{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.6649323621227887,
  "eval_steps": 500,
  "global_step": 800,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.02,
      "learning_rate": 5.208333333333334e-06,
      "loss": 0.4863,
      "step": 10
    },
    {
      "epoch": 0.04,
      "learning_rate": 1.0416666666666668e-05,
      "loss": 0.5006,
      "step": 20
    },
    {
      "epoch": 0.06,
      "learning_rate": 1.5625e-05,
      "loss": 0.4452,
      "step": 30
    },
    {
      "epoch": 0.08,
      "learning_rate": 2.0833333333333336e-05,
      "loss": 0.4253,
      "step": 40
    },
    {
      "epoch": 0.1,
      "learning_rate": 2.604166666666667e-05,
      "loss": 0.4257,
      "step": 50
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.125e-05,
      "loss": 0.4297,
      "step": 60
    },
    {
      "epoch": 0.15,
      "learning_rate": 3.6458333333333336e-05,
      "loss": 0.4049,
      "step": 70
    },
    {
      "epoch": 0.17,
      "learning_rate": 4.166666666666667e-05,
      "loss": 0.3531,
      "step": 80
    },
    {
      "epoch": 0.19,
      "learning_rate": 4.6875e-05,
      "loss": 0.3731,
      "step": 90
    },
    {
      "epoch": 0.21,
      "learning_rate": 4.999735579817769e-05,
      "loss": 0.3582,
      "step": 100
    },
    {
      "epoch": 0.23,
      "learning_rate": 4.9967614951022265e-05,
      "loss": 0.3296,
      "step": 110
    },
    {
      "epoch": 0.25,
      "learning_rate": 4.990486745229364e-05,
      "loss": 0.3353,
      "step": 120
    },
    {
      "epoch": 0.27,
      "learning_rate": 4.980919625283962e-05,
      "loss": 0.2787,
      "step": 130
    },
    {
      "epoch": 0.29,
      "learning_rate": 4.9680727827934354e-05,
      "loss": 0.3083,
      "step": 140
    },
    {
      "epoch": 0.31,
      "learning_rate": 4.951963201008076e-05,
      "loss": 0.3055,
      "step": 150
    },
    {
      "epoch": 0.33,
      "learning_rate": 4.9326121764495596e-05,
      "loss": 0.2876,
      "step": 160
    },
    {
      "epoch": 0.35,
      "learning_rate": 4.910045290757399e-05,
      "loss": 0.296,
      "step": 170
    },
    {
      "epoch": 0.37,
      "learning_rate": 4.884292376870567e-05,
      "loss": 0.2349,
      "step": 180
    },
    {
      "epoch": 0.4,
      "learning_rate": 4.8553874795889916e-05,
      "loss": 0.2837,
      "step": 190
    },
    {
      "epoch": 0.42,
      "learning_rate": 4.823368810567056e-05,
      "loss": 0.2397,
      "step": 200
    },
    {
      "epoch": 0.44,
      "learning_rate": 4.788278697798618e-05,
      "loss": 0.2827,
      "step": 210
    },
    {
      "epoch": 0.46,
      "learning_rate": 4.750163529660303e-05,
      "loss": 0.2504,
      "step": 220
    },
    {
      "epoch": 0.48,
      "learning_rate": 4.7090736935870696e-05,
      "loss": 0.2565,
      "step": 230
    },
    {
      "epoch": 0.5,
      "learning_rate": 4.665063509461097e-05,
      "loss": 0.2284,
      "step": 240
    },
    {
      "epoch": 0.52,
      "learning_rate": 4.61819115780207e-05,
      "loss": 0.2511,
      "step": 250
    },
    {
      "epoch": 0.54,
      "learning_rate": 4.568518602853776e-05,
      "loss": 0.2442,
      "step": 260
    },
    {
      "epoch": 0.56,
      "learning_rate": 4.516111510668707e-05,
      "loss": 0.2753,
      "step": 270
    },
    {
      "epoch": 0.58,
      "learning_rate": 4.4610391622989396e-05,
      "loss": 0.2495,
      "step": 280
    },
    {
      "epoch": 0.6,
      "learning_rate": 4.403374362208078e-05,
      "loss": 0.2537,
      "step": 290
    },
    {
      "epoch": 0.62,
      "learning_rate": 4.34319334202531e-05,
      "loss": 0.2774,
      "step": 300
    },
    {
      "epoch": 0.65,
      "learning_rate": 4.2805756597688276e-05,
      "loss": 0.2641,
      "step": 310
    },
    {
      "epoch": 0.67,
      "learning_rate": 4.215604094671835e-05,
      "loss": 0.2389,
      "step": 320
    },
    {
      "epoch": 0.69,
      "learning_rate": 4.148364537750172e-05,
      "loss": 0.2522,
      "step": 330
    },
    {
      "epoch": 0.71,
      "learning_rate": 4.078945878256244e-05,
      "loss": 0.2606,
      "step": 340
    },
    {
      "epoch": 0.73,
      "learning_rate": 4.007439886169324e-05,
      "loss": 0.2629,
      "step": 350
    },
    {
      "epoch": 0.75,
      "learning_rate": 3.933941090877615e-05,
      "loss": 0.2476,
      "step": 360
    },
    {
      "epoch": 0.77,
      "learning_rate": 3.858546656212425e-05,
      "loss": 0.2491,
      "step": 370
    },
    {
      "epoch": 0.79,
      "learning_rate": 3.781356251999663e-05,
      "loss": 0.2135,
      "step": 380
    },
    {
      "epoch": 0.81,
      "learning_rate": 3.702471922298469e-05,
      "loss": 0.2526,
      "step": 390
    },
    {
      "epoch": 0.83,
      "learning_rate": 3.621997950501156e-05,
      "loss": 0.2356,
      "step": 400
    },
    {
      "epoch": 0.85,
      "learning_rate": 3.540040721472798e-05,
      "loss": 0.2161,
      "step": 410
    },
    {
      "epoch": 0.87,
      "learning_rate": 3.456708580912725e-05,
      "loss": 0.2597,
      "step": 420
    },
    {
      "epoch": 0.89,
      "learning_rate": 3.372111692123827e-05,
      "loss": 0.282,
      "step": 430
    },
    {
      "epoch": 0.92,
      "learning_rate": 3.2863618903790346e-05,
      "loss": 0.2449,
      "step": 440
    },
    {
      "epoch": 0.94,
      "learning_rate": 3.1995725350774806e-05,
      "loss": 0.2086,
      "step": 450
    },
    {
      "epoch": 0.96,
      "learning_rate": 3.11185835988581e-05,
      "loss": 0.234,
      "step": 460
    },
    {
      "epoch": 0.98,
      "learning_rate": 3.02333532106273e-05,
      "loss": 0.2544,
      "step": 470
    },
    {
      "epoch": 1.0,
      "learning_rate": 2.9341204441673266e-05,
      "loss": 0.279,
      "step": 480
    },
    {
      "epoch": 1.02,
      "learning_rate": 2.844331669353777e-05,
      "loss": 0.2715,
      "step": 490
    },
    {
      "epoch": 1.04,
      "learning_rate": 2.7540876954570048e-05,
      "loss": 0.2351,
      "step": 500
    },
    {
      "epoch": 1.06,
      "learning_rate": 2.663507823075358e-05,
      "loss": 0.2275,
      "step": 510
    },
    {
      "epoch": 1.08,
      "learning_rate": 2.5727117968577784e-05,
      "loss": 0.255,
      "step": 520
    },
    {
      "epoch": 1.1,
      "learning_rate": 2.4818196472039466e-05,
      "loss": 0.2594,
      "step": 530
    },
    {
      "epoch": 1.12,
      "learning_rate": 2.3909515315866605e-05,
      "loss": 0.2336,
      "step": 540
    },
    {
      "epoch": 1.14,
      "learning_rate": 2.300227575706241e-05,
      "loss": 0.194,
      "step": 550
    },
    {
      "epoch": 1.17,
      "learning_rate": 2.2097677146869242e-05,
      "loss": 0.2508,
      "step": 560
    },
    {
      "epoch": 1.19,
      "learning_rate": 2.1196915345252084e-05,
      "loss": 0.2608,
      "step": 570
    },
    {
      "epoch": 1.21,
      "learning_rate": 2.0301181139997205e-05,
      "loss": 0.2459,
      "step": 580
    },
    {
      "epoch": 1.23,
      "learning_rate": 1.9411658672516314e-05,
      "loss": 0.2066,
      "step": 590
    },
    {
      "epoch": 1.25,
      "learning_rate": 1.852952387243698e-05,
      "loss": 0.2105,
      "step": 600
    },
    {
      "epoch": 1.27,
      "learning_rate": 1.765594290304889e-05,
      "loss": 0.2178,
      "step": 610
    },
    {
      "epoch": 1.29,
      "learning_rate": 1.6792070619660975e-05,
      "loss": 0.1902,
      "step": 620
    },
    {
      "epoch": 1.31,
      "learning_rate": 1.5939049042907462e-05,
      "loss": 0.2301,
      "step": 630
    },
    {
      "epoch": 1.33,
      "learning_rate": 1.509800584902108e-05,
      "loss": 0.2964,
      "step": 640
    },
    {
      "epoch": 1.35,
      "learning_rate": 1.427005287906928e-05,
      "loss": 0.2356,
      "step": 650
    },
    {
      "epoch": 1.37,
      "learning_rate": 1.3456284669124158e-05,
      "loss": 0.2502,
      "step": 660
    },
    {
      "epoch": 1.39,
      "learning_rate": 1.2657777003309242e-05,
      "loss": 0.2437,
      "step": 670
    },
    {
      "epoch": 1.42,
      "learning_rate": 1.1875585491636e-05,
      "loss": 0.2443,
      "step": 680
    },
    {
      "epoch": 1.44,
      "learning_rate": 1.1110744174509952e-05,
      "loss": 0.2461,
      "step": 690
    },
    {
      "epoch": 1.46,
      "learning_rate": 1.0364264155751488e-05,
      "loss": 0.221,
      "step": 700
    },
    {
      "epoch": 1.48,
      "learning_rate": 9.637132265938315e-06,
      "loss": 0.2272,
      "step": 710
    },
    {
      "epoch": 1.5,
      "learning_rate": 8.930309757836517e-06,
      "loss": 0.252,
      "step": 720
    },
    {
      "epoch": 1.52,
      "learning_rate": 8.244731035645107e-06,
      "loss": 0.1984,
      "step": 730
    },
    {
      "epoch": 1.54,
      "learning_rate": 7.581302419733632e-06,
      "loss": 0.2741,
      "step": 740
    },
    {
      "epoch": 1.56,
      "learning_rate": 6.940900948506113e-06,
      "loss": 0.2791,
      "step": 750
    },
    {
      "epoch": 1.58,
      "learning_rate": 6.324373218975105e-06,
      "loss": 0.2488,
      "step": 760
    },
    {
      "epoch": 1.6,
      "learning_rate": 5.7325342675785486e-06,
      "loss": 0.1872,
      "step": 770
    },
    {
      "epoch": 1.62,
      "learning_rate": 5.166166492719124e-06,
      "loss": 0.2079,
      "step": 780
    },
    {
      "epoch": 1.64,
      "learning_rate": 4.6260186204503094e-06,
      "loss": 0.2148,
      "step": 790
    },
    {
      "epoch": 1.66,
      "learning_rate": 4.112804714676594e-06,
      "loss": 0.2719,
      "step": 800
    }
  ],
  "logging_steps": 10,
  "max_steps": 960,
  "num_train_epochs": 2,
  "save_steps": 100,
  "total_flos": 1.790913832400978e+17,
  "trial_name": null,
  "trial_params": null
}