": 368
    },
    {
      "epoch": 2.29,
      "learning_rate": 2e-05,
      "loss": 0.1106,
      "step": 370
    },
    {
      "epoch": 2.3,
      "learning_rate": 2e-05,
      "loss": 0.1133,
      "step": 372
    },
    {
      "epoch": 2.32,
      "learning_rate": 2e-05,
      "loss": 0.1108,
      "step": 374
    },
    {
      "epoch": 2.33,
      "learning_rate": 2e-05,
      "loss": 0.1084,
      "step": 376
    },
    {
      "epoch": 2.34,
      "learning_rate": 2e-05,
      "loss": 0.1086,
      "step": 378
    },
    {
      "epoch": 2.35,
      "learning_rate": 2e-05,
      "loss": 0.1071,
      "step": 380
    },
    {
      "epoch": 2.37,
      "learning_rate": 2e-05,
      "loss": 0.1104,
      "step": 382
    },
    {
      "epoch": 2.38,
      "learning_rate": 2e-05,
      "loss": 0.1096,
      "step": 384
    },
    {
      "epoch": 2.39,
      "learning_rate": 2e-05,
      "loss": 0.1088,
      "step": 386
    },
    {
      "epoch": 2.4,
      "learning_rate": 2e-05,
      "loss": 0.1091,
      "step": 388
    },
    {
      "epoch": 2.42,
      "learning_rate": 2e-05,
      "loss": 0.1056,
      "step": 390
    },
    {
      "epoch": 2.43,
      "learning_rate": 2e-05,
      "loss": 0.1077,
      "step": 392
    },
    {
      "epoch": 2.44,
      "learning_rate": 2e-05,
      "loss": 0.1104,
      "step": 394
    },
    {
      "epoch": 2.45,
      "learning_rate": 2e-05,
      "loss": 0.1084,
      "step": 396
    },
    {
      "epoch": 2.47,
      "learning_rate": 2e-05,
      "loss": 0.1069,
      "step": 398
    },
    {
      "epoch": 2.48,
      "learning_rate": 2e-05,
      "loss": 0.1036,
      "step": 400
    },
    {
      "epoch": 2.49,
      "learning_rate": 2e-05,
      "loss": 0.1086,
      "step": 402
    },
    {
      "epoch": 2.5,
      "learning_rate": 2e-05,
      "loss": 0.1068,
      "step": 404
    },
    {
      "epoch": 2.51,
      "learning_rate": 2e-05,
      "loss": 0.1059,
      "step": 406
    },
    {
      "epoch": 2.53,
      "learning_rate": 2e-05,
      "loss": 0.1054,
      "step": 408
    },
    {
      "epoch": 2.54,
      "learning_rate": 2e-05,
      "loss": 0.1055,
      "step": 410
    },
    {
      "epoch": 2.55,
      "learning_rate": 2e-05,
      "loss": 0.1045,
      "step": 412
    },
    {
      "epoch": 2.56,
      "learning_rate": 2e-05,
      "loss": 0.1047,
      "step": 414
    },
    {
      "epoch": 2.58,
      "learning_rate": 2e-05,
      "loss": 0.1009,
      "step": 416
    },
    {
      "epoch": 2.59,
      "learning_rate": 2e-05,
      "loss": 0.1028,
      "step": 418
    },
    {
      "epoch": 2.6,
      "learning_rate": 2e-05,
      "loss": 0.1021,
      "step": 420
    },
    {
      "epoch": 2.61,
      "learning_rate": 2e-05,
      "loss": 0.1006,
      "step": 422
    },
    {
      "epoch": 2.63,
      "learning_rate": 2e-05,
      "loss": 0.1001,
      "step": 424
    },
    {
      "epoch": 2.64,
      "learning_rate": 2e-05,
      "loss": 0.1005,
      "step": 426
    },
    {
      "epoch": 2.65,
      "learning_rate": 2e-05,
      "loss": 0.0993,
      "step": 428
    },
    {
      "epoch": 2.66,
      "learning_rate": 2e-05,
      "loss": 0.0998,
      "step": 430
    },
    {
      "epoch": 2.68,
      "learning_rate": 2e-05,
      "loss": 0.1039,
      "step": 432
    },
    {
      "epoch": 2.69,
      "learning_rate": 2e-05,
      "loss": 0.0995,
      "step": 434
    },
    {
      "epoch": 2.7,
      "learning_rate": 2e-05,
      "loss": 0.0998,
      "step": 436
    },
    {
      "epoch": 2.71,
      "learning_rate": 2e-05,
      "loss": 0.0999,
      "step": 438
    },
    {
      "epoch": 2.73,
      "learning_rate": 2e-05,
      "loss": 0.1023,
      "step": 440
    },
    {
      "epoch": 2.74,
      "learning_rate": 2e-05,
      "loss": 0.1003,
      "step": 442
    },
    {
      "epoch": 2.75,
      "learning_rate": 2e-05,
      "loss": 0.0985,
      "step": 444
    },
    {
      "epoch": 2.76,
      "learning_rate": 2e-05,
      "loss": 0.0973,
      "step": 446
    },
    {
      "epoch": 2.78,
      "learning_rate": 2e-05,
      "loss": 0.0974,
      "step": 448
    },
    {
      "epoch": 2.79,
      "learning_rate": 2e-05,
      "loss": 0.0976,
      "step": 450
    },
    {
      "epoch": 2.8,
      "learning_rate": 2e-05,
      "loss": 0.0951,
      "step": 452
    },
    {
      "epoch": 2.81,
      "learning_rate": 2e-05,
      "loss": 0.0964,
      "step": 454
    },
    {
      "epoch": 2.82,
      "learning_rate": 2e-05,
      "loss": 0.0957,
      "step": 456
    },
    {
      "epoch": 2.84,
      "learning_rate": 2e-05,
      "loss": 0.0938,
      "step": 458
    },
    {
      "epoch": 2.85,
      "learning_rate": 2e-05,
      "loss": 0.0933,
      "step": 460
    },
    {
      "epoch": 2.86,
      "learning_rate": 2e-05,
      "loss": 0.0935,
      "step": 462
    },
    {
      "epoch": 2.87,
      "learning_rate": 2e-05,
      "loss": 0.0961,
      "step": 464
    },
    {
      "epoch": 2.89,
      "learning_rate": 2e-05,
      "loss": 0.0941,
      "step": 466
    },
    {
      "epoch": 2.9,
      "learning_rate": 2e-05,
      "loss": 0.0944,
      "step": 468
    },
    {
      "epoch": 2.91,
      "learning_rate": 2e-05,
      "loss": 0.0957,
      "step": 470
    },
    {
      "epoch": 2.92,
      "learning_rate": 2e-05,
      "loss": 0.0934,
      "step": 472
    },
    {
      "epoch": 2.94,
      "learning_rate": 2e-05,
      "loss": 0.094,
      "step": 474
    },
    {
      "epoch": 2.95,
      "learning_rate": 2e-05,
      "loss": 0.0966,
      "step": 476
    },
    {
      "epoch": 2.96,
      "learning_rate": 2e-05,
      "loss": 0.0929,
      "step": 478
    },
    {
      "epoch": 2.97,
      "learning_rate": 2e-05,
      "loss": 0.09,
      "step": 480
    },
    {
      "epoch": 2.99,
      "learning_rate": 2e-05,
      "loss": 0.0916,
      "step": 482
    },
    {
      "epoch": 3.0,
      "learning_rate": 2e-05,
      "loss": 0.0895,
      "step": 484
    },
    {
      "epoch": 3.01,
      "learning_rate": 2e-05,
      "loss": 0.0683,
      "step": 486
    },
    {
      "epoch": 3.02,
      "learning_rate": 2e-05,
      "loss": 0.0695,
      "step": 488
    },
    {
      "epoch": 3.04,
      "learning_rate": 2e-05,
      "loss": 0.0661,
      "step": 490
    },
    {
      "epoch": 3.05,
      "learning_rate": 2e-05,
      "loss": 0.067,
      "step": 492
    },
    {
      "epoch": 3.06,
      "learning_rate": 2e-05,
      "loss": 0.0656,
      "step": 494
    },
    {
      "epoch": 3.07,
      "learning_rate": 2e-05,
      "loss": 0.0676,
      "step": 496
    },
    {
      "epoch": 3.08,
      "learning_rate": 2e-05,
      "loss": 0.0655,
      "step": 498
    },
    {
      "epoch": 3.1,
      "learning_rate": 2e-05,
      "loss": 0.0629,
      "step": 500
    },
    {
      "epoch": 3.11,
      "learning_rate": 2e-05,
      "loss": 0.0651,
      "step": 502
    },
    {
      "epoch": 3.12,
      "learning_rate": 2e-05,
      "loss": 0.0639,
      "step": 504
    },
    {
      "epoch": 3.13,
      "learning_rate": 2e-05,
      "loss": 0.065,
      "step": 506
    },
    {
      "epoch": 3.15,
      "learning_rate": 2e-05,
      "loss": 0.065,
      "step": 508
    },
    {
      "epoch": 3.16,
      "learning_rate": 2e-05,
      "loss": 0.0648,
      "step": 510
    },
    {
      "epoch": 3.17,
      "learning_rate": 2e-05,
      "loss": 0.0639,
      "step": 512
    },
    {
      "epoch": 3.18,
      "learning_rate": 2e-05,
      "loss": 0.0632,
      "step": 514
    },
    {
      "epoch": 3.2,
      "learning_rate": 2e-05,
      "loss": 0.0615,
      "step": 516
    },
    {
      "epoch": 3.21,
      "learning_rate": 2e-05,
      "loss": 0.0621,
      "step": 518
    },
    {
      "epoch": 3.22,
      "learning_rate": 2e-05,
      "loss": 0.0635,
      "step": 520
    },
    {
      "epoch": 3.23,
      "learning_rate": 2e-05,
      "loss": 0.064,
      "step": 522
    },
    {
      "epoch": 3.25,
      "learning_rate": 2e-05,
      "loss": 0.0641,
      "step": 524
    },
    {
      "epoch": 3.26,
      "learning_rate": 2e-05,
      "loss": 0.0625,
      "step": 526
    },
    {
      "epoch": 3.27,
      "learning_rate": 2e-05,
      "loss": 0.0623,
      "step": 528
    },
    {
      "epoch": 3.28,
      "learning_rate": 2e-05,
      "loss": 0.0626,
      "step": 530
    },
    {
      "epoch": 3.3,
      "learning_rate": 2e-05,
      "loss": 0.0644,
      "step": 532
    },
    {
      "epoch": 3.31,
      "learning_rate": 2e-05,
      "loss": 0.0643,
      "step": 534
    },
    {
      "epoch": 3.32,
      "learning_rate": 2e-05,
      "loss": 0.0676,
      "step": 536
    },
    {
      "epoch": 3.33,
      "learning_rate": 2e-05,
      "loss": 0.0631,
      "step": 538
    },
    {
      "epoch": 3.34,
      "learning_rate": 2e-05,
      "loss": 0.0635,
      "step": 540
    },
    {
      "epoch": 3.36,
      "learning_rate": 2e-05,
      "loss": 0.0644,
      "step": 542
    },
    {
      "epoch": 3.37,
      "learning_rate": 2e-05,
      "loss": 0.0637,
      "step": 544
    },
    {
      "epoch": 3.38,
      "learning_rate": 2e-05,
      "loss": 0.0619,
      "step": 546
    },
    {
      "epoch": 3.39,
      "learning_rate": 2e-05,
      "loss": 0.0644,
      "step": 548
    },
    {
      "epoch": 3.41,
      "learning_rate": 2e-05,
      "loss": 0.0607,
      "step": 550
    },
    {
      "epoch": 3.42,
      "learning_rate": 2e-05,
      "loss": 0.0633,
      "step": 552
    },
    {
      "epoch": 3.43,
      "learning_rate": 2e-05,
      "loss": 0.0614,
      "step": 554
    },
    {
      "epoch": 3.44,
      "learning_rate": 2e-05,
      "loss": 0.0632,
      "step": 556
    },
    {
      "epoch": 3.46,
      "learning_rate": 2e-05,
      "loss": 0.0619,
      "step": 558
    },
    {
      "epoch": 3.47,
      "learning_rate": 2e-05,
      "loss": 0.0606,
      "step": 560
    },
    {
      "epoch": 3.48,
      "learning_rate": 2e-05,
      "loss": 0.0596,
      "step": 562
    },
    {
      "epoch": 3.49,
      "learning_rate": 2e-05,
      "loss": 0.0591,
      "step": 564
    },
    {
      "epoch": 3.51,
      "learning_rate": 2e-05,
      "loss": 0.0603,
      "step": 566
    },
    {
      "epoch": 3.52,
      "learning_rate": 2e-05,
      "loss": 0.059,
      "step": 568
    },
    {
      "epoch": 3.53,
      "learning_rate": 2e-05,
      "loss": 0.0584,
      "step": 570
    },
    {
      "epoch": 3.54,
      "learning_rate": 2e-05,
      "loss": 0.0599,
      "step": 572
    },
    {
      "epoch": 3.56,
      "learning_rate": 2e-05,
      "loss": 0.0587,
      "step": 574
    },
    {
      "epoch": 3.57,
      "learning_rate": 2e-05,
      "loss": 0.0589,
      "step": 576
    },
    {
      "epoch": 3.58,
      "learning_rate": 2e-05,
      "loss": 0.0579,
      "step": 578
    },
    {
      "epoch": 3.59,
      "learning_rate": 2e-05,
      "loss": 0.0602,
      "step": 580
    },
    {
      "epoch": 3.61,
      "learning_rate": 2e-05,
      "loss": 0.0576,
      "step": 582
    },
    {
      "epoch": 3.62,
      "learning_rate": 2e-05,
      "loss": 0.0567,
      "step": 584
    },
    {
      "epoch": 3.63,
      "learning_rate": 2e-05,
      "loss": 0.0584,
      "step": 586
    },
    {
      "epoch": 3.64,
      "learning_rate": 2e-05,
      "loss": 0.056,
      "step": 588
    },
    {
      "epoch": 3.65,
      "learning_rate": 2e-05,
      "loss": 0.0599,
      "step": 590
    },
    {
      "epoch": 3.67,
      "learning_rate": 2e-05,
      "loss": 0.058,
      "step": 592
    },
    {
      "epoch": 3.68,
      "learning_rate": 2e-05,
      "loss": 0.0582,
      "step": 594
    },
    {
      "epoch": 3.69,
      "learning_rate": 2e-05,
      "loss": 0.0558,
      "step": 596
    },
    {
      "epoch": 3.7,
      "learning_rate": 2e-05,
      "loss": 0.0575,
      "step": 598
    },
    {
      "epoch": 3.72,
      "learning_rate": 2e-05,
      "loss": 0.0548,
      "step": 600
    }
  ],
  "logging_steps": 2,
  "max_steps": 644,
  "num_train_epochs": 4,
  "save_steps": 200,
  "total_flos": 1.0201226996508262e+17,
  "trial_name": null,
  "trial_params": null
}