{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.14381040036815462,
  "eval_steps": 500,
  "global_step": 2500,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "learning_rate": 0.00019920000000000002,
      "loss": 1.0145,
      "step": 10
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001984,
      "loss": 0.9763,
      "step": 20
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001976,
      "loss": 0.9834,
      "step": 30
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001968,
      "loss": 1.0162,
      "step": 40
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.000196,
      "loss": 0.8783,
      "step": 50
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001952,
      "loss": 1.0353,
      "step": 60
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.0001944,
      "loss": 1.0571,
      "step": 70
    },
    {
      "epoch": 0.0,
      "learning_rate": 0.00019360000000000002,
      "loss": 0.9313,
      "step": 80
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001928,
      "loss": 0.9828,
      "step": 90
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000192,
      "loss": 1.0549,
      "step": 100
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001912,
      "loss": 1.0297,
      "step": 110
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001904,
      "loss": 0.8975,
      "step": 120
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001896,
      "loss": 0.9403,
      "step": 130
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001888,
      "loss": 1.0143,
      "step": 140
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.000188,
      "loss": 0.9905,
      "step": 150
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018720000000000002,
      "loss": 0.931,
      "step": 160
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018640000000000003,
      "loss": 1.0687,
      "step": 170
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001856,
      "loss": 1.1008,
      "step": 180
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018480000000000002,
      "loss": 1.0267,
      "step": 190
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018400000000000003,
      "loss": 0.9282,
      "step": 200
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001832,
      "loss": 0.9845,
      "step": 210
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018240000000000002,
      "loss": 1.0082,
      "step": 220
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018160000000000002,
      "loss": 0.9368,
      "step": 230
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.0001808,
      "loss": 1.0412,
      "step": 240
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00018,
      "loss": 0.9716,
      "step": 250
    },
    {
      "epoch": 0.01,
      "learning_rate": 0.00017920000000000002,
      "loss": 0.978,
      "step": 260
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001784,
      "loss": 1.0301,
      "step": 270
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001776,
      "loss": 1.0605,
      "step": 280
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00017680000000000001,
      "loss": 0.9383,
      "step": 290
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00017600000000000002,
      "loss": 1.0683,
      "step": 300
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001752,
      "loss": 1.0155,
      "step": 310
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001744,
      "loss": 0.9944,
      "step": 320
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00017360000000000002,
      "loss": 0.9694,
      "step": 330
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001728,
      "loss": 1.0904,
      "step": 340
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000172,
      "loss": 1.0099,
      "step": 350
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.00017120000000000001,
      "loss": 0.9639,
      "step": 360
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001704,
      "loss": 0.9827,
      "step": 370
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001696,
      "loss": 0.9929,
      "step": 380
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001688,
      "loss": 1.0025,
      "step": 390
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.000168,
      "loss": 1.0533,
      "step": 400
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001672,
      "loss": 0.9881,
      "step": 410
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001664,
      "loss": 1.0109,
      "step": 420
    },
    {
      "epoch": 0.02,
      "learning_rate": 0.0001656,
      "loss": 0.9747,
      "step": 430
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001648,
      "loss": 1.0304,
      "step": 440
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000164,
      "loss": 1.0282,
      "step": 450
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001632,
      "loss": 0.9555,
      "step": 460
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00016240000000000002,
      "loss": 1.0386,
      "step": 470
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00016160000000000002,
      "loss": 0.931,
      "step": 480
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001608,
      "loss": 0.9508,
      "step": 490
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00016,
      "loss": 0.9486,
      "step": 500
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00015920000000000002,
      "loss": 0.9454,
      "step": 510
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00015840000000000003,
      "loss": 1.0293,
      "step": 520
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001576,
      "loss": 1.0367,
      "step": 530
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00015680000000000002,
      "loss": 1.1525,
      "step": 540
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00015600000000000002,
      "loss": 1.0292,
      "step": 550
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001552,
      "loss": 1.0465,
      "step": 560
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001544,
      "loss": 0.9291,
      "step": 570
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.00015360000000000002,
      "loss": 1.0446,
      "step": 580
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.0001528,
      "loss": 0.9587,
      "step": 590
    },
    {
      "epoch": 0.03,
      "learning_rate": 0.000152,
      "loss": 0.9981,
      "step": 600
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00015120000000000002,
      "loss": 1.1322,
      "step": 610
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001504,
      "loss": 0.9966,
      "step": 620
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001496,
      "loss": 1.076,
      "step": 630
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001488,
      "loss": 0.9545,
      "step": 640
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000148,
      "loss": 0.9752,
      "step": 650
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001472,
      "loss": 0.9336,
      "step": 660
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001464,
      "loss": 0.9587,
      "step": 670
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00014560000000000002,
      "loss": 1.0586,
      "step": 680
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001448,
      "loss": 1.0869,
      "step": 690
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.000144,
      "loss": 0.9833,
      "step": 700
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001432,
      "loss": 1.0594,
      "step": 710
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001424,
      "loss": 1.1646,
      "step": 720
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001416,
      "loss": 0.954,
      "step": 730
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001408,
      "loss": 0.9845,
      "step": 740
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00014,
      "loss": 1.0662,
      "step": 750
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001392,
      "loss": 1.0889,
      "step": 760
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.0001384,
      "loss": 1.0045,
      "step": 770
    },
    {
      "epoch": 0.04,
      "learning_rate": 0.00013759999999999998,
      "loss": 0.9918,
      "step": 780
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00013680000000000002,
      "loss": 0.9599,
      "step": 790
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00013600000000000003,
      "loss": 0.9759,
      "step": 800
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001352,
      "loss": 0.9278,
      "step": 810
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00013440000000000001,
      "loss": 1.0274,
      "step": 820
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00013360000000000002,
      "loss": 1.0576,
      "step": 830
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001328,
      "loss": 0.9719,
      "step": 840
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000132,
      "loss": 0.9621,
      "step": 850
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00013120000000000002,
      "loss": 0.9488,
      "step": 860
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001304,
      "loss": 1.0323,
      "step": 870
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001296,
      "loss": 1.0105,
      "step": 880
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00012880000000000001,
      "loss": 0.9062,
      "step": 890
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00012800000000000002,
      "loss": 1.009,
      "step": 900
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001272,
      "loss": 0.9946,
      "step": 910
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001264,
      "loss": 1.0547,
      "step": 920
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.00012560000000000002,
      "loss": 0.9743,
      "step": 930
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.0001248,
      "loss": 1.0,
      "step": 940
    },
    {
      "epoch": 0.05,
      "learning_rate": 0.000124,
      "loss": 0.9044,
      "step": 950
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001232,
      "loss": 0.962,
      "step": 960
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001224,
      "loss": 1.0243,
      "step": 970
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001216,
      "loss": 1.0634,
      "step": 980
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001208,
      "loss": 1.0166,
      "step": 990
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00012,
      "loss": 0.9862,
      "step": 1000
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001192,
      "loss": 1.0362,
      "step": 1010
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001184,
      "loss": 0.966,
      "step": 1020
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001176,
      "loss": 1.0006,
      "step": 1030
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00011679999999999999,
      "loss": 1.0357,
      "step": 1040
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.000116,
      "loss": 0.913,
      "step": 1050
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001152,
      "loss": 0.9778,
      "step": 1060
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001144,
      "loss": 0.9639,
      "step": 1070
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.0001136,
      "loss": 0.9868,
      "step": 1080
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00011279999999999999,
      "loss": 0.9675,
      "step": 1090
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00011200000000000001,
      "loss": 0.9512,
      "step": 1100
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00011120000000000002,
      "loss": 0.9702,
      "step": 1110
    },
    {
      "epoch": 0.06,
      "learning_rate": 0.00011040000000000001,
      "loss": 1.0048,
      "step": 1120
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010960000000000001,
      "loss": 1.0567,
      "step": 1130
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010880000000000002,
      "loss": 1.0748,
      "step": 1140
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010800000000000001,
      "loss": 1.0211,
      "step": 1150
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010720000000000002,
      "loss": 1.0302,
      "step": 1160
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010640000000000001,
      "loss": 0.95,
      "step": 1170
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001056,
      "loss": 0.9633,
      "step": 1180
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010480000000000001,
      "loss": 0.953,
      "step": 1190
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010400000000000001,
      "loss": 1.1539,
      "step": 1200
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001032,
      "loss": 1.0204,
      "step": 1210
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010240000000000001,
      "loss": 1.0057,
      "step": 1220
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001016,
      "loss": 0.979,
      "step": 1230
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.00010080000000000001,
      "loss": 0.9177,
      "step": 1240
    },
    {
      "epoch": 0.07,
      "learning_rate": 0.0001,
      "loss": 0.9689,
      "step": 1250
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.92e-05,
      "loss": 1.0309,
      "step": 1260
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.84e-05,
      "loss": 0.9808,
      "step": 1270
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.76e-05,
      "loss": 0.9918,
      "step": 1280
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.680000000000001e-05,
      "loss": 0.9972,
      "step": 1290
    },
    {
      "epoch": 0.07,
      "learning_rate": 9.6e-05,
      "loss": 1.0096,
      "step": 1300
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.52e-05,
      "loss": 0.9938,
      "step": 1310
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.44e-05,
      "loss": 1.0777,
      "step": 1320
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.360000000000001e-05,
      "loss": 0.9023,
      "step": 1330
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.28e-05,
      "loss": 0.9703,
      "step": 1340
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.200000000000001e-05,
      "loss": 0.9471,
      "step": 1350
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.120000000000001e-05,
      "loss": 0.9502,
      "step": 1360
    },
    {
      "epoch": 0.08,
      "learning_rate": 9.04e-05,
      "loss": 0.9785,
      "step": 1370
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.960000000000001e-05,
      "loss": 0.9209,
      "step": 1380
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.88e-05,
      "loss": 0.9626,
      "step": 1390
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.800000000000001e-05,
      "loss": 0.9381,
      "step": 1400
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.72e-05,
      "loss": 0.9848,
      "step": 1410
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.64e-05,
      "loss": 0.988,
      "step": 1420
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.560000000000001e-05,
      "loss": 0.9534,
      "step": 1430
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.48e-05,
      "loss": 0.9494,
      "step": 1440
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.4e-05,
      "loss": 0.9913,
      "step": 1450
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.32e-05,
      "loss": 0.9673,
      "step": 1460
    },
    {
      "epoch": 0.08,
      "learning_rate": 8.24e-05,
      "loss": 1.0535,
      "step": 1470
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.16e-05,
      "loss": 1.0376,
      "step": 1480
    },
    {
      "epoch": 0.09,
      "learning_rate": 8.080000000000001e-05,
      "loss": 1.0549,
      "step": 1490
    },
    {
      "epoch": 0.09,
      "learning_rate": 8e-05,
      "loss": 1.1231,
      "step": 1500
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.920000000000001e-05,
      "loss": 0.9429,
      "step": 1510
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.840000000000001e-05,
      "loss": 0.9748,
      "step": 1520
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.76e-05,
      "loss": 1.1094,
      "step": 1530
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.680000000000001e-05,
      "loss": 1.0432,
      "step": 1540
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.6e-05,
      "loss": 0.9773,
      "step": 1550
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.52e-05,
      "loss": 1.0378,
      "step": 1560
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.44e-05,
      "loss": 1.0792,
      "step": 1570
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.36e-05,
      "loss": 0.9189,
      "step": 1580
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.280000000000001e-05,
      "loss": 1.0653,
      "step": 1590
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.2e-05,
      "loss": 1.021,
      "step": 1600
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.12e-05,
      "loss": 1.1548,
      "step": 1610
    },
    {
      "epoch": 0.09,
      "learning_rate": 7.04e-05,
      "loss": 1.0094,
      "step": 1620
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.96e-05,
      "loss": 0.9691,
      "step": 1630
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.879999999999999e-05,
      "loss": 0.9898,
      "step": 1640
    },
    {
      "epoch": 0.09,
      "learning_rate": 6.800000000000001e-05,
      "loss": 1.0263,
      "step": 1650
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.720000000000001e-05,
      "loss": 0.9948,
      "step": 1660
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.64e-05,
      "loss": 0.9809,
      "step": 1670
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.560000000000001e-05,
      "loss": 1.0254,
      "step": 1680
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.48e-05,
      "loss": 0.9775,
      "step": 1690
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.400000000000001e-05,
      "loss": 1.0165,
      "step": 1700
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.32e-05,
      "loss": 0.987,
      "step": 1710
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.24e-05,
      "loss": 0.9548,
      "step": 1720
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.16e-05,
      "loss": 0.982,
      "step": 1730
    },
    {
      "epoch": 0.1,
      "learning_rate": 6.08e-05,
      "loss": 1.1186,
      "step": 1740
    },
    {
      "epoch": 0.1,
      "learning_rate": 6e-05,
      "loss": 1.0086,
      "step": 1750
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.92e-05,
      "loss": 1.0121,
      "step": 1760
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.8399999999999997e-05,
      "loss": 1.0157,
      "step": 1770
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.76e-05,
      "loss": 0.9987,
      "step": 1780
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.68e-05,
      "loss": 1.0014,
      "step": 1790
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.6000000000000006e-05,
      "loss": 1.0788,
      "step": 1800
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.520000000000001e-05,
      "loss": 1.0208,
      "step": 1810
    },
    {
      "epoch": 0.1,
      "learning_rate": 5.440000000000001e-05,
      "loss": 1.0551,
      "step": 1820
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.360000000000001e-05,
      "loss": 1.0565,
      "step": 1830
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.28e-05,
      "loss": 1.0127,
      "step": 1840
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.2000000000000004e-05,
      "loss": 0.9067,
      "step": 1850
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.1200000000000004e-05,
      "loss": 0.9731,
      "step": 1860
    },
    {
      "epoch": 0.11,
      "learning_rate": 5.0400000000000005e-05,
      "loss": 1.005,
      "step": 1870
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.96e-05,
      "loss": 1.0038,
      "step": 1880
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.88e-05,
      "loss": 0.9671,
      "step": 1890
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.8e-05,
      "loss": 0.9792,
      "step": 1900
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.72e-05,
      "loss": 0.9125,
      "step": 1910
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.64e-05,
      "loss": 1.0566,
      "step": 1920
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.5600000000000004e-05,
      "loss": 0.9108,
      "step": 1930
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.4800000000000005e-05,
      "loss": 1.0017,
      "step": 1940
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.4000000000000006e-05,
      "loss": 0.96,
      "step": 1950
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.32e-05,
      "loss": 0.9096,
      "step": 1960
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.24e-05,
      "loss": 0.9037,
      "step": 1970
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.16e-05,
      "loss": 0.9881,
      "step": 1980
    },
    {
      "epoch": 0.11,
      "learning_rate": 4.08e-05,
      "loss": 0.9556,
      "step": 1990
    },
    {
      "epoch": 0.12,
      "learning_rate": 4e-05,
      "loss": 0.9572,
      "step": 2000
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.9200000000000004e-05,
      "loss": 1.0412,
      "step": 2010
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.8400000000000005e-05,
      "loss": 1.0179,
      "step": 2020
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.76e-05,
      "loss": 1.032,
      "step": 2030
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.68e-05,
      "loss": 0.9593,
      "step": 2040
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.6e-05,
      "loss": 0.9784,
      "step": 2050
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.52e-05,
      "loss": 0.9205,
      "step": 2060
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.4399999999999996e-05,
      "loss": 0.98,
      "step": 2070
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.3600000000000004e-05,
      "loss": 1.122,
      "step": 2080
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.2800000000000004e-05,
      "loss": 0.9987,
      "step": 2090
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.2000000000000005e-05,
      "loss": 0.9724,
      "step": 2100
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.12e-05,
      "loss": 1.0209,
      "step": 2110
    },
    {
      "epoch": 0.12,
      "learning_rate": 3.04e-05,
      "loss": 0.9966,
      "step": 2120
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.96e-05,
      "loss": 0.9792,
      "step": 2130
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.88e-05,
      "loss": 1.0142,
      "step": 2140
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.8000000000000003e-05,
      "loss": 0.9164,
      "step": 2150
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.7200000000000004e-05,
      "loss": 1.0151,
      "step": 2160
    },
    {
      "epoch": 0.12,
      "learning_rate": 2.64e-05,
      "loss": 0.9667,
      "step": 2170
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.5600000000000002e-05,
      "loss": 1.1009,
      "step": 2180
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.48e-05,
      "loss": 0.9313,
      "step": 2190
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.4e-05,
      "loss": 1.0617,
      "step": 2200
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.32e-05,
      "loss": 1.0034,
      "step": 2210
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.2400000000000002e-05,
      "loss": 0.9779,
      "step": 2220
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.16e-05,
      "loss": 0.9861,
      "step": 2230
    },
    {
      "epoch": 0.13,
      "learning_rate": 2.08e-05,
      "loss": 0.9929,
      "step": 2240
    },
    {
      "epoch": 0.13,
      "learning_rate": 2e-05,
      "loss": 1.0136,
      "step": 2250
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.9200000000000003e-05,
      "loss": 0.9733,
      "step": 2260
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.84e-05,
      "loss": 0.9304,
      "step": 2270
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.76e-05,
      "loss": 0.976,
      "step": 2280
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.6800000000000002e-05,
      "loss": 1.0155,
      "step": 2290
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.6000000000000003e-05,
      "loss": 0.9637,
      "step": 2300
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.52e-05,
      "loss": 0.9841,
      "step": 2310
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.44e-05,
      "loss": 0.9995,
      "step": 2320
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.3600000000000002e-05,
      "loss": 0.944,
      "step": 2330
    },
    {
      "epoch": 0.13,
      "learning_rate": 1.2800000000000001e-05,
      "loss": 1.0196,
      "step": 2340
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.2e-05,
      "loss": 1.1164,
      "step": 2350
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.1200000000000001e-05,
      "loss": 0.9235,
      "step": 2360
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.04e-05,
      "loss": 0.9698,
      "step": 2370
    },
    {
      "epoch": 0.14,
      "learning_rate": 9.600000000000001e-06,
      "loss": 1.0232,
      "step": 2380
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.8e-06,
      "loss": 0.9899,
      "step": 2390
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.000000000000001e-06,
      "loss": 0.9535,
      "step": 2400
    },
    {
      "epoch": 0.14,
      "learning_rate": 7.2e-06,
      "loss": 0.977,
      "step": 2410
    },
    {
      "epoch": 0.14,
      "learning_rate": 6.4000000000000006e-06,
      "loss": 1.0414,
      "step": 2420
    },
    {
      "epoch": 0.14,
      "learning_rate": 5.600000000000001e-06,
      "loss": 0.986,
      "step": 2430
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.800000000000001e-06,
      "loss": 1.0086,
      "step": 2440
    },
    {
      "epoch": 0.14,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.9594,
      "step": 2450
    },
    {
      "epoch": 0.14,
      "learning_rate": 3.2000000000000003e-06,
      "loss": 0.9584,
      "step": 2460
    },
    {
      "epoch": 0.14,
      "learning_rate": 2.4000000000000003e-06,
      "loss": 0.8961,
      "step": 2470
    },
    {
      "epoch": 0.14,
      "learning_rate": 1.6000000000000001e-06,
      "loss": 0.9713,
      "step": 2480
    },
    {
      "epoch": 0.14,
      "learning_rate": 8.000000000000001e-07,
      "loss": 1.0307,
      "step": 2490
    },
    {
      "epoch": 0.14,
      "learning_rate": 0.0,
      "loss": 0.9984,
      "step": 2500
    }
  ],
  "logging_steps": 10,
  "max_steps": 2500,
  "num_train_epochs": 1,
  "save_steps": 500,
  "total_flos": 1.63610722566144e+18,
  "trial_name": null,
  "trial_params": null
}