{
  "best_metric": 0.9312900900840759,
  "best_model_checkpoint": "/proj/berzelius-2023-338/users/x_boyji/model_hf/sft_medico-mixtral/checkpoint-1000",
  "epoch": 2.9991683991683993,
  "eval_steps": 500,
  "global_step": 1803,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 6.000000000000001e-07,
      "loss": 2.0272,
      "step": 3
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.2000000000000002e-06,
      "loss": 1.921,
      "step": 6
    },
    {
      "epoch": 0.01,
      "learning_rate": 1.8000000000000001e-06,
      "loss": 1.6681,
      "step": 9
    },
    {
      "epoch": 0.02,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 1.5087,
      "step": 12
    },
    {
      "epoch": 0.02,
      "learning_rate": 3e-06,
      "loss": 1.4299,
      "step": 15
    },
    {
      "epoch": 0.03,
      "learning_rate": 3.6000000000000003e-06,
      "loss": 1.3476,
      "step": 18
    },
    {
      "epoch": 0.03,
      "learning_rate": 4.2000000000000004e-06,
      "loss": 1.3558,
      "step": 21
    },
    {
      "epoch": 0.04,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.3113,
      "step": 24
    },
    {
      "epoch": 0.04,
      "learning_rate": 5.400000000000001e-06,
      "loss": 1.3135,
      "step": 27
    },
    {
      "epoch": 0.05,
      "learning_rate": 6e-06,
      "loss": 1.3068,
      "step": 30
    },
    {
      "epoch": 0.05,
      "learning_rate": 6.600000000000001e-06,
      "loss": 1.2869,
      "step": 33
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.2000000000000005e-06,
      "loss": 1.2698,
      "step": 36
    },
    {
      "epoch": 0.06,
      "learning_rate": 7.800000000000002e-06,
      "loss": 1.2347,
      "step": 39
    },
    {
      "epoch": 0.07,
      "learning_rate": 8.400000000000001e-06,
      "loss": 1.2687,
      "step": 42
    },
    {
      "epoch": 0.07,
      "learning_rate": 9e-06,
      "loss": 1.2221,
      "step": 45
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.2231,
      "step": 48
    },
    {
      "epoch": 0.08,
      "learning_rate": 1.02e-05,
      "loss": 1.1976,
      "step": 51
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.0800000000000002e-05,
      "loss": 1.1833,
      "step": 54
    },
    {
      "epoch": 0.09,
      "learning_rate": 1.14e-05,
      "loss": 1.1798,
      "step": 57
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.2e-05,
      "loss": 1.1878,
      "step": 60
    },
    {
      "epoch": 0.1,
      "learning_rate": 1.2600000000000001e-05,
      "loss": 1.1788,
      "step": 63
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.3200000000000002e-05,
      "loss": 1.1948,
      "step": 66
    },
    {
      "epoch": 0.11,
      "learning_rate": 1.38e-05,
      "loss": 1.1836,
      "step": 69
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.4400000000000001e-05,
      "loss": 1.1756,
      "step": 72
    },
    {
      "epoch": 0.12,
      "learning_rate": 1.5000000000000002e-05,
      "loss": 1.1868,
      "step": 75
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.5600000000000003e-05,
      "loss": 1.1652,
      "step": 78
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.62e-05,
      "loss": 1.1688,
      "step": 81
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 1.1783,
      "step": 84
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.7400000000000003e-05,
      "loss": 1.1799,
      "step": 87
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.8e-05,
      "loss": 1.3097,
      "step": 90
    },
    {
      "epoch": 0.15,
      "learning_rate": 1.86e-05,
      "loss": 1.2879,
      "step": 93
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 1.35,
      "step": 96
    },
    {
      "epoch": 0.16,
      "learning_rate": 1.98e-05,
      "loss": 1.2501,
      "step": 99
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.9999931938745066e-05,
      "loss": 1.1856,
      "step": 102
    },
    {
      "epoch": 0.17,
      "learning_rate": 1.999957461968997e-05,
      "loss": 1.1927,
      "step": 105
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.9998911038450304e-05,
      "loss": 1.2005,
      "step": 108
    },
    {
      "epoch": 0.18,
      "learning_rate": 1.999794121534991e-05,
      "loss": 1.1375,
      "step": 111
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9996665180092068e-05,
      "loss": 1.1645,
      "step": 114
    },
    {
      "epoch": 0.19,
      "learning_rate": 1.9995082971758574e-05,
      "loss": 1.1381,
      "step": 117
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.999319463880855e-05,
      "loss": 1.189,
      "step": 120
    },
    {
      "epoch": 0.2,
      "learning_rate": 1.9991000239076952e-05,
      "loss": 1.1539,
      "step": 123
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9988499839772805e-05,
      "loss": 1.1454,
      "step": 126
    },
    {
      "epoch": 0.21,
      "learning_rate": 1.9985693517477142e-05,
      "loss": 1.1611,
      "step": 129
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.9982581358140657e-05,
      "loss": 1.162,
      "step": 132
    },
    {
      "epoch": 0.22,
      "learning_rate": 1.997916345708108e-05,
      "loss": 1.1244,
      "step": 135
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.9975439918980246e-05,
      "loss": 1.1738,
      "step": 138
    },
    {
      "epoch": 0.23,
      "learning_rate": 1.99714108578809e-05,
      "loss": 1.1746,
      "step": 141
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9967076397183187e-05,
      "loss": 1.1332,
      "step": 144
    },
    {
      "epoch": 0.24,
      "learning_rate": 1.9962436669640906e-05,
      "loss": 1.145,
      "step": 147
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.9957491817357407e-05,
      "loss": 1.1246,
      "step": 150
    },
    {
      "epoch": 0.25,
      "learning_rate": 1.995224199178126e-05,
      "loss": 1.1561,
      "step": 153
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9946687353701607e-05,
      "loss": 1.1436,
      "step": 156
    },
    {
      "epoch": 0.26,
      "learning_rate": 1.9940828073243244e-05,
      "loss": 1.1285,
      "step": 159
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9934664329861412e-05,
      "loss": 1.1578,
      "step": 162
    },
    {
      "epoch": 0.27,
      "learning_rate": 1.9928196312336287e-05,
      "loss": 1.141,
      "step": 165
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.9921424218767215e-05,
      "loss": 1.1133,
      "step": 168
    },
    {
      "epoch": 0.28,
      "learning_rate": 1.991434825656664e-05,
      "loss": 1.114,
      "step": 171
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.9906968642453745e-05,
      "loss": 1.1049,
      "step": 174
    },
    {
      "epoch": 0.29,
      "learning_rate": 1.989928560244781e-05,
      "loss": 1.1417,
      "step": 177
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9891299371861312e-05,
      "loss": 1.1442,
      "step": 180
    },
    {
      "epoch": 0.3,
      "learning_rate": 1.9883010195292702e-05,
      "loss": 1.1164,
      "step": 183
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9874418326618906e-05,
      "loss": 1.1319,
      "step": 186
    },
    {
      "epoch": 0.31,
      "learning_rate": 1.9865524028987565e-05,
      "loss": 1.1265,
      "step": 189
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9856327574808974e-05,
      "loss": 1.1289,
      "step": 192
    },
    {
      "epoch": 0.32,
      "learning_rate": 1.9846829245747725e-05,
      "loss": 1.1381,
      "step": 195
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9837029332714094e-05,
      "loss": 1.13,
      "step": 198
    },
    {
      "epoch": 0.33,
      "learning_rate": 1.9826928135855128e-05,
      "loss": 1.134,
      "step": 201
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9816525964545447e-05,
      "loss": 1.1173,
      "step": 204
    },
    {
      "epoch": 0.34,
      "learning_rate": 1.9805823137377776e-05,
      "loss": 1.1364,
      "step": 207
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.979481998215318e-05,
      "loss": 1.1241,
      "step": 210
    },
    {
      "epoch": 0.35,
      "learning_rate": 1.978351683587103e-05,
      "loss": 1.134,
      "step": 213
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9771914044718682e-05,
      "loss": 1.1064,
      "step": 216
    },
    {
      "epoch": 0.36,
      "learning_rate": 1.9760011964060863e-05,
      "loss": 1.1324,
      "step": 219
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.974781095842881e-05,
      "loss": 1.1239,
      "step": 222
    },
    {
      "epoch": 0.37,
      "learning_rate": 1.9735311401509078e-05,
      "loss": 1.1348,
      "step": 225
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9722513676132112e-05,
      "loss": 1.1425,
      "step": 228
    },
    {
      "epoch": 0.38,
      "learning_rate": 1.9709418174260523e-05,
      "loss": 1.1224,
      "step": 231
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9696025296977067e-05,
      "loss": 1.0922,
      "step": 234
    },
    {
      "epoch": 0.39,
      "learning_rate": 1.9682335454472382e-05,
      "loss": 1.1123,
      "step": 237
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9668349066032412e-05,
      "loss": 1.0966,
      "step": 240
    },
    {
      "epoch": 0.4,
      "learning_rate": 1.9654066560025566e-05,
      "loss": 1.119,
      "step": 243
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9639488373889597e-05,
      "loss": 1.0947,
      "step": 246
    },
    {
      "epoch": 0.41,
      "learning_rate": 1.9624614954118214e-05,
      "loss": 1.0871,
      "step": 249
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9609446756247402e-05,
      "loss": 1.1088,
      "step": 252
    },
    {
      "epoch": 0.42,
      "learning_rate": 1.9593984244841455e-05,
      "loss": 1.0891,
      "step": 255
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.957822789347878e-05,
      "loss": 1.1095,
      "step": 258
    },
    {
      "epoch": 0.43,
      "learning_rate": 1.9562178184737358e-05,
      "loss": 1.1145,
      "step": 261
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.9545835610179985e-05,
      "loss": 1.1324,
      "step": 264
    },
    {
      "epoch": 0.44,
      "learning_rate": 1.952920067033921e-05,
      "loss": 1.1047,
      "step": 267
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9512273874702016e-05,
      "loss": 1.1257,
      "step": 270
    },
    {
      "epoch": 0.45,
      "learning_rate": 1.9495055741694188e-05,
      "loss": 1.1089,
      "step": 273
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9477546798664463e-05,
      "loss": 1.0767,
      "step": 276
    },
    {
      "epoch": 0.46,
      "learning_rate": 1.9459747581868366e-05,
      "loss": 1.0819,
      "step": 279
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.9441658636451794e-05,
      "loss": 1.0888,
      "step": 282
    },
    {
      "epoch": 0.47,
      "learning_rate": 1.94232805164343e-05,
      "loss": 1.1043,
      "step": 285
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9404613784692152e-05,
      "loss": 1.0753,
      "step": 288
    },
    {
      "epoch": 0.48,
      "learning_rate": 1.9385659012941072e-05,
      "loss": 1.0885,
      "step": 291
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9366416781718733e-05,
      "loss": 1.08,
      "step": 294
    },
    {
      "epoch": 0.49,
      "learning_rate": 1.9346887680366984e-05,
      "loss": 1.0894,
      "step": 297
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9327072307013794e-05,
      "loss": 1.0798,
      "step": 300
    },
    {
      "epoch": 0.5,
      "learning_rate": 1.9306971268554927e-05,
      "loss": 1.0543,
      "step": 303
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.9286585180635375e-05,
      "loss": 1.0734,
      "step": 306
    },
    {
      "epoch": 0.51,
      "learning_rate": 1.926591466763047e-05,
      "loss": 1.0822,
      "step": 309
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.9244960362626795e-05,
      "loss": 1.0944,
      "step": 312
    },
    {
      "epoch": 0.52,
      "learning_rate": 1.922372290740277e-05,
      "loss": 1.0789,
      "step": 315
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9202202952409012e-05,
      "loss": 1.0973,
      "step": 318
    },
    {
      "epoch": 0.53,
      "learning_rate": 1.9180401156748397e-05,
      "loss": 1.0952,
      "step": 321
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9158318188155882e-05,
      "loss": 1.0775,
      "step": 324
    },
    {
      "epoch": 0.54,
      "learning_rate": 1.9135954722978062e-05,
      "loss": 1.0793,
      "step": 327
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.9113311446152445e-05,
      "loss": 1.0439,
      "step": 330
    },
    {
      "epoch": 0.55,
      "learning_rate": 1.909038905118647e-05,
      "loss": 1.0889,
      "step": 333
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.906718824013628e-05,
      "loss": 1.0371,
      "step": 336
    },
    {
      "epoch": 0.56,
      "learning_rate": 1.9043709723585207e-05,
      "loss": 1.093,
      "step": 339
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.9019954220622018e-05,
      "loss": 1.0692,
      "step": 342
    },
    {
      "epoch": 0.57,
      "learning_rate": 1.8995922458818885e-05,
      "loss": 1.0658,
      "step": 345
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8971615174209104e-05,
      "loss": 1.0633,
      "step": 348
    },
    {
      "epoch": 0.58,
      "learning_rate": 1.8947033111264558e-05,
      "loss": 1.0732,
      "step": 351
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.892217702287289e-05,
      "loss": 1.0536,
      "step": 354
    },
    {
      "epoch": 0.59,
      "learning_rate": 1.889704767031449e-05,
      "loss": 1.0752,
      "step": 357
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8871645823239128e-05,
      "loss": 1.0933,
      "step": 360
    },
    {
      "epoch": 0.6,
      "learning_rate": 1.8845972259642424e-05,
      "loss": 1.0703,
      "step": 363
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8820027765841993e-05,
      "loss": 1.0722,
      "step": 366
    },
    {
      "epoch": 0.61,
      "learning_rate": 1.8793813136453364e-05,
      "loss": 1.0367,
      "step": 369
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8767329174365664e-05,
      "loss": 1.0361,
      "step": 372
    },
    {
      "epoch": 0.62,
      "learning_rate": 1.8740576690717004e-05,
      "loss": 1.0454,
      "step": 375
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8713556504869644e-05,
      "loss": 1.0725,
      "step": 378
    },
    {
      "epoch": 0.63,
      "learning_rate": 1.8686269444384905e-05,
      "loss": 1.1055,
      "step": 381
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8658716344997812e-05,
      "loss": 1.1818,
      "step": 384
    },
    {
      "epoch": 0.64,
      "learning_rate": 1.8630898050591502e-05,
      "loss": 1.101,
      "step": 387
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.860281541317138e-05,
      "loss": 1.0709,
      "step": 390
    },
    {
      "epoch": 0.65,
      "learning_rate": 1.857446929283902e-05,
      "loss": 1.0668,
      "step": 393
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8545860557765828e-05,
      "loss": 1.0396,
      "step": 396
    },
    {
      "epoch": 0.66,
      "learning_rate": 1.8516990084166444e-05,
      "loss": 1.0795,
      "step": 399
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.8487858756271918e-05,
      "loss": 1.0401,
      "step": 402
    },
    {
      "epoch": 0.67,
      "learning_rate": 1.845846746630261e-05,
      "loss": 1.0679,
      "step": 405
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.842881711444088e-05,
      "loss": 1.0702,
      "step": 408
    },
    {
      "epoch": 0.68,
      "learning_rate": 1.8398908608803518e-05,
      "loss": 1.0489,
      "step": 411
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8368742865413905e-05,
      "loss": 1.0245,
      "step": 414
    },
    {
      "epoch": 0.69,
      "learning_rate": 1.8338320808173998e-05,
      "loss": 1.0603,
      "step": 417
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8307643368836003e-05,
      "loss": 1.0493,
      "step": 420
    },
    {
      "epoch": 0.7,
      "learning_rate": 1.8276711486973838e-05,
      "loss": 1.048,
      "step": 423
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8245526109954385e-05,
      "loss": 1.078,
      "step": 426
    },
    {
      "epoch": 0.71,
      "learning_rate": 1.8214088192908433e-05,
      "loss": 1.0251,
      "step": 429
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.818239869870146e-05,
      "loss": 1.0558,
      "step": 432
    },
    {
      "epoch": 0.72,
      "learning_rate": 1.815045859790413e-05,
      "loss": 1.0433,
      "step": 435
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.811826886876255e-05,
      "loss": 1.0484,
      "step": 438
    },
    {
      "epoch": 0.73,
      "learning_rate": 1.8085830497168334e-05,
      "loss": 1.0559,
      "step": 441
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.805314447662841e-05,
      "loss": 1.0715,
      "step": 444
    },
    {
      "epoch": 0.74,
      "learning_rate": 1.8020211808234556e-05,
      "loss": 1.0418,
      "step": 447
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7987033500632785e-05,
      "loss": 1.0166,
      "step": 450
    },
    {
      "epoch": 0.75,
      "learning_rate": 1.7953610569992415e-05,
      "loss": 1.0427,
      "step": 453
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.7919944039974962e-05,
      "loss": 1.0463,
      "step": 456
    },
    {
      "epoch": 0.76,
      "learning_rate": 1.78860349417028e-05,
      "loss": 1.0035,
      "step": 459
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.7851884313727547e-05,
      "loss": 1.047,
      "step": 462
    },
    {
      "epoch": 0.77,
      "learning_rate": 1.78174932019983e-05,
      "loss": 1.0241,
      "step": 465
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7782862659829553e-05,
      "loss": 1.0308,
      "step": 468
    },
    {
      "epoch": 0.78,
      "learning_rate": 1.7747993747868985e-05,
      "loss": 1.021,
      "step": 471
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7712887534064935e-05,
      "loss": 1.0216,
      "step": 474
    },
    {
      "epoch": 0.79,
      "learning_rate": 1.7677545093633713e-05,
      "loss": 1.0228,
      "step": 477
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7641967509026667e-05,
      "loss": 1.02,
      "step": 480
    },
    {
      "epoch": 0.8,
      "learning_rate": 1.7606155869897024e-05,
      "loss": 1.0065,
      "step": 483
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.757011127306653e-05,
      "loss": 1.0353,
      "step": 486
    },
    {
      "epoch": 0.81,
      "learning_rate": 1.7533834822491832e-05,
      "loss": 1.0346,
      "step": 489
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.7497327629230707e-05,
      "loss": 1.0241,
      "step": 492
    },
    {
      "epoch": 0.82,
      "learning_rate": 1.746059081140798e-05,
      "loss": 1.0036,
      "step": 495
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7423625494181334e-05,
      "loss": 1.0195,
      "step": 498
    },
    {
      "epoch": 0.83,
      "eval_loss": 0.9461708068847656,
      "eval_runtime": 61.2738,
      "eval_samples_per_second": 7.556,
      "eval_steps_per_second": 0.947,
      "step": 500
    },
    {
      "epoch": 0.83,
      "learning_rate": 1.7386432809706795e-05,
      "loss": 1.0261,
      "step": 501
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.734901389710411e-05,
      "loss": 1.0162,
      "step": 504
    },
    {
      "epoch": 0.84,
      "learning_rate": 1.7311369902421814e-05,
      "loss": 1.0056,
      "step": 507
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.727350197860216e-05,
      "loss": 1.0025,
      "step": 510
    },
    {
      "epoch": 0.85,
      "learning_rate": 1.723541128544579e-05,
      "loss": 1.0088,
      "step": 513
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.7197098989576222e-05,
      "loss": 1.0285,
      "step": 516
    },
    {
      "epoch": 0.86,
      "learning_rate": 1.7158566264404112e-05,
      "loss": 1.0105,
      "step": 519
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.7119814290091315e-05,
      "loss": 1.0384,
      "step": 522
    },
    {
      "epoch": 0.87,
      "learning_rate": 1.708084425351476e-05,
      "loss": 1.0274,
      "step": 525
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.704165734823006e-05,
      "loss": 1.0307,
      "step": 528
    },
    {
      "epoch": 0.88,
      "learning_rate": 1.7002254774435e-05,
      "loss": 1.0112,
      "step": 531
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6962637738932746e-05,
      "loss": 1.007,
      "step": 534
    },
    {
      "epoch": 0.89,
      "learning_rate": 1.6922807455094898e-05,
      "loss": 0.9999,
      "step": 537
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.688276514282431e-05,
      "loss": 1.015,
      "step": 540
    },
    {
      "epoch": 0.9,
      "learning_rate": 1.6842512028517773e-05,
      "loss": 0.9912,
      "step": 543
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.6802049345028388e-05,
      "loss": 1.0109,
      "step": 546
    },
    {
      "epoch": 0.91,
      "learning_rate": 1.676137833162786e-05,
      "loss": 1.0079,
      "step": 549
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6720500233968522e-05,
      "loss": 1.0352,
      "step": 552
    },
    {
      "epoch": 0.92,
      "learning_rate": 1.6679416304045172e-05,
      "loss": 1.0021,
      "step": 555
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.6638127800156754e-05,
      "loss": 0.9867,
      "step": 558
    },
    {
      "epoch": 0.93,
      "learning_rate": 1.65966359868678e-05,
      "loss": 0.9974,
      "step": 561
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.65549421349697e-05,
      "loss": 1.0049,
      "step": 564
    },
    {
      "epoch": 0.94,
      "learning_rate": 1.6513047521441795e-05,
      "loss": 1.0141,
      "step": 567
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6470953429412257e-05,
      "loss": 0.9925,
      "step": 570
    },
    {
      "epoch": 0.95,
      "learning_rate": 1.6428661148118775e-05,
      "loss": 1.0206,
      "step": 573
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.638617197286911e-05,
      "loss": 1.0067,
      "step": 576
    },
    {
      "epoch": 0.96,
      "learning_rate": 1.634348720500136e-05,
      "loss": 0.9687,
      "step": 579
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6300608151844184e-05,
      "loss": 1.0217,
      "step": 582
    },
    {
      "epoch": 0.97,
      "learning_rate": 1.6257536126676684e-05,
      "loss": 0.9977,
      "step": 585
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.6214272448688228e-05,
      "loss": 0.9951,
      "step": 588
    },
    {
      "epoch": 0.98,
      "learning_rate": 1.6170818442938035e-05,
      "loss": 0.994,
      "step": 591
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.6127175440314596e-05,
      "loss": 0.9697,
      "step": 594
    },
    {
      "epoch": 0.99,
      "learning_rate": 1.6083344777494896e-05,
      "loss": 1.0084,
      "step": 597
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.6039327796903502e-05,
      "loss": 0.9991,
      "step": 600
    },
    {
      "epoch": 1.0,
      "learning_rate": 1.5995125846671417e-05,
      "loss": 0.7961,
      "step": 603
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.5950740280594815e-05,
      "loss": 0.6247,
      "step": 606
    },
    {
      "epoch": 1.01,
      "learning_rate": 1.590617245809357e-05,
      "loss": 0.6435,
      "step": 609
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5861423744169608e-05,
      "loss": 0.6282,
      "step": 612
    },
    {
      "epoch": 1.02,
      "learning_rate": 1.5816495509365123e-05,
      "loss": 0.6123,
      "step": 615
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.577138912972058e-05,
      "loss": 0.6011,
      "step": 618
    },
    {
      "epoch": 1.03,
      "learning_rate": 1.5726105986732582e-05,
      "loss": 0.6344,
      "step": 621
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.568064746731156e-05,
      "loss": 0.6108,
      "step": 624
    },
    {
      "epoch": 1.04,
      "learning_rate": 1.5635014963739278e-05,
      "loss": 0.6274,
      "step": 627
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5589209873626215e-05,
      "loss": 0.5958,
      "step": 630
    },
    {
      "epoch": 1.05,
      "learning_rate": 1.5543233599868744e-05,
      "loss": 0.6143,
      "step": 633
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5497087550606167e-05,
      "loss": 0.6181,
      "step": 636
    },
    {
      "epoch": 1.06,
      "learning_rate": 1.5450773139177587e-05,
      "loss": 0.6139,
      "step": 639
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5404291784078633e-05,
      "loss": 0.6214,
      "step": 642
    },
    {
      "epoch": 1.07,
      "learning_rate": 1.5357644908917995e-05,
      "loss": 0.5895,
      "step": 645
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.5310833942373833e-05,
      "loss": 0.5976,
      "step": 648
    },
    {
      "epoch": 1.08,
      "learning_rate": 1.526386031815003e-05,
      "loss": 0.6034,
      "step": 651
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.5216725474932246e-05,
      "loss": 0.6124,
      "step": 654
    },
    {
      "epoch": 1.09,
      "learning_rate": 1.5169430856343909e-05,
      "loss": 0.6143,
      "step": 657
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.5121977910901953e-05,
      "loss": 0.6156,
      "step": 660
    },
    {
      "epoch": 1.1,
      "learning_rate": 1.5074368091972476e-05,
      "loss": 0.622,
      "step": 663
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.5026602857726216e-05,
      "loss": 0.5983,
      "step": 666
    },
    {
      "epoch": 1.11,
      "learning_rate": 1.4978683671093905e-05,
      "loss": 0.601,
      "step": 669
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4930611999721457e-05,
      "loss": 0.5974,
      "step": 672
    },
    {
      "epoch": 1.12,
      "learning_rate": 1.4882389315925002e-05,
      "loss": 0.5977,
      "step": 675
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.483401709664582e-05,
      "loss": 0.6019,
      "step": 678
    },
    {
      "epoch": 1.13,
      "learning_rate": 1.4785496823405084e-05,
      "loss": 0.608,
      "step": 681
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.4736829982258498e-05,
      "loss": 0.6013,
      "step": 684
    },
    {
      "epoch": 1.14,
      "learning_rate": 1.468801806375077e-05,
      "loss": 0.5834,
      "step": 687
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4639062562869986e-05,
      "loss": 0.604,
      "step": 690
    },
    {
      "epoch": 1.15,
      "learning_rate": 1.4589964979001779e-05,
      "loss": 0.6028,
      "step": 693
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4540726815883445e-05,
      "loss": 0.5999,
      "step": 696
    },
    {
      "epoch": 1.16,
      "learning_rate": 1.4491349581557884e-05,
      "loss": 0.6015,
      "step": 699
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4441834788327382e-05,
      "loss": 0.6103,
      "step": 702
    },
    {
      "epoch": 1.17,
      "learning_rate": 1.4392183952707324e-05,
      "loss": 0.5958,
      "step": 705
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4342398595379738e-05,
      "loss": 0.6088,
      "step": 708
    },
    {
      "epoch": 1.18,
      "learning_rate": 1.4292480241146715e-05,
      "loss": 0.5966,
      "step": 711
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.4242430418883713e-05,
      "loss": 0.5949,
      "step": 714
    },
    {
      "epoch": 1.19,
      "learning_rate": 1.4192250661492724e-05,
      "loss": 0.5962,
      "step": 717
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.414194250585534e-05,
      "loss": 0.6081,
      "step": 720
    },
    {
      "epoch": 1.2,
      "learning_rate": 1.409150749278567e-05,
      "loss": 0.609,
      "step": 723
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.4040947166983147e-05,
      "loss": 0.6169,
      "step": 726
    },
    {
      "epoch": 1.21,
      "learning_rate": 1.3990263076985228e-05,
      "loss": 0.6158,
      "step": 729
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3939456775119962e-05,
      "loss": 0.5725,
      "step": 732
    },
    {
      "epoch": 1.22,
      "learning_rate": 1.3888529817458436e-05,
      "loss": 0.6092,
      "step": 735
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3837483763767128e-05,
      "loss": 0.6091,
      "step": 738
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.3786320177460142e-05,
      "loss": 0.629,
      "step": 741
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3735040625551303e-05,
      "loss": 0.5754,
      "step": 744
    },
    {
      "epoch": 1.24,
      "learning_rate": 1.3683646678606176e-05,
      "loss": 0.5961,
      "step": 747
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3632139910693971e-05,
      "loss": 0.5801,
      "step": 750
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.3580521899339316e-05,
      "loss": 0.6093,
      "step": 753
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3528794225473952e-05,
      "loss": 0.5903,
      "step": 756
    },
    {
      "epoch": 1.26,
      "learning_rate": 1.3476958473388315e-05,
      "loss": 0.6145,
      "step": 759
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3425016230683e-05,
      "loss": 0.6015,
      "step": 762
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.3372969088220158e-05,
      "loss": 0.5927,
      "step": 765
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.332081864007475e-05,
      "loss": 0.6049,
      "step": 768
    },
    {
      "epoch": 1.28,
      "learning_rate": 1.3268566483485734e-05,
      "loss": 0.5824,
      "step": 771
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.3216214218807152e-05,
      "loss": 0.593,
      "step": 774
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.3163763449459105e-05,
      "loss": 0.5972,
      "step": 777
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.3111215781878648e-05,
      "loss": 0.6,
      "step": 780
    },
    {
      "epoch": 1.3,
      "learning_rate": 1.3058572825470588e-05,
      "loss": 0.6063,
      "step": 783
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.3005836192558192e-05,
      "loss": 0.6135,
      "step": 786
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.2953007498333807e-05,
      "loss": 0.6082,
      "step": 789
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2900088360809396e-05,
      "loss": 0.5708,
      "step": 792
    },
    {
      "epoch": 1.32,
      "learning_rate": 1.2847080400766962e-05,
      "loss": 0.5901,
      "step": 795
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.2793985241708934e-05,
      "loss": 0.612,
      "step": 798
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.274080450980843e-05,
      "loss": 0.594,
      "step": 801
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2687539833859444e-05,
      "loss": 0.6105,
      "step": 804
    },
    {
      "epoch": 1.34,
      "learning_rate": 1.2634192845226971e-05,
      "loss": 0.6035,
      "step": 807
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.258076517779705e-05,
      "loss": 0.5969,
      "step": 810
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.2527258467926688e-05,
      "loss": 0.5916,
      "step": 813
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2473674354393797e-05,
      "loss": 0.6268,
      "step": 816
    },
    {
      "epoch": 1.36,
      "learning_rate": 1.2420014478346944e-05,
      "loss": 0.5937,
      "step": 819
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2366280483255128e-05,
      "loss": 0.5846,
      "step": 822
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.2312474014857432e-05,
      "loss": 0.6201,
      "step": 825
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.2258596721112608e-05,
      "loss": 0.5948,
      "step": 828
    },
    {
      "epoch": 1.38,
      "learning_rate": 1.2204650252148615e-05,
      "loss": 0.5824,
      "step": 831
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.2150636260212088e-05,
      "loss": 0.5968,
      "step": 834
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.209655639961771e-05,
      "loss": 0.6026,
      "step": 837
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.204241232669756e-05,
      "loss": 0.6069,
      "step": 840
    },
    {
      "epoch": 1.4,
      "learning_rate": 1.1988205699750391e-05,
      "loss": 0.5881,
      "step": 843
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1933938178990816e-05,
      "loss": 0.5851,
      "step": 846
    },
    {
      "epoch": 1.41,
      "learning_rate": 1.1879611426498488e-05,
      "loss": 0.5966,
      "step": 849
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1825227106167178e-05,
      "loss": 0.5896,
      "step": 852
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1770786883653806e-05,
      "loss": 0.6103,
      "step": 855
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1716292426327454e-05,
      "loss": 0.6132,
      "step": 858
    },
    {
      "epoch": 1.43,
      "learning_rate": 1.1661745403218276e-05,
      "loss": 0.5875,
      "step": 861
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1607147484966372e-05,
      "loss": 0.5799,
      "step": 864
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1552500343770658e-05,
      "loss": 0.6214,
      "step": 867
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.149780565333761e-05,
      "loss": 0.5999,
      "step": 870
    },
    {
      "epoch": 1.45,
      "learning_rate": 1.1443065088830019e-05,
      "loss": 0.5962,
      "step": 873
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1388280326815697e-05,
      "loss": 0.5969,
      "step": 876
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.1333453045216107e-05,
      "loss": 0.5796,
      "step": 879
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.1278584923254981e-05,
      "loss": 0.6056,
      "step": 882
    },
    {
      "epoch": 1.47,
      "learning_rate": 1.1223677641406896e-05,
      "loss": 0.5967,
      "step": 885
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.1168732881345794e-05,
      "loss": 0.5935,
      "step": 888
    },
    {
      "epoch": 1.48,
      "learning_rate": 1.1113752325893483e-05,
      "loss": 0.602,
      "step": 891
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.1058737658968102e-05,
      "loss": 0.5983,
      "step": 894
    },
    {
      "epoch": 1.49,
      "learning_rate": 1.1003690565532522e-05,
      "loss": 0.6014,
      "step": 897
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0948612731542776e-05,
      "loss": 0.5906,
      "step": 900
    },
    {
      "epoch": 1.5,
      "learning_rate": 1.0893505843896403e-05,
      "loss": 0.5803,
      "step": 903
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0838371590380765e-05,
      "loss": 0.5994,
      "step": 906
    },
    {
      "epoch": 1.51,
      "learning_rate": 1.0783211659621396e-05,
      "loss": 0.5852,
      "step": 909
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0728027741030247e-05,
      "loss": 0.6085,
      "step": 912
    },
    {
      "epoch": 1.52,
      "learning_rate": 1.0672821524753953e-05,
      "loss": 0.6101,
      "step": 915
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.061759470162209e-05,
      "loss": 0.5844,
      "step": 918
    },
    {
      "epoch": 1.53,
      "learning_rate": 1.056234896309535e-05,
      "loss": 0.6055,
      "step": 921
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.050708600121377e-05,
      "loss": 0.6062,
      "step": 924
    },
    {
      "epoch": 1.54,
      "learning_rate": 1.0451807508544891e-05,
      "loss": 0.5935,
      "step": 927
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0396515178131925e-05,
      "loss": 0.5854,
      "step": 930
    },
    {
      "epoch": 1.55,
      "learning_rate": 1.0341210703441895e-05,
      "loss": 0.5695,
      "step": 933
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0285895778313783e-05,
      "loss": 0.6004,
      "step": 936
    },
    {
      "epoch": 1.56,
      "learning_rate": 1.0230572096906634e-05,
      "loss": 0.5908,
      "step": 939
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.0175241353647678e-05,
      "loss": 0.5852,
      "step": 942
    },
    {
      "epoch": 1.57,
      "learning_rate": 1.0119905243180432e-05,
      "loss": 0.5822,
      "step": 945
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.0064565460312794e-05,
      "loss": 0.5818,
      "step": 948
    },
    {
      "epoch": 1.58,
      "learning_rate": 1.000922369996515e-05,
      "loss": 0.5944,
      "step": 951
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.953881657118438e-06,
      "loss": 0.5785,
      "step": 954
    },
    {
      "epoch": 1.59,
      "learning_rate": 9.898541026762252e-06,
      "loss": 0.5894,
      "step": 957
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.843203503842928e-06,
      "loss": 0.5947,
      "step": 960
    },
    {
      "epoch": 1.6,
      "learning_rate": 9.787870783211633e-06,
      "loss": 0.5969,
      "step": 963
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.732544559572443e-06,
      "loss": 0.588,
      "step": 966
    },
    {
      "epoch": 1.61,
      "learning_rate": 9.677226527430444e-06,
      "loss": 0.5762,
      "step": 969
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.621918381039851e-06,
      "loss": 0.5909,
      "step": 972
    },
    {
      "epoch": 1.62,
      "learning_rate": 9.56662181435209e-06,
      "loss": 0.5907,
      "step": 975
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.511338520963942e-06,
      "loss": 0.5654,
      "step": 978
    },
    {
      "epoch": 1.63,
      "learning_rate": 9.456070194065647e-06,
      "loss": 0.5793,
      "step": 981
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.400818526389063e-06,
      "loss": 0.5736,
      "step": 984
    },
    {
      "epoch": 1.64,
      "learning_rate": 9.345585210155818e-06,
      "loss": 0.5839,
      "step": 987
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.290371937025486e-06,
      "loss": 0.5605,
      "step": 990
    },
    {
      "epoch": 1.65,
      "learning_rate": 9.235180398043756e-06,
      "loss": 0.5774,
      "step": 993
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.180012283590678e-06,
      "loss": 0.5851,
      "step": 996
    },
    {
      "epoch": 1.66,
      "learning_rate": 9.12486928332884e-06,
      "loss": 0.5867,
      "step": 999
    },
    {
      "epoch": 1.66,
      "eval_loss": 0.9312900900840759,
      "eval_runtime": 61.0314,
      "eval_samples_per_second": 7.586,
      "eval_steps_per_second": 0.95,
      "step": 1000
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.069753086151657e-06,
      "loss": 0.3263,
      "step": 1002
    },
    {
      "epoch": 1.67,
      "learning_rate": 9.014665380131639e-06,
      "loss": 0.3206,
      "step": 1005
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.959607852468667e-06,
      "loss": 0.3393,
      "step": 1008
    },
    {
      "epoch": 1.68,
      "learning_rate": 8.904582189438345e-06,
      "loss": 0.3301,
      "step": 1011
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.849590076340344e-06,
      "loss": 0.3327,
      "step": 1014
    },
    {
      "epoch": 1.69,
      "learning_rate": 8.79463319744677e-06,
      "loss": 0.3337,
      "step": 1017
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.739713235950608e-06,
      "loss": 0.3504,
      "step": 1020
    },
    {
      "epoch": 1.7,
      "learning_rate": 8.684831873914146e-06,
      "loss": 0.3291,
      "step": 1023
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.629990792217464e-06,
      "loss": 0.3301,
      "step": 1026
    },
    {
      "epoch": 1.71,
      "learning_rate": 8.575191670506969e-06,
      "loss": 0.3119,
      "step": 1029
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.520436187143921e-06,
      "loss": 0.3326,
      "step": 1032
    },
    {
      "epoch": 1.72,
      "learning_rate": 8.465726019153052e-06,
      "loss": 0.3274,
      "step": 1035
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.411062842171198e-06,
      "loss": 0.3294,
      "step": 1038
    },
    {
      "epoch": 1.73,
      "learning_rate": 8.356448330395979e-06,
      "loss": 0.3375,
      "step": 1041
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.301884156534511e-06,
      "loss": 0.317,
      "step": 1044
    },
    {
      "epoch": 1.74,
      "learning_rate": 8.247371991752201e-06,
      "loss": 0.3359,
      "step": 1047
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.19291350562153e-06,
      "loss": 0.3003,
      "step": 1050
    },
    {
      "epoch": 1.75,
      "learning_rate": 8.138510366070937e-06,
      "loss": 0.3194,
      "step": 1053
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.084164239333745e-06,
      "loss": 0.3304,
      "step": 1056
    },
    {
      "epoch": 1.76,
      "learning_rate": 8.029876789897102e-06,
      "loss": 0.3219,
      "step": 1059
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.975649680451024e-06,
      "loss": 0.3059,
      "step": 1062
    },
    {
      "epoch": 1.77,
      "learning_rate": 7.921484571837465e-06,
      "loss": 0.3342,
      "step": 1065
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.867383122999435e-06,
      "loss": 0.3318,
      "step": 1068
    },
    {
      "epoch": 1.78,
      "learning_rate": 7.813346990930209e-06,
      "loss": 0.3303,
      "step": 1071
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.759377830622581e-06,
      "loss": 0.3379,
      "step": 1074
    },
    {
      "epoch": 1.79,
      "learning_rate": 7.705477295018148e-06,
      "loss": 0.3267,
      "step": 1077
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.651647034956721e-06,
      "loss": 0.3336,
      "step": 1080
    },
    {
      "epoch": 1.8,
      "learning_rate": 7.597888699125731e-06,
      "loss": 0.3346,
      "step": 1083
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.544203934009752e-06,
      "loss": 0.3449,
      "step": 1086
    },
    {
      "epoch": 1.81,
      "learning_rate": 7.4905943838400716e-06,
      "loss": 0.3323,
      "step": 1089
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.437061690544331e-06,
      "loss": 0.3114,
      "step": 1092
    },
    {
      "epoch": 1.82,
      "learning_rate": 7.383607493696235e-06,
      "loss": 0.3249,
      "step": 1095
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.330233430465334e-06,
      "loss": 0.3248,
      "step": 1098
    },
    {
      "epoch": 1.83,
      "learning_rate": 7.276941135566884e-06,
      "loss": 0.3243,
      "step": 1101
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.223732241211783e-06,
      "loss": 0.3255,
      "step": 1104
    },
    {
      "epoch": 1.84,
      "learning_rate": 7.170608377056576e-06,
      "loss": 0.3189,
      "step": 1107
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.117571170153543e-06,
      "loss": 0.3145,
      "step": 1110
    },
    {
      "epoch": 1.85,
      "learning_rate": 7.064622244900872e-06,
      "loss": 0.3178,
      "step": 1113
    },
    {
      "epoch": 1.86,
      "learning_rate": 7.011763222992891e-06,
      "loss": 0.3007,
      "step": 1116
    },
    {
      "epoch": 1.86,
      "learning_rate": 6.958995723370425e-06,
      "loss": 0.3254,
      "step": 1119
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.906321362171189e-06,
      "loss": 0.3263,
      "step": 1122
    },
    {
      "epoch": 1.87,
      "learning_rate": 6.853741752680303e-06,
      "loss": 0.3041,
      "step": 1125
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.801258505280876e-06,
      "loss": 0.329,
      "step": 1128
    },
    {
      "epoch": 1.88,
      "learning_rate": 6.74887322740469e-06,
      "loss": 0.3324,
      "step": 1131
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.696587523482951e-06,
      "loss": 0.3213,
      "step": 1134
    },
    {
      "epoch": 1.89,
      "learning_rate": 6.644402994897171e-06,
      "loss": 0.3221,
      "step": 1137
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.592321239930112e-06,
      "loss": 0.3059,
      "step": 1140
    },
    {
      "epoch": 1.9,
      "learning_rate": 6.540343853716834e-06,
      "loss": 0.3248,
      "step": 1143
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.488472428195841e-06,
      "loss": 0.3325,
      "step": 1146
    },
    {
      "epoch": 1.91,
      "learning_rate": 6.436708552060317e-06,
      "loss": 0.3132,
      "step": 1149
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.385053810709485e-06,
      "loss": 0.3205,
      "step": 1152
    },
    {
      "epoch": 1.92,
      "learning_rate": 6.333509786200032e-06,
      "loss": 0.3095,
      "step": 1155
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.282078057197671e-06,
      "loss": 0.3166,
      "step": 1158
    },
    {
      "epoch": 1.93,
      "learning_rate": 6.230760198928774e-06,
      "loss": 0.3172,
      "step": 1161
    },
    {
      "epoch": 1.94,
      "learning_rate": 6.1795577831321405e-06,
      "loss": 0.3251,
      "step": 1164
    },
    {
      "epoch": 1.94,
      "learning_rate": 6.128472378010846e-06,
      "loss": 0.3211,
      "step": 1167
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.077505548184226e-06,
      "loss": 0.3091,
      "step": 1170
    },
    {
      "epoch": 1.95,
      "learning_rate": 6.026658854639946e-06,
      "loss": 0.3341,
      "step": 1173
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.975933854686196e-06,
      "loss": 0.3016,
      "step": 1176
    },
    {
      "epoch": 1.96,
      "learning_rate": 5.925332101903994e-06,
      "loss": 0.333,
      "step": 1179
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.874855146099594e-06,
      "loss": 0.3234,
      "step": 1182
    },
    {
      "epoch": 1.97,
      "learning_rate": 5.824504533257042e-06,
      "loss": 0.3256,
      "step": 1185
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.774281805490798e-06,
      "loss": 0.3102,
      "step": 1188
    },
    {
      "epoch": 1.98,
      "learning_rate": 5.724188500998531e-06,
      "loss": 0.315,
      "step": 1191
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.674226154013988e-06,
      "loss": 0.3095,
      "step": 1194
    },
    {
      "epoch": 1.99,
      "learning_rate": 5.624396294760022e-06,
      "loss": 0.3094,
      "step": 1197
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.574700449401702e-06,
      "loss": 0.3182,
      "step": 1200
    },
    {
      "epoch": 2.0,
      "learning_rate": 5.525140139999588e-06,
      "loss": 0.2879,
      "step": 1203
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.475716884463113e-06,
      "loss": 0.2313,
      "step": 1206
    },
    {
      "epoch": 2.01,
      "learning_rate": 5.42643219650409e-06,
      "loss": 0.2257,
      "step": 1209
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.377287585590352e-06,
      "loss": 0.2328,
      "step": 1212
    },
    {
      "epoch": 2.02,
      "learning_rate": 5.328284556899513e-06,
      "loss": 0.239,
      "step": 1215
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.279424611272873e-06,
      "loss": 0.2301,
      "step": 1218
    },
    {
      "epoch": 2.03,
      "learning_rate": 5.230709245169461e-06,
      "loss": 0.2333,
      "step": 1221
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.182139950620188e-06,
      "loss": 0.2308,
      "step": 1224
    },
    {
      "epoch": 2.04,
      "learning_rate": 5.133718215182162e-06,
      "loss": 0.2562,
      "step": 1227
    },
    {
      "epoch": 2.05,
      "learning_rate": 5.085445521893113e-06,
      "loss": 0.2491,
      "step": 1230
    },
    {
      "epoch": 2.05,
      "learning_rate": 5.037323349225984e-06,
      "loss": 0.2534,
      "step": 1233
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.989353171043637e-06,
      "loss": 0.2193,
      "step": 1236
    },
    {
      "epoch": 2.06,
      "learning_rate": 4.94153645655373e-06,
      "loss": 0.2435,
      "step": 1239
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.893874670263704e-06,
      "loss": 0.2351,
      "step": 1242
    },
    {
      "epoch": 2.07,
      "learning_rate": 4.846369271935932e-06,
      "loss": 0.2399,
      "step": 1245
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.79902171654302e-06,
      "loss": 0.2506,
      "step": 1248
    },
    {
      "epoch": 2.08,
      "learning_rate": 4.7518334542232215e-06,
      "loss": 0.239,
      "step": 1251
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.704805930236044e-06,
      "loss": 0.2435,
      "step": 1254
    },
    {
      "epoch": 2.09,
      "learning_rate": 4.657940584917983e-06,
      "loss": 0.2505,
      "step": 1257
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.611238853638399e-06,
      "loss": 0.2282,
      "step": 1260
    },
    {
      "epoch": 2.1,
      "learning_rate": 4.564702166755565e-06,
      "loss": 0.2302,
      "step": 1263
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.518331949572847e-06,
      "loss": 0.2418,
      "step": 1266
    },
    {
      "epoch": 2.11,
      "learning_rate": 4.472129622295055e-06,
      "loss": 0.2361,
      "step": 1269
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.426096599984954e-06,
      "loss": 0.2533,
      "step": 1272
    },
    {
      "epoch": 2.12,
      "learning_rate": 4.380234292519917e-06,
      "loss": 0.2389,
      "step": 1275
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.33454410454874e-06,
      "loss": 0.2415,
      "step": 1278
    },
    {
      "epoch": 2.13,
      "learning_rate": 4.289027435448633e-06,
      "loss": 0.236,
      "step": 1281
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.243685679282343e-06,
      "loss": 0.2363,
      "step": 1284
    },
    {
      "epoch": 2.14,
      "learning_rate": 4.19852022475547e-06,
      "loss": 0.2401,
      "step": 1287
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.153532455173939e-06,
      "loss": 0.2428,
      "step": 1290
    },
    {
      "epoch": 2.15,
      "learning_rate": 4.1087237484016185e-06,
      "loss": 0.242,
      "step": 1293
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.064095476818133e-06,
      "loss": 0.2352,
      "step": 1296
    },
    {
      "epoch": 2.16,
      "learning_rate": 4.019649007276815e-06,
      "loss": 0.2538,
      "step": 1299
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.9753857010628615e-06,
      "loss": 0.2391,
      "step": 1302
    },
    {
      "epoch": 2.17,
      "learning_rate": 3.931306913851621e-06,
      "loss": 0.236,
      "step": 1305
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.887413995667091e-06,
      "loss": 0.2344,
      "step": 1308
    },
    {
      "epoch": 2.18,
      "learning_rate": 3.843708290840557e-06,
      "loss": 0.2529,
      "step": 1311
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.8001911379694267e-06,
      "loss": 0.2401,
      "step": 1314
    },
    {
      "epoch": 2.19,
      "learning_rate": 3.756863869876224e-06,
      "loss": 0.2384,
      "step": 1317
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.7137278135677714e-06,
      "loss": 0.2271,
      "step": 1320
    },
    {
      "epoch": 2.2,
      "learning_rate": 3.6707842901945546e-06,
      "loss": 0.2273,
      "step": 1323
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.628034615010251e-06,
      "loss": 0.2186,
      "step": 1326
    },
    {
      "epoch": 2.21,
      "learning_rate": 3.58548009733145e-06,
      "loss": 0.2526,
      "step": 1329
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.543122040497542e-06,
      "loss": 0.2402,
      "step": 1332
    },
    {
      "epoch": 2.22,
      "learning_rate": 3.5009617418308208e-06,
      "loss": 0.2327,
      "step": 1335
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.4590004925967245e-06,
      "loss": 0.2333,
      "step": 1338
    },
    {
      "epoch": 2.23,
      "learning_rate": 3.417239577964312e-06,
      "loss": 0.2345,
      "step": 1341
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.375680276966884e-06,
      "loss": 0.2385,
      "step": 1344
    },
    {
      "epoch": 2.24,
      "learning_rate": 3.3343238624628204e-06,
      "loss": 0.2247,
      "step": 1347
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.2931716010965807e-06,
      "loss": 0.2285,
      "step": 1350
    },
    {
      "epoch": 2.25,
      "learning_rate": 3.2522247532599327e-06,
      "loss": 0.251,
      "step": 1353
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.211484573053324e-06,
      "loss": 0.2324,
      "step": 1356
    },
    {
      "epoch": 2.26,
      "learning_rate": 3.1709523082474937e-06,
      "loss": 0.2279,
      "step": 1359
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.130629200245249e-06,
      "loss": 0.2438,
      "step": 1362
    },
    {
      "epoch": 2.27,
      "learning_rate": 3.090516484043432e-06,
      "loss": 0.2315,
      "step": 1365
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.0506153881951183e-06,
      "loss": 0.2468,
      "step": 1368
    },
    {
      "epoch": 2.28,
      "learning_rate": 3.010927134771966e-06,
      "loss": 0.2221,
      "step": 1371
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.971452939326802e-06,
      "loss": 0.2515,
      "step": 1374
    },
    {
      "epoch": 2.29,
      "learning_rate": 2.9321940108563907e-06,
      "loss": 0.2159,
      "step": 1377
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.8931515517644017e-06,
      "loss": 0.2418,
      "step": 1380
    },
    {
      "epoch": 2.3,
      "learning_rate": 2.854326757824576e-06,
      "loss": 0.2311,
      "step": 1383
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.815720818144123e-06,
      "loss": 0.2677,
      "step": 1386
    },
    {
      "epoch": 2.31,
      "learning_rate": 2.7773349151272745e-06,
      "loss": 0.2474,
      "step": 1389
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.7391702244390984e-06,
      "loss": 0.2315,
      "step": 1392
    },
    {
      "epoch": 2.32,
      "learning_rate": 2.7012279149694685e-06,
      "loss": 0.2319,
      "step": 1395
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.663509148797281e-06,
      "loss": 0.2272,
      "step": 1398
    },
    {
      "epoch": 2.33,
      "learning_rate": 2.6260150811548434e-06,
      "loss": 0.2451,
      "step": 1401
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.588746860392516e-06,
      "loss": 0.2426,
      "step": 1404
    },
    {
      "epoch": 2.34,
      "learning_rate": 2.5517056279435183e-06,
      "loss": 0.238,
      "step": 1407
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.514892518288988e-06,
      "loss": 0.2351,
      "step": 1410
    },
    {
      "epoch": 2.35,
      "learning_rate": 2.47830865892323e-06,
      "loss": 0.2278,
      "step": 1413
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.441955170319168e-06,
      "loss": 0.2369,
      "step": 1416
    },
    {
      "epoch": 2.36,
      "learning_rate": 2.405833165894055e-06,
      "loss": 0.2356,
      "step": 1419
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.3699437519753444e-06,
      "loss": 0.2281,
      "step": 1422
    },
    {
      "epoch": 2.37,
      "learning_rate": 2.334288027766828e-06,
      "loss": 0.2253,
      "step": 1425
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2988670853149574e-06,
      "loss": 0.2432,
      "step": 1428
    },
    {
      "epoch": 2.38,
      "learning_rate": 2.2636820094754077e-06,
      "loss": 0.2365,
      "step": 1431
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.2287338778798283e-06,
      "loss": 0.2324,
      "step": 1434
    },
    {
      "epoch": 2.39,
      "learning_rate": 2.1940237609028735e-06,
      "loss": 0.2435,
      "step": 1437
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.159552721629381e-06,
      "loss": 0.2275,
      "step": 1440
    },
    {
      "epoch": 2.4,
      "learning_rate": 2.125321815821846e-06,
      "loss": 0.2411,
      "step": 1443
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0913320918880643e-06,
      "loss": 0.2582,
      "step": 1446
    },
    {
      "epoch": 2.41,
      "learning_rate": 2.0575845908490267e-06,
      "loss": 0.2433,
      "step": 1449
    },
    {
      "epoch": 2.42,
      "learning_rate": 2.0240803463070425e-06,
      "loss": 0.2225,
      "step": 1452
    },
    {
      "epoch": 2.42,
      "learning_rate": 1.990820384414075e-06,
      "loss": 0.2155,
      "step": 1455
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9578057238403124e-06,
      "loss": 0.2397,
      "step": 1458
    },
    {
      "epoch": 2.43,
      "learning_rate": 1.9250373757429753e-06,
      "loss": 0.2453,
      "step": 1461
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8925163437353465e-06,
      "loss": 0.2407,
      "step": 1464
    },
    {
      "epoch": 2.44,
      "learning_rate": 1.8602436238560218e-06,
      "loss": 0.2512,
      "step": 1467
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.828220204538419e-06,
      "loss": 0.2458,
      "step": 1470
    },
    {
      "epoch": 2.45,
      "learning_rate": 1.7964470665804923e-06,
      "loss": 0.2314,
      "step": 1473
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.7649251831147018e-06,
      "loss": 0.2339,
      "step": 1476
    },
    {
      "epoch": 2.46,
      "learning_rate": 1.733655519578209e-06,
      "loss": 0.2383,
      "step": 1479
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.7026390336832933e-06,
      "loss": 0.2307,
      "step": 1482
    },
    {
      "epoch": 2.47,
      "learning_rate": 1.671876675388039e-06,
      "loss": 0.2364,
      "step": 1485
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6413693868672332e-06,
      "loss": 0.2278,
      "step": 1488
    },
    {
      "epoch": 2.48,
      "learning_rate": 1.6111181024834999e-06,
      "loss": 0.2282,
      "step": 1491
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5811237487586973e-06,
      "loss": 0.2436,
      "step": 1494
    },
    {
      "epoch": 2.49,
      "learning_rate": 1.5513872443455358e-06,
      "loss": 0.2206,
      "step": 1497
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.5219094999994322e-06,
      "loss": 0.2405,
      "step": 1500
    },
    {
      "epoch": 2.5,
      "eval_loss": 0.9922696948051453,
      "eval_runtime": 61.3926,
      "eval_samples_per_second": 7.542,
      "eval_steps_per_second": 0.945,
      "step": 1500
    },
    {
      "epoch": 2.5,
      "learning_rate": 1.4926914185506315e-06,
      "loss": 0.2409,
      "step": 1503
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4637338948765467e-06,
      "loss": 0.2313,
      "step": 1506
    },
    {
      "epoch": 2.51,
      "learning_rate": 1.4350378158743484e-06,
      "loss": 0.2355,
      "step": 1509
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.4066040604338104e-06,
      "loss": 0.2309,
      "step": 1512
    },
    {
      "epoch": 2.52,
      "learning_rate": 1.3784334994103787e-06,
      "loss": 0.2225,
      "step": 1515
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.3505269955985122e-06,
      "loss": 0.2376,
      "step": 1518
    },
    {
      "epoch": 2.53,
      "learning_rate": 1.322885403705254e-06,
      "loss": 0.2168,
      "step": 1521
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2955095703240438e-06,
      "loss": 0.2493,
      "step": 1524
    },
    {
      "epoch": 2.54,
      "learning_rate": 1.2684003339088024e-06,
      "loss": 0.2473,
      "step": 1527
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.2415585247482498e-06,
      "loss": 0.2251,
      "step": 1530
    },
    {
      "epoch": 2.55,
      "learning_rate": 1.21498496494046e-06,
      "loss": 0.2266,
      "step": 1533
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1886804683677068e-06,
      "loss": 0.2196,
      "step": 1536
    },
    {
      "epoch": 2.56,
      "learning_rate": 1.1626458406715192e-06,
      "loss": 0.2298,
      "step": 1539
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1368818792280089e-06,
      "loss": 0.2332,
      "step": 1542
    },
    {
      "epoch": 2.57,
      "learning_rate": 1.1113893731234537e-06,
      "loss": 0.2366,
      "step": 1545
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0861691031301313e-06,
      "loss": 0.2327,
      "step": 1548
    },
    {
      "epoch": 2.58,
      "learning_rate": 1.0612218416823926e-06,
      "loss": 0.2244,
      "step": 1551
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.0365483528530206e-06,
      "loss": 0.2349,
      "step": 1554
    },
    {
      "epoch": 2.59,
      "learning_rate": 1.012149392329822e-06,
      "loss": 0.248,
      "step": 1557
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.88025707392475e-07,
      "loss": 0.226,
      "step": 1560
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.641780368896592e-07,
      "loss": 0.2437,
      "step": 1563
    },
    {
      "epoch": 2.6,
      "learning_rate": 9.406071112164061e-07,
      "loss": 0.2188,
      "step": 1566
    },
    {
      "epoch": 2.61,
      "learning_rate": 9.173136522917458e-07,
      "loss": 0.2368,
      "step": 1569
    },
    {
      "epoch": 2.61,
      "learning_rate": 8.942983735365885e-07,
      "loss": 0.2409,
      "step": 1572
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.715619798518715e-07,
      "loss": 0.2313,
      "step": 1575
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.491051675969774e-07,
      "loss": 0.2467,
      "step": 1578
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.269286245684038e-07,
      "loss": 0.2276,
      "step": 1581
    },
    {
      "epoch": 2.63,
      "learning_rate": 8.050330299786913e-07,
      "loss": 0.2377,
      "step": 1584
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.834190544356312e-07,
      "loss": 0.2311,
      "step": 1587
    },
    {
      "epoch": 2.64,
      "learning_rate": 7.620873599217215e-07,
      "loss": 0.2366,
      "step": 1590
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.410385997738867e-07,
      "loss": 0.2264,
      "step": 1593
    },
    {
      "epoch": 2.65,
      "learning_rate": 7.202734186634785e-07,
      "loss": 0.2235,
      "step": 1596
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.997924525765209e-07,
      "loss": 0.2264,
      "step": 1599
    },
    {
      "epoch": 2.66,
      "learning_rate": 6.79596328794242e-07,
      "loss": 0.2267,
      "step": 1602
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.596856658738504e-07,
      "loss": 0.223,
      "step": 1605
    },
    {
      "epoch": 2.67,
      "learning_rate": 6.40061073629602e-07,
      "loss": 0.2284,
      "step": 1608
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.20723153114109e-07,
      "loss": 0.2485,
      "step": 1611
    },
    {
      "epoch": 2.68,
      "learning_rate": 6.016724965999487e-07,
      "loss": 0.2402,
      "step": 1614
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.829096875615048e-07,
      "loss": 0.2168,
      "step": 1617
    },
    {
      "epoch": 2.69,
      "learning_rate": 5.644353006571124e-07,
      "loss": 0.2476,
      "step": 1620
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.462499017114508e-07,
      "loss": 0.2362,
      "step": 1623
    },
    {
      "epoch": 2.7,
      "learning_rate": 5.283540476982108e-07,
      "loss": 0.2446,
      "step": 1626
    },
    {
      "epoch": 2.71,
      "learning_rate": 5.107482867230462e-07,
      "loss": 0.2244,
      "step": 1629
    },
    {
      "epoch": 2.71,
      "learning_rate": 4.934331580067719e-07,
      "loss": 0.2285,
      "step": 1632
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.7640919186886294e-07,
      "loss": 0.2382,
      "step": 1635
    },
    {
      "epoch": 2.72,
      "learning_rate": 4.596769097112042e-07,
      "loss": 0.2369,
      "step": 1638
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.4323682400212675e-07,
      "loss": 0.2276,
      "step": 1641
    },
    {
      "epoch": 2.73,
      "learning_rate": 4.2708943826070157e-07,
      "loss": 0.2217,
      "step": 1644
    },
    {
      "epoch": 2.74,
      "learning_rate": 4.1123524704133276e-07,
      "loss": 0.2245,
      "step": 1647
    },
    {
      "epoch": 2.74,
      "learning_rate": 3.956747359185964e-07,
      "loss": 0.251,
      "step": 1650
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.804083814723791e-07,
      "loss": 0.2263,
      "step": 1653
    },
    {
      "epoch": 2.75,
      "learning_rate": 3.6543665127327765e-07,
      "loss": 0.2521,
      "step": 1656
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.5076000386827903e-07,
      "loss": 0.2253,
      "step": 1659
    },
    {
      "epoch": 2.76,
      "learning_rate": 3.3637888876671076e-07,
      "loss": 0.2405,
      "step": 1662
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.2229374642648193e-07,
      "loss": 0.2062,
      "step": 1665
    },
    {
      "epoch": 2.77,
      "learning_rate": 3.085050082405916e-07,
      "loss": 0.2409,
      "step": 1668
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.9501309652391063e-07,
      "loss": 0.2245,
      "step": 1671
    },
    {
      "epoch": 2.78,
      "learning_rate": 2.818184245002542e-07,
      "loss": 0.2338,
      "step": 1674
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.689213962897186e-07,
      "loss": 0.2306,
      "step": 1677
    },
    {
      "epoch": 2.79,
      "learning_rate": 2.5632240689631216e-07,
      "loss": 0.2255,
      "step": 1680
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.4402184219584735e-07,
      "loss": 0.224,
      "step": 1683
    },
    {
      "epoch": 2.8,
      "learning_rate": 2.3202007892413447e-07,
      "loss": 0.2341,
      "step": 1686
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.2031748466543214e-07,
      "loss": 0.244,
      "step": 1689
    },
    {
      "epoch": 2.81,
      "learning_rate": 2.0891441784119725e-07,
      "loss": 0.2269,
      "step": 1692
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.9781122769910065e-07,
      "loss": 0.2213,
      "step": 1695
    },
    {
      "epoch": 2.82,
      "learning_rate": 1.8700825430233216e-07,
      "loss": 0.2259,
      "step": 1698
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.765058285191923e-07,
      "loss": 0.2355,
      "step": 1701
    },
    {
      "epoch": 2.83,
      "learning_rate": 1.663042720129504e-07,
      "loss": 0.2335,
      "step": 1704
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.5640389723199478e-07,
      "loss": 0.252,
      "step": 1707
    },
    {
      "epoch": 2.84,
      "learning_rate": 1.468050074002658e-07,
      "loss": 0.223,
      "step": 1710
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.375078965079657e-07,
      "loss": 0.2249,
      "step": 1713
    },
    {
      "epoch": 2.85,
      "learning_rate": 1.2851284930255447e-07,
      "loss": 0.249,
      "step": 1716
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1982014128003373e-07,
      "loss": 0.2375,
      "step": 1719
    },
    {
      "epoch": 2.86,
      "learning_rate": 1.1143003867650437e-07,
      "loss": 0.2182,
      "step": 1722
    },
    {
      "epoch": 2.87,
      "learning_rate": 1.0334279846001106e-07,
      "loss": 0.2249,
      "step": 1725
    },
    {
      "epoch": 2.87,
      "learning_rate": 9.555866832267502e-08,
      "loss": 0.2259,
      "step": 1728
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.807788667311023e-08,
      "loss": 0.2455,
      "step": 1731
    },
    {
      "epoch": 2.88,
      "learning_rate": 8.090068262911366e-08,
      "loss": 0.2314,
      "step": 1734
    },
    {
      "epoch": 2.89,
      "learning_rate": 7.402727601065529e-08,
      "loss": 0.2351,
      "step": 1737
    },
    {
      "epoch": 2.89,
      "learning_rate": 6.745787733314469e-08,
      "loss": 0.2203,
      "step": 1740
    },
    {
      "epoch": 2.9,
      "learning_rate": 6.11926878009772e-08,
      "loss": 0.2377,
      "step": 1743
    },
    {
      "epoch": 2.9,
      "learning_rate": 5.5231899301380065e-08,
      "loss": 0.23,
      "step": 1746
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.957569439853038e-08,
      "loss": 0.2136,
      "step": 1749
    },
    {
      "epoch": 2.91,
      "learning_rate": 4.422424632796185e-08,
      "loss": 0.2231,
      "step": 1752
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.917771899126677e-08,
      "loss": 0.2499,
      "step": 1755
    },
    {
      "epoch": 2.92,
      "learning_rate": 3.4436266951067834e-08,
      "loss": 0.2348,
      "step": 1758
    },
    {
      "epoch": 2.93,
      "learning_rate": 3.000003542628749e-08,
      "loss": 0.2341,
      "step": 1761
    },
    {
      "epoch": 2.93,
      "learning_rate": 2.586916028770259e-08,
      "loss": 0.2424,
      "step": 1764
    },
    {
      "epoch": 2.94,
      "learning_rate": 2.2043768053777726e-08,
      "loss": 0.2495,
      "step": 1767
    },
    {
      "epoch": 2.94,
      "learning_rate": 1.852397588679722e-08,
      "loss": 0.2498,
      "step": 1770
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.5309891589267988e-08,
      "loss": 0.2203,
      "step": 1773
    },
    {
      "epoch": 2.95,
      "learning_rate": 1.2401613600627749e-08,
      "loss": 0.2216,
      "step": 1776
    },
    {
      "epoch": 2.96,
      "learning_rate": 9.799230994220754e-09,
      "loss": 0.2279,
      "step": 1779
    },
    {
      "epoch": 2.96,
      "learning_rate": 7.502823474576648e-09,
      "loss": 0.2379,
      "step": 1782
    },
    {
      "epoch": 2.97,
      "learning_rate": 5.512461374964639e-09,
      "loss": 0.2306,
      "step": 1785
    },
    {
      "epoch": 2.97,
      "learning_rate": 3.828205655244111e-09,
      "loss": 0.2546,
      "step": 1788
    },
    {
      "epoch": 2.98,
      "learning_rate": 2.4501078999916716e-09,
      "loss": 0.2396,
      "step": 1791
    },
    {
      "epoch": 2.98,
      "learning_rate": 1.3782103169246441e-09,
      "loss": 0.2226,
      "step": 1794
    },
    {
      "epoch": 2.99,
      "learning_rate": 6.125457356120912e-10,
      "loss": 0.2357,
      "step": 1797
    },
    {
      "epoch": 2.99,
      "learning_rate": 1.5313760645896403e-10,
      "loss": 0.2264,
      "step": 1800
    },
    {
      "epoch": 3.0,
      "learning_rate": 0.0,
      "loss": 0.236,
      "step": 1803
    },
    {
      "epoch": 3.0,
      "step": 1803,
      "total_flos": 1.9808467849371766e+20,
      "train_loss": 0.11453570330466949,
      "train_runtime": 197727.0226,
      "train_samples_per_second": 7.005,
      "train_steps_per_second": 0.009
    }
  ],
  "logging_steps": 3,
  "max_steps": 1803,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 500,
  "total_flos": 1.9808467849371766e+20,
  "train_batch_size": 96,
  "trial_name": null,
  "trial_params": null
}