{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.7790697674418605,
  "global_step": 1300,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.01,
      "eval_loss": 2.1321089267730713,
      "eval_runtime": 5.3447,
      "eval_samples_per_second": 7.11,
      "eval_steps_per_second": 1.871,
      "step": 5
    },
    {
      "epoch": 0.03,
      "eval_loss": 1.7755846977233887,
      "eval_runtime": 6.2618,
      "eval_samples_per_second": 6.069,
      "eval_steps_per_second": 1.597,
      "step": 10
    },
    {
      "epoch": 0.04,
      "eval_loss": 1.661117672920227,
      "eval_runtime": 5.1457,
      "eval_samples_per_second": 7.385,
      "eval_steps_per_second": 1.943,
      "step": 15
    },
    {
      "epoch": 0.06,
      "eval_loss": 1.6071234941482544,
      "eval_runtime": 6.3033,
      "eval_samples_per_second": 6.029,
      "eval_steps_per_second": 1.586,
      "step": 20
    },
    {
      "epoch": 0.07,
      "eval_loss": 1.5763286352157593,
      "eval_runtime": 5.1664,
      "eval_samples_per_second": 7.355,
      "eval_steps_per_second": 1.936,
      "step": 25
    },
    {
      "epoch": 0.09,
      "eval_loss": 1.54438316822052,
      "eval_runtime": 12.625,
      "eval_samples_per_second": 3.01,
      "eval_steps_per_second": 0.792,
      "step": 30
    },
    {
      "epoch": 0.1,
      "eval_loss": 1.529874563217163,
      "eval_runtime": 5.1855,
      "eval_samples_per_second": 7.328,
      "eval_steps_per_second": 1.928,
      "step": 35
    },
    {
      "epoch": 0.12,
      "eval_loss": 1.511551856994629,
      "eval_runtime": 5.1943,
      "eval_samples_per_second": 7.316,
      "eval_steps_per_second": 1.925,
      "step": 40
    },
    {
      "epoch": 0.13,
      "eval_loss": 1.499739170074463,
      "eval_runtime": 5.1845,
      "eval_samples_per_second": 7.33,
      "eval_steps_per_second": 1.929,
      "step": 45
    },
    {
      "epoch": 0.15,
      "eval_loss": 1.482990026473999,
      "eval_runtime": 5.191,
      "eval_samples_per_second": 7.32,
      "eval_steps_per_second": 1.926,
      "step": 50
    },
    {
      "epoch": 0.16,
      "eval_loss": 1.4811052083969116,
      "eval_runtime": 5.2095,
      "eval_samples_per_second": 7.294,
      "eval_steps_per_second": 1.92,
      "step": 55
    },
    {
      "epoch": 0.17,
      "eval_loss": 1.470807671546936,
      "eval_runtime": 5.1982,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 60
    },
    {
      "epoch": 0.19,
      "eval_loss": 1.4635131359100342,
      "eval_runtime": 5.22,
      "eval_samples_per_second": 7.28,
      "eval_steps_per_second": 1.916,
      "step": 65
    },
    {
      "epoch": 0.2,
      "eval_loss": 1.4525114297866821,
      "eval_runtime": 5.22,
      "eval_samples_per_second": 7.28,
      "eval_steps_per_second": 1.916,
      "step": 70
    },
    {
      "epoch": 0.22,
      "eval_loss": 1.4475783109664917,
      "eval_runtime": 5.2098,
      "eval_samples_per_second": 7.294,
      "eval_steps_per_second": 1.919,
      "step": 75
    },
    {
      "epoch": 0.23,
      "eval_loss": 1.4456068277359009,
      "eval_runtime": 5.2045,
      "eval_samples_per_second": 7.301,
      "eval_steps_per_second": 1.921,
      "step": 80
    },
    {
      "epoch": 0.25,
      "eval_loss": 1.4422187805175781,
      "eval_runtime": 5.2064,
      "eval_samples_per_second": 7.299,
      "eval_steps_per_second": 1.921,
      "step": 85
    },
    {
      "epoch": 0.26,
      "eval_loss": 1.4375722408294678,
      "eval_runtime": 5.2062,
      "eval_samples_per_second": 7.299,
      "eval_steps_per_second": 1.921,
      "step": 90
    },
    {
      "epoch": 0.28,
      "eval_loss": 1.435781717300415,
      "eval_runtime": 5.2273,
      "eval_samples_per_second": 7.27,
      "eval_steps_per_second": 1.913,
      "step": 95
    },
    {
      "epoch": 0.29,
      "learning_rate": 9.854651162790698e-06,
      "loss": 1.5857,
      "step": 100
    },
    {
      "epoch": 0.29,
      "eval_loss": 1.4250125885009766,
      "eval_runtime": 5.2302,
      "eval_samples_per_second": 7.266,
      "eval_steps_per_second": 1.912,
      "step": 100
    },
    {
      "epoch": 0.31,
      "eval_loss": 1.4228482246398926,
      "eval_runtime": 5.1419,
      "eval_samples_per_second": 7.39,
      "eval_steps_per_second": 1.945,
      "step": 105
    },
    {
      "epoch": 0.32,
      "eval_loss": 1.41592538356781,
      "eval_runtime": 5.1705,
      "eval_samples_per_second": 7.349,
      "eval_steps_per_second": 1.934,
      "step": 110
    },
    {
      "epoch": 0.33,
      "eval_loss": 1.4095944166183472,
      "eval_runtime": 5.1762,
      "eval_samples_per_second": 7.341,
      "eval_steps_per_second": 1.932,
      "step": 115
    },
    {
      "epoch": 0.35,
      "eval_loss": 1.405040979385376,
      "eval_runtime": 5.1734,
      "eval_samples_per_second": 7.345,
      "eval_steps_per_second": 1.933,
      "step": 120
    },
    {
      "epoch": 0.36,
      "eval_loss": 1.4017434120178223,
      "eval_runtime": 5.1936,
      "eval_samples_per_second": 7.317,
      "eval_steps_per_second": 1.925,
      "step": 125
    },
    {
      "epoch": 0.38,
      "eval_loss": 1.3985930681228638,
      "eval_runtime": 5.1856,
      "eval_samples_per_second": 7.328,
      "eval_steps_per_second": 1.928,
      "step": 130
    },
    {
      "epoch": 0.39,
      "eval_loss": 1.3949148654937744,
      "eval_runtime": 6.2986,
      "eval_samples_per_second": 6.033,
      "eval_steps_per_second": 1.588,
      "step": 135
    },
    {
      "epoch": 0.41,
      "eval_loss": 1.3919830322265625,
      "eval_runtime": 5.193,
      "eval_samples_per_second": 7.318,
      "eval_steps_per_second": 1.926,
      "step": 140
    },
    {
      "epoch": 0.42,
      "eval_loss": 1.389172077178955,
      "eval_runtime": 5.1972,
      "eval_samples_per_second": 7.312,
      "eval_steps_per_second": 1.924,
      "step": 145
    },
    {
      "epoch": 0.44,
      "eval_loss": 1.3876895904541016,
      "eval_runtime": 5.215,
      "eval_samples_per_second": 7.287,
      "eval_steps_per_second": 1.918,
      "step": 150
    },
    {
      "epoch": 0.45,
      "eval_loss": 1.387901782989502,
      "eval_runtime": 5.2122,
      "eval_samples_per_second": 7.291,
      "eval_steps_per_second": 1.919,
      "step": 155
    },
    {
      "epoch": 0.47,
      "eval_loss": 1.3818577527999878,
      "eval_runtime": 5.2068,
      "eval_samples_per_second": 7.298,
      "eval_steps_per_second": 1.921,
      "step": 160
    },
    {
      "epoch": 0.48,
      "eval_loss": 1.3780750036239624,
      "eval_runtime": 5.2225,
      "eval_samples_per_second": 7.276,
      "eval_steps_per_second": 1.915,
      "step": 165
    },
    {
      "epoch": 0.49,
      "eval_loss": 1.374315857887268,
      "eval_runtime": 5.2084,
      "eval_samples_per_second": 7.296,
      "eval_steps_per_second": 1.92,
      "step": 170
    },
    {
      "epoch": 0.51,
      "eval_loss": 1.372006893157959,
      "eval_runtime": 5.209,
      "eval_samples_per_second": 7.295,
      "eval_steps_per_second": 1.92,
      "step": 175
    },
    {
      "epoch": 0.52,
      "eval_loss": 1.3736425638198853,
      "eval_runtime": 5.2311,
      "eval_samples_per_second": 7.264,
      "eval_steps_per_second": 1.912,
      "step": 180
    },
    {
      "epoch": 0.54,
      "eval_loss": 1.3681703805923462,
      "eval_runtime": 5.2032,
      "eval_samples_per_second": 7.303,
      "eval_steps_per_second": 1.922,
      "step": 185
    },
    {
      "epoch": 0.55,
      "eval_loss": 1.3626172542572021,
      "eval_runtime": 5.2168,
      "eval_samples_per_second": 7.284,
      "eval_steps_per_second": 1.917,
      "step": 190
    },
    {
      "epoch": 0.57,
      "eval_loss": 1.3621835708618164,
      "eval_runtime": 5.2269,
      "eval_samples_per_second": 7.27,
      "eval_steps_per_second": 1.913,
      "step": 195
    },
    {
      "epoch": 0.58,
      "learning_rate": 9.709302325581395e-06,
      "loss": 1.4034,
      "step": 200
    },
    {
      "epoch": 0.58,
      "eval_loss": 1.364233374595642,
      "eval_runtime": 5.3231,
      "eval_samples_per_second": 7.139,
      "eval_steps_per_second": 1.879,
      "step": 200
    },
    {
      "epoch": 0.6,
      "eval_loss": 1.3565268516540527,
      "eval_runtime": 5.1533,
      "eval_samples_per_second": 7.374,
      "eval_steps_per_second": 1.941,
      "step": 205
    },
    {
      "epoch": 0.61,
      "eval_loss": 1.35232412815094,
      "eval_runtime": 14.9101,
      "eval_samples_per_second": 2.549,
      "eval_steps_per_second": 0.671,
      "step": 210
    },
    {
      "epoch": 0.62,
      "eval_loss": 1.3503801822662354,
      "eval_runtime": 5.1561,
      "eval_samples_per_second": 7.37,
      "eval_steps_per_second": 1.939,
      "step": 215
    },
    {
      "epoch": 0.64,
      "eval_loss": 1.3449465036392212,
      "eval_runtime": 5.1639,
      "eval_samples_per_second": 7.359,
      "eval_steps_per_second": 1.937,
      "step": 220
    },
    {
      "epoch": 0.65,
      "eval_loss": 1.3519093990325928,
      "eval_runtime": 5.1765,
      "eval_samples_per_second": 7.341,
      "eval_steps_per_second": 1.932,
      "step": 225
    },
    {
      "epoch": 0.67,
      "eval_loss": 1.352266550064087,
      "eval_runtime": 17.2964,
      "eval_samples_per_second": 2.197,
      "eval_steps_per_second": 0.578,
      "step": 230
    },
    {
      "epoch": 0.68,
      "eval_loss": 1.3380497694015503,
      "eval_runtime": 5.1906,
      "eval_samples_per_second": 7.321,
      "eval_steps_per_second": 1.927,
      "step": 235
    },
    {
      "epoch": 0.7,
      "eval_loss": 1.340211033821106,
      "eval_runtime": 6.4362,
      "eval_samples_per_second": 5.904,
      "eval_steps_per_second": 1.554,
      "step": 240
    },
    {
      "epoch": 0.71,
      "eval_loss": 1.335599660873413,
      "eval_runtime": 5.1978,
      "eval_samples_per_second": 7.311,
      "eval_steps_per_second": 1.924,
      "step": 245
    },
    {
      "epoch": 0.73,
      "eval_loss": 1.3313099145889282,
      "eval_runtime": 5.1628,
      "eval_samples_per_second": 7.36,
      "eval_steps_per_second": 1.937,
      "step": 250
    },
    {
      "epoch": 0.74,
      "eval_loss": 1.334537386894226,
      "eval_runtime": 7.3524,
      "eval_samples_per_second": 5.168,
      "eval_steps_per_second": 1.36,
      "step": 255
    },
    {
      "epoch": 0.76,
      "eval_loss": 1.3304086923599243,
      "eval_runtime": 5.1618,
      "eval_samples_per_second": 7.362,
      "eval_steps_per_second": 1.937,
      "step": 260
    },
    {
      "epoch": 0.77,
      "eval_loss": 1.3271808624267578,
      "eval_runtime": 5.1719,
      "eval_samples_per_second": 7.347,
      "eval_steps_per_second": 1.934,
      "step": 265
    },
    {
      "epoch": 0.78,
      "eval_loss": 1.324944019317627,
      "eval_runtime": 5.1801,
      "eval_samples_per_second": 7.336,
      "eval_steps_per_second": 1.93,
      "step": 270
    },
    {
      "epoch": 0.8,
      "eval_loss": 1.3218427896499634,
      "eval_runtime": 5.1851,
      "eval_samples_per_second": 7.329,
      "eval_steps_per_second": 1.929,
      "step": 275
    },
    {
      "epoch": 0.81,
      "eval_loss": 1.3217554092407227,
      "eval_runtime": 5.2075,
      "eval_samples_per_second": 7.297,
      "eval_steps_per_second": 1.92,
      "step": 280
    },
    {
      "epoch": 0.83,
      "eval_loss": 1.3211233615875244,
      "eval_runtime": 5.198,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 285
    },
    {
      "epoch": 0.84,
      "eval_loss": 1.3166344165802002,
      "eval_runtime": 6.1478,
      "eval_samples_per_second": 6.181,
      "eval_steps_per_second": 1.627,
      "step": 290
    },
    {
      "epoch": 0.86,
      "eval_loss": 1.3159008026123047,
      "eval_runtime": 5.2103,
      "eval_samples_per_second": 7.293,
      "eval_steps_per_second": 1.919,
      "step": 295
    },
    {
      "epoch": 0.87,
      "learning_rate": 9.563953488372094e-06,
      "loss": 1.3513,
      "step": 300
    },
    {
      "epoch": 0.87,
      "eval_loss": 1.3075319528579712,
      "eval_runtime": 5.2012,
      "eval_samples_per_second": 7.306,
      "eval_steps_per_second": 1.923,
      "step": 300
    },
    {
      "epoch": 0.89,
      "eval_loss": 1.3096569776535034,
      "eval_runtime": 5.144,
      "eval_samples_per_second": 7.387,
      "eval_steps_per_second": 1.944,
      "step": 305
    },
    {
      "epoch": 0.9,
      "eval_loss": 1.3026896715164185,
      "eval_runtime": 5.1657,
      "eval_samples_per_second": 7.356,
      "eval_steps_per_second": 1.936,
      "step": 310
    },
    {
      "epoch": 0.92,
      "eval_loss": 1.2962864637374878,
      "eval_runtime": 5.1706,
      "eval_samples_per_second": 7.349,
      "eval_steps_per_second": 1.934,
      "step": 315
    },
    {
      "epoch": 0.93,
      "eval_loss": 1.294777512550354,
      "eval_runtime": 5.1816,
      "eval_samples_per_second": 7.334,
      "eval_steps_per_second": 1.93,
      "step": 320
    },
    {
      "epoch": 0.94,
      "eval_loss": 1.2898850440979004,
      "eval_runtime": 6.0458,
      "eval_samples_per_second": 6.285,
      "eval_steps_per_second": 1.654,
      "step": 325
    },
    {
      "epoch": 0.96,
      "eval_loss": 1.2885679006576538,
      "eval_runtime": 5.1898,
      "eval_samples_per_second": 7.322,
      "eval_steps_per_second": 1.927,
      "step": 330
    },
    {
      "epoch": 0.97,
      "eval_loss": 1.288139820098877,
      "eval_runtime": 5.1979,
      "eval_samples_per_second": 7.311,
      "eval_steps_per_second": 1.924,
      "step": 335
    },
    {
      "epoch": 0.99,
      "eval_loss": 1.281440258026123,
      "eval_runtime": 5.1891,
      "eval_samples_per_second": 7.323,
      "eval_steps_per_second": 1.927,
      "step": 340
    },
    {
      "epoch": 1.0,
      "eval_loss": 1.2789274454116821,
      "eval_runtime": 5.2144,
      "eval_samples_per_second": 7.288,
      "eval_steps_per_second": 1.918,
      "step": 345
    },
    {
      "epoch": 1.02,
      "eval_loss": 1.273104190826416,
      "eval_runtime": 5.1983,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 350
    },
    {
      "epoch": 1.03,
      "eval_loss": 1.2735443115234375,
      "eval_runtime": 5.2215,
      "eval_samples_per_second": 7.278,
      "eval_steps_per_second": 1.915,
      "step": 355
    },
    {
      "epoch": 1.05,
      "eval_loss": 1.2782071828842163,
      "eval_runtime": 5.6853,
      "eval_samples_per_second": 6.684,
      "eval_steps_per_second": 1.759,
      "step": 360
    },
    {
      "epoch": 1.06,
      "eval_loss": 1.2720850706100464,
      "eval_runtime": 5.222,
      "eval_samples_per_second": 7.277,
      "eval_steps_per_second": 1.915,
      "step": 365
    },
    {
      "epoch": 1.08,
      "eval_loss": 1.2730863094329834,
      "eval_runtime": 5.2055,
      "eval_samples_per_second": 7.3,
      "eval_steps_per_second": 1.921,
      "step": 370
    },
    {
      "epoch": 1.09,
      "eval_loss": 1.2692456245422363,
      "eval_runtime": 7.9591,
      "eval_samples_per_second": 4.774,
      "eval_steps_per_second": 1.256,
      "step": 375
    },
    {
      "epoch": 1.1,
      "eval_loss": 1.2623660564422607,
      "eval_runtime": 5.1922,
      "eval_samples_per_second": 7.319,
      "eval_steps_per_second": 1.926,
      "step": 380
    },
    {
      "epoch": 1.12,
      "eval_loss": 1.264005422592163,
      "eval_runtime": 5.2073,
      "eval_samples_per_second": 7.297,
      "eval_steps_per_second": 1.92,
      "step": 385
    },
    {
      "epoch": 1.13,
      "eval_loss": 1.264898419380188,
      "eval_runtime": 5.1964,
      "eval_samples_per_second": 7.313,
      "eval_steps_per_second": 1.924,
      "step": 390
    },
    {
      "epoch": 1.15,
      "eval_loss": 1.2602505683898926,
      "eval_runtime": 5.2183,
      "eval_samples_per_second": 7.282,
      "eval_steps_per_second": 1.916,
      "step": 395
    },
    {
      "epoch": 1.16,
      "learning_rate": 9.418604651162791e-06,
      "loss": 1.2432,
      "step": 400
    },
    {
      "epoch": 1.16,
      "eval_loss": 1.2597508430480957,
      "eval_runtime": 5.2008,
      "eval_samples_per_second": 7.307,
      "eval_steps_per_second": 1.923,
      "step": 400
    },
    {
      "epoch": 1.18,
      "eval_loss": 1.256094217300415,
      "eval_runtime": 5.1612,
      "eval_samples_per_second": 7.363,
      "eval_steps_per_second": 1.938,
      "step": 405
    },
    {
      "epoch": 1.19,
      "eval_loss": 1.2484766244888306,
      "eval_runtime": 5.1459,
      "eval_samples_per_second": 7.384,
      "eval_steps_per_second": 1.943,
      "step": 410
    },
    {
      "epoch": 1.21,
      "eval_loss": 1.2463735342025757,
      "eval_runtime": 5.1578,
      "eval_samples_per_second": 7.367,
      "eval_steps_per_second": 1.939,
      "step": 415
    },
    {
      "epoch": 1.22,
      "eval_loss": 1.244008183479309,
      "eval_runtime": 5.1699,
      "eval_samples_per_second": 7.35,
      "eval_steps_per_second": 1.934,
      "step": 420
    },
    {
      "epoch": 1.24,
      "eval_loss": 1.2463507652282715,
      "eval_runtime": 5.5355,
      "eval_samples_per_second": 6.865,
      "eval_steps_per_second": 1.807,
      "step": 425
    },
    {
      "epoch": 1.25,
      "eval_loss": 1.23945152759552,
      "eval_runtime": 5.1986,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 430
    },
    {
      "epoch": 1.26,
      "eval_loss": 1.2383465766906738,
      "eval_runtime": 5.2615,
      "eval_samples_per_second": 7.222,
      "eval_steps_per_second": 1.901,
      "step": 435
    },
    {
      "epoch": 1.28,
      "eval_loss": 1.240387201309204,
      "eval_runtime": 5.2049,
      "eval_samples_per_second": 7.301,
      "eval_steps_per_second": 1.921,
      "step": 440
    },
    {
      "epoch": 1.29,
      "eval_loss": 1.2330259084701538,
      "eval_runtime": 5.1959,
      "eval_samples_per_second": 7.313,
      "eval_steps_per_second": 1.925,
      "step": 445
    },
    {
      "epoch": 1.31,
      "eval_loss": 1.2376052141189575,
      "eval_runtime": 5.1991,
      "eval_samples_per_second": 7.309,
      "eval_steps_per_second": 1.923,
      "step": 450
    },
    {
      "epoch": 1.32,
      "eval_loss": 1.2420060634613037,
      "eval_runtime": 14.5358,
      "eval_samples_per_second": 2.614,
      "eval_steps_per_second": 0.688,
      "step": 455
    },
    {
      "epoch": 1.34,
      "eval_loss": 1.2296521663665771,
      "eval_runtime": 7.0505,
      "eval_samples_per_second": 5.39,
      "eval_steps_per_second": 1.418,
      "step": 460
    },
    {
      "epoch": 1.35,
      "eval_loss": 1.226386547088623,
      "eval_runtime": 5.1853,
      "eval_samples_per_second": 7.328,
      "eval_steps_per_second": 1.929,
      "step": 465
    },
    {
      "epoch": 1.37,
      "eval_loss": 1.2350796461105347,
      "eval_runtime": 5.1891,
      "eval_samples_per_second": 7.323,
      "eval_steps_per_second": 1.927,
      "step": 470
    },
    {
      "epoch": 1.38,
      "eval_loss": 1.2271411418914795,
      "eval_runtime": 5.187,
      "eval_samples_per_second": 7.326,
      "eval_steps_per_second": 1.928,
      "step": 475
    },
    {
      "epoch": 1.4,
      "eval_loss": 1.2260706424713135,
      "eval_runtime": 5.1939,
      "eval_samples_per_second": 7.316,
      "eval_steps_per_second": 1.925,
      "step": 480
    },
    {
      "epoch": 1.41,
      "eval_loss": 1.2366853952407837,
      "eval_runtime": 5.2167,
      "eval_samples_per_second": 7.284,
      "eval_steps_per_second": 1.917,
      "step": 485
    },
    {
      "epoch": 1.42,
      "eval_loss": 1.2304943799972534,
      "eval_runtime": 5.2057,
      "eval_samples_per_second": 7.3,
      "eval_steps_per_second": 1.921,
      "step": 490
    },
    {
      "epoch": 1.44,
      "eval_loss": 1.219275951385498,
      "eval_runtime": 5.1983,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 495
    },
    {
      "epoch": 1.45,
      "learning_rate": 9.273255813953488e-06,
      "loss": 1.1933,
      "step": 500
    },
    {
      "epoch": 1.45,
      "eval_loss": 1.220719575881958,
      "eval_runtime": 8.1461,
      "eval_samples_per_second": 4.665,
      "eval_steps_per_second": 1.228,
      "step": 500
    },
    {
      "epoch": 1.47,
      "eval_loss": 1.2208986282348633,
      "eval_runtime": 5.1444,
      "eval_samples_per_second": 7.387,
      "eval_steps_per_second": 1.944,
      "step": 505
    },
    {
      "epoch": 1.48,
      "eval_loss": 1.2238763570785522,
      "eval_runtime": 5.1637,
      "eval_samples_per_second": 7.359,
      "eval_steps_per_second": 1.937,
      "step": 510
    },
    {
      "epoch": 1.5,
      "eval_loss": 1.219651699066162,
      "eval_runtime": 5.177,
      "eval_samples_per_second": 7.34,
      "eval_steps_per_second": 1.932,
      "step": 515
    },
    {
      "epoch": 1.51,
      "eval_loss": 1.2128971815109253,
      "eval_runtime": 5.1833,
      "eval_samples_per_second": 7.331,
      "eval_steps_per_second": 1.929,
      "step": 520
    },
    {
      "epoch": 1.53,
      "eval_loss": 1.2134791612625122,
      "eval_runtime": 6.8202,
      "eval_samples_per_second": 5.572,
      "eval_steps_per_second": 1.466,
      "step": 525
    },
    {
      "epoch": 1.54,
      "eval_loss": 1.2144546508789062,
      "eval_runtime": 5.1795,
      "eval_samples_per_second": 7.337,
      "eval_steps_per_second": 1.931,
      "step": 530
    },
    {
      "epoch": 1.56,
      "eval_loss": 1.2091519832611084,
      "eval_runtime": 5.1715,
      "eval_samples_per_second": 7.348,
      "eval_steps_per_second": 1.934,
      "step": 535
    },
    {
      "epoch": 1.57,
      "eval_loss": 1.2074944972991943,
      "eval_runtime": 5.1783,
      "eval_samples_per_second": 7.338,
      "eval_steps_per_second": 1.931,
      "step": 540
    },
    {
      "epoch": 1.58,
      "eval_loss": 1.2082455158233643,
      "eval_runtime": 5.1756,
      "eval_samples_per_second": 7.342,
      "eval_steps_per_second": 1.932,
      "step": 545
    },
    {
      "epoch": 1.6,
      "eval_loss": 1.2062183618545532,
      "eval_runtime": 5.183,
      "eval_samples_per_second": 7.332,
      "eval_steps_per_second": 1.929,
      "step": 550
    },
    {
      "epoch": 1.61,
      "eval_loss": 1.2009626626968384,
      "eval_runtime": 5.6305,
      "eval_samples_per_second": 6.749,
      "eval_steps_per_second": 1.776,
      "step": 555
    },
    {
      "epoch": 1.63,
      "eval_loss": 1.200234293937683,
      "eval_runtime": 5.2064,
      "eval_samples_per_second": 7.299,
      "eval_steps_per_second": 1.921,
      "step": 560
    },
    {
      "epoch": 1.64,
      "eval_loss": 1.1972352266311646,
      "eval_runtime": 5.3956,
      "eval_samples_per_second": 7.043,
      "eval_steps_per_second": 1.853,
      "step": 565
    },
    {
      "epoch": 1.66,
      "eval_loss": 1.199525237083435,
      "eval_runtime": 5.2172,
      "eval_samples_per_second": 7.284,
      "eval_steps_per_second": 1.917,
      "step": 570
    },
    {
      "epoch": 1.67,
      "eval_loss": 1.1978554725646973,
      "eval_runtime": 5.5745,
      "eval_samples_per_second": 6.817,
      "eval_steps_per_second": 1.794,
      "step": 575
    },
    {
      "epoch": 1.69,
      "eval_loss": 1.1939880847930908,
      "eval_runtime": 5.2146,
      "eval_samples_per_second": 7.287,
      "eval_steps_per_second": 1.918,
      "step": 580
    },
    {
      "epoch": 1.7,
      "eval_loss": 1.1886717081069946,
      "eval_runtime": 5.2166,
      "eval_samples_per_second": 7.284,
      "eval_steps_per_second": 1.917,
      "step": 585
    },
    {
      "epoch": 1.72,
      "eval_loss": 1.1879463195800781,
      "eval_runtime": 5.2066,
      "eval_samples_per_second": 7.298,
      "eval_steps_per_second": 1.921,
      "step": 590
    },
    {
      "epoch": 1.73,
      "eval_loss": 1.1874936819076538,
      "eval_runtime": 5.2058,
      "eval_samples_per_second": 7.3,
      "eval_steps_per_second": 1.921,
      "step": 595
    },
    {
      "epoch": 1.74,
      "learning_rate": 9.127906976744186e-06,
      "loss": 1.1598,
      "step": 600
    },
    {
      "epoch": 1.74,
      "eval_loss": 1.184380054473877,
      "eval_runtime": 5.2949,
      "eval_samples_per_second": 7.177,
      "eval_steps_per_second": 1.889,
      "step": 600
    },
    {
      "epoch": 1.76,
      "eval_loss": 1.1831018924713135,
      "eval_runtime": 5.1436,
      "eval_samples_per_second": 7.388,
      "eval_steps_per_second": 1.944,
      "step": 605
    },
    {
      "epoch": 1.77,
      "eval_loss": 1.1859960556030273,
      "eval_runtime": 5.1652,
      "eval_samples_per_second": 7.357,
      "eval_steps_per_second": 1.936,
      "step": 610
    },
    {
      "epoch": 1.79,
      "eval_loss": 1.186689853668213,
      "eval_runtime": 5.1749,
      "eval_samples_per_second": 7.343,
      "eval_steps_per_second": 1.932,
      "step": 615
    },
    {
      "epoch": 1.8,
      "eval_loss": 1.1849379539489746,
      "eval_runtime": 5.8205,
      "eval_samples_per_second": 6.529,
      "eval_steps_per_second": 1.718,
      "step": 620
    },
    {
      "epoch": 1.82,
      "eval_loss": 1.1795214414596558,
      "eval_runtime": 5.1765,
      "eval_samples_per_second": 7.341,
      "eval_steps_per_second": 1.932,
      "step": 625
    },
    {
      "epoch": 1.83,
      "eval_loss": 1.1802735328674316,
      "eval_runtime": 7.6116,
      "eval_samples_per_second": 4.992,
      "eval_steps_per_second": 1.314,
      "step": 630
    },
    {
      "epoch": 1.85,
      "eval_loss": 1.179026484489441,
      "eval_runtime": 5.2022,
      "eval_samples_per_second": 7.305,
      "eval_steps_per_second": 1.922,
      "step": 635
    },
    {
      "epoch": 1.86,
      "eval_loss": 1.1783987283706665,
      "eval_runtime": 5.1881,
      "eval_samples_per_second": 7.324,
      "eval_steps_per_second": 1.927,
      "step": 640
    },
    {
      "epoch": 1.88,
      "eval_loss": 1.1762430667877197,
      "eval_runtime": 5.2146,
      "eval_samples_per_second": 7.287,
      "eval_steps_per_second": 1.918,
      "step": 645
    },
    {
      "epoch": 1.89,
      "eval_loss": 1.180951476097107,
      "eval_runtime": 5.1987,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 650
    },
    {
      "epoch": 1.9,
      "eval_loss": 1.175215244293213,
      "eval_runtime": 5.222,
      "eval_samples_per_second": 7.277,
      "eval_steps_per_second": 1.915,
      "step": 655
    },
    {
      "epoch": 1.92,
      "eval_loss": 1.1738007068634033,
      "eval_runtime": 5.2057,
      "eval_samples_per_second": 7.3,
      "eval_steps_per_second": 1.921,
      "step": 660
    },
    {
      "epoch": 1.93,
      "eval_loss": 1.1722891330718994,
      "eval_runtime": 5.1993,
      "eval_samples_per_second": 7.309,
      "eval_steps_per_second": 1.923,
      "step": 665
    },
    {
      "epoch": 1.95,
      "eval_loss": 1.1736668348312378,
      "eval_runtime": 5.202,
      "eval_samples_per_second": 7.305,
      "eval_steps_per_second": 1.922,
      "step": 670
    },
    {
      "epoch": 1.96,
      "eval_loss": 1.1691397428512573,
      "eval_runtime": 5.2043,
      "eval_samples_per_second": 7.302,
      "eval_steps_per_second": 1.922,
      "step": 675
    },
    {
      "epoch": 1.98,
      "eval_loss": 1.1707478761672974,
      "eval_runtime": 5.2058,
      "eval_samples_per_second": 7.3,
      "eval_steps_per_second": 1.921,
      "step": 680
    },
    {
      "epoch": 1.99,
      "eval_loss": 1.17750084400177,
      "eval_runtime": 5.2084,
      "eval_samples_per_second": 7.296,
      "eval_steps_per_second": 1.92,
      "step": 685
    },
    {
      "epoch": 2.01,
      "eval_loss": 1.1693485975265503,
      "eval_runtime": 5.2105,
      "eval_samples_per_second": 7.293,
      "eval_steps_per_second": 1.919,
      "step": 690
    },
    {
      "epoch": 2.02,
      "eval_loss": 1.174012541770935,
      "eval_runtime": 8.4066,
      "eval_samples_per_second": 4.52,
      "eval_steps_per_second": 1.19,
      "step": 695
    },
    {
      "epoch": 2.03,
      "learning_rate": 8.982558139534884e-06,
      "loss": 1.1203,
      "step": 700
    },
    {
      "epoch": 2.03,
      "eval_loss": 1.1877881288528442,
      "eval_runtime": 5.1939,
      "eval_samples_per_second": 7.316,
      "eval_steps_per_second": 1.925,
      "step": 700
    },
    {
      "epoch": 2.05,
      "eval_loss": 1.1876509189605713,
      "eval_runtime": 5.1556,
      "eval_samples_per_second": 7.371,
      "eval_steps_per_second": 1.94,
      "step": 705
    },
    {
      "epoch": 2.06,
      "eval_loss": 1.1674307584762573,
      "eval_runtime": 5.149,
      "eval_samples_per_second": 7.38,
      "eval_steps_per_second": 1.942,
      "step": 710
    },
    {
      "epoch": 2.08,
      "eval_loss": 1.167423129081726,
      "eval_runtime": 5.1595,
      "eval_samples_per_second": 7.365,
      "eval_steps_per_second": 1.938,
      "step": 715
    },
    {
      "epoch": 2.09,
      "eval_loss": 1.177182674407959,
      "eval_runtime": 8.3224,
      "eval_samples_per_second": 4.566,
      "eval_steps_per_second": 1.202,
      "step": 720
    },
    {
      "epoch": 2.11,
      "eval_loss": 1.1748428344726562,
      "eval_runtime": 5.1694,
      "eval_samples_per_second": 7.351,
      "eval_steps_per_second": 1.934,
      "step": 725
    },
    {
      "epoch": 2.12,
      "eval_loss": 1.1770687103271484,
      "eval_runtime": 6.4645,
      "eval_samples_per_second": 5.878,
      "eval_steps_per_second": 1.547,
      "step": 730
    },
    {
      "epoch": 2.14,
      "eval_loss": 1.1737899780273438,
      "eval_runtime": 5.1852,
      "eval_samples_per_second": 7.329,
      "eval_steps_per_second": 1.929,
      "step": 735
    },
    {
      "epoch": 2.15,
      "eval_loss": 1.164935827255249,
      "eval_runtime": 5.1999,
      "eval_samples_per_second": 7.308,
      "eval_steps_per_second": 1.923,
      "step": 740
    },
    {
      "epoch": 2.17,
      "eval_loss": 1.1611236333847046,
      "eval_runtime": 5.1943,
      "eval_samples_per_second": 7.316,
      "eval_steps_per_second": 1.925,
      "step": 745
    },
    {
      "epoch": 2.18,
      "eval_loss": 1.16485595703125,
      "eval_runtime": 5.2046,
      "eval_samples_per_second": 7.301,
      "eval_steps_per_second": 1.921,
      "step": 750
    },
    {
      "epoch": 2.19,
      "eval_loss": 1.1628483533859253,
      "eval_runtime": 5.1977,
      "eval_samples_per_second": 7.311,
      "eval_steps_per_second": 1.924,
      "step": 755
    },
    {
      "epoch": 2.21,
      "eval_loss": 1.1594696044921875,
      "eval_runtime": 5.1698,
      "eval_samples_per_second": 7.35,
      "eval_steps_per_second": 1.934,
      "step": 760
    },
    {
      "epoch": 2.22,
      "eval_loss": 1.1594704389572144,
      "eval_runtime": 5.1839,
      "eval_samples_per_second": 7.33,
      "eval_steps_per_second": 1.929,
      "step": 765
    },
    {
      "epoch": 2.24,
      "eval_loss": 1.1618760824203491,
      "eval_runtime": 5.1971,
      "eval_samples_per_second": 7.312,
      "eval_steps_per_second": 1.924,
      "step": 770
    },
    {
      "epoch": 2.25,
      "eval_loss": 1.1532135009765625,
      "eval_runtime": 5.1859,
      "eval_samples_per_second": 7.327,
      "eval_steps_per_second": 1.928,
      "step": 775
    },
    {
      "epoch": 2.27,
      "eval_loss": 1.148254156112671,
      "eval_runtime": 5.1949,
      "eval_samples_per_second": 7.315,
      "eval_steps_per_second": 1.925,
      "step": 780
    },
    {
      "epoch": 2.28,
      "eval_loss": 1.1526851654052734,
      "eval_runtime": 5.2153,
      "eval_samples_per_second": 7.286,
      "eval_steps_per_second": 1.917,
      "step": 785
    },
    {
      "epoch": 2.3,
      "eval_loss": 1.1553150415420532,
      "eval_runtime": 7.2932,
      "eval_samples_per_second": 5.21,
      "eval_steps_per_second": 1.371,
      "step": 790
    },
    {
      "epoch": 2.31,
      "eval_loss": 1.1568715572357178,
      "eval_runtime": 5.2156,
      "eval_samples_per_second": 7.286,
      "eval_steps_per_second": 1.917,
      "step": 795
    },
    {
      "epoch": 2.33,
      "learning_rate": 8.837209302325582e-06,
      "loss": 1.0236,
      "step": 800
    },
    {
      "epoch": 2.33,
      "eval_loss": 1.1593568325042725,
      "eval_runtime": 15.4054,
      "eval_samples_per_second": 2.467,
      "eval_steps_per_second": 0.649,
      "step": 800
    },
    {
      "epoch": 2.34,
      "eval_loss": 1.1579800844192505,
      "eval_runtime": 5.1382,
      "eval_samples_per_second": 7.396,
      "eval_steps_per_second": 1.946,
      "step": 805
    },
    {
      "epoch": 2.35,
      "eval_loss": 1.1529394388198853,
      "eval_runtime": 5.1435,
      "eval_samples_per_second": 7.388,
      "eval_steps_per_second": 1.944,
      "step": 810
    },
    {
      "epoch": 2.37,
      "eval_loss": 1.1465649604797363,
      "eval_runtime": 5.1543,
      "eval_samples_per_second": 7.372,
      "eval_steps_per_second": 1.94,
      "step": 815
    },
    {
      "epoch": 2.38,
      "eval_loss": 1.1523265838623047,
      "eval_runtime": 5.1668,
      "eval_samples_per_second": 7.355,
      "eval_steps_per_second": 1.935,
      "step": 820
    },
    {
      "epoch": 2.4,
      "eval_loss": 1.1473236083984375,
      "eval_runtime": 5.1929,
      "eval_samples_per_second": 7.318,
      "eval_steps_per_second": 1.926,
      "step": 825
    },
    {
      "epoch": 2.41,
      "eval_loss": 1.1426836252212524,
      "eval_runtime": 5.1805,
      "eval_samples_per_second": 7.335,
      "eval_steps_per_second": 1.93,
      "step": 830
    },
    {
      "epoch": 2.43,
      "eval_loss": 1.1456120014190674,
      "eval_runtime": 5.1879,
      "eval_samples_per_second": 7.325,
      "eval_steps_per_second": 1.928,
      "step": 835
    },
    {
      "epoch": 2.44,
      "eval_loss": 1.1546478271484375,
      "eval_runtime": 6.9177,
      "eval_samples_per_second": 5.493,
      "eval_steps_per_second": 1.446,
      "step": 840
    },
    {
      "epoch": 2.46,
      "eval_loss": 1.150542140007019,
      "eval_runtime": 5.2006,
      "eval_samples_per_second": 7.307,
      "eval_steps_per_second": 1.923,
      "step": 845
    },
    {
      "epoch": 2.47,
      "eval_loss": 1.1451619863510132,
      "eval_runtime": 5.209,
      "eval_samples_per_second": 7.295,
      "eval_steps_per_second": 1.92,
      "step": 850
    },
    {
      "epoch": 2.49,
      "eval_loss": 1.1448664665222168,
      "eval_runtime": 5.1984,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 855
    },
    {
      "epoch": 2.5,
      "eval_loss": 1.142190933227539,
      "eval_runtime": 5.2014,
      "eval_samples_per_second": 7.306,
      "eval_steps_per_second": 1.923,
      "step": 860
    },
    {
      "epoch": 2.51,
      "eval_loss": 1.1360749006271362,
      "eval_runtime": 5.213,
      "eval_samples_per_second": 7.289,
      "eval_steps_per_second": 1.918,
      "step": 865
    },
    {
      "epoch": 2.53,
      "eval_loss": 1.1361669301986694,
      "eval_runtime": 5.225,
      "eval_samples_per_second": 7.273,
      "eval_steps_per_second": 1.914,
      "step": 870
    },
    {
      "epoch": 2.54,
      "eval_loss": 1.152858018875122,
      "eval_runtime": 5.2072,
      "eval_samples_per_second": 7.298,
      "eval_steps_per_second": 1.92,
      "step": 875
    },
    {
      "epoch": 2.56,
      "eval_loss": 1.1414388418197632,
      "eval_runtime": 5.2264,
      "eval_samples_per_second": 7.271,
      "eval_steps_per_second": 1.913,
      "step": 880
    },
    {
      "epoch": 2.57,
      "eval_loss": 1.1355433464050293,
      "eval_runtime": 5.2274,
      "eval_samples_per_second": 7.269,
      "eval_steps_per_second": 1.913,
      "step": 885
    },
    {
      "epoch": 2.59,
      "eval_loss": 1.140533447265625,
      "eval_runtime": 5.2091,
      "eval_samples_per_second": 7.295,
      "eval_steps_per_second": 1.92,
      "step": 890
    },
    {
      "epoch": 2.6,
      "eval_loss": 1.1398884057998657,
      "eval_runtime": 5.2089,
      "eval_samples_per_second": 7.295,
      "eval_steps_per_second": 1.92,
      "step": 895
    },
    {
      "epoch": 2.62,
      "learning_rate": 8.69186046511628e-06,
      "loss": 1.005,
      "step": 900
    },
    {
      "epoch": 2.62,
      "eval_loss": 1.1313834190368652,
      "eval_runtime": 7.9714,
      "eval_samples_per_second": 4.767,
      "eval_steps_per_second": 1.254,
      "step": 900
    },
    {
      "epoch": 2.63,
      "eval_loss": 1.1292839050292969,
      "eval_runtime": 5.1391,
      "eval_samples_per_second": 7.394,
      "eval_steps_per_second": 1.946,
      "step": 905
    },
    {
      "epoch": 2.65,
      "eval_loss": 1.1339421272277832,
      "eval_runtime": 5.1448,
      "eval_samples_per_second": 7.386,
      "eval_steps_per_second": 1.944,
      "step": 910
    },
    {
      "epoch": 2.66,
      "eval_loss": 1.1322623491287231,
      "eval_runtime": 5.1737,
      "eval_samples_per_second": 7.345,
      "eval_steps_per_second": 1.933,
      "step": 915
    },
    {
      "epoch": 2.67,
      "eval_loss": 1.1225138902664185,
      "eval_runtime": 5.1822,
      "eval_samples_per_second": 7.333,
      "eval_steps_per_second": 1.93,
      "step": 920
    },
    {
      "epoch": 2.69,
      "eval_loss": 1.117431640625,
      "eval_runtime": 5.1739,
      "eval_samples_per_second": 7.345,
      "eval_steps_per_second": 1.933,
      "step": 925
    },
    {
      "epoch": 2.7,
      "eval_loss": 1.1242172718048096,
      "eval_runtime": 7.2847,
      "eval_samples_per_second": 5.216,
      "eval_steps_per_second": 1.373,
      "step": 930
    },
    {
      "epoch": 2.72,
      "eval_loss": 1.1318007707595825,
      "eval_runtime": 5.4547,
      "eval_samples_per_second": 6.966,
      "eval_steps_per_second": 1.833,
      "step": 935
    },
    {
      "epoch": 2.73,
      "eval_loss": 1.1215720176696777,
      "eval_runtime": 5.1989,
      "eval_samples_per_second": 7.309,
      "eval_steps_per_second": 1.923,
      "step": 940
    },
    {
      "epoch": 2.75,
      "eval_loss": 1.1199665069580078,
      "eval_runtime": 5.1934,
      "eval_samples_per_second": 7.317,
      "eval_steps_per_second": 1.926,
      "step": 945
    },
    {
      "epoch": 2.76,
      "eval_loss": 1.1283185482025146,
      "eval_runtime": 5.2047,
      "eval_samples_per_second": 7.301,
      "eval_steps_per_second": 1.921,
      "step": 950
    },
    {
      "epoch": 2.78,
      "eval_loss": 1.1272627115249634,
      "eval_runtime": 5.2028,
      "eval_samples_per_second": 7.304,
      "eval_steps_per_second": 1.922,
      "step": 955
    },
    {
      "epoch": 2.79,
      "eval_loss": 1.116969347000122,
      "eval_runtime": 5.2048,
      "eval_samples_per_second": 7.301,
      "eval_steps_per_second": 1.921,
      "step": 960
    },
    {
      "epoch": 2.81,
      "eval_loss": 1.1125539541244507,
      "eval_runtime": 5.2168,
      "eval_samples_per_second": 7.284,
      "eval_steps_per_second": 1.917,
      "step": 965
    },
    {
      "epoch": 2.82,
      "eval_loss": 1.1195616722106934,
      "eval_runtime": 5.1986,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 970
    },
    {
      "epoch": 2.83,
      "eval_loss": 1.1192817687988281,
      "eval_runtime": 5.2203,
      "eval_samples_per_second": 7.279,
      "eval_steps_per_second": 1.916,
      "step": 975
    },
    {
      "epoch": 2.85,
      "eval_loss": 1.1130948066711426,
      "eval_runtime": 9.4606,
      "eval_samples_per_second": 4.017,
      "eval_steps_per_second": 1.057,
      "step": 980
    },
    {
      "epoch": 2.86,
      "eval_loss": 1.1146584749221802,
      "eval_runtime": 5.2013,
      "eval_samples_per_second": 7.306,
      "eval_steps_per_second": 1.923,
      "step": 985
    },
    {
      "epoch": 2.88,
      "eval_loss": 1.118652582168579,
      "eval_runtime": 5.201,
      "eval_samples_per_second": 7.306,
      "eval_steps_per_second": 1.923,
      "step": 990
    },
    {
      "epoch": 2.89,
      "eval_loss": 1.1133984327316284,
      "eval_runtime": 5.2035,
      "eval_samples_per_second": 7.303,
      "eval_steps_per_second": 1.922,
      "step": 995
    },
    {
      "epoch": 2.91,
      "learning_rate": 8.546511627906978e-06,
      "loss": 1.0003,
      "step": 1000
    },
    {
      "epoch": 2.91,
      "eval_loss": 1.1122275590896606,
      "eval_runtime": 5.5704,
      "eval_samples_per_second": 6.822,
      "eval_steps_per_second": 1.795,
      "step": 1000
    },
    {
      "epoch": 2.92,
      "eval_loss": 1.1242709159851074,
      "eval_runtime": 5.1423,
      "eval_samples_per_second": 7.39,
      "eval_steps_per_second": 1.945,
      "step": 1005
    },
    {
      "epoch": 2.94,
      "eval_loss": 1.1163734197616577,
      "eval_runtime": 5.159,
      "eval_samples_per_second": 7.366,
      "eval_steps_per_second": 1.938,
      "step": 1010
    },
    {
      "epoch": 2.95,
      "eval_loss": 1.1083950996398926,
      "eval_runtime": 5.1716,
      "eval_samples_per_second": 7.348,
      "eval_steps_per_second": 1.934,
      "step": 1015
    },
    {
      "epoch": 2.97,
      "eval_loss": 1.1112630367279053,
      "eval_runtime": 5.1839,
      "eval_samples_per_second": 7.33,
      "eval_steps_per_second": 1.929,
      "step": 1020
    },
    {
      "epoch": 2.98,
      "eval_loss": 1.1141904592514038,
      "eval_runtime": 5.1752,
      "eval_samples_per_second": 7.343,
      "eval_steps_per_second": 1.932,
      "step": 1025
    },
    {
      "epoch": 2.99,
      "eval_loss": 1.109470248222351,
      "eval_runtime": 5.1964,
      "eval_samples_per_second": 7.313,
      "eval_steps_per_second": 1.924,
      "step": 1030
    },
    {
      "epoch": 3.01,
      "eval_loss": 1.114129662513733,
      "eval_runtime": 5.1874,
      "eval_samples_per_second": 7.325,
      "eval_steps_per_second": 1.928,
      "step": 1035
    },
    {
      "epoch": 3.02,
      "eval_loss": 1.1438744068145752,
      "eval_runtime": 6.8038,
      "eval_samples_per_second": 5.585,
      "eval_steps_per_second": 1.47,
      "step": 1040
    },
    {
      "epoch": 3.04,
      "eval_loss": 1.1356358528137207,
      "eval_runtime": 5.186,
      "eval_samples_per_second": 7.327,
      "eval_steps_per_second": 1.928,
      "step": 1045
    },
    {
      "epoch": 3.05,
      "eval_loss": 1.1280732154846191,
      "eval_runtime": 5.1897,
      "eval_samples_per_second": 7.322,
      "eval_steps_per_second": 1.927,
      "step": 1050
    },
    {
      "epoch": 3.07,
      "eval_loss": 1.130995750427246,
      "eval_runtime": 5.1866,
      "eval_samples_per_second": 7.327,
      "eval_steps_per_second": 1.928,
      "step": 1055
    },
    {
      "epoch": 3.08,
      "eval_loss": 1.1382439136505127,
      "eval_runtime": 5.2111,
      "eval_samples_per_second": 7.292,
      "eval_steps_per_second": 1.919,
      "step": 1060
    },
    {
      "epoch": 3.1,
      "eval_loss": 1.142386555671692,
      "eval_runtime": 5.1981,
      "eval_samples_per_second": 7.31,
      "eval_steps_per_second": 1.924,
      "step": 1065
    },
    {
      "epoch": 3.11,
      "eval_loss": 1.126651406288147,
      "eval_runtime": 5.2225,
      "eval_samples_per_second": 7.276,
      "eval_steps_per_second": 1.915,
      "step": 1070
    },
    {
      "epoch": 3.12,
      "eval_loss": 1.1165131330490112,
      "eval_runtime": 5.2221,
      "eval_samples_per_second": 7.277,
      "eval_steps_per_second": 1.915,
      "step": 1075
    },
    {
      "epoch": 3.14,
      "eval_loss": 1.1212615966796875,
      "eval_runtime": 5.2824,
      "eval_samples_per_second": 7.194,
      "eval_steps_per_second": 1.893,
      "step": 1080
    },
    {
      "epoch": 3.15,
      "eval_loss": 1.134791374206543,
      "eval_runtime": 5.2245,
      "eval_samples_per_second": 7.273,
      "eval_steps_per_second": 1.914,
      "step": 1085
    },
    {
      "epoch": 3.17,
      "eval_loss": 1.1270241737365723,
      "eval_runtime": 5.3762,
      "eval_samples_per_second": 7.068,
      "eval_steps_per_second": 1.86,
      "step": 1090
    },
    {
      "epoch": 3.18,
      "eval_loss": 1.130293846130371,
      "eval_runtime": 5.2263,
      "eval_samples_per_second": 7.271,
      "eval_steps_per_second": 1.913,
      "step": 1095
    },
    {
      "epoch": 3.2,
      "learning_rate": 8.401162790697675e-06,
      "loss": 0.9125,
      "step": 1100
    },
    {
      "epoch": 3.2,
      "eval_loss": 1.1280182600021362,
      "eval_runtime": 5.2073,
      "eval_samples_per_second": 7.297,
      "eval_steps_per_second": 1.92,
      "step": 1100
    },
    {
      "epoch": 3.21,
      "eval_loss": 1.1197805404663086,
      "eval_runtime": 5.1419,
      "eval_samples_per_second": 7.39,
      "eval_steps_per_second": 1.945,
      "step": 1105
    },
    {
      "epoch": 3.23,
      "eval_loss": 1.1133369207382202,
      "eval_runtime": 5.1481,
      "eval_samples_per_second": 7.381,
      "eval_steps_per_second": 1.942,
      "step": 1110
    },
    {
      "epoch": 3.24,
      "eval_loss": 1.1289596557617188,
      "eval_runtime": 5.1768,
      "eval_samples_per_second": 7.34,
      "eval_steps_per_second": 1.932,
      "step": 1115
    },
    {
      "epoch": 3.26,
      "eval_loss": 1.1368350982666016,
      "eval_runtime": 5.1857,
      "eval_samples_per_second": 7.328,
      "eval_steps_per_second": 1.928,
      "step": 1120
    },
    {
      "epoch": 3.27,
      "eval_loss": 1.1226954460144043,
      "eval_runtime": 5.4937,
      "eval_samples_per_second": 6.917,
      "eval_steps_per_second": 1.82,
      "step": 1125
    },
    {
      "epoch": 3.28,
      "eval_loss": 1.1116739511489868,
      "eval_runtime": 5.1807,
      "eval_samples_per_second": 7.335,
      "eval_steps_per_second": 1.93,
      "step": 1130
    },
    {
      "epoch": 3.3,
      "eval_loss": 1.1072471141815186,
      "eval_runtime": 5.89,
      "eval_samples_per_second": 6.452,
      "eval_steps_per_second": 1.698,
      "step": 1135
    },
    {
      "epoch": 3.31,
      "eval_loss": 1.1166975498199463,
      "eval_runtime": 5.2072,
      "eval_samples_per_second": 7.298,
      "eval_steps_per_second": 1.92,
      "step": 1140
    },
    {
      "epoch": 3.33,
      "eval_loss": 1.119249939918518,
      "eval_runtime": 5.7107,
      "eval_samples_per_second": 6.654,
      "eval_steps_per_second": 1.751,
      "step": 1145
    },
    {
      "epoch": 3.34,
      "eval_loss": 1.1184831857681274,
      "eval_runtime": 5.2137,
      "eval_samples_per_second": 7.289,
      "eval_steps_per_second": 1.918,
      "step": 1150
    },
    {
      "epoch": 3.36,
      "eval_loss": 1.1111912727355957,
      "eval_runtime": 5.9367,
      "eval_samples_per_second": 6.401,
      "eval_steps_per_second": 1.684,
      "step": 1155
    },
    {
      "epoch": 3.37,
      "eval_loss": 1.105357050895691,
      "eval_runtime": 5.2105,
      "eval_samples_per_second": 7.293,
      "eval_steps_per_second": 1.919,
      "step": 1160
    },
    {
      "epoch": 3.39,
      "eval_loss": 1.1098227500915527,
      "eval_runtime": 5.2189,
      "eval_samples_per_second": 7.281,
      "eval_steps_per_second": 1.916,
      "step": 1165
    },
    {
      "epoch": 3.4,
      "eval_loss": 1.1147123575210571,
      "eval_runtime": 5.2081,
      "eval_samples_per_second": 7.296,
      "eval_steps_per_second": 1.92,
      "step": 1170
    },
    {
      "epoch": 3.42,
      "eval_loss": 1.1088005304336548,
      "eval_runtime": 7.6087,
      "eval_samples_per_second": 4.994,
      "eval_steps_per_second": 1.314,
      "step": 1175
    },
    {
      "epoch": 3.43,
      "eval_loss": 1.1164624691009521,
      "eval_runtime": 5.2125,
      "eval_samples_per_second": 7.29,
      "eval_steps_per_second": 1.918,
      "step": 1180
    },
    {
      "epoch": 3.44,
      "eval_loss": 1.121812105178833,
      "eval_runtime": 5.2191,
      "eval_samples_per_second": 7.281,
      "eval_steps_per_second": 1.916,
      "step": 1185
    },
    {
      "epoch": 3.46,
      "eval_loss": 1.1162291765213013,
      "eval_runtime": 5.2203,
      "eval_samples_per_second": 7.279,
      "eval_steps_per_second": 1.916,
      "step": 1190
    },
    {
      "epoch": 3.47,
      "eval_loss": 1.1079984903335571,
      "eval_runtime": 5.2071,
      "eval_samples_per_second": 7.298,
      "eval_steps_per_second": 1.92,
      "step": 1195
    },
    {
      "epoch": 3.49,
      "learning_rate": 8.255813953488374e-06,
      "loss": 0.8748,
      "step": 1200
    },
    {
      "epoch": 3.49,
      "eval_loss": 1.105452060699463,
      "eval_runtime": 5.2094,
      "eval_samples_per_second": 7.294,
      "eval_steps_per_second": 1.92,
      "step": 1200
    },
    {
      "epoch": 3.5,
      "eval_loss": 1.106619119644165,
      "eval_runtime": 5.1432,
      "eval_samples_per_second": 7.388,
      "eval_steps_per_second": 1.944,
      "step": 1205
    },
    {
      "epoch": 3.52,
      "eval_loss": 1.106793761253357,
      "eval_runtime": 5.1604,
      "eval_samples_per_second": 7.364,
      "eval_steps_per_second": 1.938,
      "step": 1210
    },
    {
      "epoch": 3.53,
      "eval_loss": 1.1173338890075684,
      "eval_runtime": 5.1605,
      "eval_samples_per_second": 7.364,
      "eval_steps_per_second": 1.938,
      "step": 1215
    },
    {
      "epoch": 3.55,
      "eval_loss": 1.114292025566101,
      "eval_runtime": 5.1855,
      "eval_samples_per_second": 7.328,
      "eval_steps_per_second": 1.928,
      "step": 1220
    },
    {
      "epoch": 3.56,
      "eval_loss": 1.1030842065811157,
      "eval_runtime": 6.5892,
      "eval_samples_per_second": 5.767,
      "eval_steps_per_second": 1.518,
      "step": 1225
    },
    {
      "epoch": 3.58,
      "eval_loss": 1.1062043905258179,
      "eval_runtime": 5.1841,
      "eval_samples_per_second": 7.33,
      "eval_steps_per_second": 1.929,
      "step": 1230
    },
    {
      "epoch": 3.59,
      "eval_loss": 1.1063551902770996,
      "eval_runtime": 5.1918,
      "eval_samples_per_second": 7.319,
      "eval_steps_per_second": 1.926,
      "step": 1235
    },
    {
      "epoch": 3.6,
      "eval_loss": 1.1138606071472168,
      "eval_runtime": 5.2119,
      "eval_samples_per_second": 7.291,
      "eval_steps_per_second": 1.919,
      "step": 1240
    },
    {
      "epoch": 3.62,
      "eval_loss": 1.110581874847412,
      "eval_runtime": 5.1939,
      "eval_samples_per_second": 7.316,
      "eval_steps_per_second": 1.925,
      "step": 1245
    },
    {
      "epoch": 3.63,
      "eval_loss": 1.100100040435791,
      "eval_runtime": 5.1966,
      "eval_samples_per_second": 7.312,
      "eval_steps_per_second": 1.924,
      "step": 1250
    },
    {
      "epoch": 3.65,
      "eval_loss": 1.0952939987182617,
      "eval_runtime": 5.1974,
      "eval_samples_per_second": 7.311,
      "eval_steps_per_second": 1.924,
      "step": 1255
    },
    {
      "epoch": 3.66,
      "eval_loss": 1.0998646020889282,
      "eval_runtime": 5.19,
      "eval_samples_per_second": 7.322,
      "eval_steps_per_second": 1.927,
      "step": 1260
    },
    {
      "epoch": 3.68,
      "eval_loss": 1.1082687377929688,
      "eval_runtime": 5.4201,
      "eval_samples_per_second": 7.011,
      "eval_steps_per_second": 1.845,
      "step": 1265
    },
    {
      "epoch": 3.69,
      "eval_loss": 1.1146087646484375,
      "eval_runtime": 5.1949,
      "eval_samples_per_second": 7.315,
      "eval_steps_per_second": 1.925,
      "step": 1270
    },
    {
      "epoch": 3.71,
      "eval_loss": 1.1090608835220337,
      "eval_runtime": 5.2206,
      "eval_samples_per_second": 7.279,
      "eval_steps_per_second": 1.915,
      "step": 1275
    },
    {
      "epoch": 3.72,
      "eval_loss": 1.1007745265960693,
      "eval_runtime": 5.2062,
      "eval_samples_per_second": 7.299,
      "eval_steps_per_second": 1.921,
      "step": 1280
    },
    {
      "epoch": 3.74,
      "eval_loss": 1.0957188606262207,
      "eval_runtime": 5.2196,
      "eval_samples_per_second": 7.28,
      "eval_steps_per_second": 1.916,
      "step": 1285
    },
    {
      "epoch": 3.75,
      "eval_loss": 1.1012167930603027,
      "eval_runtime": 5.2206,
      "eval_samples_per_second": 7.279,
      "eval_steps_per_second": 1.915,
      "step": 1290
    },
    {
      "epoch": 3.76,
      "eval_loss": 1.09498929977417,
      "eval_runtime": 5.2178,
      "eval_samples_per_second": 7.283,
      "eval_steps_per_second": 1.917,
      "step": 1295
    },
    {
      "epoch": 3.78,
      "learning_rate": 8.110465116279071e-06,
      "loss": 0.8599,
      "step": 1300
    },
    {
      "epoch": 3.78,
      "eval_loss": 1.0983973741531372,
      "eval_runtime": 5.203,
      "eval_samples_per_second": 7.303,
      "eval_steps_per_second": 1.922,
      "step": 1300
    }
  ],
  "max_steps": 6880,
  "num_train_epochs": 20,
  "total_flos": 5658055802880000.0,
  "trial_name": null,
  "trial_params": null
}